Mercurial > hg > octave-nkf > gnulib-hg
annotate lib/regcomp.c @ 12570:3ed9d84fee81
regcomp: recognize ill-formed { } expressions
* lib/regcomp.c (parse_dup_op): From glibc:
http://sourceware.org/git/gitweb.cgi?p=glibc.git;a=commitdiff;h=a87cd2894cb
author | Ulrich Drepper <drepper@redhat.com> |
---|---|
date | Mon, 04 Jan 2010 10:51:34 +0100 |
parents | 51ea446bf1f8 |
children | 64d47f001127 |
rev | line source |
---|---|
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1 /* Extended regular expression matching and search library. |
12559
c2cbabec01dd
update nearly all FSF copyright year lists to include 2010
Jim Meyering <meyering@redhat.com>
parents:
12351
diff
changeset
|
2 Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 Free |
c2cbabec01dd
update nearly all FSF copyright year lists to include 2010
Jim Meyering <meyering@redhat.com>
parents:
12351
diff
changeset
|
3 Software Foundation, Inc. |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
4 This file is part of the GNU C Library. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
5 Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
6 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
7 This program is free software; you can redistribute it and/or modify |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
8 it under the terms of the GNU General Public License as published by |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
9 the Free Software Foundation; either version 2, or (at your option) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
10 any later version. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
11 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
12 This program is distributed in the hope that it will be useful, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
13 but WITHOUT ANY WARRANTY; without even the implied warranty of |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
15 GNU General Public License for more details. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
16 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
17 You should have received a copy of the GNU General Public License along |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
18 with this program; if not, write to the Free Software Foundation, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
19 Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
20 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
21 static reg_errcode_t re_compile_internal (regex_t *preg, const char * pattern, |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
22 size_t length, reg_syntax_t syntax); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
23 static void re_compile_fastmap_iter (regex_t *bufp, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
24 const re_dfastate_t *init_state, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
25 char *fastmap); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
26 static reg_errcode_t init_dfa (re_dfa_t *dfa, size_t pat_len); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
27 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
28 static void free_charset (re_charset_t *cset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
29 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
30 static void free_workarea_compile (regex_t *preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
31 static reg_errcode_t create_initial_state (re_dfa_t *dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
32 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
33 static void optimize_utf8 (re_dfa_t *dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
34 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
35 static reg_errcode_t analyze (regex_t *preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
36 static reg_errcode_t preorder (bin_tree_t *root, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
37 reg_errcode_t (fn (void *, bin_tree_t *)), |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
38 void *extra); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
39 static reg_errcode_t postorder (bin_tree_t *root, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
40 reg_errcode_t (fn (void *, bin_tree_t *)), |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
41 void *extra); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
42 static reg_errcode_t optimize_subexps (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
43 static reg_errcode_t lower_subexps (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
44 static bin_tree_t *lower_subexp (reg_errcode_t *err, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
45 bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
46 static reg_errcode_t calc_first (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
47 static reg_errcode_t calc_next (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
48 static reg_errcode_t link_nfa_nodes (void *extra, bin_tree_t *node); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
49 static Idx duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint); |
6185
6b09f7f6ba73
* lib/regcomp.c (search_duplicated_node): Make first pointer arg
Paul Eggert <eggert@cs.ucla.edu>
parents:
6184
diff
changeset
|
50 static Idx search_duplicated_node (const re_dfa_t *dfa, Idx org_node, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
51 unsigned int constraint); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
52 static reg_errcode_t calc_eclosure (re_dfa_t *dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
53 static reg_errcode_t calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
54 Idx node, bool root); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
55 static reg_errcode_t calc_inveclosure (re_dfa_t *dfa); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
56 static Idx fetch_number (re_string_t *input, re_token_t *token, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
57 reg_syntax_t syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
58 static int peek_token (re_token_t *token, re_string_t *input, |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
59 reg_syntax_t syntax) internal_function; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
60 static bin_tree_t *parse (re_string_t *regexp, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
61 reg_syntax_t syntax, reg_errcode_t *err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
62 static bin_tree_t *parse_reg_exp (re_string_t *regexp, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
63 re_token_t *token, reg_syntax_t syntax, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
64 Idx nest, reg_errcode_t *err); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
65 static bin_tree_t *parse_branch (re_string_t *regexp, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
66 re_token_t *token, reg_syntax_t syntax, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
67 Idx nest, reg_errcode_t *err); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
68 static bin_tree_t *parse_expression (re_string_t *regexp, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
69 re_token_t *token, reg_syntax_t syntax, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
70 Idx nest, reg_errcode_t *err); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
71 static bin_tree_t *parse_sub_exp (re_string_t *regexp, regex_t *preg, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
72 re_token_t *token, reg_syntax_t syntax, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
73 Idx nest, reg_errcode_t *err); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
74 static bin_tree_t *parse_dup_op (bin_tree_t *dup_elem, re_string_t *regexp, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
75 re_dfa_t *dfa, re_token_t *token, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
76 reg_syntax_t syntax, reg_errcode_t *err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
77 static bin_tree_t *parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
78 re_token_t *token, reg_syntax_t syntax, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
79 reg_errcode_t *err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
80 static reg_errcode_t parse_bracket_element (bracket_elem_t *elem, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
81 re_string_t *regexp, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
82 re_token_t *token, int token_len, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
83 re_dfa_t *dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
84 reg_syntax_t syntax, |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
85 bool accept_hyphen); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
86 static reg_errcode_t parse_bracket_symbol (bracket_elem_t *elem, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
87 re_string_t *regexp, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
88 re_token_t *token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
89 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
90 static reg_errcode_t build_equiv_class (bitset_t sbcset, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
91 re_charset_t *mbcset, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
92 Idx *equiv_class_alloc, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
93 const unsigned char *name); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
94 static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
95 bitset_t sbcset, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
96 re_charset_t *mbcset, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
97 Idx *char_class_alloc, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
98 const unsigned char *class_name, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
99 reg_syntax_t syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
100 #else /* not RE_ENABLE_I18N */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
101 static reg_errcode_t build_equiv_class (bitset_t sbcset, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
102 const unsigned char *name); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
103 static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
104 bitset_t sbcset, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
105 const unsigned char *class_name, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
106 reg_syntax_t syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
107 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
108 static bin_tree_t *build_charclass_op (re_dfa_t *dfa, |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
109 RE_TRANSLATE_TYPE trans, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
110 const unsigned char *class_name, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
111 const unsigned char *extra, |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
112 bool non_match, reg_errcode_t *err); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
113 static bin_tree_t *create_tree (re_dfa_t *dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
114 bin_tree_t *left, bin_tree_t *right, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
115 re_token_type_t type); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
116 static bin_tree_t *create_token_tree (re_dfa_t *dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
117 bin_tree_t *left, bin_tree_t *right, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
118 const re_token_t *token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
119 static bin_tree_t *duplicate_tree (const bin_tree_t *src, re_dfa_t *dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
120 static void free_token (re_token_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
121 static reg_errcode_t free_tree (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
122 static reg_errcode_t mark_opt_subexp (void *extra, bin_tree_t *node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
123 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
124 /* This table gives an error message for each of the error codes listed |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
125 in regex.h. Obviously the order here has to be same as there. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
126 POSIX doesn't require that we do anything for REG_NOERROR, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
127 but why not be nice? */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
128 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
129 static const char __re_error_msgid[] = |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
130 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
131 #define REG_NOERROR_IDX 0 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
132 gettext_noop ("Success") /* REG_NOERROR */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
133 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
134 #define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
135 gettext_noop ("No match") /* REG_NOMATCH */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
136 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
137 #define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
138 gettext_noop ("Invalid regular expression") /* REG_BADPAT */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
139 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
140 #define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
141 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
142 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
143 #define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
144 gettext_noop ("Invalid character class name") /* REG_ECTYPE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
145 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
146 #define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
147 gettext_noop ("Trailing backslash") /* REG_EESCAPE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
148 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
149 #define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
150 gettext_noop ("Invalid back reference") /* REG_ESUBREG */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
151 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
152 #define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
153 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
154 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
155 #define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
156 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
157 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
158 #define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
159 gettext_noop ("Unmatched \\{") /* REG_EBRACE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
160 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
161 #define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
162 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
163 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
164 #define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
165 gettext_noop ("Invalid range end") /* REG_ERANGE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
166 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
167 #define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
168 gettext_noop ("Memory exhausted") /* REG_ESPACE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
169 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
170 #define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
171 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
172 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
173 #define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
174 gettext_noop ("Premature end of regular expression") /* REG_EEND */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
175 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
176 #define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
177 gettext_noop ("Regular expression too big") /* REG_ESIZE */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
178 "\0" |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
179 #define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big") |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
180 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
181 }; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
182 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
183 static const size_t __re_error_msgid_idx[] = |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
184 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
185 REG_NOERROR_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
186 REG_NOMATCH_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
187 REG_BADPAT_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
188 REG_ECOLLATE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
189 REG_ECTYPE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
190 REG_EESCAPE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
191 REG_ESUBREG_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
192 REG_EBRACK_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
193 REG_EPAREN_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
194 REG_EBRACE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
195 REG_BADBR_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
196 REG_ERANGE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
197 REG_ESPACE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
198 REG_BADRPT_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
199 REG_EEND_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
200 REG_ESIZE_IDX, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
201 REG_ERPAREN_IDX |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
202 }; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
203 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
204 /* Entry points for GNU code. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
205 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
206 /* re_compile_pattern is the GNU regular expression compiler: it |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
207 compiles PATTERN (of length LENGTH) and puts the result in BUFP. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
208 Returns 0 if the pattern was valid, otherwise an error string. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
209 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
210 Assumes the `allocated' (and perhaps `buffer') and `translate' fields |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
211 are set in BUFP on entry. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
212 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
213 #ifdef _LIBC |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
214 const char * |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
215 re_compile_pattern (pattern, length, bufp) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
216 const char *pattern; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
217 size_t length; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
218 struct re_pattern_buffer *bufp; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
219 #else /* size_t might promote */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
220 const char * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
221 re_compile_pattern (const char *pattern, size_t length, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
222 struct re_pattern_buffer *bufp) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
223 #endif |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
224 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
225 reg_errcode_t ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
226 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
227 /* And GNU code determines whether or not to get register information |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
228 by passing null for the REGS argument to re_match, etc., not by |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
229 setting no_sub, unless RE_NO_SUB is set. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
230 bufp->no_sub = !!(re_syntax_options & RE_NO_SUB); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
231 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
232 /* Match anchors at newline. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
233 bufp->newline_anchor = 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
234 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
235 ret = re_compile_internal (bufp, pattern, length, re_syntax_options); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
236 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
237 if (!ret) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
238 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
239 return gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
240 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
241 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
242 weak_alias (__re_compile_pattern, re_compile_pattern) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
243 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
244 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
245 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
246 also be assigned to arbitrarily: each pattern buffer stores its own |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
247 syntax, so it can be changed between regex compilations. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
248 /* This has no initializer because initialized variables in Emacs |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
249 become read-only after dumping. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
250 reg_syntax_t re_syntax_options; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
251 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
252 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
253 /* Specify the precise syntax of regexps for compilation. This provides |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
254 for compatibility for various utilities which historically have |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
255 different, incompatible syntaxes. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
256 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
257 The argument SYNTAX is a bit mask comprised of the various bits |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
258 defined in regex.h. We return the old syntax. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
259 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
260 reg_syntax_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
261 re_set_syntax (syntax) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
262 reg_syntax_t syntax; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
263 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
264 reg_syntax_t ret = re_syntax_options; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
265 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
266 re_syntax_options = syntax; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
267 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
268 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
269 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
270 weak_alias (__re_set_syntax, re_set_syntax) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
271 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
272 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
273 int |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
274 re_compile_fastmap (bufp) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
275 struct re_pattern_buffer *bufp; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
276 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
277 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
278 char *fastmap = bufp->fastmap; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
279 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
280 memset (fastmap, '\0', sizeof (char) * SBC_MAX); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
281 re_compile_fastmap_iter (bufp, dfa->init_state, fastmap); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
282 if (dfa->init_state != dfa->init_state_word) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
283 re_compile_fastmap_iter (bufp, dfa->init_state_word, fastmap); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
284 if (dfa->init_state != dfa->init_state_nl) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
285 re_compile_fastmap_iter (bufp, dfa->init_state_nl, fastmap); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
286 if (dfa->init_state != dfa->init_state_begbuf) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
287 re_compile_fastmap_iter (bufp, dfa->init_state_begbuf, fastmap); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
288 bufp->fastmap_accurate = 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
289 return 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
290 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
291 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
292 weak_alias (__re_compile_fastmap, re_compile_fastmap) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
293 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
294 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
295 static inline void |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
296 __attribute ((always_inline)) |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
297 re_set_fastmap (char *fastmap, bool icase, int ch) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
298 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
299 fastmap[ch] = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
300 if (icase) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
301 fastmap[tolower (ch)] = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
302 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
303 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
304 /* Helper function for re_compile_fastmap. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
305 Compile fastmap for the initial_state INIT_STATE. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
306 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
307 static void |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
308 re_compile_fastmap_iter (regex_t *bufp, const re_dfastate_t *init_state, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
309 char *fastmap) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
310 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
311 re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
312 Idx node_cnt; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
313 bool icase = (dfa->mb_cur_max == 1 && (bufp->syntax & RE_ICASE)); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
314 for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
315 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
316 Idx node = init_state->nodes.elems[node_cnt]; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
317 re_token_type_t type = dfa->nodes[node].type; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
318 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
319 if (type == CHARACTER) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
320 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
321 re_set_fastmap (fastmap, icase, dfa->nodes[node].opr.c); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
322 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
323 if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
324 { |
6119
c3bf2ea44695
Make regex safe for g++. This fixes one real bug (an "err"
Paul Eggert <eggert@cs.ucla.edu>
parents:
6104
diff
changeset
|
325 unsigned char buf[MB_LEN_MAX]; |
c3bf2ea44695
Make regex safe for g++. This fixes one real bug (an "err"
Paul Eggert <eggert@cs.ucla.edu>
parents:
6104
diff
changeset
|
326 unsigned char *p; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
327 wchar_t wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
328 mbstate_t state; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
329 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
330 p = buf; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
331 *p++ = dfa->nodes[node].opr.c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
332 while (++node < dfa->nodes_len |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
333 && dfa->nodes[node].type == CHARACTER |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
334 && dfa->nodes[node].mb_partial) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
335 *p++ = dfa->nodes[node].opr.c; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
336 memset (&state, '\0', sizeof (state)); |
10998
cc7a1af3872f
regex: replace mbrtowc with __mbrtowc.
Paolo Bonzini <bonzini@gnu.org>
parents:
10078
diff
changeset
|
337 if (__mbrtowc (&wc, (const char *) buf, p - buf, |
cc7a1af3872f
regex: replace mbrtowc with __mbrtowc.
Paolo Bonzini <bonzini@gnu.org>
parents:
10078
diff
changeset
|
338 &state) == p - buf |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
339 && (__wcrtomb ((char *) buf, towlower (wc), &state) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
340 != (size_t) -1)) |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
341 re_set_fastmap (fastmap, false, buf[0]); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
342 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
343 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
344 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
345 else if (type == SIMPLE_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
346 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
347 int i, ch; |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
348 for (i = 0, ch = 0; i < BITSET_WORDS; ++i) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
349 { |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
350 int j; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
351 bitset_word_t w = dfa->nodes[node].opr.sbcset[i]; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
352 for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
353 if (w & ((bitset_word_t) 1 << j)) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
354 re_set_fastmap (fastmap, icase, ch); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
355 } |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
356 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
357 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
358 else if (type == COMPLEX_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
359 { |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
360 re_charset_t *cset = dfa->nodes[node].opr.mbcset; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
361 Idx i; |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
362 |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
363 # ifdef _LIBC |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
364 /* See if we have to try all bytes which start multiple collation |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
365 elements. |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
366 e.g. In da_DK, we want to catch 'a' since "aa" is a valid |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
367 collation element, and don't catch 'b' since 'b' is |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
368 the only collation element which starts from 'b' (and |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
369 it is caught by SIMPLE_BRACKET). */ |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
370 if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0 |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
371 && (cset->ncoll_syms || cset->nranges)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
372 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
373 const int32_t *table = (const int32_t *) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
374 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
375 for (i = 0; i < SBC_MAX; ++i) |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
376 if (table[i] < 0) |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
377 re_set_fastmap (fastmap, icase, i); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
378 } |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
379 # endif /* _LIBC */ |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
380 |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
381 /* See if we have to start the match at all multibyte characters, |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
382 i.e. where we would not find an invalid sequence. This only |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
383 applies to multibyte character sets; for single byte character |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
384 sets, the SIMPLE_BRACKET again suffices. */ |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
385 if (dfa->mb_cur_max > 1 |
12351
0b80ccdc9aa4
regex: Fix fastmap for multibyte character ranges.
Paolo Bonzini <bonzini@gnu.org>
parents:
11000
diff
changeset
|
386 && (cset->nchar_classes || cset->non_match || cset->nranges |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
387 # ifdef _LIBC |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
388 || cset->nequiv_classes |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
389 # endif /* _LIBC */ |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
390 )) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
391 { |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
392 unsigned char c = 0; |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
393 do |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
394 { |
11000
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
395 mbstate_t mbs; |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
396 memset (&mbs, 0, sizeof (mbs)); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
397 if (__mbrtowc (NULL, (char *) &c, 1, &mbs) == (size_t) -2) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
398 re_set_fastmap (fastmap, false, (int) c); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
399 } |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
400 while (++c != 0); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
401 } |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
402 |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
403 else |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
404 { |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
405 /* ... Else catch all bytes which can start the mbchars. */ |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
406 for (i = 0; i < cset->nmbchars; ++i) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
407 { |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
408 char buf[256]; |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
409 mbstate_t state; |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
410 memset (&state, '\0', sizeof (state)); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
411 if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
412 re_set_fastmap (fastmap, icase, *(unsigned char *) buf); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
413 if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
414 { |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
415 if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
416 != (size_t) -1) |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
417 re_set_fastmap (fastmap, false, *(unsigned char *) buf); |
683b379e6760
regex: fix glibc bug 9697
Paolo Bonzini <bonzini@gnu.org>
parents:
10998
diff
changeset
|
418 } |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
419 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
420 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
421 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
422 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
423 else if (type == OP_PERIOD |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
424 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
425 || type == OP_UTF8_PERIOD |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
426 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
427 || type == END_OF_RE) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
428 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
429 memset (fastmap, '\1', sizeof (char) * SBC_MAX); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
430 if (type == END_OF_RE) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
431 bufp->can_be_null = 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
432 return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
433 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
434 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
435 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
436 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
437 /* Entry point for POSIX code. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
438 /* regcomp takes a regular expression as a string and compiles it. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
439 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
440 PREG is a regex_t *. We do not expect any fields to be initialized, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
441 since POSIX says we shouldn't. Thus, we set |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
442 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
443 `buffer' to the compiled pattern; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
444 `used' to the length of the compiled pattern; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
445 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
446 REG_EXTENDED bit in CFLAGS is set; otherwise, to |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
447 RE_SYNTAX_POSIX_BASIC; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
448 `newline_anchor' to REG_NEWLINE being set in CFLAGS; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
449 `fastmap' to an allocated space for the fastmap; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
450 `fastmap_accurate' to zero; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
451 `re_nsub' to the number of subexpressions in PATTERN. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
452 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
453 PATTERN is the address of the pattern string. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
454 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
455 CFLAGS is a series of bits which affect compilation. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
456 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
457 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
458 use POSIX basic syntax. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
459 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
460 If REG_NEWLINE is set, then . and [^...] don't match newline. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
461 Also, regexec will try a match beginning after every newline. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
462 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
463 If REG_ICASE is set, then we considers upper- and lowercase |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
464 versions of letters to be equivalent when matching. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
465 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
466 If REG_NOSUB is set, then when PREG is passed to regexec, that |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
467 routine will report only success or failure, and nothing about the |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
468 registers. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
469 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
470 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
471 the return codes and their meanings.) */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
472 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
473 int |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
474 regcomp (preg, pattern, cflags) |
8045
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
475 regex_t *_Restrict_ preg; |
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
476 const char *_Restrict_ pattern; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
477 int cflags; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
478 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
479 reg_errcode_t ret; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
480 reg_syntax_t syntax = ((cflags & REG_EXTENDED) ? RE_SYNTAX_POSIX_EXTENDED |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
481 : RE_SYNTAX_POSIX_BASIC); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
482 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
483 preg->buffer = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
484 preg->allocated = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
485 preg->used = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
486 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
487 /* Try to allocate space for the fastmap. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
488 preg->fastmap = re_malloc (char, SBC_MAX); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
489 if (BE (preg->fastmap == NULL, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
490 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
491 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
492 syntax |= (cflags & REG_ICASE) ? RE_ICASE : 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
493 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
494 /* If REG_NEWLINE is set, newlines are treated differently. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
495 if (cflags & REG_NEWLINE) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
496 { /* REG_NEWLINE implies neither . nor [^...] match newline. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
497 syntax &= ~RE_DOT_NEWLINE; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
498 syntax |= RE_HAT_LISTS_NOT_NEWLINE; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
499 /* It also changes the matching behavior. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
500 preg->newline_anchor = 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
501 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
502 else |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
503 preg->newline_anchor = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
504 preg->no_sub = !!(cflags & REG_NOSUB); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
505 preg->translate = NULL; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
506 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
507 ret = re_compile_internal (preg, pattern, strlen (pattern), syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
508 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
509 /* POSIX doesn't distinguish between an unmatched open-group and an |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
510 unmatched close-group: both are REG_EPAREN. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
511 if (ret == REG_ERPAREN) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
512 ret = REG_EPAREN; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
513 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
514 /* We have already checked preg->fastmap != NULL. */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
515 if (BE (ret == REG_NOERROR, 1)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
516 /* Compute the fastmap now, since regexec cannot modify the pattern |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
517 buffer. This function never fails in this implementation. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
518 (void) re_compile_fastmap (preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
519 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
520 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
521 /* Some error occurred while compiling the expression. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
522 re_free (preg->fastmap); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
523 preg->fastmap = NULL; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
524 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
525 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
526 return (int) ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
527 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
528 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
529 weak_alias (__regcomp, regcomp) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
530 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
531 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
532 /* Returns a message corresponding to an error code, ERRCODE, returned |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
533 from either regcomp or regexec. We don't use PREG here. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
534 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
535 #ifdef _LIBC |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
536 size_t |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
537 regerror (errcode, preg, errbuf, errbuf_size) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
538 int errcode; |
8045
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
539 const regex_t *_Restrict_ preg; |
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
540 char *_Restrict_ errbuf; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
541 size_t errbuf_size; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
542 #else /* size_t might promote */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
543 size_t |
8045
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
544 regerror (int errcode, const regex_t *_Restrict_ preg, |
7dcf8a1f2f5e
* lib/regex.h (_Restrict_): Renamed from __restrict, to avoid
Paul Eggert <eggert@cs.ucla.edu>
parents:
7694
diff
changeset
|
545 char *_Restrict_ errbuf, size_t errbuf_size) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
546 #endif |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
547 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
548 const char *msg; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
549 size_t msg_size; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
550 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
551 if (BE (errcode < 0 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
552 || errcode >= (int) (sizeof (__re_error_msgid_idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
553 / sizeof (__re_error_msgid_idx[0])), 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
554 /* Only error codes returned by the rest of the code should be passed |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
555 to this routine. If we are given anything else, or if other regex |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
556 code generates an invalid error code, then the program has a bug. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
557 Dump core so we can fix it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
558 abort (); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
559 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
560 msg = gettext (__re_error_msgid + __re_error_msgid_idx[errcode]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
561 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
562 msg_size = strlen (msg) + 1; /* Includes the null. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
563 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
564 if (BE (errbuf_size != 0, 1)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
565 { |
8073
eaa00773406b
Avoid mempcpy in the regex code, as the string.h mempcpy stuff
Paul Eggert <eggert@cs.ucla.edu>
parents:
8045
diff
changeset
|
566 size_t cpy_size = msg_size; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
567 if (BE (msg_size > errbuf_size, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
568 { |
8073
eaa00773406b
Avoid mempcpy in the regex code, as the string.h mempcpy stuff
Paul Eggert <eggert@cs.ucla.edu>
parents:
8045
diff
changeset
|
569 cpy_size = errbuf_size - 1; |
eaa00773406b
Avoid mempcpy in the regex code, as the string.h mempcpy stuff
Paul Eggert <eggert@cs.ucla.edu>
parents:
8045
diff
changeset
|
570 errbuf[cpy_size] = '\0'; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
571 } |
8073
eaa00773406b
Avoid mempcpy in the regex code, as the string.h mempcpy stuff
Paul Eggert <eggert@cs.ucla.edu>
parents:
8045
diff
changeset
|
572 memcpy (errbuf, msg, cpy_size); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
573 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
574 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
575 return msg_size; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
576 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
577 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
578 weak_alias (__regerror, regerror) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
579 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
580 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
581 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
582 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
583 /* This static array is used for the map to single-byte characters when |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
584 UTF-8 is used. Otherwise we would allocate memory just to initialize |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
585 it the same all the time. UTF-8 is the preferred encoding so this is |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
586 a worthwhile optimization. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
587 static const bitset_t utf8_sb_map = |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
588 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
589 /* Set the first 128 bits. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
590 # if 4 * BITSET_WORD_BITS < ASCII_CHARS |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
591 # error "bitset_word_t is narrower than 32 bits" |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
592 # elif 3 * BITSET_WORD_BITS < ASCII_CHARS |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
593 BITSET_WORD_MAX, BITSET_WORD_MAX, BITSET_WORD_MAX, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
594 # elif 2 * BITSET_WORD_BITS < ASCII_CHARS |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
595 BITSET_WORD_MAX, BITSET_WORD_MAX, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
596 # elif 1 * BITSET_WORD_BITS < ASCII_CHARS |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
597 BITSET_WORD_MAX, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
598 # endif |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
599 (BITSET_WORD_MAX |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
600 >> (SBC_MAX % BITSET_WORD_BITS == 0 |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
601 ? 0 |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
602 : BITSET_WORD_BITS - SBC_MAX % BITSET_WORD_BITS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
603 }; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
604 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
605 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
606 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
607 static void |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
608 free_dfa_content (re_dfa_t *dfa) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
609 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
610 Idx i, j; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
611 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
612 if (dfa->nodes) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
613 for (i = 0; i < dfa->nodes_len; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
614 free_token (dfa->nodes + i); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
615 re_free (dfa->nexts); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
616 for (i = 0; i < dfa->nodes_len; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
617 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
618 if (dfa->eclosures != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
619 re_node_set_free (dfa->eclosures + i); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
620 if (dfa->inveclosures != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
621 re_node_set_free (dfa->inveclosures + i); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
622 if (dfa->edests != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
623 re_node_set_free (dfa->edests + i); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
624 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
625 re_free (dfa->edests); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
626 re_free (dfa->eclosures); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
627 re_free (dfa->inveclosures); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
628 re_free (dfa->nodes); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
629 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
630 if (dfa->state_table) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
631 for (i = 0; i <= dfa->state_hash_mask; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
632 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
633 struct re_state_table_entry *entry = dfa->state_table + i; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
634 for (j = 0; j < entry->num; ++j) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
635 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
636 re_dfastate_t *state = entry->array[j]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
637 free_state (state); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
638 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
639 re_free (entry->array); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
640 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
641 re_free (dfa->state_table); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
642 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
643 if (dfa->sb_char != utf8_sb_map) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
644 re_free (dfa->sb_char); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
645 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
646 re_free (dfa->subexp_map); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
647 #ifdef DEBUG |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
648 re_free (dfa->re_str); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
649 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
650 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
651 re_free (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
652 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
653 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
654 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
655 /* Free dynamically allocated space used by PREG. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
656 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
657 void |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
658 regfree (preg) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
659 regex_t *preg; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
660 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
661 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
662 if (BE (dfa != NULL, 1)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
663 free_dfa_content (dfa); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
664 preg->buffer = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
665 preg->allocated = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
666 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
667 re_free (preg->fastmap); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
668 preg->fastmap = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
669 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
670 re_free (preg->translate); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
671 preg->translate = NULL; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
672 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
673 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
674 weak_alias (__regfree, regfree) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
675 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
676 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
677 /* Entry points compatible with 4.2 BSD regex library. We don't define |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
678 them unless specifically requested. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
679 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
680 #if defined _REGEX_RE_COMP || defined _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
681 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
682 /* BSD has one and only one pattern buffer. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
683 static struct re_pattern_buffer re_comp_buf; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
684 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
685 char * |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
686 # ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
687 /* Make these definitions weak in libc, so POSIX programs can redefine |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
688 these names if they don't use our functions, and still use |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
689 regcomp/regexec above without link errors. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
690 weak_function |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
691 # endif |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
692 re_comp (s) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
693 const char *s; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
694 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
695 reg_errcode_t ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
696 char *fastmap; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
697 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
698 if (!s) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
699 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
700 if (!re_comp_buf.buffer) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
701 return gettext ("No previous regular expression"); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
702 return 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
703 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
704 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
705 if (re_comp_buf.buffer) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
706 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
707 fastmap = re_comp_buf.fastmap; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
708 re_comp_buf.fastmap = NULL; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
709 __regfree (&re_comp_buf); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
710 memset (&re_comp_buf, '\0', sizeof (re_comp_buf)); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
711 re_comp_buf.fastmap = fastmap; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
712 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
713 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
714 if (re_comp_buf.fastmap == NULL) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
715 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
716 re_comp_buf.fastmap = (char *) malloc (SBC_MAX); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
717 if (re_comp_buf.fastmap == NULL) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
718 return (char *) gettext (__re_error_msgid |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
719 + __re_error_msgid_idx[(int) REG_ESPACE]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
720 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
721 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
722 /* Since `re_exec' always passes NULL for the `regs' argument, we |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
723 don't need to initialize the pattern buffer fields which affect it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
724 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
725 /* Match anchors at newlines. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
726 re_comp_buf.newline_anchor = 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
727 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
728 ret = re_compile_internal (&re_comp_buf, s, strlen (s), re_syntax_options); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
729 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
730 if (!ret) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
731 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
732 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
733 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
734 return (char *) gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
735 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
736 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
737 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
738 libc_freeres_fn (free_mem) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
739 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
740 __regfree (&re_comp_buf); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
741 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
742 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
743 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
744 #endif /* _REGEX_RE_COMP */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
745 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
746 /* Internal entry point. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
747 Compile the regular expression PATTERN, whose length is LENGTH. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
748 SYNTAX indicate regular expression's syntax. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
749 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
750 static reg_errcode_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
751 re_compile_internal (regex_t *preg, const char * pattern, size_t length, |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
752 reg_syntax_t syntax) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
753 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
754 reg_errcode_t err = REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
755 re_dfa_t *dfa; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
756 re_string_t regexp; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
757 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
758 /* Initialize the pattern buffer. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
759 preg->fastmap_accurate = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
760 preg->syntax = syntax; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
761 preg->not_bol = preg->not_eol = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
762 preg->used = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
763 preg->re_nsub = 0; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
764 preg->can_be_null = 0; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
765 preg->regs_allocated = REGS_UNALLOCATED; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
766 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
767 /* Initialize the dfa. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
768 dfa = (re_dfa_t *) preg->buffer; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
769 if (BE (preg->allocated < sizeof (re_dfa_t), 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
770 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
771 /* If zero allocated, but buffer is non-null, try to realloc |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
772 enough space. This loses if buffer's address is bogus, but |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
773 that is the user's responsibility. If ->buffer is NULL this |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
774 is a simple allocation. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
775 dfa = re_realloc (preg->buffer, re_dfa_t, 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
776 if (dfa == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
777 return REG_ESPACE; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
778 preg->allocated = sizeof (re_dfa_t); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
779 preg->buffer = (unsigned char *) dfa; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
780 } |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
781 preg->used = sizeof (re_dfa_t); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
782 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
783 err = init_dfa (dfa, length); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
784 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
785 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
786 free_dfa_content (dfa); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
787 preg->buffer = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
788 preg->allocated = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
789 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
790 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
791 #ifdef DEBUG |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
792 /* Note: length+1 will not overflow since it is checked in init_dfa. */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
793 dfa->re_str = re_malloc (char, length + 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
794 strncpy (dfa->re_str, pattern, length + 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
795 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
796 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
797 __libc_lock_init (dfa->lock); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
798 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
799 err = re_string_construct (®exp, pattern, length, preg->translate, |
10078
f47c913858de
Fix violation of <stdbool.h> replacement in regex.
Eric Blake <ebb9@byu.net>
parents:
10075
diff
changeset
|
800 (syntax & RE_ICASE) != 0, dfa); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
801 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
802 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
803 re_compile_internal_free_return: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
804 free_workarea_compile (preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
805 re_string_destruct (®exp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
806 free_dfa_content (dfa); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
807 preg->buffer = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
808 preg->allocated = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
809 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
810 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
811 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
812 /* Parse the regular expression, and build a structure tree. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
813 preg->re_nsub = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
814 dfa->str_tree = parse (®exp, preg, syntax, &err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
815 if (BE (dfa->str_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
816 goto re_compile_internal_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
817 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
818 /* Analyze the tree and create the nfa. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
819 err = analyze (preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
820 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
821 goto re_compile_internal_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
822 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
823 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
824 /* If possible, do searching in single byte encoding to speed things up. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
825 if (dfa->is_utf8 && !(syntax & RE_ICASE) && preg->translate == NULL) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
826 optimize_utf8 (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
827 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
828 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
829 /* Then create the initial state of the dfa. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
830 err = create_initial_state (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
831 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
832 /* Release work areas. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
833 free_workarea_compile (preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
834 re_string_destruct (®exp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
835 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
836 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
837 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
838 free_dfa_content (dfa); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
839 preg->buffer = NULL; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
840 preg->allocated = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
841 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
842 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
843 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
844 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
845 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
846 /* Initialize DFA. We use the length of the regular expression PAT_LEN |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
847 as the initial length of some arrays. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
848 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
849 static reg_errcode_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
850 init_dfa (re_dfa_t *dfa, size_t pat_len) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
851 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
852 __re_size_t table_size; |
12567
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
853 #ifndef _LIBC |
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
854 char *codeset_name; |
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
855 #endif |
6733
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
856 #ifdef RE_ENABLE_I18N |
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
857 size_t max_i18n_object_size = MAX (sizeof (wchar_t), sizeof (wctype_t)); |
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
858 #else |
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
859 size_t max_i18n_object_size = 0; |
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
860 #endif |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
861 size_t max_object_size = |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
862 MAX (sizeof (struct re_state_table_entry), |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
863 MAX (sizeof (re_token_t), |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
864 MAX (sizeof (re_node_set), |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
865 MAX (sizeof (regmatch_t), |
6733
1c9a307d93bd
* regcomp.c (init_dfa): Don't use wchar_t or wctype_t if RE_ENABLE_I18N
Paul Eggert <eggert@cs.ucla.edu>
parents:
6729
diff
changeset
|
866 max_i18n_object_size)))); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
867 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
868 memset (dfa, '\0', sizeof (re_dfa_t)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
869 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
870 /* Force allocation of str_tree_storage the first time. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
871 dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
872 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
873 /* Avoid overflows. The extra "/ 2" is for the table_size doubling |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
874 calculation below, and for similar doubling calculations |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
875 elsewhere. And it's <= rather than <, because some of the |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
876 doubling calculations add 1 afterwards. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
877 if (BE (SIZE_MAX / max_object_size / 2 <= pat_len, 0)) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
878 return REG_ESPACE; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
879 |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
880 dfa->nodes_alloc = pat_len + 1; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
881 dfa->nodes = re_malloc (re_token_t, dfa->nodes_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
882 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
883 /* table_size = 2 ^ ceil(log pat_len) */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
884 for (table_size = 1; ; table_size <<= 1) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
885 if (table_size > pat_len) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
886 break; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
887 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
888 dfa->state_table = calloc (sizeof (struct re_state_table_entry), table_size); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
889 dfa->state_hash_mask = table_size - 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
890 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
891 dfa->mb_cur_max = MB_CUR_MAX; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
892 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
893 if (dfa->mb_cur_max == 6 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
894 && strcmp (_NL_CURRENT (LC_CTYPE, _NL_CTYPE_CODESET_NAME), "UTF-8") == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
895 dfa->is_utf8 = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
896 dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
897 != 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
898 #else |
12567
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
899 codeset_name = nl_langinfo (CODESET); |
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
900 if (strcasecmp (codeset_name, "UTF-8") == 0 |
ceb1562f60a5
regcomp: sync from glibc; always use nl_langinfo
Jim Meyering <meyering@redhat.com>
parents:
12559
diff
changeset
|
901 || strcasecmp (codeset_name, "UTF8") == 0) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
902 dfa->is_utf8 = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
903 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
904 /* We check exhaustively in the loop below if this charset is a |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
905 superset of ASCII. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
906 dfa->map_notascii = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
907 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
908 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
909 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
910 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
911 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
912 if (dfa->is_utf8) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
913 dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
914 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
915 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
916 int i, j, ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
917 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
918 dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
919 if (BE (dfa->sb_char == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
920 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
921 |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
922 /* Set the bits corresponding to single byte chars. */ |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
923 for (i = 0, ch = 0; i < BITSET_WORDS; ++i) |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
924 for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
925 { |
5972
aa260da0bbbe
* config/srclist.txt: Comment out regcomp.c, since we have a porting fix
Paul Eggert <eggert@cs.ucla.edu>
parents:
5968
diff
changeset
|
926 wint_t wch = __btowc (ch); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
927 if (wch != WEOF) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
928 dfa->sb_char[i] |= (bitset_word_t) 1 << j; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
929 # ifndef _LIBC |
5972
aa260da0bbbe
* config/srclist.txt: Comment out regcomp.c, since we have a porting fix
Paul Eggert <eggert@cs.ucla.edu>
parents:
5968
diff
changeset
|
930 if (isascii (ch) && wch != ch) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
931 dfa->map_notascii = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
932 # endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
933 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
934 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
935 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
936 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
937 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
938 if (BE (dfa->nodes == NULL || dfa->state_table == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
939 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
940 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
941 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
942 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
943 /* Initialize WORD_CHAR table, which indicate which character is |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
944 "word". In this case "word" means that it is the word construction |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
945 character used by some operators like "\<", "\>", etc. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
946 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
947 static void |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
948 internal_function |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
949 init_word_char (re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
950 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
951 int i, j, ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
952 dfa->word_ops_used = 1; |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
953 for (i = 0, ch = 0; i < BITSET_WORDS; ++i) |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
954 for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
955 if (isalnum (ch) || ch == '_') |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
956 dfa->word_char[i] |= (bitset_word_t) 1 << j; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
957 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
958 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
959 /* Free the work area which are only used while compiling. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
960 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
961 static void |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
962 free_workarea_compile (regex_t *preg) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
963 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
964 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
965 bin_tree_storage_t *storage, *next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
966 for (storage = dfa->str_tree_storage; storage; storage = next) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
967 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
968 next = storage->next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
969 re_free (storage); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
970 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
971 dfa->str_tree_storage = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
972 dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
973 dfa->str_tree = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
974 re_free (dfa->org_indices); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
975 dfa->org_indices = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
976 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
977 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
978 /* Create initial states for all contexts. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
979 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
980 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
981 create_initial_state (re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
982 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
983 Idx first, i; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
984 reg_errcode_t err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
985 re_node_set init_nodes; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
986 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
987 /* Initial states have the epsilon closure of the node which is |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
988 the first node of the regular expression. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
989 first = dfa->str_tree->first->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
990 dfa->init_node = first; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
991 err = re_node_set_init_copy (&init_nodes, dfa->eclosures + first); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
992 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
993 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
994 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
995 /* The back-references which are in initial states can epsilon transit, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
996 since in this case all of the subexpressions can be null. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
997 Then we add epsilon closures of the nodes which are the next nodes of |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
998 the back-references. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
999 if (dfa->nbackref > 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1000 for (i = 0; i < init_nodes.nelem; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1001 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1002 Idx node_idx = init_nodes.elems[i]; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1003 re_token_type_t type = dfa->nodes[node_idx].type; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1004 |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1005 Idx clexp_idx; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1006 if (type != OP_BACK_REF) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1007 continue; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1008 for (clexp_idx = 0; clexp_idx < init_nodes.nelem; ++clexp_idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1009 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1010 re_token_t *clexp_node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1011 clexp_node = dfa->nodes + init_nodes.elems[clexp_idx]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1012 if (clexp_node->type == OP_CLOSE_SUBEXP |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1013 && clexp_node->opr.idx == dfa->nodes[node_idx].opr.idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1014 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1015 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1016 if (clexp_idx == init_nodes.nelem) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1017 continue; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1018 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1019 if (type == OP_BACK_REF) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1020 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1021 Idx dest_idx = dfa->edests[node_idx].elems[0]; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1022 if (!re_node_set_contains (&init_nodes, dest_idx)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1023 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1024 re_node_set_merge (&init_nodes, dfa->eclosures + dest_idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1025 i = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1026 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1027 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1028 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1029 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1030 /* It must be the first time to invoke acquire_state. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1031 dfa->init_state = re_acquire_state_context (&err, dfa, &init_nodes, 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1032 /* We don't check ERR here, since the initial state must not be NULL. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1033 if (BE (dfa->init_state == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1034 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1035 if (dfa->init_state->has_constraint) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1036 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1037 dfa->init_state_word = re_acquire_state_context (&err, dfa, &init_nodes, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1038 CONTEXT_WORD); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1039 dfa->init_state_nl = re_acquire_state_context (&err, dfa, &init_nodes, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1040 CONTEXT_NEWLINE); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1041 dfa->init_state_begbuf = re_acquire_state_context (&err, dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1042 &init_nodes, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1043 CONTEXT_NEWLINE |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1044 | CONTEXT_BEGBUF); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1045 if (BE (dfa->init_state_word == NULL || dfa->init_state_nl == NULL |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1046 || dfa->init_state_begbuf == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1047 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1048 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1049 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1050 dfa->init_state_word = dfa->init_state_nl |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1051 = dfa->init_state_begbuf = dfa->init_state; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1052 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1053 re_node_set_free (&init_nodes); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1054 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1055 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1056 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1057 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1058 /* If it is possible to do searching in single byte encoding instead of UTF-8 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1059 to speed things up, set dfa->mb_cur_max to 1, clear is_utf8 and change |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1060 DFA nodes where needed. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1061 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1062 static void |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1063 optimize_utf8 (re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1064 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1065 Idx node; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1066 int i; |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1067 bool mb_chars = false; |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1068 bool has_period = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1069 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1070 for (node = 0; node < dfa->nodes_len; ++node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1071 switch (dfa->nodes[node].type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1072 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1073 case CHARACTER: |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1074 if (dfa->nodes[node].opr.c >= ASCII_CHARS) |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1075 mb_chars = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1076 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1077 case ANCHOR: |
9494
7cd817e07a16
Fix a 4-year-old used-uninitialized bug in regcomp.c.
Jim Meyering <meyering@redhat.com>
parents:
8153
diff
changeset
|
1078 switch (dfa->nodes[node].opr.ctx_type) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1079 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1080 case LINE_FIRST: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1081 case LINE_LAST: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1082 case BUF_FIRST: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1083 case BUF_LAST: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1084 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1085 default: |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1086 /* Word anchors etc. cannot be handled. It's okay to test |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1087 opr.ctx_type since constraints (for all DFA nodes) are |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1088 created by ORing one or more opr.ctx_type values. */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1089 return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1090 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1091 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1092 case OP_PERIOD: |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1093 has_period = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1094 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1095 case OP_BACK_REF: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1096 case OP_ALT: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1097 case END_OF_RE: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1098 case OP_DUP_ASTERISK: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1099 case OP_OPEN_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1100 case OP_CLOSE_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1101 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1102 case COMPLEX_BRACKET: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1103 return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1104 case SIMPLE_BRACKET: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1105 /* Just double check. */ |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1106 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1107 int rshift = (ASCII_CHARS % BITSET_WORD_BITS == 0 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1108 ? 0 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1109 : BITSET_WORD_BITS - ASCII_CHARS % BITSET_WORD_BITS); |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1110 for (i = ASCII_CHARS / BITSET_WORD_BITS; i < BITSET_WORDS; ++i) |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1111 { |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1112 if (dfa->nodes[node].opr.sbcset[i] >> rshift != 0) |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1113 return; |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1114 rshift = 0; |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1115 } |
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1116 } |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1117 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1118 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1119 abort (); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1120 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1121 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1122 if (mb_chars || has_period) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1123 for (node = 0; node < dfa->nodes_len; ++node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1124 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1125 if (dfa->nodes[node].type == CHARACTER |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1126 && dfa->nodes[node].opr.c >= ASCII_CHARS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1127 dfa->nodes[node].mb_partial = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1128 else if (dfa->nodes[node].type == OP_PERIOD) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1129 dfa->nodes[node].type = OP_UTF8_PERIOD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1130 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1131 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1132 /* The search can be in single byte locale. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1133 dfa->mb_cur_max = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1134 dfa->is_utf8 = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1135 dfa->has_mb_node = dfa->nbackref > 0 || has_period; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1136 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1137 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1138 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1139 /* Analyze the structure tree, and calculate "first", "next", "edest", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1140 "eclosure", and "inveclosure". */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1141 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1142 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1143 analyze (regex_t *preg) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1144 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1145 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1146 reg_errcode_t ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1147 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1148 /* Allocate arrays. */ |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1149 dfa->nexts = re_malloc (Idx, dfa->nodes_alloc); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1150 dfa->org_indices = re_malloc (Idx, dfa->nodes_alloc); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1151 dfa->edests = re_malloc (re_node_set, dfa->nodes_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1152 dfa->eclosures = re_malloc (re_node_set, dfa->nodes_alloc); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1153 if (BE (dfa->nexts == NULL || dfa->org_indices == NULL || dfa->edests == NULL |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1154 || dfa->eclosures == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1155 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1156 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1157 dfa->subexp_map = re_malloc (Idx, preg->re_nsub); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1158 if (dfa->subexp_map != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1159 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1160 Idx i; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1161 for (i = 0; i < preg->re_nsub; i++) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1162 dfa->subexp_map[i] = i; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1163 preorder (dfa->str_tree, optimize_subexps, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1164 for (i = 0; i < preg->re_nsub; i++) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1165 if (dfa->subexp_map[i] != i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1166 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1167 if (i == preg->re_nsub) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1168 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1169 free (dfa->subexp_map); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1170 dfa->subexp_map = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1171 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1172 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1173 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1174 ret = postorder (dfa->str_tree, lower_subexps, preg); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1175 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1176 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1177 ret = postorder (dfa->str_tree, calc_first, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1178 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1179 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1180 preorder (dfa->str_tree, calc_next, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1181 ret = preorder (dfa->str_tree, link_nfa_nodes, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1182 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1183 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1184 ret = calc_eclosure (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1185 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1186 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1187 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1188 /* We only need this during the prune_impossible_nodes pass in regexec.c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1189 skip it if p_i_n will not run, as calc_inveclosure can be quadratic. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1190 if ((!preg->no_sub && preg->re_nsub > 0 && dfa->has_plural_match) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1191 || dfa->nbackref) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1192 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1193 dfa->inveclosures = re_malloc (re_node_set, dfa->nodes_len); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1194 if (BE (dfa->inveclosures == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1195 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1196 ret = calc_inveclosure (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1197 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1198 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1199 return ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1200 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1201 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1202 /* Our parse trees are very unbalanced, so we cannot use a stack to |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1203 implement parse tree visits. Instead, we use parent pointers and |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1204 some hairy code in these two functions. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1205 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1206 postorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1207 void *extra) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1208 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1209 bin_tree_t *node, *prev; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1210 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1211 for (node = root; ; ) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1212 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1213 /* Descend down the tree, preferably to the left (or to the right |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1214 if that's the only child). */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1215 while (node->left || node->right) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1216 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1217 node = node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1218 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1219 node = node->right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1220 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1221 do |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1222 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1223 reg_errcode_t err = fn (extra, node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1224 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1225 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1226 if (node->parent == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1227 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1228 prev = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1229 node = node->parent; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1230 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1231 /* Go up while we have a node that is reached from the right. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1232 while (node->right == prev || node->right == NULL); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1233 node = node->right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1234 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1235 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1236 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1237 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1238 preorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1239 void *extra) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1240 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1241 bin_tree_t *node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1242 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1243 for (node = root; ; ) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1244 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1245 reg_errcode_t err = fn (extra, node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1246 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1247 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1248 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1249 /* Go to the left node, or up and to the right. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1250 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1251 node = node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1252 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1253 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1254 bin_tree_t *prev = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1255 while (node->right == prev || node->right == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1256 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1257 prev = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1258 node = node->parent; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1259 if (!node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1260 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1261 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1262 node = node->right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1263 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1264 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1265 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1266 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1267 /* Optimization pass: if a SUBEXP is entirely contained, strip it and tell |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1268 re_search_internal to map the inner one's opr.idx to this one's. Adjust |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1269 backreferences as well. Requires a preorder visit. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1270 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1271 optimize_subexps (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1272 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1273 re_dfa_t *dfa = (re_dfa_t *) extra; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1274 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1275 if (node->token.type == OP_BACK_REF && dfa->subexp_map) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1276 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1277 int idx = node->token.opr.idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1278 node->token.opr.idx = dfa->subexp_map[idx]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1279 dfa->used_bkref_map |= 1 << node->token.opr.idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1280 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1281 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1282 else if (node->token.type == SUBEXP |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1283 && node->left && node->left->token.type == SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1284 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1285 Idx other_idx = node->left->token.opr.idx; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1286 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1287 node->left = node->left->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1288 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1289 node->left->parent = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1290 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1291 dfa->subexp_map[other_idx] = dfa->subexp_map[node->token.opr.idx]; |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
1292 if (other_idx < BITSET_WORD_BITS) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1293 dfa->used_bkref_map &= ~((bitset_word_t) 1 << other_idx); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1294 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1295 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1296 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1297 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1298 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1299 /* Lowering pass: Turn each SUBEXP node into the appropriate concatenation |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1300 of OP_OPEN_SUBEXP, the body of the SUBEXP (if any) and OP_CLOSE_SUBEXP. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1301 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1302 lower_subexps (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1303 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1304 regex_t *preg = (regex_t *) extra; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1305 reg_errcode_t err = REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1306 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1307 if (node->left && node->left->token.type == SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1308 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1309 node->left = lower_subexp (&err, preg, node->left); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1310 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1311 node->left->parent = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1312 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1313 if (node->right && node->right->token.type == SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1314 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1315 node->right = lower_subexp (&err, preg, node->right); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1316 if (node->right) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1317 node->right->parent = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1318 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1319 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1320 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1321 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1322 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1323 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1324 lower_subexp (reg_errcode_t *err, regex_t *preg, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1325 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1326 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1327 bin_tree_t *body = node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1328 bin_tree_t *op, *cls, *tree1, *tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1329 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1330 if (preg->no_sub |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1331 /* We do not optimize empty subexpressions, because otherwise we may |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1332 have bad CONCAT nodes with NULL children. This is obviously not |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1333 very common, so we do not lose much. An example that triggers |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1334 this case is the sed "script" /\(\)/x. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1335 && node->left != NULL |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1336 && (node->token.opr.idx >= BITSET_WORD_BITS |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1337 || !(dfa->used_bkref_map |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1338 & ((bitset_word_t) 1 << node->token.opr.idx)))) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1339 return node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1340 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1341 /* Convert the SUBEXP node to the concatenation of an |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1342 OP_OPEN_SUBEXP, the contents, and an OP_CLOSE_SUBEXP. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1343 op = create_tree (dfa, NULL, NULL, OP_OPEN_SUBEXP); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1344 cls = create_tree (dfa, NULL, NULL, OP_CLOSE_SUBEXP); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1345 tree1 = body ? create_tree (dfa, body, cls, CONCAT) : cls; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1346 tree = create_tree (dfa, op, tree1, CONCAT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1347 if (BE (tree == NULL || tree1 == NULL || op == NULL || cls == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1348 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1349 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1350 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1351 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1352 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1353 op->token.opr.idx = cls->token.opr.idx = node->token.opr.idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1354 op->token.opt_subexp = cls->token.opt_subexp = node->token.opt_subexp; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1355 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1356 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1357 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1358 /* Pass 1 in building the NFA: compute FIRST and create unlinked automaton |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1359 nodes. Requires a postorder visit. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1360 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1361 calc_first (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1362 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1363 re_dfa_t *dfa = (re_dfa_t *) extra; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1364 if (node->token.type == CONCAT) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1365 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1366 node->first = node->left->first; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1367 node->node_idx = node->left->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1368 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1369 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1370 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1371 node->first = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1372 node->node_idx = re_dfa_add_node (dfa, node->token); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1373 if (BE (node->node_idx == REG_MISSING, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1374 return REG_ESPACE; |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1375 if (node->token.type == ANCHOR) |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1376 dfa->nodes[node->node_idx].constraint = node->token.opr.ctx_type; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1377 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1378 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1379 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1380 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1381 /* Pass 2: compute NEXT on the tree. Preorder visit. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1382 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1383 calc_next (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1384 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1385 switch (node->token.type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1386 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1387 case OP_DUP_ASTERISK: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1388 node->left->next = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1389 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1390 case CONCAT: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1391 node->left->next = node->right->first; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1392 node->right->next = node->next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1393 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1394 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1395 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1396 node->left->next = node->next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1397 if (node->right) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1398 node->right->next = node->next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1399 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1400 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1401 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1402 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1403 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1404 /* Pass 3: link all DFA nodes to their NEXT node (any order will do). */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1405 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1406 link_nfa_nodes (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1407 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1408 re_dfa_t *dfa = (re_dfa_t *) extra; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1409 Idx idx = node->node_idx; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1410 reg_errcode_t err = REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1411 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1412 switch (node->token.type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1413 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1414 case CONCAT: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1415 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1416 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1417 case END_OF_RE: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1418 assert (node->next == NULL); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1419 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1420 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1421 case OP_DUP_ASTERISK: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1422 case OP_ALT: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1423 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1424 Idx left, right; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1425 dfa->has_plural_match = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1426 if (node->left != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1427 left = node->left->first->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1428 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1429 left = node->next->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1430 if (node->right != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1431 right = node->right->first->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1432 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1433 right = node->next->node_idx; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1434 assert (REG_VALID_INDEX (left)); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1435 assert (REG_VALID_INDEX (right)); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1436 err = re_node_set_init_2 (dfa->edests + idx, left, right); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1437 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1438 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1439 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1440 case ANCHOR: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1441 case OP_OPEN_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1442 case OP_CLOSE_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1443 err = re_node_set_init_1 (dfa->edests + idx, node->next->node_idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1444 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1445 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1446 case OP_BACK_REF: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1447 dfa->nexts[idx] = node->next->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1448 if (node->token.type == OP_BACK_REF) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1449 re_node_set_init_1 (dfa->edests + idx, dfa->nexts[idx]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1450 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1451 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1452 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1453 assert (!IS_EPSILON_NODE (node->token.type)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1454 dfa->nexts[idx] = node->next->node_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1455 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1456 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1457 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1458 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1459 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1460 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1461 /* Duplicate the epsilon closure of the node ROOT_NODE. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1462 Note that duplicated nodes have constraint INIT_CONSTRAINT in addition |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1463 to their own constraint. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1464 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1465 static reg_errcode_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1466 internal_function |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1467 duplicate_node_closure (re_dfa_t *dfa, Idx top_org_node, Idx top_clone_node, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1468 Idx root_node, unsigned int init_constraint) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1469 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1470 Idx org_node, clone_node; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1471 bool ok; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1472 unsigned int constraint = init_constraint; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1473 for (org_node = top_org_node, clone_node = top_clone_node;;) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1474 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1475 Idx org_dest, clone_dest; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1476 if (dfa->nodes[org_node].type == OP_BACK_REF) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1477 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1478 /* If the back reference epsilon-transit, its destination must |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1479 also have the constraint. Then duplicate the epsilon closure |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1480 of the destination of the back reference, and store it in |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1481 edests of the back reference. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1482 org_dest = dfa->nexts[org_node]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1483 re_node_set_empty (dfa->edests + clone_node); |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1484 clone_dest = duplicate_node (dfa, org_dest, constraint); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1485 if (BE (clone_dest == REG_MISSING, 0)) |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1486 return REG_ESPACE; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1487 dfa->nexts[clone_node] = dfa->nexts[org_node]; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1488 ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1489 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1490 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1491 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1492 else if (dfa->edests[org_node].nelem == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1493 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1494 /* In case of the node can't epsilon-transit, don't duplicate the |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1495 destination and store the original destination as the |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1496 destination of the node. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1497 dfa->nexts[clone_node] = dfa->nexts[org_node]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1498 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1499 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1500 else if (dfa->edests[org_node].nelem == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1501 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1502 /* In case of the node can epsilon-transit, and it has only one |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1503 destination. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1504 org_dest = dfa->edests[org_node].elems[0]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1505 re_node_set_empty (dfa->edests + clone_node); |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1506 /* If the node is root_node itself, it means the epsilon closure |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1507 has a loop. Then tie it to the destination of the root_node. */ |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1508 if (org_node == root_node && clone_node != org_node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1509 { |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1510 ok = re_node_set_insert (dfa->edests + clone_node, org_dest); |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1511 if (BE (! ok, 0)) |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1512 return REG_ESPACE; |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1513 break; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1514 } |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1515 /* In case the node has another constraint, append it. */ |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1516 constraint |= dfa->nodes[org_node].constraint; |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1517 clone_dest = duplicate_node (dfa, org_dest, constraint); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1518 if (BE (clone_dest == REG_MISSING, 0)) |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1519 return REG_ESPACE; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1520 ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1521 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1522 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1523 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1524 else /* dfa->edests[org_node].nelem == 2 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1525 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1526 /* In case of the node can epsilon-transit, and it has two |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1527 destinations. In the bin_tree_t and DFA, that's '|' and '*'. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1528 org_dest = dfa->edests[org_node].elems[0]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1529 re_node_set_empty (dfa->edests + clone_node); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1530 /* Search for a duplicated node which satisfies the constraint. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1531 clone_dest = search_duplicated_node (dfa, org_dest, constraint); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1532 if (clone_dest == REG_MISSING) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1533 { |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1534 /* There is no such duplicated node, create a new one. */ |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1535 reg_errcode_t err; |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1536 clone_dest = duplicate_node (dfa, org_dest, constraint); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1537 if (BE (clone_dest == REG_MISSING, 0)) |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1538 return REG_ESPACE; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1539 ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1540 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1541 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1542 err = duplicate_node_closure (dfa, org_dest, clone_dest, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1543 root_node, constraint); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1544 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1545 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1546 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1547 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1548 { |
12569
51ea446bf1f8
regcomp: fix typo in comment
Jim Meyering <meyering@redhat.com>
parents:
12568
diff
changeset
|
1549 /* There is a duplicated node which satisfies the constraint, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1550 use it to avoid infinite loop. */ |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1551 ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1552 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1553 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1554 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1555 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1556 org_dest = dfa->edests[org_node].elems[1]; |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1557 clone_dest = duplicate_node (dfa, org_dest, constraint); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1558 if (BE (clone_dest == REG_MISSING, 0)) |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1559 return REG_ESPACE; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1560 ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1561 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1562 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1563 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1564 org_node = org_dest; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1565 clone_node = clone_dest; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1566 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1567 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1568 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1569 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1570 /* Search for a node which is duplicated from the node ORG_NODE, and |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1571 satisfies the constraint CONSTRAINT. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1572 |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1573 static Idx |
6185
6b09f7f6ba73
* lib/regcomp.c (search_duplicated_node): Make first pointer arg
Paul Eggert <eggert@cs.ucla.edu>
parents:
6184
diff
changeset
|
1574 search_duplicated_node (const re_dfa_t *dfa, Idx org_node, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1575 unsigned int constraint) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1576 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1577 Idx idx; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1578 for (idx = dfa->nodes_len - 1; dfa->nodes[idx].duplicated && idx > 0; --idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1579 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1580 if (org_node == dfa->org_indices[idx] |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1581 && constraint == dfa->nodes[idx].constraint) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1582 return idx; /* Found. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1583 } |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1584 return REG_MISSING; /* Not found. */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1585 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1586 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1587 /* Duplicate the node whose index is ORG_IDX and set the constraint CONSTRAINT. |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1588 Return the index of the new node, or REG_MISSING if insufficient storage is |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1589 available. */ |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1590 |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1591 static Idx |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1592 duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1593 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1594 Idx dup_idx = re_dfa_add_node (dfa, dfa->nodes[org_idx]); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1595 if (BE (dup_idx != REG_MISSING, 1)) |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1596 { |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1597 dfa->nodes[dup_idx].constraint = constraint; |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1598 dfa->nodes[dup_idx].constraint |= dfa->nodes[org_idx].constraint; |
6071
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1599 dfa->nodes[dup_idx].duplicated = 1; |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1600 |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1601 /* Store the index of the original node. */ |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1602 dfa->org_indices[dup_idx] = org_idx; |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1603 } |
c1760162e42f
(duplicate_node): Return new index, not an error code,
Paul Eggert <eggert@cs.ucla.edu>
parents:
5972
diff
changeset
|
1604 return dup_idx; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1605 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1606 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1607 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1608 calc_inveclosure (re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1609 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1610 Idx src, idx; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1611 bool ok; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1612 for (idx = 0; idx < dfa->nodes_len; ++idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1613 re_node_set_init_empty (dfa->inveclosures + idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1614 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1615 for (src = 0; src < dfa->nodes_len; ++src) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1616 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1617 Idx *elems = dfa->eclosures[src].elems; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1618 for (idx = 0; idx < dfa->eclosures[src].nelem; ++idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1619 { |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1620 ok = re_node_set_insert_last (dfa->inveclosures + elems[idx], src); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1621 if (BE (! ok, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1622 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1623 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1624 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1625 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1626 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1627 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1628 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1629 /* Calculate "eclosure" for all the node in DFA. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1630 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1631 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1632 calc_eclosure (re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1633 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1634 Idx node_idx; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1635 bool incomplete; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1636 #ifdef DEBUG |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1637 assert (dfa->nodes_len > 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1638 #endif |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1639 incomplete = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1640 /* For each nodes, calculate epsilon closure. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1641 for (node_idx = 0; ; ++node_idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1642 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1643 reg_errcode_t err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1644 re_node_set eclosure_elem; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1645 if (node_idx == dfa->nodes_len) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1646 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1647 if (!incomplete) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1648 break; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1649 incomplete = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1650 node_idx = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1651 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1652 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1653 #ifdef DEBUG |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1654 assert (dfa->eclosures[node_idx].nelem != REG_MISSING); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1655 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1656 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1657 /* If we have already calculated, skip it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1658 if (dfa->eclosures[node_idx].nelem != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1659 continue; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1660 /* Calculate epsilon closure of `node_idx'. */ |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1661 err = calc_eclosure_iter (&eclosure_elem, dfa, node_idx, true); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1662 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1663 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1664 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1665 if (dfa->eclosures[node_idx].nelem == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1666 { |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1667 incomplete = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1668 re_node_set_free (&eclosure_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1669 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1670 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1671 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1672 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1673 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1674 /* Calculate epsilon closure of NODE. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1675 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1676 static reg_errcode_t |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1677 calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, Idx node, bool root) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1678 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1679 reg_errcode_t err; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1680 Idx i; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1681 bool incomplete; |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1682 bool ok; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1683 re_node_set eclosure; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1684 incomplete = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1685 err = re_node_set_alloc (&eclosure, dfa->edests[node].nelem + 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1686 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1687 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1688 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1689 /* This indicates that we are calculating this node now. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1690 We reference this value to avoid infinite loop. */ |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1691 dfa->eclosures[node].nelem = REG_MISSING; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1692 |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1693 /* If the current node has constraints, duplicate all nodes |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1694 since they must inherit the constraints. */ |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1695 if (dfa->nodes[node].constraint |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1696 && dfa->edests[node].nelem |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1697 && !dfa->nodes[dfa->edests[node].elems[0]].duplicated) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1698 { |
10075
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1699 err = duplicate_node_closure (dfa, node, node, node, |
cc7bfc9f7fc1
optimize double anchors such as ^$
Paolo Bonzini <bonzini@gnu.org>
parents:
9494
diff
changeset
|
1700 dfa->nodes[node].constraint); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1701 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1702 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1703 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1704 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1705 /* Expand each epsilon destination nodes. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1706 if (IS_EPSILON_NODE(dfa->nodes[node].type)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1707 for (i = 0; i < dfa->edests[node].nelem; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1708 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1709 re_node_set eclosure_elem; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1710 Idx edest = dfa->edests[node].elems[i]; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1711 /* If calculating the epsilon closure of `edest' is in progress, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1712 return intermediate result. */ |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
1713 if (dfa->eclosures[edest].nelem == REG_MISSING) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1714 { |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1715 incomplete = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1716 continue; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1717 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1718 /* If we haven't calculated the epsilon closure of `edest' yet, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1719 calculate now. Otherwise use calculated epsilon closure. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1720 if (dfa->eclosures[edest].nelem == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1721 { |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1722 err = calc_eclosure_iter (&eclosure_elem, dfa, edest, false); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1723 if (BE (err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1724 return err; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1725 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1726 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1727 eclosure_elem = dfa->eclosures[edest]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1728 /* Merge the epsilon closure of `edest'. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1729 re_node_set_merge (&eclosure, &eclosure_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1730 /* If the epsilon closure of `edest' is incomplete, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1731 the epsilon closure of this node is also incomplete. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1732 if (dfa->eclosures[edest].nelem == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1733 { |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1734 incomplete = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1735 re_node_set_free (&eclosure_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1736 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1737 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1738 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1739 /* Epsilon closures include itself. */ |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1740 ok = re_node_set_insert (&eclosure, node); |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1741 if (BE (! ok, 0)) |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
1742 return REG_ESPACE; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1743 if (incomplete && !root) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1744 dfa->eclosures[node].nelem = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1745 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1746 dfa->eclosures[node] = eclosure; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1747 *new_set = eclosure; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1748 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1749 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1750 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1751 /* Functions for token which are used in the parser. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1752 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1753 /* Fetch a token from INPUT. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1754 We must not use this function inside bracket expressions. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1755 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1756 static void |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1757 internal_function |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1758 fetch_token (re_token_t *result, re_string_t *input, reg_syntax_t syntax) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1759 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1760 re_string_skip_bytes (input, peek_token (result, input, syntax)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1761 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1762 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1763 /* Peek a token from INPUT, and return the length of the token. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1764 We must not use this function inside bracket expressions. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1765 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1766 static int |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1767 internal_function |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
1768 peek_token (re_token_t *token, re_string_t *input, reg_syntax_t syntax) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1769 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1770 unsigned char c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1771 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1772 if (re_string_eoi (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1773 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1774 token->type = END_OF_RE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1775 return 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1776 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1777 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1778 c = re_string_peek_byte (input, 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1779 token->opr.c = c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1780 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1781 token->word_char = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1782 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1783 token->mb_partial = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1784 if (input->mb_cur_max > 1 && |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1785 !re_string_first_byte (input, re_string_cur_idx (input))) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1786 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1787 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1788 token->mb_partial = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1789 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1790 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1791 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1792 if (c == '\\') |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1793 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1794 unsigned char c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1795 if (re_string_cur_idx (input) + 1 >= re_string_length (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1796 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1797 token->type = BACK_SLASH; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1798 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1799 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1800 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1801 c2 = re_string_peek_byte_case (input, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1802 token->opr.c = c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1803 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1804 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1805 if (input->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1806 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1807 wint_t wc = re_string_wchar_at (input, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1808 re_string_cur_idx (input) + 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1809 token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1810 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1811 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1812 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1813 token->word_char = IS_WORD_CHAR (c2) != 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1814 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1815 switch (c2) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1816 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1817 case '|': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1818 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_NO_BK_VBAR)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1819 token->type = OP_ALT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1820 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1821 case '1': case '2': case '3': case '4': case '5': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1822 case '6': case '7': case '8': case '9': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1823 if (!(syntax & RE_NO_BK_REFS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1824 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1825 token->type = OP_BACK_REF; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1826 token->opr.idx = c2 - '1'; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1827 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1828 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1829 case '<': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1830 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1831 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1832 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1833 token->opr.ctx_type = WORD_FIRST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1834 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1835 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1836 case '>': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1837 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1838 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1839 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1840 token->opr.ctx_type = WORD_LAST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1841 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1842 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1843 case 'b': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1844 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1845 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1846 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1847 token->opr.ctx_type = WORD_DELIM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1848 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1849 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1850 case 'B': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1851 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1852 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1853 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1854 token->opr.ctx_type = NOT_WORD_DELIM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1855 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1856 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1857 case 'w': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1858 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1859 token->type = OP_WORD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1860 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1861 case 'W': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1862 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1863 token->type = OP_NOTWORD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1864 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1865 case 's': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1866 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1867 token->type = OP_SPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1868 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1869 case 'S': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1870 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1871 token->type = OP_NOTSPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1872 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1873 case '`': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1874 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1875 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1876 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1877 token->opr.ctx_type = BUF_FIRST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1878 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1879 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1880 case '\'': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1881 if (!(syntax & RE_NO_GNU_OPS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1882 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1883 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1884 token->opr.ctx_type = BUF_LAST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1885 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1886 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1887 case '(': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1888 if (!(syntax & RE_NO_BK_PARENS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1889 token->type = OP_OPEN_SUBEXP; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1890 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1891 case ')': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1892 if (!(syntax & RE_NO_BK_PARENS)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1893 token->type = OP_CLOSE_SUBEXP; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1894 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1895 case '+': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1896 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1897 token->type = OP_DUP_PLUS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1898 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1899 case '?': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1900 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1901 token->type = OP_DUP_QUESTION; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1902 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1903 case '{': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1904 if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1905 token->type = OP_OPEN_DUP_NUM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1906 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1907 case '}': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1908 if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1909 token->type = OP_CLOSE_DUP_NUM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1910 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1911 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1912 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1913 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1914 return 2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1915 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1916 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1917 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1918 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1919 if (input->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1920 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1921 wint_t wc = re_string_wchar_at (input, re_string_cur_idx (input)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1922 token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1923 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1924 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1925 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1926 token->word_char = IS_WORD_CHAR (token->opr.c); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1927 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1928 switch (c) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1929 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1930 case '\n': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1931 if (syntax & RE_NEWLINE_ALT) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1932 token->type = OP_ALT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1933 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1934 case '|': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1935 if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_NO_BK_VBAR)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1936 token->type = OP_ALT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1937 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1938 case '*': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1939 token->type = OP_DUP_ASTERISK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1940 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1941 case '+': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1942 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1943 token->type = OP_DUP_PLUS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1944 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1945 case '?': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1946 if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1947 token->type = OP_DUP_QUESTION; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1948 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1949 case '{': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1950 if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1951 token->type = OP_OPEN_DUP_NUM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1952 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1953 case '}': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1954 if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1955 token->type = OP_CLOSE_DUP_NUM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1956 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1957 case '(': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1958 if (syntax & RE_NO_BK_PARENS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1959 token->type = OP_OPEN_SUBEXP; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1960 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1961 case ')': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1962 if (syntax & RE_NO_BK_PARENS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1963 token->type = OP_CLOSE_SUBEXP; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1964 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1965 case '[': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1966 token->type = OP_OPEN_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1967 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1968 case '.': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1969 token->type = OP_PERIOD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1970 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1971 case '^': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1972 if (!(syntax & (RE_CONTEXT_INDEP_ANCHORS | RE_CARET_ANCHORS_HERE)) && |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1973 re_string_cur_idx (input) != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1974 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1975 char prev = re_string_peek_byte (input, -1); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1976 if (!(syntax & RE_NEWLINE_ALT) || prev != '\n') |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1977 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1978 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1979 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1980 token->opr.ctx_type = LINE_FIRST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1981 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1982 case '$': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
1983 if (!(syntax & RE_CONTEXT_INDEP_ANCHORS) && |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1984 re_string_cur_idx (input) + 1 != re_string_length (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1985 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1986 re_token_t next; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1987 re_string_skip_bytes (input, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1988 peek_token (&next, input, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1989 re_string_skip_bytes (input, -1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1990 if (next.type != OP_ALT && next.type != OP_CLOSE_SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1991 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1992 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1993 token->type = ANCHOR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1994 token->opr.ctx_type = LINE_LAST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1995 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1996 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1997 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1998 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
1999 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2000 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2001 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2002 /* Peek a token from INPUT, and return the length of the token. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2003 We must not use this function out of bracket expressions. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2004 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2005 static int |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2006 internal_function |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2007 peek_token_bracket (re_token_t *token, re_string_t *input, reg_syntax_t syntax) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2008 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2009 unsigned char c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2010 if (re_string_eoi (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2011 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2012 token->type = END_OF_RE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2013 return 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2014 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2015 c = re_string_peek_byte (input, 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2016 token->opr.c = c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2017 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2018 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2019 if (input->mb_cur_max > 1 && |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2020 !re_string_first_byte (input, re_string_cur_idx (input))) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2021 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2022 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2023 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2024 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2025 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2026 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2027 if (c == '\\' && (syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2028 && re_string_cur_idx (input) + 1 < re_string_length (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2029 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2030 /* In this case, '\' escape a character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2031 unsigned char c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2032 re_string_skip_bytes (input, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2033 c2 = re_string_peek_byte (input, 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2034 token->opr.c = c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2035 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2036 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2037 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2038 if (c == '[') /* '[' is a special char in a bracket exps. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2039 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2040 unsigned char c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2041 int token_len; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2042 if (re_string_cur_idx (input) + 1 < re_string_length (input)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2043 c2 = re_string_peek_byte (input, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2044 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2045 c2 = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2046 token->opr.c = c2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2047 token_len = 2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2048 switch (c2) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2049 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2050 case '.': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2051 token->type = OP_OPEN_COLL_ELEM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2052 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2053 case '=': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2054 token->type = OP_OPEN_EQUIV_CLASS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2055 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2056 case ':': |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2057 if (syntax & RE_CHAR_CLASSES) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2058 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2059 token->type = OP_OPEN_CHAR_CLASS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2060 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2061 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2062 /* else fall through. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2063 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2064 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2065 token->opr.c = c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2066 token_len = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2067 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2068 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2069 return token_len; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2070 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2071 switch (c) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2072 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2073 case '-': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2074 token->type = OP_CHARSET_RANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2075 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2076 case ']': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2077 token->type = OP_CLOSE_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2078 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2079 case '^': |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2080 token->type = OP_NON_MATCH_LIST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2081 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2082 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2083 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2084 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2085 return 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2086 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2087 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2088 /* Functions for parser. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2089 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2090 /* Entry point of the parser. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2091 Parse the regular expression REGEXP and return the structure tree. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2092 If an error is occured, ERR is set by error code, and return NULL. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2093 This function build the following tree, from regular expression <reg_exp>: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2094 CAT |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2095 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2096 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2097 <reg_exp> EOR |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2098 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2099 CAT means concatenation. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2100 EOR means end of regular expression. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2101 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2102 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2103 parse (re_string_t *regexp, regex_t *preg, reg_syntax_t syntax, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2104 reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2105 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2106 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2107 bin_tree_t *tree, *eor, *root; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2108 re_token_t current_token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2109 dfa->syntax = syntax; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2110 fetch_token (¤t_token, regexp, syntax | RE_CARET_ANCHORS_HERE); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2111 tree = parse_reg_exp (regexp, preg, ¤t_token, syntax, 0, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2112 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2113 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2114 eor = create_tree (dfa, NULL, NULL, END_OF_RE); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2115 if (tree != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2116 root = create_tree (dfa, tree, eor, CONCAT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2117 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2118 root = eor; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2119 if (BE (eor == NULL || root == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2120 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2121 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2122 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2123 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2124 return root; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2125 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2126 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2127 /* This function build the following tree, from regular expression |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2128 <branch1>|<branch2>: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2129 ALT |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2130 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2131 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2132 <branch1> <branch2> |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2133 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2134 ALT means alternative, which represents the operator `|'. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2135 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2136 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2137 parse_reg_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2138 reg_syntax_t syntax, Idx nest, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2139 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2140 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2141 bin_tree_t *tree, *branch = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2142 tree = parse_branch (regexp, preg, token, syntax, nest, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2143 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2144 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2145 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2146 while (token->type == OP_ALT) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2147 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2148 fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2149 if (token->type != OP_ALT && token->type != END_OF_RE |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2150 && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2151 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2152 branch = parse_branch (regexp, preg, token, syntax, nest, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2153 if (BE (*err != REG_NOERROR && branch == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2154 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2155 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2156 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2157 branch = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2158 tree = create_tree (dfa, tree, branch, OP_ALT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2159 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2160 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2161 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2162 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2163 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2164 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2165 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2166 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2167 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2168 /* This function build the following tree, from regular expression |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2169 <exp1><exp2>: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2170 CAT |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2171 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2172 / \ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2173 <exp1> <exp2> |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2174 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2175 CAT means concatenation. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2176 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2177 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2178 parse_branch (re_string_t *regexp, regex_t *preg, re_token_t *token, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2179 reg_syntax_t syntax, Idx nest, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2180 { |
7694
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2181 bin_tree_t *tree, *expr; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2182 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2183 tree = parse_expression (regexp, preg, token, syntax, nest, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2184 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2185 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2186 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2187 while (token->type != OP_ALT && token->type != END_OF_RE |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2188 && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2189 { |
7694
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2190 expr = parse_expression (regexp, preg, token, syntax, nest, err); |
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2191 if (BE (*err != REG_NOERROR && expr == NULL, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2192 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2193 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2194 } |
7694
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2195 if (tree != NULL && expr != NULL) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2196 { |
7694
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2197 tree = create_tree (dfa, tree, expr, CONCAT); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2198 if (tree == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2199 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2200 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2201 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2202 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2203 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2204 else if (tree == NULL) |
7694
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2205 tree = expr; |
c818925b8298
* lib/regcomp.c (parse_branch): Rename local, exp->expr, to avoid
Jim Meyering <jim@meyering.net>
parents:
6733
diff
changeset
|
2206 /* Otherwise expr == NULL, we don't need to create new tree. */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2207 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2208 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2209 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2210 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2211 /* This function build the following tree, from regular expression a*: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2212 * |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2213 | |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2214 a |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2215 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2216 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2217 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2218 parse_expression (re_string_t *regexp, regex_t *preg, re_token_t *token, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2219 reg_syntax_t syntax, Idx nest, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2220 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2221 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2222 bin_tree_t *tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2223 switch (token->type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2224 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2225 case CHARACTER: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2226 tree = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2227 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2228 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2229 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2230 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2231 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2232 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2233 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2234 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2235 while (!re_string_eoi (regexp) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2236 && !re_string_first_byte (regexp, re_string_cur_idx (regexp))) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2237 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2238 bin_tree_t *mbc_remain; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2239 fetch_token (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2240 mbc_remain = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2241 tree = create_tree (dfa, tree, mbc_remain, CONCAT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2242 if (BE (mbc_remain == NULL || tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2243 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2244 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2245 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2246 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2247 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2248 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2249 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2250 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2251 case OP_OPEN_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2252 tree = parse_sub_exp (regexp, preg, token, syntax, nest + 1, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2253 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2254 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2255 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2256 case OP_OPEN_BRACKET: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2257 tree = parse_bracket_exp (regexp, dfa, token, syntax, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2258 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2259 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2260 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2261 case OP_BACK_REF: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2262 if (!BE (dfa->completed_bkref_map & (1 << token->opr.idx), 1)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2263 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2264 *err = REG_ESUBREG; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2265 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2266 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2267 dfa->used_bkref_map |= 1 << token->opr.idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2268 tree = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2269 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2270 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2271 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2272 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2273 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2274 ++dfa->nbackref; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2275 dfa->has_mb_node = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2276 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2277 case OP_OPEN_DUP_NUM: |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2278 if (syntax & RE_CONTEXT_INVALID_DUP) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2279 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2280 *err = REG_BADRPT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2281 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2282 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2283 /* FALLTHROUGH */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2284 case OP_DUP_ASTERISK: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2285 case OP_DUP_PLUS: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2286 case OP_DUP_QUESTION: |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2287 if (syntax & RE_CONTEXT_INVALID_OPS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2288 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2289 *err = REG_BADRPT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2290 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2291 } |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2292 else if (syntax & RE_CONTEXT_INDEP_OPS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2293 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2294 fetch_token (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2295 return parse_expression (regexp, preg, token, syntax, nest, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2296 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2297 /* else fall through */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2298 case OP_CLOSE_SUBEXP: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2299 if ((token->type == OP_CLOSE_SUBEXP) && |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2300 !(syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2301 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2302 *err = REG_ERPAREN; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2303 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2304 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2305 /* else fall through */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2306 case OP_CLOSE_DUP_NUM: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2307 /* We treat it as a normal character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2308 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2309 /* Then we can these characters as normal characters. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2310 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2311 /* mb_partial and word_char bits should be initialized already |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2312 by peek_token. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2313 tree = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2314 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2315 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2316 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2317 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2318 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2319 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2320 case ANCHOR: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2321 if ((token->opr.ctx_type |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2322 & (WORD_DELIM | NOT_WORD_DELIM | WORD_FIRST | WORD_LAST)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2323 && dfa->word_ops_used == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2324 init_word_char (dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2325 if (token->opr.ctx_type == WORD_DELIM |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2326 || token->opr.ctx_type == NOT_WORD_DELIM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2327 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2328 bin_tree_t *tree_first, *tree_last; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2329 if (token->opr.ctx_type == WORD_DELIM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2330 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2331 token->opr.ctx_type = WORD_FIRST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2332 tree_first = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2333 token->opr.ctx_type = WORD_LAST; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2334 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2335 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2336 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2337 token->opr.ctx_type = INSIDE_WORD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2338 tree_first = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2339 token->opr.ctx_type = INSIDE_NOTWORD; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2340 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2341 tree_last = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2342 tree = create_tree (dfa, tree_first, tree_last, OP_ALT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2343 if (BE (tree_first == NULL || tree_last == NULL || tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2344 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2345 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2346 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2347 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2348 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2349 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2350 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2351 tree = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2352 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2353 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2354 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2355 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2356 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2357 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2358 /* We must return here, since ANCHORs can't be followed |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2359 by repetition operators. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2360 eg. RE"^*" is invalid or "<ANCHOR(^)><CHAR(*)>", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2361 it must not be "<ANCHOR(^)><REPEAT(*)>". */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2362 fetch_token (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2363 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2364 case OP_PERIOD: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2365 tree = create_token_tree (dfa, NULL, NULL, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2366 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2367 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2368 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2369 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2370 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2371 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2372 dfa->has_mb_node = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2373 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2374 case OP_WORD: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2375 case OP_NOTWORD: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2376 tree = build_charclass_op (dfa, regexp->trans, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2377 (const unsigned char *) "alnum", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2378 (const unsigned char *) "_", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2379 token->type == OP_NOTWORD, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2380 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2381 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2382 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2383 case OP_SPACE: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2384 case OP_NOTSPACE: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2385 tree = build_charclass_op (dfa, regexp->trans, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2386 (const unsigned char *) "space", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2387 (const unsigned char *) "", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2388 token->type == OP_NOTSPACE, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2389 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2390 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2391 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2392 case OP_ALT: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2393 case END_OF_RE: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2394 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2395 case BACK_SLASH: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2396 *err = REG_EESCAPE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2397 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2398 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2399 /* Must not happen? */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2400 #ifdef DEBUG |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2401 assert (0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2402 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2403 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2404 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2405 fetch_token (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2406 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2407 while (token->type == OP_DUP_ASTERISK || token->type == OP_DUP_PLUS |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2408 || token->type == OP_DUP_QUESTION || token->type == OP_OPEN_DUP_NUM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2409 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2410 tree = parse_dup_op (tree, regexp, dfa, token, syntax, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2411 if (BE (*err != REG_NOERROR && tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2412 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2413 /* In BRE consecutive duplications are not allowed. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2414 if ((syntax & RE_CONTEXT_INVALID_DUP) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2415 && (token->type == OP_DUP_ASTERISK |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2416 || token->type == OP_OPEN_DUP_NUM)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2417 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2418 *err = REG_BADRPT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2419 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2420 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2421 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2422 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2423 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2424 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2425 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2426 /* This function build the following tree, from regular expression |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2427 (<reg_exp>): |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2428 SUBEXP |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2429 | |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2430 <reg_exp> |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2431 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2432 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2433 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2434 parse_sub_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2435 reg_syntax_t syntax, Idx nest, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2436 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2437 re_dfa_t *dfa = (re_dfa_t *) preg->buffer; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2438 bin_tree_t *tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2439 size_t cur_nsub; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2440 cur_nsub = preg->re_nsub++; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2441 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2442 fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2443 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2444 /* The subexpression may be a null string. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2445 if (token->type == OP_CLOSE_SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2446 tree = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2447 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2448 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2449 tree = parse_reg_exp (regexp, preg, token, syntax, nest, err); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2450 if (BE (*err == REG_NOERROR && token->type != OP_CLOSE_SUBEXP, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2451 *err = REG_EPAREN; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2452 if (BE (*err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2453 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2454 } |
6171
5862ee08bfc1
* lib/regcomp.c (re_compile_fastmap_iter, init_dfa, init_word_char):
Paul Eggert <eggert@cs.ucla.edu>
parents:
6170
diff
changeset
|
2455 |
5862ee08bfc1
* lib/regcomp.c (re_compile_fastmap_iter, init_dfa, init_word_char):
Paul Eggert <eggert@cs.ucla.edu>
parents:
6170
diff
changeset
|
2456 if (cur_nsub <= '9' - '1') |
5862ee08bfc1
* lib/regcomp.c (re_compile_fastmap_iter, init_dfa, init_word_char):
Paul Eggert <eggert@cs.ucla.edu>
parents:
6170
diff
changeset
|
2457 dfa->completed_bkref_map |= 1 << cur_nsub; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2458 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2459 tree = create_tree (dfa, tree, NULL, SUBEXP); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2460 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2461 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2462 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2463 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2464 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2465 tree->token.opr.idx = cur_nsub; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2466 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2467 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2468 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2469 /* This function parse repetition operators like "*", "+", "{1,3}" etc. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2470 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2471 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2472 parse_dup_op (bin_tree_t *elem, re_string_t *regexp, re_dfa_t *dfa, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2473 re_token_t *token, reg_syntax_t syntax, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2474 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2475 bin_tree_t *tree = NULL, *old_tree = NULL; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2476 Idx i, start, end, start_idx = re_string_cur_idx (regexp); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2477 re_token_t start_token = *token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2478 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2479 if (token->type == OP_OPEN_DUP_NUM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2480 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2481 end = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2482 start = fetch_number (regexp, token, syntax); |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2483 if (start == REG_MISSING) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2484 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2485 if (token->type == CHARACTER && token->opr.c == ',') |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2486 start = 0; /* We treat "{,m}" as "{0,m}". */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2487 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2488 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2489 *err = REG_BADBR; /* <re>{} is invalid. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2490 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2491 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2492 } |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2493 if (BE (start != REG_ERROR, 1)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2494 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2495 /* We treat "{n}" as "{n,n}". */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2496 end = ((token->type == OP_CLOSE_DUP_NUM) ? start |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2497 : ((token->type == CHARACTER && token->opr.c == ',') |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2498 ? fetch_number (regexp, token, syntax) : REG_ERROR)); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2499 } |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2500 if (BE (start == REG_ERROR || end == REG_ERROR, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2501 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2502 /* Invalid sequence. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2503 if (BE (!(syntax & RE_INVALID_INTERVAL_ORD), 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2504 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2505 if (token->type == END_OF_RE) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2506 *err = REG_EBRACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2507 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2508 *err = REG_BADBR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2509 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2510 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2511 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2512 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2513 /* If the syntax bit is set, rollback. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2514 re_string_set_index (regexp, start_idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2515 *token = start_token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2516 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2517 /* mb_partial and word_char bits should be already initialized by |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2518 peek_token. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2519 return elem; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2520 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2521 |
12570
3ed9d84fee81
regcomp: recognize ill-formed { } expressions
Ulrich Drepper <drepper@redhat.com>
parents:
12569
diff
changeset
|
2522 if (BE ((end != REG_MISSING && start > end) |
3ed9d84fee81
regcomp: recognize ill-formed { } expressions
Ulrich Drepper <drepper@redhat.com>
parents:
12569
diff
changeset
|
2523 || token->type != OP_CLOSE_DUP_NUM, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2524 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2525 /* First number greater than second. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2526 *err = REG_BADBR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2527 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2528 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2529 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2530 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2531 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2532 start = (token->type == OP_DUP_PLUS) ? 1 : 0; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2533 end = (token->type == OP_DUP_QUESTION) ? 1 : REG_MISSING; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2534 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2535 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2536 fetch_token (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2537 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2538 if (BE (elem == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2539 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2540 if (BE (start == 0 && end == 0, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2541 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2542 postorder (elem, free_tree, NULL); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2543 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2544 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2545 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2546 /* Extract "<re>{n,m}" to "<re><re>...<re><re>{0,<m-n>}". */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2547 if (BE (start > 0, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2548 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2549 tree = elem; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2550 for (i = 2; i <= start; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2551 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2552 elem = duplicate_tree (elem, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2553 tree = create_tree (dfa, tree, elem, CONCAT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2554 if (BE (elem == NULL || tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2555 goto parse_dup_op_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2556 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2557 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2558 if (start == end) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2559 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2560 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2561 /* Duplicate ELEM before it is marked optional. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2562 elem = duplicate_tree (elem, dfa); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2563 old_tree = tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2564 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2565 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2566 old_tree = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2567 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2568 if (elem->token.type == SUBEXP) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2569 postorder (elem, mark_opt_subexp, (void *) (long) elem->token.opr.idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2570 |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2571 tree = create_tree (dfa, elem, NULL, |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2572 (end == REG_MISSING ? OP_DUP_ASTERISK : OP_ALT)); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2573 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2574 goto parse_dup_op_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2575 |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2576 /* This loop is actually executed only when end != REG_MISSING, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2577 to rewrite <re>{0,n} as (<re>(<re>...<re>?)?)?... We have |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2578 already created the start+1-th copy. */ |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2579 if ((Idx) -1 < 0 || end != REG_MISSING) |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2580 for (i = start + 2; i <= end; ++i) |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2581 { |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2582 elem = duplicate_tree (elem, dfa); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2583 tree = create_tree (dfa, tree, elem, CONCAT); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2584 if (BE (elem == NULL || tree == NULL, 0)) |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2585 goto parse_dup_op_espace; |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2586 |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2587 tree = create_tree (dfa, tree, NULL, OP_ALT); |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2588 if (BE (tree == NULL, 0)) |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2589 goto parse_dup_op_espace; |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2590 } |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2591 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2592 if (old_tree) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2593 tree = create_tree (dfa, old_tree, tree, CONCAT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2594 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2595 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2596 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2597 parse_dup_op_espace: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2598 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2599 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2600 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2601 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2602 /* Size of the names for collating symbol/equivalence_class/character_class. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2603 I'm not sure, but maybe enough. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2604 #define BRACKET_NAME_BUF_SIZE 32 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2605 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2606 #ifndef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2607 /* Local function for parse_bracket_exp only used in case of NOT _LIBC. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2608 Build the range expression which starts from START_ELEM, and ends |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2609 at END_ELEM. The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2610 RANGE_ALLOC is the allocated size of mbcset->range_starts, and |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2611 mbcset->range_ends, is a pointer argument sinse we may |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2612 update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2613 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2614 static reg_errcode_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2615 internal_function |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2616 # ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2617 build_range_exp (bitset_t sbcset, re_charset_t *mbcset, Idx *range_alloc, |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2618 bracket_elem_t *start_elem, bracket_elem_t *end_elem) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2619 # else /* not RE_ENABLE_I18N */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2620 build_range_exp (bitset_t sbcset, bracket_elem_t *start_elem, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2621 bracket_elem_t *end_elem) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2622 # endif /* not RE_ENABLE_I18N */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2623 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2624 unsigned int start_ch, end_ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2625 /* Equivalence Classes and Character Classes can't be a range start/end. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2626 if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2627 || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2628 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2629 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2630 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2631 /* We can handle no multi character collating elements without libc |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2632 support. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2633 if (BE ((start_elem->type == COLL_SYM |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2634 && strlen ((char *) start_elem->opr.name) > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2635 || (end_elem->type == COLL_SYM |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2636 && strlen ((char *) end_elem->opr.name) > 1), 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2637 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2638 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2639 # ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2640 { |
5972
aa260da0bbbe
* config/srclist.txt: Comment out regcomp.c, since we have a porting fix
Paul Eggert <eggert@cs.ucla.edu>
parents:
5968
diff
changeset
|
2641 wchar_t wc; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2642 wint_t start_wc; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2643 wint_t end_wc; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2644 wchar_t cmp_buf[6] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'}; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2645 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2646 start_ch = ((start_elem->type == SB_CHAR) ? start_elem->opr.ch |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2647 : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2648 : 0)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2649 end_ch = ((end_elem->type == SB_CHAR) ? end_elem->opr.ch |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2650 : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2651 : 0)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2652 start_wc = ((start_elem->type == SB_CHAR || start_elem->type == COLL_SYM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2653 ? __btowc (start_ch) : start_elem->opr.wch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2654 end_wc = ((end_elem->type == SB_CHAR || end_elem->type == COLL_SYM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2655 ? __btowc (end_ch) : end_elem->opr.wch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2656 if (start_wc == WEOF || end_wc == WEOF) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2657 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2658 cmp_buf[0] = start_wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2659 cmp_buf[4] = end_wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2660 if (wcscoll (cmp_buf, cmp_buf + 4) > 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2661 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2662 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2663 /* Got valid collation sequence values, add them as a new entry. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2664 However, for !_LIBC we have no collation elements: if the |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2665 character set is single byte, the single byte character set |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2666 that we build below suffices. parse_bracket_exp passes |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2667 no MBCSET if dfa->mb_cur_max == 1. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2668 if (mbcset) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2669 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2670 /* Check the space of the arrays. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2671 if (BE (*range_alloc == mbcset->nranges, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2672 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2673 /* There is not enough space, need realloc. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2674 wchar_t *new_array_start, *new_array_end; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2675 Idx new_nranges; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2676 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2677 /* +1 in case of mbcset->nranges is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2678 new_nranges = 2 * mbcset->nranges + 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2679 /* Use realloc since mbcset->range_starts and mbcset->range_ends |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2680 are NULL if *range_alloc == 0. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2681 new_array_start = re_realloc (mbcset->range_starts, wchar_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2682 new_nranges); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2683 new_array_end = re_realloc (mbcset->range_ends, wchar_t, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2684 new_nranges); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2685 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2686 if (BE (new_array_start == NULL || new_array_end == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2687 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2688 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2689 mbcset->range_starts = new_array_start; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2690 mbcset->range_ends = new_array_end; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2691 *range_alloc = new_nranges; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2692 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2693 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2694 mbcset->range_starts[mbcset->nranges] = start_wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2695 mbcset->range_ends[mbcset->nranges++] = end_wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2696 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2697 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2698 /* Build the table for single byte characters. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2699 for (wc = 0; wc < SBC_MAX; ++wc) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2700 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2701 cmp_buf[2] = wc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2702 if (wcscoll (cmp_buf, cmp_buf + 2) <= 0 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2703 && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2704 bitset_set (sbcset, wc); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2705 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2706 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2707 # else /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2708 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2709 unsigned int ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2710 start_ch = ((start_elem->type == SB_CHAR ) ? start_elem->opr.ch |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2711 : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2712 : 0)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2713 end_ch = ((end_elem->type == SB_CHAR ) ? end_elem->opr.ch |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2714 : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2715 : 0)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2716 if (start_ch > end_ch) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2717 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2718 /* Build the table for single byte characters. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2719 for (ch = 0; ch < SBC_MAX; ++ch) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2720 if (start_ch <= ch && ch <= end_ch) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2721 bitset_set (sbcset, ch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2722 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2723 # endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2724 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2725 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2726 #endif /* not _LIBC */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2727 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2728 #ifndef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2729 /* Helper function for parse_bracket_exp only used in case of NOT _LIBC.. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2730 Build the collating element which is represented by NAME. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2731 The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2732 COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2733 pointer argument since we may update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2734 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2735 static reg_errcode_t |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2736 internal_function |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2737 build_collating_symbol (bitset_t sbcset, |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2738 # ifdef RE_ENABLE_I18N |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2739 re_charset_t *mbcset, Idx *coll_sym_alloc, |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2740 # endif |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2741 const unsigned char *name) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2742 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2743 size_t name_len = strlen ((const char *) name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2744 if (BE (name_len != 1, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2745 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2746 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2747 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2748 bitset_set (sbcset, name[0]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2749 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2750 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2751 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2752 #endif /* not _LIBC */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2753 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2754 /* This function parse bracket expression like "[abc]", "[a-c]", |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2755 "[[.a-a.]]" etc. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2756 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2757 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2758 parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, re_token_t *token, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
2759 reg_syntax_t syntax, reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2760 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2761 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2762 const unsigned char *collseqmb; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2763 const char *collseqwc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2764 uint32_t nrules; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2765 int32_t table_size; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2766 const int32_t *symb_table; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2767 const unsigned char *extra; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2768 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2769 /* Local function for parse_bracket_exp used in _LIBC environement. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2770 Seek the collating symbol entry correspondings to NAME. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2771 Return the index of the symbol in the SYMB_TABLE. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2772 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2773 auto inline int32_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2774 __attribute ((always_inline)) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2775 seek_collating_symbol_entry (name, name_len) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2776 const unsigned char *name; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2777 size_t name_len; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2778 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2779 int32_t hash = elem_hash ((const char *) name, name_len); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2780 int32_t elem = hash % table_size; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2781 if (symb_table[2 * elem] != 0) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2782 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2783 int32_t second = hash % (table_size - 2) + 1; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2784 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2785 do |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2786 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2787 /* First compare the hashing value. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2788 if (symb_table[2 * elem] == hash |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2789 /* Compare the length of the name. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2790 && name_len == extra[symb_table[2 * elem + 1]] |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2791 /* Compare the name. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2792 && memcmp (name, &extra[symb_table[2 * elem + 1] + 1], |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2793 name_len) == 0) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2794 { |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2795 /* Yep, this is the entry. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2796 break; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2797 } |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2798 |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2799 /* Next entry. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2800 elem += second; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2801 } |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2802 while (symb_table[2 * elem] != 0); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2803 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2804 return elem; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2805 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2806 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2807 /* Local function for parse_bracket_exp used in _LIBC environement. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2808 Look up the collation sequence value of BR_ELEM. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2809 Return the value if succeeded, UINT_MAX otherwise. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2810 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2811 auto inline unsigned int |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2812 __attribute ((always_inline)) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2813 lookup_collation_sequence_value (br_elem) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2814 bracket_elem_t *br_elem; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2815 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2816 if (br_elem->type == SB_CHAR) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2817 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2818 /* |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2819 if (MB_CUR_MAX == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2820 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2821 if (nrules == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2822 return collseqmb[br_elem->opr.ch]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2823 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2824 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2825 wint_t wc = __btowc (br_elem->opr.ch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2826 return __collseq_table_lookup (collseqwc, wc); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2827 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2828 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2829 else if (br_elem->type == MB_CHAR) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2830 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2831 return __collseq_table_lookup (collseqwc, br_elem->opr.wch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2832 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2833 else if (br_elem->type == COLL_SYM) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2834 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2835 size_t sym_name_len = strlen ((char *) br_elem->opr.name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2836 if (nrules != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2837 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2838 int32_t elem, idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2839 elem = seek_collating_symbol_entry (br_elem->opr.name, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2840 sym_name_len); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2841 if (symb_table[2 * elem] != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2842 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2843 /* We found the entry. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2844 idx = symb_table[2 * elem + 1]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2845 /* Skip the name of collating element name. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2846 idx += 1 + extra[idx]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2847 /* Skip the byte sequence of the collating element. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2848 idx += 1 + extra[idx]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2849 /* Adjust for the alignment. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2850 idx = (idx + 3) & ~3; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2851 /* Skip the multibyte collation sequence value. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2852 idx += sizeof (unsigned int); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2853 /* Skip the wide char sequence of the collating element. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2854 idx += sizeof (unsigned int) * |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2855 (1 + *(unsigned int *) (extra + idx)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2856 /* Return the collation sequence value. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2857 return *(unsigned int *) (extra + idx); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2858 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2859 else if (symb_table[2 * elem] == 0 && sym_name_len == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2860 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2861 /* No valid character. Match it as a single byte |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2862 character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2863 return collseqmb[br_elem->opr.name[0]]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2864 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2865 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2866 else if (sym_name_len == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2867 return collseqmb[br_elem->opr.name[0]]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2868 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2869 return UINT_MAX; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2870 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2871 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2872 /* Local function for parse_bracket_exp used in _LIBC environement. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2873 Build the range expression which starts from START_ELEM, and ends |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2874 at END_ELEM. The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2875 RANGE_ALLOC is the allocated size of mbcset->range_starts, and |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2876 mbcset->range_ends, is a pointer argument sinse we may |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2877 update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2878 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2879 auto inline reg_errcode_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2880 __attribute ((always_inline)) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2881 build_range_exp (sbcset, mbcset, range_alloc, start_elem, end_elem) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2882 re_charset_t *mbcset; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2883 Idx *range_alloc; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2884 bitset_t sbcset; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2885 bracket_elem_t *start_elem, *end_elem; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2886 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2887 unsigned int ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2888 uint32_t start_collseq; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2889 uint32_t end_collseq; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2890 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2891 /* Equivalence Classes and Character Classes can't be a range |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2892 start/end. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2893 if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2894 || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2895 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2896 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2897 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2898 start_collseq = lookup_collation_sequence_value (start_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2899 end_collseq = lookup_collation_sequence_value (end_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2900 /* Check start/end collation sequence values. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2901 if (BE (start_collseq == UINT_MAX || end_collseq == UINT_MAX, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2902 return REG_ECOLLATE; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2903 if (BE ((syntax & RE_NO_EMPTY_RANGES) && start_collseq > end_collseq, 0)) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2904 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2905 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2906 /* Got valid collation sequence values, add them as a new entry. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2907 However, if we have no collation elements, and the character set |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2908 is single byte, the single byte character set that we |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2909 build below suffices. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2910 if (nrules > 0 || dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2911 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2912 /* Check the space of the arrays. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2913 if (BE (*range_alloc == mbcset->nranges, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2914 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2915 /* There is not enough space, need realloc. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2916 uint32_t *new_array_start; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2917 uint32_t *new_array_end; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
2918 Idx new_nranges; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2919 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2920 /* +1 in case of mbcset->nranges is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2921 new_nranges = 2 * mbcset->nranges + 1; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2922 new_array_start = re_realloc (mbcset->range_starts, uint32_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2923 new_nranges); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2924 new_array_end = re_realloc (mbcset->range_ends, uint32_t, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2925 new_nranges); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2926 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2927 if (BE (new_array_start == NULL || new_array_end == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2928 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2929 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2930 mbcset->range_starts = new_array_start; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2931 mbcset->range_ends = new_array_end; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2932 *range_alloc = new_nranges; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2933 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2934 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2935 mbcset->range_starts[mbcset->nranges] = start_collseq; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2936 mbcset->range_ends[mbcset->nranges++] = end_collseq; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2937 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2938 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2939 /* Build the table for single byte characters. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2940 for (ch = 0; ch < SBC_MAX; ch++) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2941 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2942 uint32_t ch_collseq; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2943 /* |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2944 if (MB_CUR_MAX == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2945 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2946 if (nrules == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2947 ch_collseq = collseqmb[ch]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2948 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2949 ch_collseq = __collseq_table_lookup (collseqwc, __btowc (ch)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2950 if (start_collseq <= ch_collseq && ch_collseq <= end_collseq) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2951 bitset_set (sbcset, ch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2952 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2953 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2954 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2955 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2956 /* Local function for parse_bracket_exp used in _LIBC environement. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2957 Build the collating element which is represented by NAME. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2958 The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2959 COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2960 pointer argument sinse we may update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2961 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2962 auto inline reg_errcode_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2963 __attribute ((always_inline)) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2964 build_collating_symbol (sbcset, mbcset, coll_sym_alloc, name) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2965 re_charset_t *mbcset; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2966 Idx *coll_sym_alloc; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2967 bitset_t sbcset; |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2968 const unsigned char *name; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2969 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2970 int32_t elem, idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2971 size_t name_len = strlen ((const char *) name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2972 if (nrules != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2973 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2974 elem = seek_collating_symbol_entry (name, name_len); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2975 if (symb_table[2 * elem] != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2976 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2977 /* We found the entry. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2978 idx = symb_table[2 * elem + 1]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2979 /* Skip the name of collating element name. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2980 idx += 1 + extra[idx]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2981 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2982 else if (symb_table[2 * elem] == 0 && name_len == 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2983 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2984 /* No valid character, treat it as a normal |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2985 character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2986 bitset_set (sbcset, name[0]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2987 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2988 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2989 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2990 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2991 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2992 /* Got valid collation sequence, add it as a new entry. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2993 /* Check the space of the arrays. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2994 if (BE (*coll_sym_alloc == mbcset->ncoll_syms, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2995 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2996 /* Not enough, realloc it. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2997 /* +1 in case of mbcset->ncoll_syms is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
2998 Idx new_coll_sym_alloc = 2 * mbcset->ncoll_syms + 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
2999 /* Use realloc since mbcset->coll_syms is NULL |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3000 if *alloc == 0. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3001 int32_t *new_coll_syms = re_realloc (mbcset->coll_syms, int32_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3002 new_coll_sym_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3003 if (BE (new_coll_syms == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3004 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3005 mbcset->coll_syms = new_coll_syms; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3006 *coll_sym_alloc = new_coll_sym_alloc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3007 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3008 mbcset->coll_syms[mbcset->ncoll_syms++] = idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3009 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3010 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3011 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3012 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3013 if (BE (name_len != 1, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3014 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3015 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3016 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3017 bitset_set (sbcset, name[0]); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3018 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3019 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3020 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3021 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3022 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3023 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3024 re_token_t br_token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3025 re_bitset_ptr_t sbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3026 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3027 re_charset_t *mbcset; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3028 Idx coll_sym_alloc = 0, range_alloc = 0, mbchar_alloc = 0; |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3029 Idx equiv_class_alloc = 0, char_class_alloc = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3030 #endif /* not RE_ENABLE_I18N */ |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3031 bool non_match = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3032 bin_tree_t *work_tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3033 int token_len; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3034 bool first_round = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3035 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3036 collseqmb = (const unsigned char *) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3037 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3038 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3039 if (nrules) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3040 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3041 /* |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3042 if (MB_CUR_MAX > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3043 */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3044 collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3045 table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3046 symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3047 _NL_COLLATE_SYMB_TABLEMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3048 extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3049 _NL_COLLATE_SYMB_EXTRAMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3050 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3051 #endif |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3052 sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3053 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3054 mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3055 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3056 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3057 if (BE (sbcset == NULL || mbcset == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3058 #else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3059 if (BE (sbcset == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3060 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3061 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3062 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3063 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3064 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3065 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3066 token_len = peek_token_bracket (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3067 if (BE (token->type == END_OF_RE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3068 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3069 *err = REG_BADPAT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3070 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3071 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3072 if (token->type == OP_NON_MATCH_LIST) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3073 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3074 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3075 mbcset->non_match = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3076 #endif /* not RE_ENABLE_I18N */ |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3077 non_match = true; |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3078 if (syntax & RE_HAT_LISTS_NOT_NEWLINE) |
8110
8428e492271f
2007-02-05 Paolo Bonzini <bonzini@gnu.org>
Paolo Bonzini <bonzini@gnu.org>
parents:
8073
diff
changeset
|
3079 bitset_set (sbcset, '\n'); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3080 re_string_skip_bytes (regexp, token_len); /* Skip a token. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3081 token_len = peek_token_bracket (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3082 if (BE (token->type == END_OF_RE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3083 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3084 *err = REG_BADPAT; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3085 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3086 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3087 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3088 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3089 /* We treat the first ']' as a normal character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3090 if (token->type == OP_CLOSE_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3091 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3092 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3093 while (1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3094 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3095 bracket_elem_t start_elem, end_elem; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3096 unsigned char start_name_buf[BRACKET_NAME_BUF_SIZE]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3097 unsigned char end_name_buf[BRACKET_NAME_BUF_SIZE]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3098 reg_errcode_t ret; |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3099 int token_len2 = 0; |
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3100 bool is_range_exp = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3101 re_token_t token2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3102 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3103 start_elem.opr.name = start_name_buf; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3104 ret = parse_bracket_element (&start_elem, regexp, token, token_len, dfa, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3105 syntax, first_round); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3106 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3107 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3108 *err = ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3109 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3110 } |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3111 first_round = false; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3112 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3113 /* Get information about the next token. We need it in any case. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3114 token_len = peek_token_bracket (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3115 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3116 /* Do not check for ranges if we know they are not allowed. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3117 if (start_elem.type != CHAR_CLASS && start_elem.type != EQUIV_CLASS) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3118 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3119 if (BE (token->type == END_OF_RE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3120 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3121 *err = REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3122 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3123 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3124 if (token->type == OP_CHARSET_RANGE) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3125 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3126 re_string_skip_bytes (regexp, token_len); /* Skip '-'. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3127 token_len2 = peek_token_bracket (&token2, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3128 if (BE (token2.type == END_OF_RE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3129 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3130 *err = REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3131 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3132 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3133 if (token2.type == OP_CLOSE_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3134 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3135 /* We treat the last '-' as a normal character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3136 re_string_skip_bytes (regexp, -token_len); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3137 token->type = CHARACTER; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3138 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3139 else |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3140 is_range_exp = true; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3141 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3142 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3143 |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3144 if (is_range_exp == true) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3145 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3146 end_elem.opr.name = end_name_buf; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3147 ret = parse_bracket_element (&end_elem, regexp, &token2, token_len2, |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3148 dfa, syntax, true); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3149 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3150 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3151 *err = ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3152 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3153 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3154 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3155 token_len = peek_token_bracket (token, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3156 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3157 #ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3158 *err = build_range_exp (sbcset, mbcset, &range_alloc, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3159 &start_elem, &end_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3160 #else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3161 # ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3162 *err = build_range_exp (sbcset, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3163 dfa->mb_cur_max > 1 ? mbcset : NULL, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3164 &range_alloc, &start_elem, &end_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3165 # else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3166 *err = build_range_exp (sbcset, &start_elem, &end_elem); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3167 # endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3168 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3169 if (BE (*err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3170 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3171 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3172 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3173 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3174 switch (start_elem.type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3175 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3176 case SB_CHAR: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3177 bitset_set (sbcset, start_elem.opr.ch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3178 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3179 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3180 case MB_CHAR: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3181 /* Check whether the array has enough space. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3182 if (BE (mbchar_alloc == mbcset->nmbchars, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3183 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3184 wchar_t *new_mbchars; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3185 /* Not enough, realloc it. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3186 /* +1 in case of mbcset->nmbchars is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3187 mbchar_alloc = 2 * mbcset->nmbchars + 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3188 /* Use realloc since array is NULL if *alloc == 0. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3189 new_mbchars = re_realloc (mbcset->mbchars, wchar_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3190 mbchar_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3191 if (BE (new_mbchars == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3192 goto parse_bracket_exp_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3193 mbcset->mbchars = new_mbchars; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3194 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3195 mbcset->mbchars[mbcset->nmbchars++] = start_elem.opr.wch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3196 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3197 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3198 case EQUIV_CLASS: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3199 *err = build_equiv_class (sbcset, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3200 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3201 mbcset, &equiv_class_alloc, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3202 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3203 start_elem.opr.name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3204 if (BE (*err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3205 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3206 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3207 case COLL_SYM: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3208 *err = build_collating_symbol (sbcset, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3209 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3210 mbcset, &coll_sym_alloc, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3211 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3212 start_elem.opr.name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3213 if (BE (*err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3214 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3215 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3216 case CHAR_CLASS: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3217 *err = build_charclass (regexp->trans, sbcset, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3218 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3219 mbcset, &char_class_alloc, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3220 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3221 start_elem.opr.name, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3222 if (BE (*err != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3223 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3224 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3225 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3226 assert (0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3227 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3228 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3229 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3230 if (BE (token->type == END_OF_RE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3231 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3232 *err = REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3233 goto parse_bracket_exp_free_return; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3234 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3235 if (token->type == OP_CLOSE_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3236 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3237 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3238 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3239 re_string_skip_bytes (regexp, token_len); /* Skip a token. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3240 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3241 /* If it is non-matching list. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3242 if (non_match) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3243 bitset_not (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3244 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3245 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3246 /* Ensure only single byte characters are set. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3247 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3248 bitset_mask (sbcset, dfa->sb_char); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3249 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3250 if (mbcset->nmbchars || mbcset->ncoll_syms || mbcset->nequiv_classes |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3251 || mbcset->nranges || (dfa->mb_cur_max > 1 && (mbcset->nchar_classes |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3252 || mbcset->non_match))) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3253 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3254 bin_tree_t *mbc_tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3255 int sbc_idx; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3256 /* Build a tree for complex bracket. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3257 dfa->has_mb_node = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3258 br_token.type = COMPLEX_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3259 br_token.opr.mbcset = mbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3260 mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3261 if (BE (mbc_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3262 goto parse_bracket_exp_espace; |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
3263 for (sbc_idx = 0; sbc_idx < BITSET_WORDS; ++sbc_idx) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3264 if (sbcset[sbc_idx]) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3265 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3266 /* If there are no bits set in sbcset, there is no point |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3267 of having both SIMPLE_BRACKET and COMPLEX_BRACKET. */ |
6214
afb93b90dcb8
Change bitset word type from unsigned int to unsigned long int,
Paul Eggert <eggert@cs.ucla.edu>
parents:
6206
diff
changeset
|
3268 if (sbc_idx < BITSET_WORDS) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3269 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3270 /* Build a tree for simple bracket. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3271 br_token.type = SIMPLE_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3272 br_token.opr.sbcset = sbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3273 work_tree = create_token_tree (dfa, NULL, NULL, &br_token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3274 if (BE (work_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3275 goto parse_bracket_exp_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3276 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3277 /* Then join them by ALT node. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3278 work_tree = create_tree (dfa, work_tree, mbc_tree, OP_ALT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3279 if (BE (work_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3280 goto parse_bracket_exp_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3281 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3282 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3283 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3284 re_free (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3285 work_tree = mbc_tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3286 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3287 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3288 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3289 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3290 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3291 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3292 free_charset (mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3293 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3294 /* Build a tree for simple bracket. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3295 br_token.type = SIMPLE_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3296 br_token.opr.sbcset = sbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3297 work_tree = create_token_tree (dfa, NULL, NULL, &br_token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3298 if (BE (work_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3299 goto parse_bracket_exp_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3300 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3301 return work_tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3302 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3303 parse_bracket_exp_espace: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3304 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3305 parse_bracket_exp_free_return: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3306 re_free (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3307 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3308 free_charset (mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3309 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3310 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3311 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3312 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3313 /* Parse an element in the bracket expression. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3314 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3315 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3316 parse_bracket_element (bracket_elem_t *elem, re_string_t *regexp, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3317 re_token_t *token, int token_len, re_dfa_t *dfa, |
6195
25eaa608fc4e
Use bool where appropriate.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6185
diff
changeset
|
3318 reg_syntax_t syntax, bool accept_hyphen) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3319 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3320 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3321 int cur_char_size; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3322 cur_char_size = re_string_char_size_at (regexp, re_string_cur_idx (regexp)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3323 if (cur_char_size > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3324 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3325 elem->type = MB_CHAR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3326 elem->opr.wch = re_string_wchar_at (regexp, re_string_cur_idx (regexp)); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3327 re_string_skip_bytes (regexp, cur_char_size); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3328 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3329 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3330 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3331 re_string_skip_bytes (regexp, token_len); /* Skip a token. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3332 if (token->type == OP_OPEN_COLL_ELEM || token->type == OP_OPEN_CHAR_CLASS |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3333 || token->type == OP_OPEN_EQUIV_CLASS) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3334 return parse_bracket_symbol (elem, regexp, token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3335 if (BE (token->type == OP_CHARSET_RANGE, 0) && !accept_hyphen) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3336 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3337 /* A '-' must only appear as anything but a range indicator before |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3338 the closing bracket. Everything else is an error. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3339 re_token_t token2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3340 (void) peek_token_bracket (&token2, regexp, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3341 if (token2.type != OP_CLOSE_BRACKET) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3342 /* The actual error value is not standardized since this whole |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3343 case is undefined. But ERANGE makes good sense. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3344 return REG_ERANGE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3345 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3346 elem->type = SB_CHAR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3347 elem->opr.ch = token->opr.c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3348 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3349 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3350 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3351 /* Parse a bracket symbol in the bracket expression. Bracket symbols are |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3352 such as [:<character_class>:], [.<collating_element>.], and |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3353 [=<equivalent_class>=]. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3354 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3355 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3356 parse_bracket_symbol (bracket_elem_t *elem, re_string_t *regexp, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3357 re_token_t *token) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3358 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3359 unsigned char ch, delim = token->opr.c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3360 int i = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3361 if (re_string_eoi(regexp)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3362 return REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3363 for (;; ++i) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3364 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3365 if (i >= BRACKET_NAME_BUF_SIZE) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3366 return REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3367 if (token->type == OP_OPEN_CHAR_CLASS) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3368 ch = re_string_fetch_byte_case (regexp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3369 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3370 ch = re_string_fetch_byte (regexp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3371 if (re_string_eoi(regexp)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3372 return REG_EBRACK; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3373 if (ch == delim && re_string_peek_byte (regexp, 0) == ']') |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3374 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3375 elem->opr.name[i] = ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3376 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3377 re_string_skip_bytes (regexp, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3378 elem->opr.name[i] = '\0'; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3379 switch (token->type) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3380 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3381 case OP_OPEN_COLL_ELEM: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3382 elem->type = COLL_SYM; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3383 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3384 case OP_OPEN_EQUIV_CLASS: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3385 elem->type = EQUIV_CLASS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3386 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3387 case OP_OPEN_CHAR_CLASS: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3388 elem->type = CHAR_CLASS; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3389 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3390 default: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3391 break; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3392 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3393 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3394 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3395 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3396 /* Helper function for parse_bracket_exp. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3397 Build the equivalence class which is represented by NAME. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3398 The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3399 EQUIV_CLASS_ALLOC is the allocated size of mbcset->equiv_classes, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3400 is a pointer argument sinse we may update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3401 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3402 static reg_errcode_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3403 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3404 build_equiv_class (bitset_t sbcset, re_charset_t *mbcset, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3405 Idx *equiv_class_alloc, const unsigned char *name) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3406 #else /* not RE_ENABLE_I18N */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3407 build_equiv_class (bitset_t sbcset, const unsigned char *name) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3408 #endif /* not RE_ENABLE_I18N */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3409 { |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3410 #ifdef _LIBC |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3411 uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3412 if (nrules != 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3413 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3414 const int32_t *table, *indirect; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3415 const unsigned char *weights, *extra, *cp; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3416 unsigned char char_buf[2]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3417 int32_t idx1, idx2; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3418 unsigned int ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3419 size_t len; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3420 /* This #include defines a local function! */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3421 # include <locale/weight.h> |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3422 /* Calculate the index for equivalence class. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3423 cp = name; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3424 table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3425 weights = (const unsigned char *) _NL_CURRENT (LC_COLLATE, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3426 _NL_COLLATE_WEIGHTMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3427 extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3428 _NL_COLLATE_EXTRAMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3429 indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3430 _NL_COLLATE_INDIRECTMB); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3431 idx1 = findidx (&cp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3432 if (BE (idx1 == 0 || cp < name + strlen ((const char *) name), 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3433 /* This isn't a valid character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3434 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3435 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3436 /* Build single byte matcing table for this equivalence class. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3437 char_buf[1] = (unsigned char) '\0'; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3438 len = weights[idx1]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3439 for (ch = 0; ch < SBC_MAX; ++ch) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3440 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3441 char_buf[0] = ch; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3442 cp = char_buf; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3443 idx2 = findidx (&cp); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3444 /* |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3445 idx2 = table[ch]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3446 */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3447 if (idx2 == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3448 /* This isn't a valid character. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3449 continue; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3450 if (len == weights[idx2]) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3451 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3452 int cnt = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3453 while (cnt <= len && |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3454 weights[idx1 + 1 + cnt] == weights[idx2 + 1 + cnt]) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3455 ++cnt; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3456 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3457 if (cnt > len) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3458 bitset_set (sbcset, ch); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3459 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3460 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3461 /* Check whether the array has enough space. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3462 if (BE (*equiv_class_alloc == mbcset->nequiv_classes, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3463 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3464 /* Not enough, realloc it. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3465 /* +1 in case of mbcset->nequiv_classes is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3466 Idx new_equiv_class_alloc = 2 * mbcset->nequiv_classes + 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3467 /* Use realloc since the array is NULL if *alloc == 0. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3468 int32_t *new_equiv_classes = re_realloc (mbcset->equiv_classes, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3469 int32_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3470 new_equiv_class_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3471 if (BE (new_equiv_classes == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3472 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3473 mbcset->equiv_classes = new_equiv_classes; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3474 *equiv_class_alloc = new_equiv_class_alloc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3475 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3476 mbcset->equiv_classes[mbcset->nequiv_classes++] = idx1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3477 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3478 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3479 #endif /* _LIBC */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3480 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3481 if (BE (strlen ((const char *) name) != 1, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3482 return REG_ECOLLATE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3483 bitset_set (sbcset, *name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3484 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3485 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3486 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3487 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3488 /* Helper function for parse_bracket_exp. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3489 Build the character class which is represented by NAME. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3490 The result are written to MBCSET and SBCSET. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3491 CHAR_CLASS_ALLOC is the allocated size of mbcset->char_classes, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3492 is a pointer argument sinse we may update it. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3493 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3494 static reg_errcode_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3495 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3496 build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3497 re_charset_t *mbcset, Idx *char_class_alloc, |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3498 const unsigned char *class_name, reg_syntax_t syntax) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3499 #else /* not RE_ENABLE_I18N */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3500 build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3501 const unsigned char *class_name, reg_syntax_t syntax) |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3502 #endif /* not RE_ENABLE_I18N */ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3503 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3504 int i; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3505 const char *name = (const char *) class_name; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3506 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3507 /* In case of REG_ICASE "upper" and "lower" match the both of |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3508 upper and lower cases. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3509 if ((syntax & RE_ICASE) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3510 && (strcmp (name, "upper") == 0 || strcmp (name, "lower") == 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3511 name = "alpha"; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3512 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3513 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3514 /* Check the space of the arrays. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3515 if (BE (*char_class_alloc == mbcset->nchar_classes, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3516 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3517 /* Not enough, realloc it. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3518 /* +1 in case of mbcset->nchar_classes is 0. */ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3519 Idx new_char_class_alloc = 2 * mbcset->nchar_classes + 1; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3520 /* Use realloc since array is NULL if *alloc == 0. */ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3521 wctype_t *new_char_classes = re_realloc (mbcset->char_classes, wctype_t, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3522 new_char_class_alloc); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3523 if (BE (new_char_classes == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3524 return REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3525 mbcset->char_classes = new_char_classes; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3526 *char_class_alloc = new_char_class_alloc; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3527 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3528 mbcset->char_classes[mbcset->nchar_classes++] = __wctype (name); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3529 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3530 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3531 #define BUILD_CHARCLASS_LOOP(ctype_func) \ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3532 do { \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3533 if (BE (trans != NULL, 0)) \ |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3534 { \ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3535 for (i = 0; i < SBC_MAX; ++i) \ |
6729
c5495b5c3f32
Fix space-tab problem. From Jim Meyering.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6726
diff
changeset
|
3536 if (ctype_func (i)) \ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3537 bitset_set (sbcset, trans[i]); \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3538 } \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3539 else \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3540 { \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3541 for (i = 0; i < SBC_MAX; ++i) \ |
6729
c5495b5c3f32
Fix space-tab problem. From Jim Meyering.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6726
diff
changeset
|
3542 if (ctype_func (i)) \ |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3543 bitset_set (sbcset, i); \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3544 } \ |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3545 } while (0) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3546 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3547 if (strcmp (name, "alnum") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3548 BUILD_CHARCLASS_LOOP (isalnum); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3549 else if (strcmp (name, "cntrl") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3550 BUILD_CHARCLASS_LOOP (iscntrl); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3551 else if (strcmp (name, "lower") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3552 BUILD_CHARCLASS_LOOP (islower); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3553 else if (strcmp (name, "space") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3554 BUILD_CHARCLASS_LOOP (isspace); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3555 else if (strcmp (name, "alpha") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3556 BUILD_CHARCLASS_LOOP (isalpha); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3557 else if (strcmp (name, "digit") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3558 BUILD_CHARCLASS_LOOP (isdigit); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3559 else if (strcmp (name, "print") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3560 BUILD_CHARCLASS_LOOP (isprint); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3561 else if (strcmp (name, "upper") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3562 BUILD_CHARCLASS_LOOP (isupper); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3563 else if (strcmp (name, "blank") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3564 BUILD_CHARCLASS_LOOP (isblank); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3565 else if (strcmp (name, "graph") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3566 BUILD_CHARCLASS_LOOP (isgraph); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3567 else if (strcmp (name, "punct") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3568 BUILD_CHARCLASS_LOOP (ispunct); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3569 else if (strcmp (name, "xdigit") == 0) |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3570 BUILD_CHARCLASS_LOOP (isxdigit); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3571 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3572 return REG_ECTYPE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3573 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3574 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3575 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3576 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3577 static bin_tree_t * |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3578 build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans, |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3579 const unsigned char *class_name, |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3580 const unsigned char *extra, bool non_match, |
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3581 reg_errcode_t *err) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3582 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3583 re_bitset_ptr_t sbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3584 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3585 re_charset_t *mbcset; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3586 Idx alloc = 0; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3587 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3588 reg_errcode_t ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3589 re_token_t br_token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3590 bin_tree_t *tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3591 |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3592 sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3593 #ifdef RE_ENABLE_I18N |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3594 mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3595 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3596 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3597 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3598 if (BE (sbcset == NULL || mbcset == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3599 #else /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3600 if (BE (sbcset == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3601 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3602 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3603 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3604 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3605 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3606 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3607 if (non_match) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3608 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3609 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3610 mbcset->non_match = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3611 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3612 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3613 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3614 /* We don't care the syntax in this case. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3615 ret = build_charclass (trans, sbcset, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3616 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3617 mbcset, &alloc, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3618 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3619 class_name, 0); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3620 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3621 if (BE (ret != REG_NOERROR, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3622 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3623 re_free (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3624 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3625 free_charset (mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3626 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3627 *err = ret; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3628 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3629 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3630 /* \w match '_' also. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3631 for (; *extra; extra++) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3632 bitset_set (sbcset, *extra); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3633 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3634 /* If it is non-matching list. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3635 if (non_match) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3636 bitset_not (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3637 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3638 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3639 /* Ensure only single byte characters are set. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3640 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3641 bitset_mask (sbcset, dfa->sb_char); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3642 #endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3643 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3644 /* Build a tree for simple bracket. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3645 br_token.type = SIMPLE_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3646 br_token.opr.sbcset = sbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3647 tree = create_token_tree (dfa, NULL, NULL, &br_token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3648 if (BE (tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3649 goto build_word_op_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3650 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3651 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3652 if (dfa->mb_cur_max > 1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3653 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3654 bin_tree_t *mbc_tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3655 /* Build a tree for complex bracket. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3656 br_token.type = COMPLEX_BRACKET; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3657 br_token.opr.mbcset = mbcset; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3658 dfa->has_mb_node = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3659 mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3660 if (BE (mbc_tree == NULL, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3661 goto build_word_op_espace; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3662 /* Then join them by ALT node. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3663 tree = create_tree (dfa, tree, mbc_tree, OP_ALT); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3664 if (BE (mbc_tree != NULL, 1)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3665 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3666 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3667 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3668 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3669 free_charset (mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3670 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3671 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3672 #else /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3673 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3674 #endif /* not RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3675 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3676 build_word_op_espace: |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3677 re_free (sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3678 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3679 free_charset (mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3680 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3681 *err = REG_ESPACE; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3682 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3683 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3684 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3685 /* This is intended for the expressions like "a{1,3}". |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3686 Fetch a number from `input', and return the number. |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3687 Return REG_MISSING if the number field is empty like "{,1}". |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3688 Return REG_ERROR if an error occurred. */ |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3689 |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3690 static Idx |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3691 fetch_number (re_string_t *input, re_token_t *token, reg_syntax_t syntax) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3692 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3693 Idx num = REG_MISSING; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3694 unsigned char c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3695 while (1) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3696 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3697 fetch_token (token, input, syntax); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3698 c = token->opr.c; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3699 if (BE (token->type == END_OF_RE, 0)) |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3700 return REG_ERROR; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3701 if (token->type == OP_CLOSE_DUP_NUM || c == ',') |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3702 break; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3703 num = ((token->type != CHARACTER || c < '0' || '9' < c |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3704 || num == REG_ERROR) |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3705 ? REG_ERROR |
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3706 : ((num == REG_MISSING) ? c - '0' : num * 10 + c - '0')); |
6726
af9abbcedfbd
Merge regex changes from libc, removing some of our
Paul Eggert <eggert@cs.ucla.edu>
parents:
6236
diff
changeset
|
3707 num = (num > RE_DUP_MAX) ? REG_ERROR : num; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3708 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3709 return num; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3710 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3711 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3712 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3713 static void |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3714 free_charset (re_charset_t *cset) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3715 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3716 re_free (cset->mbchars); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3717 # ifdef _LIBC |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3718 re_free (cset->coll_syms); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3719 re_free (cset->equiv_classes); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3720 re_free (cset->range_starts); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3721 re_free (cset->range_ends); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3722 # endif |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3723 re_free (cset->char_classes); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3724 re_free (cset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3725 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3726 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3727 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3728 /* Functions for binary tree operation. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3729 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3730 /* Create a tree node. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3731 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3732 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3733 create_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3734 re_token_type_t type) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3735 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3736 re_token_t t; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3737 t.type = type; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3738 return create_token_tree (dfa, left, right, &t); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3739 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3740 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3741 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3742 create_token_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, |
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3743 const re_token_t *token) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3744 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3745 bin_tree_t *tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3746 if (BE (dfa->str_tree_storage_idx == BIN_TREE_STORAGE_SIZE, 0)) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3747 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3748 bin_tree_storage_t *storage = re_malloc (bin_tree_storage_t, 1); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3749 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3750 if (storage == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3751 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3752 storage->next = dfa->str_tree_storage; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3753 dfa->str_tree_storage = storage; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3754 dfa->str_tree_storage_idx = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3755 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3756 tree = &dfa->str_tree_storage->data[dfa->str_tree_storage_idx++]; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3757 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3758 tree->parent = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3759 tree->left = left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3760 tree->right = right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3761 tree->token = *token; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3762 tree->token.duplicated = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3763 tree->token.opt_subexp = 0; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3764 tree->first = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3765 tree->next = NULL; |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3766 tree->node_idx = REG_MISSING; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3767 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3768 if (left != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3769 left->parent = tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3770 if (right != NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3771 right->parent = tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3772 return tree; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3773 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3774 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3775 /* Mark the tree SRC as an optional subexpression. |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3776 To be called from preorder or postorder. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3777 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3778 static reg_errcode_t |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3779 mark_opt_subexp (void *extra, bin_tree_t *node) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3780 { |
6184
f1728546eca4
On 64-bit hosts (where size_t is 64 bits and int is 32 bits), the
Paul Eggert <eggert@cs.ucla.edu>
parents:
6174
diff
changeset
|
3781 Idx idx = (Idx) (long) extra; |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3782 if (node->token.type == SUBEXP && node->token.opr.idx == idx) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3783 node->token.opt_subexp = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3784 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3785 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3786 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3787 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3788 /* Free the allocated memory inside NODE. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3789 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3790 static void |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3791 free_token (re_token_t *node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3792 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3793 #ifdef RE_ENABLE_I18N |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3794 if (node->type == COMPLEX_BRACKET && node->duplicated == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3795 free_charset (node->opr.mbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3796 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3797 #endif /* RE_ENABLE_I18N */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3798 if (node->type == SIMPLE_BRACKET && node->duplicated == 0) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3799 re_free (node->opr.sbcset); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3800 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3801 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3802 /* Worker function for tree walking. Free the allocated memory inside NODE |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3803 and its children. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3804 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3805 static reg_errcode_t |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3806 free_tree (void *extra, bin_tree_t *node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3807 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3808 free_token (&node->token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3809 return REG_NOERROR; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3810 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3811 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3812 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3813 /* Duplicate the node SRC, and return new node. This is a preorder |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3814 visit similar to the one implemented by the generic visitor, but |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3815 we need more infrastructure to maintain two parallel trees --- so, |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3816 it's easier to duplicate. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3817 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3818 static bin_tree_t * |
6076
e2dd51f6e259
* config/srclist.txt: Add glibc bugs 1220, 1221, 1222.
Paul Eggert <eggert@cs.ucla.edu>
parents:
6071
diff
changeset
|
3819 duplicate_tree (const bin_tree_t *root, re_dfa_t *dfa) |
5968
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3820 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3821 const bin_tree_t *node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3822 bin_tree_t *dup_root; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3823 bin_tree_t **p_new = &dup_root, *dup_node = root->parent; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3824 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3825 for (node = root; ; ) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3826 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3827 /* Create a new tree and link it back to the current parent. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3828 *p_new = create_token_tree (dfa, NULL, NULL, &node->token); |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3829 if (*p_new == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3830 return NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3831 (*p_new)->parent = dup_node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3832 (*p_new)->token.duplicated = 1; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3833 dup_node = *p_new; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3834 |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3835 /* Go to the left node, or up and to the right. */ |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3836 if (node->left) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3837 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3838 node = node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3839 p_new = &dup_node->left; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3840 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3841 else |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3842 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3843 const bin_tree_t *prev = NULL; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3844 while (node->right == prev || node->right == NULL) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3845 { |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3846 prev = node; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3847 node = node->parent; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3848 dup_node = dup_node->parent; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3849 if (!node) |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3850 return dup_root; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3851 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3852 node = node->right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3853 p_new = &dup_node->right; |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3854 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3855 } |
541fed6ae301
* modules/regex (Files): Add lib/regex_internal.c,
Paul Eggert <eggert@cs.ucla.edu>
parents:
diff
changeset
|
3856 } |