2
|
1 /* Extended regular expression matching and search library, |
|
2 version 0.11. |
|
3 (Implements POSIX draft P10003.2/D11.2, except for |
|
4 internationalization features.) |
|
5 |
|
6 Copyright (C) 1985, 89, 90, 91, 92 Free Software Foundation, Inc. |
|
7 |
|
8 This program is free software; you can redistribute it and/or modify |
|
9 it under the terms of the GNU General Public License as published by |
|
10 the Free Software Foundation; either version 2, or (at your option) |
|
11 any later version. |
|
12 |
|
13 This program is distributed in the hope that it will be useful, |
|
14 but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
16 GNU General Public License for more details. |
|
17 |
|
18 You should have received a copy of the GNU General Public License |
|
19 along with this program; if not, write to the Free Software |
|
20 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ |
|
21 |
|
22 /* AIX requires this to be the first thing in the file. */ |
|
23 #if defined (_AIX) && !defined (REGEX_MALLOC) |
|
24 #pragma alloca |
|
25 #endif |
|
26 |
|
27 #define _GNU_SOURCE |
|
28 |
|
29 /* We need this for `regex.h', and perhaps for the Emacs include files. */ |
|
30 #include <sys/types.h> |
|
31 |
|
32 /* The `emacs' switch turns on certain matching commands |
|
33 that make sense only in Emacs. */ |
|
34 #ifdef emacs |
|
35 |
|
36 #include "config.h" |
|
37 #include "lisp.h" |
|
38 #include "buffer.h" |
|
39 #include "syntax.h" |
|
40 |
|
41 /* Emacs uses `NULL' as a predicate. */ |
|
42 #undef NULL |
|
43 |
|
44 #else /* not emacs */ |
|
45 |
|
46 /* We used to test for `BSTRING' here, but only GCC and Emacs define |
|
47 `BSTRING', as far as I know, and neither of them use this code. */ |
|
48 #if USG || STDC_HEADERS |
|
49 #include <string.h> |
|
50 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n)) |
|
51 #define bcopy(s, d, n) memcpy ((d), (s), (n)) |
|
52 #define bzero(s, n) memset ((s), 0, (n)) |
|
53 #else |
|
54 #include <strings.h> |
|
55 #endif |
|
56 |
|
57 #ifdef STDC_HEADERS |
|
58 #include <stdlib.h> |
|
59 #else |
|
60 char *malloc (); |
|
61 char *realloc (); |
|
62 #endif |
|
63 |
|
64 |
|
65 /* Define the syntax stuff for \<, \>, etc. */ |
|
66 |
|
67 /* This must be nonzero for the wordchar and notwordchar pattern |
|
68 commands in re_match_2. */ |
|
69 #ifndef Sword |
|
70 #define Sword 1 |
|
71 #endif |
|
72 |
|
73 #ifdef SYNTAX_TABLE |
|
74 |
|
75 extern char *re_syntax_table; |
|
76 |
|
77 #else /* not SYNTAX_TABLE */ |
|
78 |
|
79 /* How many characters in the character set. */ |
|
80 #define CHAR_SET_SIZE 256 |
|
81 |
|
82 static char re_syntax_table[CHAR_SET_SIZE]; |
|
83 |
|
84 static void |
|
85 init_syntax_once () |
|
86 { |
|
87 register int c; |
|
88 static int done = 0; |
|
89 |
|
90 if (done) |
|
91 return; |
|
92 |
|
93 bzero (re_syntax_table, sizeof re_syntax_table); |
|
94 |
|
95 for (c = 'a'; c <= 'z'; c++) |
|
96 re_syntax_table[c] = Sword; |
|
97 |
|
98 for (c = 'A'; c <= 'Z'; c++) |
|
99 re_syntax_table[c] = Sword; |
|
100 |
|
101 for (c = '0'; c <= '9'; c++) |
|
102 re_syntax_table[c] = Sword; |
|
103 |
|
104 re_syntax_table['_'] = Sword; |
|
105 |
|
106 done = 1; |
|
107 } |
|
108 |
|
109 #endif /* not SYNTAX_TABLE */ |
|
110 |
|
111 #define SYNTAX(c) re_syntax_table[c] |
|
112 |
|
113 #endif /* not emacs */ |
|
114 |
|
115 /* Get the interface, including the syntax bits. */ |
|
116 #include "regex.h" |
|
117 |
|
118 |
|
119 /* isalpha etc. are used for the character classes. */ |
|
120 #include <ctype.h> |
|
121 #ifndef isgraph |
|
122 #define isgraph(c) (isprint (c) && !isspace (c)) |
|
123 #endif |
|
124 #ifndef isblank |
|
125 #define isblank(c) ((c) == ' ' || (c) == '\t') |
|
126 #endif |
|
127 |
|
128 #ifndef NULL |
|
129 #define NULL 0 |
|
130 #endif |
|
131 |
|
132 /* We remove any previous definition of `SIGN_EXTEND_CHAR', |
|
133 since ours (we hope) works properly with all combinations of |
|
134 machines, compilers, `char' and `unsigned char' argument types. |
|
135 (Per Bothner suggested the basic approach.) */ |
|
136 #undef SIGN_EXTEND_CHAR |
|
137 #if __STDC__ |
22
|
138 #ifndef VMS |
2
|
139 #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) |
22
|
140 #else /* On VMS, VAXC doesn't recognize `signed' before `char' */ |
|
141 #define SIGN_EXTEND_CHAR(c) ((char) (c)) |
|
142 #endif /* VMS */ |
2
|
143 #else |
|
144 /* As in Harbison and Steele. */ |
|
145 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) |
|
146 #endif |
|
147 |
|
148 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we |
|
149 use `alloca' instead of `malloc'. This is because using malloc in |
|
150 re_search* or re_match* could cause memory leaks when C-g is used in |
|
151 Emacs; also, malloc is slower and causes storage fragmentation. On |
|
152 the other hand, malloc is more portable, and easier to debug. |
|
153 |
|
154 Because we sometimes use alloca, some routines have to be macros, |
|
155 not functions -- `alloca'-allocated space disappears at the end of the |
|
156 function it is called in. */ |
|
157 |
|
158 #ifdef REGEX_MALLOC |
|
159 |
|
160 #define REGEX_ALLOCATE malloc |
|
161 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize) |
|
162 |
|
163 #else /* not REGEX_MALLOC */ |
|
164 |
|
165 /* Emacs already defines alloca, sometimes. */ |
|
166 #ifndef alloca |
|
167 |
|
168 /* Make alloca work the best possible way. */ |
|
169 #ifdef __GNUC__ |
|
170 #define alloca __builtin_alloca |
|
171 #else /* not __GNUC__ */ |
|
172 #if HAVE_ALLOCA_H |
|
173 #include <alloca.h> |
|
174 #else /* not __GNUC__ or HAVE_ALLOCA_H */ |
|
175 #ifndef _AIX /* Already did AIX, up at the top. */ |
|
176 char *alloca (); |
|
177 #endif /* not _AIX */ |
|
178 #endif /* not HAVE_ALLOCA_H */ |
|
179 #endif /* not __GNUC__ */ |
|
180 |
|
181 #endif /* not alloca */ |
|
182 |
|
183 #define REGEX_ALLOCATE alloca |
|
184 |
|
185 /* Assumes a `char *destination' variable. */ |
|
186 #define REGEX_REALLOCATE(source, osize, nsize) \ |
|
187 (destination = (char *) alloca (nsize), \ |
|
188 bcopy (source, destination, osize), \ |
|
189 destination) |
|
190 |
|
191 #endif /* not REGEX_MALLOC */ |
|
192 |
|
193 |
|
194 /* True if `size1' is non-NULL and PTR is pointing anywhere inside |
|
195 `string1' or just past its end. This works if PTR is NULL, which is |
|
196 a good thing. */ |
|
197 #define FIRST_STRING_P(ptr) \ |
|
198 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) |
|
199 |
|
200 /* (Re)Allocate N items of type T using malloc, or fail. */ |
|
201 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) |
|
202 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) |
|
203 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) |
|
204 |
|
205 #define BYTEWIDTH 8 /* In bits. */ |
|
206 |
|
207 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) |
|
208 |
|
209 #define MAX(a, b) ((a) > (b) ? (a) : (b)) |
|
210 #define MIN(a, b) ((a) < (b) ? (a) : (b)) |
|
211 |
|
212 typedef char boolean; |
|
213 #define false 0 |
|
214 #define true 1 |
|
215 |
|
216 /* These are the command codes that appear in compiled regular |
|
217 expressions. Some opcodes are followed by argument bytes. A |
|
218 command code can specify any interpretation whatsoever for its |
|
219 arguments. Zero bytes may appear in the compiled regular expression. |
|
220 |
|
221 The value of `exactn' is needed in search.c (search_buffer) in Emacs. |
|
222 So regex.h defines a symbol `RE_EXACTN_VALUE' to be 1; the value of |
|
223 `exactn' we use here must also be 1. */ |
|
224 |
|
225 typedef enum |
|
226 { |
|
227 no_op = 0, |
|
228 |
|
229 /* Followed by one byte giving n, then by n literal bytes. */ |
|
230 exactn = 1, |
|
231 |
|
232 /* Matches any (more or less) character. */ |
|
233 anychar, |
|
234 |
|
235 /* Matches any one char belonging to specified set. First |
|
236 following byte is number of bitmap bytes. Then come bytes |
|
237 for a bitmap saying which chars are in. Bits in each byte |
|
238 are ordered low-bit-first. A character is in the set if its |
|
239 bit is 1. A character too large to have a bit in the map is |
|
240 automatically not in the set. */ |
|
241 charset, |
|
242 |
|
243 /* Same parameters as charset, but match any character that is |
|
244 not one of those specified. */ |
|
245 charset_not, |
|
246 |
|
247 /* Start remembering the text that is matched, for storing in a |
|
248 register. Followed by one byte with the register number, in |
|
249 the range 0 to one less than the pattern buffer's re_nsub |
|
250 field. Then followed by one byte with the number of groups |
|
251 inner to this one. (This last has to be part of the |
|
252 start_memory only because we need it in the on_failure_jump |
|
253 of re_match_2.) */ |
|
254 start_memory, |
|
255 |
|
256 /* Stop remembering the text that is matched and store it in a |
|
257 memory register. Followed by one byte with the register |
|
258 number, in the range 0 to one less than `re_nsub' in the |
|
259 pattern buffer, and one byte with the number of inner groups, |
|
260 just like `start_memory'. (We need the number of inner |
|
261 groups here because we don't have any easy way of finding the |
|
262 corresponding start_memory when we're at a stop_memory.) */ |
|
263 stop_memory, |
|
264 |
|
265 /* Match a duplicate of something remembered. Followed by one |
|
266 byte containing the register number. */ |
|
267 duplicate, |
|
268 |
|
269 /* Fail unless at beginning of line. */ |
|
270 begline, |
|
271 |
|
272 /* Fail unless at end of line. */ |
|
273 endline, |
|
274 |
|
275 /* Succeeds if at beginning of buffer (if emacs) or at beginning |
|
276 of string to be matched (if not). */ |
|
277 begbuf, |
|
278 |
|
279 /* Analogously, for end of buffer/string. */ |
|
280 endbuf, |
|
281 |
|
282 /* Followed by two byte relative address to which to jump. */ |
|
283 jump, |
|
284 |
|
285 /* Same as jump, but marks the end of an alternative. */ |
|
286 jump_past_alt, |
|
287 |
|
288 /* Followed by two-byte relative address of place to resume at |
|
289 in case of failure. */ |
|
290 on_failure_jump, |
|
291 |
|
292 /* Like on_failure_jump, but pushes a placeholder instead of the |
|
293 current string position when executed. */ |
|
294 on_failure_keep_string_jump, |
|
295 |
|
296 /* Throw away latest failure point and then jump to following |
|
297 two-byte relative address. */ |
|
298 pop_failure_jump, |
|
299 |
|
300 /* Change to pop_failure_jump if know won't have to backtrack to |
|
301 match; otherwise change to jump. This is used to jump |
|
302 back to the beginning of a repeat. If what follows this jump |
|
303 clearly won't match what the repeat does, such that we can be |
|
304 sure that there is no use backtracking out of repetitions |
|
305 already matched, then we change it to a pop_failure_jump. |
|
306 Followed by two-byte address. */ |
|
307 maybe_pop_jump, |
|
308 |
|
309 /* Jump to following two-byte address, and push a dummy failure |
|
310 point. This failure point will be thrown away if an attempt |
|
311 is made to use it for a failure. A `+' construct makes this |
|
312 before the first repeat. Also used as an intermediary kind |
|
313 of jump when compiling an alternative. */ |
|
314 dummy_failure_jump, |
|
315 |
|
316 /* Push a dummy failure point and continue. Used at the end of |
|
317 alternatives. */ |
|
318 push_dummy_failure, |
|
319 |
|
320 /* Followed by two-byte relative address and two-byte number n. |
|
321 After matching N times, jump to the address upon failure. */ |
|
322 succeed_n, |
|
323 |
|
324 /* Followed by two-byte relative address, and two-byte number n. |
|
325 Jump to the address N times, then fail. */ |
|
326 jump_n, |
|
327 |
|
328 /* Set the following two-byte relative address to the |
|
329 subsequent two-byte number. The address *includes* the two |
|
330 bytes of number. */ |
|
331 set_number_at, |
|
332 |
|
333 wordchar, /* Matches any word-constituent character. */ |
|
334 notwordchar, /* Matches any char that is not a word-constituent. */ |
|
335 |
|
336 wordbeg, /* Succeeds if at word beginning. */ |
|
337 wordend, /* Succeeds if at word end. */ |
|
338 |
|
339 wordbound, /* Succeeds if at a word boundary. */ |
|
340 notwordbound /* Succeeds if not at a word boundary. */ |
|
341 |
|
342 #ifdef emacs |
|
343 ,before_dot, /* Succeeds if before point. */ |
|
344 at_dot, /* Succeeds if at point. */ |
|
345 after_dot, /* Succeeds if after point. */ |
|
346 |
|
347 /* Matches any character whose syntax is specified. Followed by |
|
348 a byte which contains a syntax code, e.g., Sword. */ |
|
349 syntaxspec, |
|
350 |
|
351 /* Matches any character whose syntax is not that specified. */ |
|
352 notsyntaxspec |
|
353 #endif /* emacs */ |
|
354 } re_opcode_t; |
|
355 |
|
356 /* Common operations on the compiled pattern. */ |
|
357 |
|
358 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */ |
|
359 |
|
360 #define STORE_NUMBER(destination, number) \ |
|
361 do { \ |
|
362 (destination)[0] = (number) & 0377; \ |
|
363 (destination)[1] = (number) >> 8; \ |
|
364 } while (0) |
|
365 |
|
366 /* Same as STORE_NUMBER, except increment DESTINATION to |
|
367 the byte after where the number is stored. Therefore, DESTINATION |
|
368 must be an lvalue. */ |
|
369 |
|
370 #define STORE_NUMBER_AND_INCR(destination, number) \ |
|
371 do { \ |
|
372 STORE_NUMBER (destination, number); \ |
|
373 (destination) += 2; \ |
|
374 } while (0) |
|
375 |
|
376 /* Put into DESTINATION a number stored in two contiguous bytes starting |
|
377 at SOURCE. */ |
|
378 |
|
379 #define EXTRACT_NUMBER(destination, source) \ |
|
380 do { \ |
|
381 (destination) = *(source) & 0377; \ |
|
382 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ |
|
383 } while (0) |
|
384 |
|
385 #ifdef DEBUG |
|
386 static void |
|
387 extract_number (dest, source) |
|
388 int *dest; |
|
389 unsigned char *source; |
|
390 { |
|
391 int temp = SIGN_EXTEND_CHAR (*(source + 1)); |
|
392 *dest = *source & 0377; |
|
393 *dest += temp << 8; |
|
394 } |
|
395 |
|
396 #ifndef EXTRACT_MACROS /* To debug the macros. */ |
|
397 #undef EXTRACT_NUMBER |
|
398 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) |
|
399 #endif /* not EXTRACT_MACROS */ |
|
400 |
|
401 #endif /* DEBUG */ |
|
402 |
|
403 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. |
|
404 SOURCE must be an lvalue. */ |
|
405 |
|
406 #define EXTRACT_NUMBER_AND_INCR(destination, source) \ |
|
407 do { \ |
|
408 EXTRACT_NUMBER (destination, source); \ |
|
409 (source) += 2; \ |
|
410 } while (0) |
|
411 |
|
412 #ifdef DEBUG |
|
413 static void |
|
414 extract_number_and_incr (destination, source) |
|
415 int *destination; |
|
416 unsigned char **source; |
|
417 { |
|
418 extract_number (destination, *source); |
|
419 *source += 2; |
|
420 } |
|
421 |
|
422 #ifndef EXTRACT_MACROS |
|
423 #undef EXTRACT_NUMBER_AND_INCR |
|
424 #define EXTRACT_NUMBER_AND_INCR(dest, src) \ |
|
425 extract_number_and_incr (&dest, &src) |
|
426 #endif /* not EXTRACT_MACROS */ |
|
427 |
|
428 #endif /* DEBUG */ |
|
429 |
|
430 /* If DEBUG is defined, Regex prints many voluminous messages about what |
|
431 it is doing (if the variable `debug' is nonzero). If linked with the |
|
432 main program in `iregex.c', you can enter patterns and strings |
|
433 interactively. And if linked with the main program in `main.c' and |
|
434 the other test files, you can run the already-written tests. */ |
|
435 |
|
436 #ifdef DEBUG |
|
437 |
|
438 /* We use standard I/O for debugging. */ |
|
439 #include <stdio.h> |
|
440 |
|
441 /* It is useful to test things that ``must'' be true when debugging. */ |
|
442 #include <assert.h> |
|
443 |
|
444 static int debug = 0; |
|
445 |
|
446 #define DEBUG_STATEMENT(e) e |
|
447 #define DEBUG_PRINT1(x) if (debug) printf (x) |
|
448 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) |
|
449 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) |
|
450 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ |
|
451 if (debug) print_partial_compiled_pattern (s, e) |
|
452 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ |
|
453 if (debug) print_double_string (w, s1, sz1, s2, sz2) |
|
454 |
|
455 |
|
456 extern void printchar (); |
|
457 |
|
458 /* Print the fastmap in human-readable form. */ |
|
459 |
|
460 void |
|
461 print_fastmap (fastmap) |
|
462 char *fastmap; |
|
463 { |
|
464 unsigned was_a_range = 0; |
|
465 unsigned i = 0; |
|
466 |
|
467 while (i < (1 << BYTEWIDTH)) |
|
468 { |
|
469 if (fastmap[i++]) |
|
470 { |
|
471 was_a_range = 0; |
|
472 printchar (i - 1); |
|
473 while (i < (1 << BYTEWIDTH) && fastmap[i]) |
|
474 { |
|
475 was_a_range = 1; |
|
476 i++; |
|
477 } |
|
478 if (was_a_range) |
|
479 { |
|
480 printf ("-"); |
|
481 printchar (i - 1); |
|
482 } |
|
483 } |
|
484 } |
|
485 putchar ('\n'); |
|
486 } |
|
487 |
|
488 |
|
489 /* Print a compiled pattern string in human-readable form, starting at |
|
490 the START pointer into it and ending just before the pointer END. */ |
|
491 |
|
492 void |
|
493 print_partial_compiled_pattern (start, end) |
|
494 unsigned char *start; |
|
495 unsigned char *end; |
|
496 { |
|
497 int mcnt, mcnt2; |
|
498 unsigned char *p = start; |
|
499 unsigned char *pend = end; |
|
500 |
|
501 if (start == NULL) |
|
502 { |
|
503 printf ("(null)\n"); |
|
504 return; |
|
505 } |
|
506 |
|
507 /* Loop over pattern commands. */ |
|
508 while (p < pend) |
|
509 { |
|
510 switch ((re_opcode_t) *p++) |
|
511 { |
|
512 case no_op: |
|
513 printf ("/no_op"); |
|
514 break; |
|
515 |
|
516 case exactn: |
|
517 mcnt = *p++; |
|
518 printf ("/exactn/%d", mcnt); |
|
519 do |
|
520 { |
|
521 putchar ('/'); |
|
522 printchar (*p++); |
|
523 } |
|
524 while (--mcnt); |
|
525 break; |
|
526 |
|
527 case start_memory: |
|
528 mcnt = *p++; |
|
529 printf ("/start_memory/%d/%d", mcnt, *p++); |
|
530 break; |
|
531 |
|
532 case stop_memory: |
|
533 mcnt = *p++; |
|
534 printf ("/stop_memory/%d/%d", mcnt, *p++); |
|
535 break; |
|
536 |
|
537 case duplicate: |
|
538 printf ("/duplicate/%d", *p++); |
|
539 break; |
|
540 |
|
541 case anychar: |
|
542 printf ("/anychar"); |
|
543 break; |
|
544 |
|
545 case charset: |
|
546 case charset_not: |
|
547 { |
|
548 register int c; |
|
549 |
|
550 printf ("/charset%s", |
|
551 (re_opcode_t) *(p - 1) == charset_not ? "_not" : ""); |
|
552 |
|
553 assert (p + *p < pend); |
|
554 |
|
555 for (c = 0; c < *p; c++) |
|
556 { |
|
557 unsigned bit; |
|
558 unsigned char map_byte = p[1 + c]; |
|
559 |
|
560 putchar ('/'); |
|
561 |
|
562 for (bit = 0; bit < BYTEWIDTH; bit++) |
|
563 if (map_byte & (1 << bit)) |
|
564 printchar (c * BYTEWIDTH + bit); |
|
565 } |
|
566 p += 1 + *p; |
|
567 break; |
|
568 } |
|
569 |
|
570 case begline: |
|
571 printf ("/begline"); |
|
572 break; |
|
573 |
|
574 case endline: |
|
575 printf ("/endline"); |
|
576 break; |
|
577 |
|
578 case on_failure_jump: |
|
579 extract_number_and_incr (&mcnt, &p); |
|
580 printf ("/on_failure_jump/0/%d", mcnt); |
|
581 break; |
|
582 |
|
583 case on_failure_keep_string_jump: |
|
584 extract_number_and_incr (&mcnt, &p); |
|
585 printf ("/on_failure_keep_string_jump/0/%d", mcnt); |
|
586 break; |
|
587 |
|
588 case dummy_failure_jump: |
|
589 extract_number_and_incr (&mcnt, &p); |
|
590 printf ("/dummy_failure_jump/0/%d", mcnt); |
|
591 break; |
|
592 |
|
593 case push_dummy_failure: |
|
594 printf ("/push_dummy_failure"); |
|
595 break; |
|
596 |
|
597 case maybe_pop_jump: |
|
598 extract_number_and_incr (&mcnt, &p); |
|
599 printf ("/maybe_pop_jump/0/%d", mcnt); |
|
600 break; |
|
601 |
|
602 case pop_failure_jump: |
|
603 extract_number_and_incr (&mcnt, &p); |
|
604 printf ("/pop_failure_jump/0/%d", mcnt); |
|
605 break; |
|
606 |
|
607 case jump_past_alt: |
|
608 extract_number_and_incr (&mcnt, &p); |
|
609 printf ("/jump_past_alt/0/%d", mcnt); |
|
610 break; |
|
611 |
|
612 case jump: |
|
613 extract_number_and_incr (&mcnt, &p); |
|
614 printf ("/jump/0/%d", mcnt); |
|
615 break; |
|
616 |
|
617 case succeed_n: |
|
618 extract_number_and_incr (&mcnt, &p); |
|
619 extract_number_and_incr (&mcnt2, &p); |
|
620 printf ("/succeed_n/0/%d/0/%d", mcnt, mcnt2); |
|
621 break; |
|
622 |
|
623 case jump_n: |
|
624 extract_number_and_incr (&mcnt, &p); |
|
625 extract_number_and_incr (&mcnt2, &p); |
|
626 printf ("/jump_n/0/%d/0/%d", mcnt, mcnt2); |
|
627 break; |
|
628 |
|
629 case set_number_at: |
|
630 extract_number_and_incr (&mcnt, &p); |
|
631 extract_number_and_incr (&mcnt2, &p); |
|
632 printf ("/set_number_at/0/%d/0/%d", mcnt, mcnt2); |
|
633 break; |
|
634 |
|
635 case wordbound: |
|
636 printf ("/wordbound"); |
|
637 break; |
|
638 |
|
639 case notwordbound: |
|
640 printf ("/notwordbound"); |
|
641 break; |
|
642 |
|
643 case wordbeg: |
|
644 printf ("/wordbeg"); |
|
645 break; |
|
646 |
|
647 case wordend: |
|
648 printf ("/wordend"); |
|
649 |
|
650 #ifdef emacs |
|
651 case before_dot: |
|
652 printf ("/before_dot"); |
|
653 break; |
|
654 |
|
655 case at_dot: |
|
656 printf ("/at_dot"); |
|
657 break; |
|
658 |
|
659 case after_dot: |
|
660 printf ("/after_dot"); |
|
661 break; |
|
662 |
|
663 case syntaxspec: |
|
664 printf ("/syntaxspec"); |
|
665 mcnt = *p++; |
|
666 printf ("/%d", mcnt); |
|
667 break; |
|
668 |
|
669 case notsyntaxspec: |
|
670 printf ("/notsyntaxspec"); |
|
671 mcnt = *p++; |
|
672 printf ("/%d", mcnt); |
|
673 break; |
|
674 #endif /* emacs */ |
|
675 |
|
676 case wordchar: |
|
677 printf ("/wordchar"); |
|
678 break; |
|
679 |
|
680 case notwordchar: |
|
681 printf ("/notwordchar"); |
|
682 break; |
|
683 |
|
684 case begbuf: |
|
685 printf ("/begbuf"); |
|
686 break; |
|
687 |
|
688 case endbuf: |
|
689 printf ("/endbuf"); |
|
690 break; |
|
691 |
|
692 default: |
|
693 printf ("?%d", *(p-1)); |
|
694 } |
|
695 } |
|
696 printf ("/\n"); |
|
697 } |
|
698 |
|
699 |
|
700 void |
|
701 print_compiled_pattern (bufp) |
|
702 struct re_pattern_buffer *bufp; |
|
703 { |
|
704 unsigned char *buffer = bufp->buffer; |
|
705 |
|
706 print_partial_compiled_pattern (buffer, buffer + bufp->used); |
|
707 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated); |
|
708 |
|
709 if (bufp->fastmap_accurate && bufp->fastmap) |
|
710 { |
|
711 printf ("fastmap: "); |
|
712 print_fastmap (bufp->fastmap); |
|
713 } |
|
714 |
|
715 printf ("re_nsub: %d\t", bufp->re_nsub); |
|
716 printf ("regs_alloc: %d\t", bufp->regs_allocated); |
|
717 printf ("can_be_null: %d\t", bufp->can_be_null); |
|
718 printf ("newline_anchor: %d\n", bufp->newline_anchor); |
|
719 printf ("no_sub: %d\t", bufp->no_sub); |
|
720 printf ("not_bol: %d\t", bufp->not_bol); |
|
721 printf ("not_eol: %d\t", bufp->not_eol); |
|
722 printf ("syntax: %d\n", bufp->syntax); |
|
723 /* Perhaps we should print the translate table? */ |
|
724 } |
|
725 |
|
726 |
|
727 void |
|
728 print_double_string (where, string1, size1, string2, size2) |
|
729 const char *where; |
|
730 const char *string1; |
|
731 const char *string2; |
|
732 int size1; |
|
733 int size2; |
|
734 { |
|
735 unsigned this_char; |
|
736 |
|
737 if (where == NULL) |
|
738 printf ("(null)"); |
|
739 else |
|
740 { |
|
741 if (FIRST_STRING_P (where)) |
|
742 { |
|
743 for (this_char = where - string1; this_char < size1; this_char++) |
|
744 printchar (string1[this_char]); |
|
745 |
|
746 where = string2; |
|
747 } |
|
748 |
|
749 for (this_char = where - string2; this_char < size2; this_char++) |
|
750 printchar (string2[this_char]); |
|
751 } |
|
752 } |
|
753 |
|
754 #else /* not DEBUG */ |
|
755 |
|
756 #undef assert |
|
757 #define assert(e) |
|
758 |
|
759 #define DEBUG_STATEMENT(e) |
|
760 #define DEBUG_PRINT1(x) |
|
761 #define DEBUG_PRINT2(x1, x2) |
|
762 #define DEBUG_PRINT3(x1, x2, x3) |
|
763 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) |
|
764 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) |
|
765 |
|
766 #endif /* not DEBUG */ |
|
767 |
|
768 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can |
|
769 also be assigned to arbitrarily: each pattern buffer stores its own |
|
770 syntax, so it can be changed between regex compilations. */ |
|
771 reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS; |
|
772 |
|
773 |
|
774 /* Specify the precise syntax of regexps for compilation. This provides |
|
775 for compatibility for various utilities which historically have |
|
776 different, incompatible syntaxes. |
|
777 |
|
778 The argument SYNTAX is a bit mask comprised of the various bits |
|
779 defined in regex.h. We return the old syntax. */ |
|
780 |
|
781 reg_syntax_t |
|
782 re_set_syntax (syntax) |
|
783 reg_syntax_t syntax; |
|
784 { |
|
785 reg_syntax_t ret = re_syntax_options; |
|
786 |
|
787 re_syntax_options = syntax; |
|
788 return ret; |
|
789 } |
|
790 |
|
791 /* This table gives an error message for each of the error codes listed |
|
792 in regex.h. Obviously the order here has to be same as there. */ |
|
793 |
|
794 static const char *re_error_msg[] = |
|
795 { NULL, /* REG_NOERROR */ |
|
796 "No match", /* REG_NOMATCH */ |
|
797 "Invalid regular expression", /* REG_BADPAT */ |
|
798 "Invalid collation character", /* REG_ECOLLATE */ |
|
799 "Invalid character class name", /* REG_ECTYPE */ |
|
800 "Trailing backslash", /* REG_EESCAPE */ |
|
801 "Invalid back reference", /* REG_ESUBREG */ |
|
802 "Unmatched [ or [^", /* REG_EBRACK */ |
|
803 "Unmatched ( or \\(", /* REG_EPAREN */ |
|
804 "Unmatched \\{", /* REG_EBRACE */ |
|
805 "Invalid content of \\{\\}", /* REG_BADBR */ |
|
806 "Invalid range end", /* REG_ERANGE */ |
|
807 "Memory exhausted", /* REG_ESPACE */ |
|
808 "Invalid preceding regular expression", /* REG_BADRPT */ |
|
809 "Premature end of regular expression", /* REG_EEND */ |
|
810 "Regular expression too big", /* REG_ESIZE */ |
|
811 "Unmatched ) or \\)", /* REG_ERPAREN */ |
|
812 }; |
|
813 |
|
814 /* Subroutine declarations and macros for regex_compile. */ |
|
815 |
|
816 static void store_op1 (), store_op2 (); |
|
817 static void insert_op1 (), insert_op2 (); |
|
818 static boolean at_begline_loc_p (), at_endline_loc_p (); |
|
819 static boolean group_in_compile_stack (); |
|
820 static reg_errcode_t compile_range (); |
|
821 |
|
822 /* Fetch the next character in the uncompiled pattern---translating it |
|
823 if necessary. Also cast from a signed character in the constant |
|
824 string passed to us by the user to an unsigned char that we can use |
|
825 as an array index (in, e.g., `translate'). */ |
|
826 #define PATFETCH(c) \ |
|
827 do {if (p == pend) return REG_EEND; \ |
|
828 c = (unsigned char) *p++; \ |
|
829 if (translate) c = translate[c]; \ |
|
830 } while (0) |
|
831 |
|
832 /* Fetch the next character in the uncompiled pattern, with no |
|
833 translation. */ |
|
834 #define PATFETCH_RAW(c) \ |
|
835 do {if (p == pend) return REG_EEND; \ |
|
836 c = (unsigned char) *p++; \ |
|
837 } while (0) |
|
838 |
|
839 /* Go backwards one character in the pattern. */ |
|
840 #define PATUNFETCH p-- |
|
841 |
|
842 |
|
843 /* If `translate' is non-null, return translate[D], else just D. We |
|
844 cast the subscript to translate because some data is declared as |
|
845 `char *', to avoid warnings when a string constant is passed. But |
|
846 when we use a character as a subscript we must make it unsigned. */ |
|
847 #define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d)) |
|
848 |
|
849 |
|
850 /* Macros for outputting the compiled pattern into `buffer'. */ |
|
851 |
|
852 /* If the buffer isn't allocated when it comes in, use this. */ |
|
853 #define INIT_BUF_SIZE 32 |
|
854 |
|
855 /* Make sure we have at least N more bytes of space in buffer. */ |
|
856 #define GET_BUFFER_SPACE(n) \ |
|
857 while (b - bufp->buffer + (n) > bufp->allocated) \ |
|
858 EXTEND_BUFFER () |
|
859 |
|
860 /* Make sure we have one more byte of buffer space and then add C to it. */ |
|
861 #define BUF_PUSH(c) \ |
|
862 do { \ |
|
863 GET_BUFFER_SPACE (1); \ |
|
864 *b++ = (unsigned char) (c); \ |
|
865 } while (0) |
|
866 |
|
867 |
|
868 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */ |
|
869 #define BUF_PUSH_2(c1, c2) \ |
|
870 do { \ |
|
871 GET_BUFFER_SPACE (2); \ |
|
872 *b++ = (unsigned char) (c1); \ |
|
873 *b++ = (unsigned char) (c2); \ |
|
874 } while (0) |
|
875 |
|
876 |
|
877 /* As with BUF_PUSH_2, except for three bytes. */ |
|
878 #define BUF_PUSH_3(c1, c2, c3) \ |
|
879 do { \ |
|
880 GET_BUFFER_SPACE (3); \ |
|
881 *b++ = (unsigned char) (c1); \ |
|
882 *b++ = (unsigned char) (c2); \ |
|
883 *b++ = (unsigned char) (c3); \ |
|
884 } while (0) |
|
885 |
|
886 |
|
887 /* Store a jump with opcode OP at LOC to location TO. We store a |
|
888 relative address offset by the three bytes the jump itself occupies. */ |
|
889 #define STORE_JUMP(op, loc, to) \ |
|
890 store_op1 (op, loc, (to) - (loc) - 3) |
|
891 |
|
892 /* Likewise, for a two-argument jump. */ |
|
893 #define STORE_JUMP2(op, loc, to, arg) \ |
|
894 store_op2 (op, loc, (to) - (loc) - 3, arg) |
|
895 |
|
896 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ |
|
897 #define INSERT_JUMP(op, loc, to) \ |
|
898 insert_op1 (op, loc, (to) - (loc) - 3, b) |
|
899 |
|
900 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ |
|
901 #define INSERT_JUMP2(op, loc, to, arg) \ |
|
902 insert_op2 (op, loc, (to) - (loc) - 3, arg, b) |
|
903 |
|
904 |
|
905 /* This is not an arbitrary limit: the arguments which represent offsets |
|
906 into the pattern are two bytes long. So if 2^16 bytes turns out to |
|
907 be too small, many things would have to change. */ |
|
908 #define MAX_BUF_SIZE (1L << 16) |
|
909 |
|
910 |
|
911 /* Extend the buffer by twice its current size via realloc and |
|
912 reset the pointers that pointed into the old block to point to the |
|
913 correct places in the new one. If extending the buffer results in it |
|
914 being larger than MAX_BUF_SIZE, then flag memory exhausted. */ |
|
915 #define EXTEND_BUFFER() \ |
|
916 do { \ |
|
917 unsigned char *old_buffer = bufp->buffer; \ |
|
918 if (bufp->allocated == MAX_BUF_SIZE) \ |
|
919 return REG_ESIZE; \ |
|
920 bufp->allocated <<= 1; \ |
|
921 if (bufp->allocated > MAX_BUF_SIZE) \ |
|
922 bufp->allocated = MAX_BUF_SIZE; \ |
|
923 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\ |
|
924 if (bufp->buffer == NULL) \ |
|
925 return REG_ESPACE; \ |
|
926 /* If the buffer moved, move all the pointers into it. */ \ |
|
927 if (old_buffer != bufp->buffer) \ |
|
928 { \ |
|
929 b = (b - old_buffer) + bufp->buffer; \ |
|
930 begalt = (begalt - old_buffer) + bufp->buffer; \ |
|
931 if (fixup_alt_jump) \ |
|
932 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ |
|
933 if (laststart) \ |
|
934 laststart = (laststart - old_buffer) + bufp->buffer; \ |
|
935 if (pending_exact) \ |
|
936 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ |
|
937 } \ |
|
938 } while (0) |
|
939 |
|
940 |
|
941 /* Since we have one byte reserved for the register number argument to |
|
942 {start,stop}_memory, the maximum number of groups we can report |
|
943 things about is what fits in that byte. */ |
|
944 #define MAX_REGNUM 255 |
|
945 |
|
946 /* But patterns can have more than `MAX_REGNUM' registers. We just |
|
947 ignore the excess. */ |
|
948 typedef unsigned regnum_t; |
|
949 |
|
950 |
|
951 /* Macros for the compile stack. */ |
|
952 |
|
953 /* Since offsets can go either forwards or backwards, this type needs to |
|
954 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ |
|
955 typedef int pattern_offset_t; |
|
956 |
|
957 typedef struct |
|
958 { |
|
959 pattern_offset_t begalt_offset; |
|
960 pattern_offset_t fixup_alt_jump; |
|
961 pattern_offset_t inner_group_offset; |
|
962 pattern_offset_t laststart_offset; |
|
963 regnum_t regnum; |
|
964 } compile_stack_elt_t; |
|
965 |
|
966 |
|
967 typedef struct |
|
968 { |
|
969 compile_stack_elt_t *stack; |
|
970 unsigned size; |
|
971 unsigned avail; /* Offset of next open position. */ |
|
972 } compile_stack_type; |
|
973 |
|
974 |
|
975 #define INIT_COMPILE_STACK_SIZE 32 |
|
976 |
|
977 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) |
|
978 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) |
|
979 |
|
980 /* The next available element. */ |
|
981 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) |
|
982 |
|
983 |
|
984 /* Set the bit for character C in a list. */ |
|
985 #define SET_LIST_BIT(c) \ |
|
986 (b[((unsigned char) (c)) / BYTEWIDTH] \ |
|
987 |= 1 << (((unsigned char) c) % BYTEWIDTH)) |
|
988 |
|
989 |
|
990 /* Get the next unsigned number in the uncompiled pattern. */ |
|
991 #define GET_UNSIGNED_NUMBER(num) \ |
|
992 { if (p != pend) \ |
|
993 { \ |
|
994 PATFETCH (c); \ |
|
995 while (isdigit (c)) \ |
|
996 { \ |
|
997 if (num < 0) \ |
|
998 num = 0; \ |
|
999 num = num * 10 + c - '0'; \ |
|
1000 if (p == pend) \ |
|
1001 break; \ |
|
1002 PATFETCH (c); \ |
|
1003 } \ |
|
1004 } \ |
|
1005 } |
|
1006 |
|
1007 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ |
|
1008 |
|
1009 #define IS_CHAR_CLASS(string) \ |
|
1010 (STREQ (string, "alpha") || STREQ (string, "upper") \ |
|
1011 || STREQ (string, "lower") || STREQ (string, "digit") \ |
|
1012 || STREQ (string, "alnum") || STREQ (string, "xdigit") \ |
|
1013 || STREQ (string, "space") || STREQ (string, "print") \ |
|
1014 || STREQ (string, "punct") || STREQ (string, "graph") \ |
|
1015 || STREQ (string, "cntrl") || STREQ (string, "blank")) |
|
1016 |
|
1017 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. |
|
1018 Returns one of error codes defined in `regex.h', or zero for success. |
|
1019 |
|
1020 Assumes the `allocated' (and perhaps `buffer') and `translate' |
|
1021 fields are set in BUFP on entry. |
|
1022 |
|
1023 If it succeeds, results are put in BUFP (if it returns an error, the |
|
1024 contents of BUFP are undefined): |
|
1025 `buffer' is the compiled pattern; |
|
1026 `syntax' is set to SYNTAX; |
|
1027 `used' is set to the length of the compiled pattern; |
|
1028 `fastmap_accurate' is set to zero; |
|
1029 `re_nsub' is set to the number of groups in PATTERN; |
|
1030 `not_bol' and `not_eol' are set to zero. |
|
1031 |
|
1032 The `fastmap' and `newline_anchor' fields are neither |
|
1033 examined nor set. */ |
|
1034 |
|
1035 static reg_errcode_t |
|
1036 regex_compile (pattern, size, syntax, bufp) |
|
1037 const char *pattern; |
|
1038 int size; |
|
1039 reg_syntax_t syntax; |
|
1040 struct re_pattern_buffer *bufp; |
|
1041 { |
|
1042 /* We fetch characters from PATTERN here. Even though PATTERN is |
|
1043 `char *' (i.e., signed), we declare these variables as unsigned, so |
|
1044 they can be reliably used as array indices. */ |
|
1045 register unsigned char c, c1; |
|
1046 |
|
1047 /* A random tempory spot in PATTERN. */ |
|
1048 const char *p1; |
|
1049 |
|
1050 /* Points to the end of the buffer, where we should append. */ |
|
1051 register unsigned char *b; |
|
1052 |
|
1053 /* Keeps track of unclosed groups. */ |
|
1054 compile_stack_type compile_stack; |
|
1055 |
|
1056 /* Points to the current (ending) position in the pattern. */ |
|
1057 const char *p = pattern; |
|
1058 const char *pend = pattern + size; |
|
1059 |
|
1060 /* How to translate the characters in the pattern. */ |
|
1061 char *translate = bufp->translate; |
|
1062 |
|
1063 /* Address of the count-byte of the most recently inserted `exactn' |
|
1064 command. This makes it possible to tell if a new exact-match |
|
1065 character can be added to that command or if the character requires |
|
1066 a new `exactn' command. */ |
|
1067 unsigned char *pending_exact = 0; |
|
1068 |
|
1069 /* Address of start of the most recently finished expression. |
|
1070 This tells, e.g., postfix * where to find the start of its |
|
1071 operand. Reset at the beginning of groups and alternatives. */ |
|
1072 unsigned char *laststart = 0; |
|
1073 |
|
1074 /* Address of beginning of regexp, or inside of last group. */ |
|
1075 unsigned char *begalt; |
|
1076 |
|
1077 /* Place in the uncompiled pattern (i.e., the {) to |
|
1078 which to go back if the interval is invalid. */ |
|
1079 const char *beg_interval; |
|
1080 |
|
1081 /* Address of the place where a forward jump should go to the end of |
|
1082 the containing expression. Each alternative of an `or' -- except the |
|
1083 last -- ends with a forward jump of this sort. */ |
|
1084 unsigned char *fixup_alt_jump = 0; |
|
1085 |
|
1086 /* Counts open-groups as they are encountered. Remembered for the |
|
1087 matching close-group on the compile stack, so the same register |
|
1088 number is put in the stop_memory as the start_memory. */ |
|
1089 regnum_t regnum = 0; |
|
1090 |
|
1091 #ifdef DEBUG |
|
1092 DEBUG_PRINT1 ("\nCompiling pattern: "); |
|
1093 if (debug) |
|
1094 { |
|
1095 unsigned debug_count; |
|
1096 |
|
1097 for (debug_count = 0; debug_count < size; debug_count++) |
|
1098 printchar (pattern[debug_count]); |
|
1099 putchar ('\n'); |
|
1100 } |
|
1101 #endif /* DEBUG */ |
|
1102 |
|
1103 /* Initialize the compile stack. */ |
|
1104 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); |
|
1105 if (compile_stack.stack == NULL) |
|
1106 return REG_ESPACE; |
|
1107 |
|
1108 compile_stack.size = INIT_COMPILE_STACK_SIZE; |
|
1109 compile_stack.avail = 0; |
|
1110 |
|
1111 /* Initialize the pattern buffer. */ |
|
1112 bufp->syntax = syntax; |
|
1113 bufp->fastmap_accurate = 0; |
|
1114 bufp->not_bol = bufp->not_eol = 0; |
|
1115 |
|
1116 /* Set `used' to zero, so that if we return an error, the pattern |
|
1117 printer (for debugging) will think there's no pattern. We reset it |
|
1118 at the end. */ |
|
1119 bufp->used = 0; |
|
1120 |
|
1121 /* Always count groups, whether or not bufp->no_sub is set. */ |
|
1122 bufp->re_nsub = 0; |
|
1123 |
|
1124 #if !defined (emacs) && !defined (SYNTAX_TABLE) |
|
1125 /* Initialize the syntax table. */ |
|
1126 init_syntax_once (); |
|
1127 #endif |
|
1128 |
|
1129 if (bufp->allocated == 0) |
|
1130 { |
|
1131 if (bufp->buffer) |
|
1132 { /* If zero allocated, but buffer is non-null, try to realloc |
|
1133 enough space. This loses if buffer's address is bogus, but |
|
1134 that is the user's responsibility. */ |
|
1135 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); |
|
1136 } |
|
1137 else |
|
1138 { /* Caller did not allocate a buffer. Do it for them. */ |
|
1139 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); |
|
1140 } |
|
1141 if (!bufp->buffer) return REG_ESPACE; |
|
1142 |
|
1143 bufp->allocated = INIT_BUF_SIZE; |
|
1144 } |
|
1145 |
|
1146 begalt = b = bufp->buffer; |
|
1147 |
|
1148 /* Loop through the uncompiled pattern until we're at the end. */ |
|
1149 while (p != pend) |
|
1150 { |
|
1151 PATFETCH (c); |
|
1152 |
|
1153 switch (c) |
|
1154 { |
|
1155 case '^': |
|
1156 { |
|
1157 if ( /* If at start of pattern, it's an operator. */ |
|
1158 p == pattern + 1 |
|
1159 /* If context independent, it's an operator. */ |
|
1160 || syntax & RE_CONTEXT_INDEP_ANCHORS |
|
1161 /* Otherwise, depends on what's come before. */ |
|
1162 || at_begline_loc_p (pattern, p, syntax)) |
|
1163 BUF_PUSH (begline); |
|
1164 else |
|
1165 goto normal_char; |
|
1166 } |
|
1167 break; |
|
1168 |
|
1169 |
|
1170 case '$': |
|
1171 { |
|
1172 if ( /* If at end of pattern, it's an operator. */ |
|
1173 p == pend |
|
1174 /* If context independent, it's an operator. */ |
|
1175 || syntax & RE_CONTEXT_INDEP_ANCHORS |
|
1176 /* Otherwise, depends on what's next. */ |
|
1177 || at_endline_loc_p (p, pend, syntax)) |
|
1178 BUF_PUSH (endline); |
|
1179 else |
|
1180 goto normal_char; |
|
1181 } |
|
1182 break; |
|
1183 |
|
1184 |
|
1185 case '+': |
|
1186 case '?': |
|
1187 if ((syntax & RE_BK_PLUS_QM) |
|
1188 || (syntax & RE_LIMITED_OPS)) |
|
1189 goto normal_char; |
|
1190 handle_plus: |
|
1191 case '*': |
|
1192 /* If there is no previous pattern... */ |
|
1193 if (!laststart) |
|
1194 { |
|
1195 if (syntax & RE_CONTEXT_INVALID_OPS) |
|
1196 return REG_BADRPT; |
|
1197 else if (!(syntax & RE_CONTEXT_INDEP_OPS)) |
|
1198 goto normal_char; |
|
1199 } |
|
1200 |
|
1201 { |
|
1202 /* Are we optimizing this jump? */ |
|
1203 boolean keep_string_p = false; |
|
1204 |
|
1205 /* 1 means zero (many) matches is allowed. */ |
|
1206 char zero_times_ok = 0, many_times_ok = 0; |
|
1207 |
|
1208 /* If there is a sequence of repetition chars, collapse it |
|
1209 down to just one (the right one). We can't combine |
|
1210 interval operators with these because of, e.g., `a{2}*', |
|
1211 which should only match an even number of `a's. */ |
|
1212 |
|
1213 for (;;) |
|
1214 { |
|
1215 zero_times_ok |= c != '+'; |
|
1216 many_times_ok |= c != '?'; |
|
1217 |
|
1218 if (p == pend) |
|
1219 break; |
|
1220 |
|
1221 PATFETCH (c); |
|
1222 |
|
1223 if (c == '*' |
|
1224 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) |
|
1225 ; |
|
1226 |
|
1227 else if (syntax & RE_BK_PLUS_QM && c == '\\') |
|
1228 { |
|
1229 if (p == pend) return REG_EESCAPE; |
|
1230 |
|
1231 PATFETCH (c1); |
|
1232 if (!(c1 == '+' || c1 == '?')) |
|
1233 { |
|
1234 PATUNFETCH; |
|
1235 PATUNFETCH; |
|
1236 break; |
|
1237 } |
|
1238 |
|
1239 c = c1; |
|
1240 } |
|
1241 else |
|
1242 { |
|
1243 PATUNFETCH; |
|
1244 break; |
|
1245 } |
|
1246 |
|
1247 /* If we get here, we found another repeat character. */ |
|
1248 } |
|
1249 |
|
1250 /* Star, etc. applied to an empty pattern is equivalent |
|
1251 to an empty pattern. */ |
|
1252 if (!laststart) |
|
1253 break; |
|
1254 |
|
1255 /* Now we know whether or not zero matches is allowed |
|
1256 and also whether or not two or more matches is allowed. */ |
|
1257 if (many_times_ok) |
|
1258 { /* More than one repetition is allowed, so put in at the |
|
1259 end a backward relative jump from `b' to before the next |
|
1260 jump we're going to put in below (which jumps from |
|
1261 laststart to after this jump). |
|
1262 |
|
1263 But if we are at the `*' in the exact sequence `.*\n', |
|
1264 insert an unconditional jump backwards to the ., |
|
1265 instead of the beginning of the loop. This way we only |
|
1266 push a failure point once, instead of every time |
|
1267 through the loop. */ |
|
1268 assert (p - 1 > pattern); |
|
1269 |
|
1270 /* Allocate the space for the jump. */ |
|
1271 GET_BUFFER_SPACE (3); |
|
1272 |
|
1273 /* We know we are not at the first character of the pattern, |
|
1274 because laststart was nonzero. And we've already |
|
1275 incremented `p', by the way, to be the character after |
|
1276 the `*'. Do we have to do something analogous here |
|
1277 for null bytes, because of RE_DOT_NOT_NULL? */ |
|
1278 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') |
|
1279 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n') |
|
1280 && !(syntax & RE_DOT_NEWLINE)) |
|
1281 { /* We have .*\n. */ |
|
1282 STORE_JUMP (jump, b, laststart); |
|
1283 keep_string_p = true; |
|
1284 } |
|
1285 else |
|
1286 /* Anything else. */ |
|
1287 STORE_JUMP (maybe_pop_jump, b, laststart - 3); |
|
1288 |
|
1289 /* We've added more stuff to the buffer. */ |
|
1290 b += 3; |
|
1291 } |
|
1292 |
|
1293 /* On failure, jump from laststart to b + 3, which will be the |
|
1294 end of the buffer after this jump is inserted. */ |
|
1295 GET_BUFFER_SPACE (3); |
|
1296 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump |
|
1297 : on_failure_jump, |
|
1298 laststart, b + 3); |
|
1299 pending_exact = 0; |
|
1300 b += 3; |
|
1301 |
|
1302 if (!zero_times_ok) |
|
1303 { |
|
1304 /* At least one repetition is required, so insert a |
|
1305 `dummy_failure_jump' before the initial |
|
1306 `on_failure_jump' instruction of the loop. This |
|
1307 effects a skip over that instruction the first time |
|
1308 we hit that loop. */ |
|
1309 GET_BUFFER_SPACE (3); |
|
1310 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); |
|
1311 b += 3; |
|
1312 } |
|
1313 } |
|
1314 break; |
|
1315 |
|
1316 |
|
1317 case '.': |
|
1318 laststart = b; |
|
1319 BUF_PUSH (anychar); |
|
1320 break; |
|
1321 |
|
1322 |
|
1323 case '[': |
|
1324 { |
|
1325 boolean had_char_class = false; |
|
1326 |
|
1327 if (p == pend) return REG_EBRACK; |
|
1328 |
|
1329 /* Ensure that we have enough space to push a charset: the |
|
1330 opcode, the length count, and the bitset; 34 bytes in all. */ |
|
1331 GET_BUFFER_SPACE (34); |
|
1332 |
|
1333 laststart = b; |
|
1334 |
|
1335 /* We test `*p == '^' twice, instead of using an if |
|
1336 statement, so we only need one BUF_PUSH. */ |
|
1337 BUF_PUSH (*p == '^' ? charset_not : charset); |
|
1338 if (*p == '^') |
|
1339 p++; |
|
1340 |
|
1341 /* Remember the first position in the bracket expression. */ |
|
1342 p1 = p; |
|
1343 |
|
1344 /* Push the number of bytes in the bitmap. */ |
|
1345 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); |
|
1346 |
|
1347 /* Clear the whole map. */ |
|
1348 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); |
|
1349 |
|
1350 /* charset_not matches newline according to a syntax bit. */ |
|
1351 if ((re_opcode_t) b[-2] == charset_not |
|
1352 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
|
1353 SET_LIST_BIT ('\n'); |
|
1354 |
|
1355 /* Read in characters and ranges, setting map bits. */ |
|
1356 for (;;) |
|
1357 { |
|
1358 if (p == pend) return REG_EBRACK; |
|
1359 |
|
1360 PATFETCH (c); |
|
1361 |
|
1362 /* \ might escape characters inside [...] and [^...]. */ |
|
1363 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') |
|
1364 { |
|
1365 if (p == pend) return REG_EESCAPE; |
|
1366 |
|
1367 PATFETCH (c1); |
|
1368 SET_LIST_BIT (c1); |
|
1369 continue; |
|
1370 } |
|
1371 |
|
1372 /* Could be the end of the bracket expression. If it's |
|
1373 not (i.e., when the bracket expression is `[]' so |
|
1374 far), the ']' character bit gets set way below. */ |
|
1375 if (c == ']' && p != p1 + 1) |
|
1376 break; |
|
1377 |
|
1378 /* Look ahead to see if it's a range when the last thing |
|
1379 was a character class. */ |
|
1380 if (had_char_class && c == '-' && *p != ']') |
|
1381 return REG_ERANGE; |
|
1382 |
|
1383 /* Look ahead to see if it's a range when the last thing |
|
1384 was a character: if this is a hyphen not at the |
|
1385 beginning or the end of a list, then it's the range |
|
1386 operator. */ |
|
1387 if (c == '-' |
|
1388 && !(p - 2 >= pattern && p[-2] == '[') |
|
1389 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
|
1390 && *p != ']') |
|
1391 { |
|
1392 reg_errcode_t ret |
|
1393 = compile_range (&p, pend, translate, syntax, b); |
|
1394 if (ret != REG_NOERROR) return ret; |
|
1395 } |
|
1396 |
|
1397 else if (p[0] == '-' && p[1] != ']') |
|
1398 { /* This handles ranges made up of characters only. */ |
|
1399 reg_errcode_t ret; |
|
1400 |
|
1401 /* Move past the `-'. */ |
|
1402 PATFETCH (c1); |
|
1403 |
|
1404 ret = compile_range (&p, pend, translate, syntax, b); |
|
1405 if (ret != REG_NOERROR) return ret; |
|
1406 } |
|
1407 |
|
1408 /* See if we're at the beginning of a possible character |
|
1409 class. */ |
|
1410 |
|
1411 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') |
|
1412 { /* Leave room for the null. */ |
|
1413 char str[CHAR_CLASS_MAX_LENGTH + 1]; |
|
1414 |
|
1415 PATFETCH (c); |
|
1416 c1 = 0; |
|
1417 |
|
1418 /* If pattern is `[[:'. */ |
|
1419 if (p == pend) return REG_EBRACK; |
|
1420 |
|
1421 for (;;) |
|
1422 { |
|
1423 PATFETCH (c); |
|
1424 if (c == ':' || c == ']' || p == pend |
|
1425 || c1 == CHAR_CLASS_MAX_LENGTH) |
|
1426 break; |
|
1427 str[c1++] = c; |
|
1428 } |
|
1429 str[c1] = '\0'; |
|
1430 |
|
1431 /* If isn't a word bracketed by `[:' and:`]': |
|
1432 undo the ending character, the letters, and leave |
|
1433 the leading `:' and `[' (but set bits for them). */ |
|
1434 if (c == ':' && *p == ']') |
|
1435 { |
|
1436 int ch; |
|
1437 boolean is_alnum = STREQ (str, "alnum"); |
|
1438 boolean is_alpha = STREQ (str, "alpha"); |
|
1439 boolean is_blank = STREQ (str, "blank"); |
|
1440 boolean is_cntrl = STREQ (str, "cntrl"); |
|
1441 boolean is_digit = STREQ (str, "digit"); |
|
1442 boolean is_graph = STREQ (str, "graph"); |
|
1443 boolean is_lower = STREQ (str, "lower"); |
|
1444 boolean is_print = STREQ (str, "print"); |
|
1445 boolean is_punct = STREQ (str, "punct"); |
|
1446 boolean is_space = STREQ (str, "space"); |
|
1447 boolean is_upper = STREQ (str, "upper"); |
|
1448 boolean is_xdigit = STREQ (str, "xdigit"); |
|
1449 |
|
1450 if (!IS_CHAR_CLASS (str)) return REG_ECTYPE; |
|
1451 |
|
1452 /* Throw away the ] at the end of the character |
|
1453 class. */ |
|
1454 PATFETCH (c); |
|
1455 |
|
1456 if (p == pend) return REG_EBRACK; |
|
1457 |
|
1458 for (ch = 0; ch < 1 << BYTEWIDTH; ch++) |
|
1459 { |
|
1460 if ( (is_alnum && isalnum (ch)) |
|
1461 || (is_alpha && isalpha (ch)) |
|
1462 || (is_blank && isblank (ch)) |
|
1463 || (is_cntrl && iscntrl (ch)) |
|
1464 || (is_digit && isdigit (ch)) |
|
1465 || (is_graph && isgraph (ch)) |
|
1466 || (is_lower && islower (ch)) |
|
1467 || (is_print && isprint (ch)) |
|
1468 || (is_punct && ispunct (ch)) |
|
1469 || (is_space && isspace (ch)) |
|
1470 || (is_upper && isupper (ch)) |
|
1471 || (is_xdigit && isxdigit (ch))) |
|
1472 SET_LIST_BIT (ch); |
|
1473 } |
|
1474 had_char_class = true; |
|
1475 } |
|
1476 else |
|
1477 { |
|
1478 c1++; |
|
1479 while (c1--) |
|
1480 PATUNFETCH; |
|
1481 SET_LIST_BIT ('['); |
|
1482 SET_LIST_BIT (':'); |
|
1483 had_char_class = false; |
|
1484 } |
|
1485 } |
|
1486 else |
|
1487 { |
|
1488 had_char_class = false; |
|
1489 SET_LIST_BIT (c); |
|
1490 } |
|
1491 } |
|
1492 |
|
1493 /* Discard any (non)matching list bytes that are all 0 at the |
|
1494 end of the map. Decrease the map-length byte too. */ |
|
1495 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) |
|
1496 b[-1]--; |
|
1497 b += b[-1]; |
|
1498 } |
|
1499 break; |
|
1500 |
|
1501 |
|
1502 case '(': |
|
1503 if (syntax & RE_NO_BK_PARENS) |
|
1504 goto handle_open; |
|
1505 else |
|
1506 goto normal_char; |
|
1507 |
|
1508 |
|
1509 case ')': |
|
1510 if (syntax & RE_NO_BK_PARENS) |
|
1511 goto handle_close; |
|
1512 else |
|
1513 goto normal_char; |
|
1514 |
|
1515 |
|
1516 case '\n': |
|
1517 if (syntax & RE_NEWLINE_ALT) |
|
1518 goto handle_alt; |
|
1519 else |
|
1520 goto normal_char; |
|
1521 |
|
1522 |
|
1523 case '|': |
|
1524 if (syntax & RE_NO_BK_VBAR) |
|
1525 goto handle_alt; |
|
1526 else |
|
1527 goto normal_char; |
|
1528 |
|
1529 |
|
1530 case '{': |
|
1531 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) |
|
1532 goto handle_interval; |
|
1533 else |
|
1534 goto normal_char; |
|
1535 |
|
1536 |
|
1537 case '\\': |
|
1538 if (p == pend) return REG_EESCAPE; |
|
1539 |
|
1540 /* Do not translate the character after the \, so that we can |
|
1541 distinguish, e.g., \B from \b, even if we normally would |
|
1542 translate, e.g., B to b. */ |
|
1543 PATFETCH_RAW (c); |
|
1544 |
|
1545 switch (c) |
|
1546 { |
|
1547 case '(': |
|
1548 if (syntax & RE_NO_BK_PARENS) |
|
1549 goto normal_backslash; |
|
1550 |
|
1551 handle_open: |
|
1552 bufp->re_nsub++; |
|
1553 regnum++; |
|
1554 |
|
1555 if (COMPILE_STACK_FULL) |
|
1556 { |
|
1557 RETALLOC (compile_stack.stack, compile_stack.size << 1, |
|
1558 compile_stack_elt_t); |
|
1559 if (compile_stack.stack == NULL) return REG_ESPACE; |
|
1560 |
|
1561 compile_stack.size <<= 1; |
|
1562 } |
|
1563 |
|
1564 /* These are the values to restore when we hit end of this |
|
1565 group. They are all relative offsets, so that if the |
|
1566 whole pattern moves because of realloc, they will still |
|
1567 be valid. */ |
|
1568 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; |
|
1569 COMPILE_STACK_TOP.fixup_alt_jump |
|
1570 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; |
|
1571 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; |
|
1572 COMPILE_STACK_TOP.regnum = regnum; |
|
1573 |
|
1574 /* We will eventually replace the 0 with the number of |
|
1575 groups inner to this one. But do not push a |
|
1576 start_memory for groups beyond the last one we can |
|
1577 represent in the compiled pattern. */ |
|
1578 if (regnum <= MAX_REGNUM) |
|
1579 { |
|
1580 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; |
|
1581 BUF_PUSH_3 (start_memory, regnum, 0); |
|
1582 } |
|
1583 |
|
1584 compile_stack.avail++; |
|
1585 |
|
1586 fixup_alt_jump = 0; |
|
1587 laststart = 0; |
|
1588 begalt = b; |
|
1589 break; |
|
1590 |
|
1591 |
|
1592 case ')': |
|
1593 if (syntax & RE_NO_BK_PARENS) goto normal_backslash; |
|
1594 |
|
1595 if (COMPILE_STACK_EMPTY) |
|
1596 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
|
1597 goto normal_backslash; |
|
1598 else |
|
1599 return REG_ERPAREN; |
|
1600 |
|
1601 handle_close: |
|
1602 if (fixup_alt_jump) |
|
1603 { /* Push a dummy failure point at the end of the |
|
1604 alternative for a possible future |
|
1605 `pop_failure_jump' to pop. See comments at |
|
1606 `push_dummy_failure' in `re_match_2'. */ |
|
1607 BUF_PUSH (push_dummy_failure); |
|
1608 |
|
1609 /* We allocated space for this jump when we assigned |
|
1610 to `fixup_alt_jump', in the `handle_alt' case below. */ |
|
1611 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); |
|
1612 } |
|
1613 |
|
1614 /* See similar code for backslashed left paren above. */ |
|
1615 if (COMPILE_STACK_EMPTY) |
|
1616 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
|
1617 goto normal_char; |
|
1618 else |
|
1619 return REG_ERPAREN; |
|
1620 |
|
1621 /* Since we just checked for an empty stack above, this |
|
1622 ``can't happen''. */ |
|
1623 assert (compile_stack.avail != 0); |
|
1624 { |
|
1625 /* We don't just want to restore into `regnum', because |
|
1626 later groups should continue to be numbered higher, |
|
1627 as in `(ab)c(de)' -- the second group is #2. */ |
|
1628 regnum_t this_group_regnum; |
|
1629 |
|
1630 compile_stack.avail--; |
|
1631 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; |
|
1632 fixup_alt_jump |
|
1633 = COMPILE_STACK_TOP.fixup_alt_jump |
|
1634 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 |
|
1635 : 0; |
|
1636 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; |
|
1637 this_group_regnum = COMPILE_STACK_TOP.regnum; |
|
1638 |
|
1639 /* We're at the end of the group, so now we know how many |
|
1640 groups were inside this one. */ |
|
1641 if (this_group_regnum <= MAX_REGNUM) |
|
1642 { |
|
1643 unsigned char *inner_group_loc |
|
1644 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; |
|
1645 |
|
1646 *inner_group_loc = regnum - this_group_regnum; |
|
1647 BUF_PUSH_3 (stop_memory, this_group_regnum, |
|
1648 regnum - this_group_regnum); |
|
1649 } |
|
1650 } |
|
1651 break; |
|
1652 |
|
1653 |
|
1654 case '|': /* `\|'. */ |
|
1655 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) |
|
1656 goto normal_backslash; |
|
1657 handle_alt: |
|
1658 if (syntax & RE_LIMITED_OPS) |
|
1659 goto normal_char; |
|
1660 |
|
1661 /* Insert before the previous alternative a jump which |
|
1662 jumps to this alternative if the former fails. */ |
|
1663 GET_BUFFER_SPACE (3); |
|
1664 INSERT_JUMP (on_failure_jump, begalt, b + 6); |
|
1665 pending_exact = 0; |
|
1666 b += 3; |
|
1667 |
|
1668 /* The alternative before this one has a jump after it |
|
1669 which gets executed if it gets matched. Adjust that |
|
1670 jump so it will jump to this alternative's analogous |
|
1671 jump (put in below, which in turn will jump to the next |
|
1672 (if any) alternative's such jump, etc.). The last such |
|
1673 jump jumps to the correct final destination. A picture: |
|
1674 _____ _____ |
|
1675 | | | | |
|
1676 | v | v |
|
1677 a | b | c |
|
1678 |
|
1679 If we are at `b,' then fixup_alt_jump right now points to a |
|
1680 three-byte space after `a.' We'll put in the jump, set |
|
1681 fixup_alt_jump to right after `b,' and leave behind three |
|
1682 bytes which we'll fill in when we get to after `c.' */ |
|
1683 |
|
1684 if (fixup_alt_jump) |
|
1685 STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
|
1686 |
|
1687 /* Mark and leave space for a jump after this alternative, |
|
1688 to be filled in later either by next alternative or |
|
1689 when know we're at the end of a series of alternatives. */ |
|
1690 fixup_alt_jump = b; |
|
1691 GET_BUFFER_SPACE (3); |
|
1692 b += 3; |
|
1693 |
|
1694 laststart = 0; |
|
1695 begalt = b; |
|
1696 break; |
|
1697 |
|
1698 |
|
1699 case '{': |
|
1700 /* If \{ is a literal. */ |
|
1701 if (!(syntax & RE_INTERVALS) |
|
1702 /* If we're at `\{' and it's not the open-interval |
|
1703 operator. */ |
|
1704 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
|
1705 || (p - 2 == pattern && p == pend)) |
|
1706 goto normal_backslash; |
|
1707 |
|
1708 handle_interval: |
|
1709 { |
|
1710 /* If got here, then the syntax allows intervals. */ |
|
1711 |
|
1712 /* At least (most) this many matches must be made. */ |
|
1713 int lower_bound = -1, upper_bound = -1; |
|
1714 |
|
1715 beg_interval = p - 1; |
|
1716 |
|
1717 if (p == pend) |
|
1718 { |
|
1719 if (syntax & RE_NO_BK_BRACES) |
|
1720 goto unfetch_interval; |
|
1721 else |
|
1722 return REG_EBRACE; |
|
1723 } |
|
1724 |
|
1725 GET_UNSIGNED_NUMBER (lower_bound); |
|
1726 |
|
1727 if (c == ',') |
|
1728 { |
|
1729 GET_UNSIGNED_NUMBER (upper_bound); |
|
1730 if (upper_bound < 0) upper_bound = RE_DUP_MAX; |
|
1731 } |
|
1732 else |
|
1733 /* Interval such as `{1}' => match exactly once. */ |
|
1734 upper_bound = lower_bound; |
|
1735 |
|
1736 if (lower_bound < 0 || upper_bound > RE_DUP_MAX |
|
1737 || lower_bound > upper_bound) |
|
1738 { |
|
1739 if (syntax & RE_NO_BK_BRACES) |
|
1740 goto unfetch_interval; |
|
1741 else |
|
1742 return REG_BADBR; |
|
1743 } |
|
1744 |
|
1745 if (!(syntax & RE_NO_BK_BRACES)) |
|
1746 { |
|
1747 if (c != '\\') return REG_EBRACE; |
|
1748 |
|
1749 PATFETCH (c); |
|
1750 } |
|
1751 |
|
1752 if (c != '}') |
|
1753 { |
|
1754 if (syntax & RE_NO_BK_BRACES) |
|
1755 goto unfetch_interval; |
|
1756 else |
|
1757 return REG_BADBR; |
|
1758 } |
|
1759 |
|
1760 /* We just parsed a valid interval. */ |
|
1761 |
|
1762 /* If it's invalid to have no preceding re. */ |
|
1763 if (!laststart) |
|
1764 { |
|
1765 if (syntax & RE_CONTEXT_INVALID_OPS) |
|
1766 return REG_BADRPT; |
|
1767 else if (syntax & RE_CONTEXT_INDEP_OPS) |
|
1768 laststart = b; |
|
1769 else |
|
1770 goto unfetch_interval; |
|
1771 } |
|
1772 |
|
1773 /* If the upper bound is zero, don't want to succeed at |
|
1774 all; jump from `laststart' to `b + 3', which will be |
|
1775 the end of the buffer after we insert the jump. */ |
|
1776 if (upper_bound == 0) |
|
1777 { |
|
1778 GET_BUFFER_SPACE (3); |
|
1779 INSERT_JUMP (jump, laststart, b + 3); |
|
1780 b += 3; |
|
1781 } |
|
1782 |
|
1783 /* Otherwise, we have a nontrivial interval. When |
|
1784 we're all done, the pattern will look like: |
|
1785 set_number_at <jump count> <upper bound> |
|
1786 set_number_at <succeed_n count> <lower bound> |
|
1787 succeed_n <after jump addr> <succed_n count> |
|
1788 <body of loop> |
|
1789 jump_n <succeed_n addr> <jump count> |
|
1790 (The upper bound and `jump_n' are omitted if |
|
1791 `upper_bound' is 1, though.) */ |
|
1792 else |
|
1793 { /* If the upper bound is > 1, we need to insert |
|
1794 more at the end of the loop. */ |
|
1795 unsigned nbytes = 10 + (upper_bound > 1) * 10; |
|
1796 |
|
1797 GET_BUFFER_SPACE (nbytes); |
|
1798 |
|
1799 /* Initialize lower bound of the `succeed_n', even |
|
1800 though it will be set during matching by its |
|
1801 attendant `set_number_at' (inserted next), |
|
1802 because `re_compile_fastmap' needs to know. |
|
1803 Jump to the `jump_n' we might insert below. */ |
|
1804 INSERT_JUMP2 (succeed_n, laststart, |
|
1805 b + 5 + (upper_bound > 1) * 5, |
|
1806 lower_bound); |
|
1807 b += 5; |
|
1808 |
|
1809 /* Code to initialize the lower bound. Insert |
|
1810 before the `succeed_n'. The `5' is the last two |
|
1811 bytes of this `set_number_at', plus 3 bytes of |
|
1812 the following `succeed_n'. */ |
|
1813 insert_op2 (set_number_at, laststart, 5, lower_bound, b); |
|
1814 b += 5; |
|
1815 |
|
1816 if (upper_bound > 1) |
|
1817 { /* More than one repetition is allowed, so |
|
1818 append a backward jump to the `succeed_n' |
|
1819 that starts this interval. |
|
1820 |
|
1821 When we've reached this during matching, |
|
1822 we'll have matched the interval once, so |
|
1823 jump back only `upper_bound - 1' times. */ |
|
1824 STORE_JUMP2 (jump_n, b, laststart + 5, |
|
1825 upper_bound - 1); |
|
1826 b += 5; |
|
1827 |
|
1828 /* The location we want to set is the second |
|
1829 parameter of the `jump_n'; that is `b-2' as |
|
1830 an absolute address. `laststart' will be |
|
1831 the `set_number_at' we're about to insert; |
|
1832 `laststart+3' the number to set, the source |
|
1833 for the relative address. But we are |
|
1834 inserting into the middle of the pattern -- |
|
1835 so everything is getting moved up by 5. |
|
1836 Conclusion: (b - 2) - (laststart + 3) + 5, |
|
1837 i.e., b - laststart. |
|
1838 |
|
1839 We insert this at the beginning of the loop |
|
1840 so that if we fail during matching, we'll |
|
1841 reinitialize the bounds. */ |
|
1842 insert_op2 (set_number_at, laststart, b - laststart, |
|
1843 upper_bound - 1, b); |
|
1844 b += 5; |
|
1845 } |
|
1846 } |
|
1847 pending_exact = 0; |
|
1848 beg_interval = NULL; |
|
1849 } |
|
1850 break; |
|
1851 |
|
1852 unfetch_interval: |
|
1853 /* If an invalid interval, match the characters as literals. */ |
|
1854 assert (beg_interval); |
|
1855 p = beg_interval; |
|
1856 beg_interval = NULL; |
|
1857 |
|
1858 /* normal_char and normal_backslash need `c'. */ |
|
1859 PATFETCH (c); |
|
1860 |
|
1861 if (!(syntax & RE_NO_BK_BRACES)) |
|
1862 { |
|
1863 if (p > pattern && p[-1] == '\\') |
|
1864 goto normal_backslash; |
|
1865 } |
|
1866 goto normal_char; |
|
1867 |
|
1868 #ifdef emacs |
|
1869 /* There is no way to specify the before_dot and after_dot |
|
1870 operators. rms says this is ok. --karl */ |
|
1871 case '=': |
|
1872 BUF_PUSH (at_dot); |
|
1873 break; |
|
1874 |
|
1875 case 's': |
|
1876 laststart = b; |
|
1877 PATFETCH (c); |
|
1878 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); |
|
1879 break; |
|
1880 |
|
1881 case 'S': |
|
1882 laststart = b; |
|
1883 PATFETCH (c); |
|
1884 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); |
|
1885 break; |
|
1886 #endif /* emacs */ |
|
1887 |
|
1888 |
|
1889 case 'w': |
|
1890 laststart = b; |
|
1891 BUF_PUSH (wordchar); |
|
1892 break; |
|
1893 |
|
1894 |
|
1895 case 'W': |
|
1896 laststart = b; |
|
1897 BUF_PUSH (notwordchar); |
|
1898 break; |
|
1899 |
|
1900 |
|
1901 case '<': |
|
1902 BUF_PUSH (wordbeg); |
|
1903 break; |
|
1904 |
|
1905 case '>': |
|
1906 BUF_PUSH (wordend); |
|
1907 break; |
|
1908 |
|
1909 case 'b': |
|
1910 BUF_PUSH (wordbound); |
|
1911 break; |
|
1912 |
|
1913 case 'B': |
|
1914 BUF_PUSH (notwordbound); |
|
1915 break; |
|
1916 |
|
1917 case '`': |
|
1918 BUF_PUSH (begbuf); |
|
1919 break; |
|
1920 |
|
1921 case '\'': |
|
1922 BUF_PUSH (endbuf); |
|
1923 break; |
|
1924 |
|
1925 case '1': case '2': case '3': case '4': case '5': |
|
1926 case '6': case '7': case '8': case '9': |
|
1927 if (syntax & RE_NO_BK_REFS) |
|
1928 goto normal_char; |
|
1929 |
|
1930 c1 = c - '0'; |
|
1931 |
|
1932 if (c1 > regnum) |
|
1933 return REG_ESUBREG; |
|
1934 |
|
1935 /* Can't back reference to a subexpression if inside of it. */ |
|
1936 if (group_in_compile_stack (compile_stack, c1)) |
|
1937 goto normal_char; |
|
1938 |
|
1939 laststart = b; |
|
1940 BUF_PUSH_2 (duplicate, c1); |
|
1941 break; |
|
1942 |
|
1943 |
|
1944 case '+': |
|
1945 case '?': |
|
1946 if (syntax & RE_BK_PLUS_QM) |
|
1947 goto handle_plus; |
|
1948 else |
|
1949 goto normal_backslash; |
|
1950 |
|
1951 default: |
|
1952 normal_backslash: |
|
1953 /* You might think it would be useful for \ to mean |
|
1954 not to translate; but if we don't translate it |
|
1955 it will never match anything. */ |
|
1956 c = TRANSLATE (c); |
|
1957 goto normal_char; |
|
1958 } |
|
1959 break; |
|
1960 |
|
1961 |
|
1962 default: |
|
1963 /* Expects the character in `c'. */ |
|
1964 normal_char: |
|
1965 /* If no exactn currently being built. */ |
|
1966 if (!pending_exact |
|
1967 |
|
1968 /* If last exactn not at current position. */ |
|
1969 || pending_exact + *pending_exact + 1 != b |
|
1970 |
|
1971 /* We have only one byte following the exactn for the count. */ |
|
1972 || *pending_exact == (1 << BYTEWIDTH) - 1 |
|
1973 |
|
1974 /* If followed by a repetition operator. */ |
|
1975 || *p == '*' || *p == '^' |
|
1976 || ((syntax & RE_BK_PLUS_QM) |
|
1977 ? *p == '\\' && (p[1] == '+' || p[1] == '?') |
|
1978 : (*p == '+' || *p == '?')) |
|
1979 || ((syntax & RE_INTERVALS) |
|
1980 && ((syntax & RE_NO_BK_BRACES) |
|
1981 ? *p == '{' |
|
1982 : (p[0] == '\\' && p[1] == '{')))) |
|
1983 { |
|
1984 /* Start building a new exactn. */ |
|
1985 |
|
1986 laststart = b; |
|
1987 |
|
1988 BUF_PUSH_2 (exactn, 0); |
|
1989 pending_exact = b - 1; |
|
1990 } |
|
1991 |
|
1992 BUF_PUSH (c); |
|
1993 (*pending_exact)++; |
|
1994 break; |
|
1995 } /* switch (c) */ |
|
1996 } /* while p != pend */ |
|
1997 |
|
1998 |
|
1999 /* Through the pattern now. */ |
|
2000 |
|
2001 if (fixup_alt_jump) |
|
2002 STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
|
2003 |
|
2004 if (!COMPILE_STACK_EMPTY) |
|
2005 return REG_EPAREN; |
|
2006 |
|
2007 free (compile_stack.stack); |
|
2008 |
|
2009 /* We have succeeded; set the length of the buffer. */ |
|
2010 bufp->used = b - bufp->buffer; |
|
2011 |
|
2012 #ifdef DEBUG |
|
2013 if (debug) |
|
2014 { |
|
2015 DEBUG_PRINT1 ("\nCompiled pattern: "); |
|
2016 print_compiled_pattern (bufp); |
|
2017 } |
|
2018 #endif /* DEBUG */ |
|
2019 |
|
2020 return REG_NOERROR; |
|
2021 } /* regex_compile */ |
|
2022 |
|
2023 /* Subroutines for `regex_compile'. */ |
|
2024 |
|
2025 /* Store OP at LOC followed by two-byte integer parameter ARG. */ |
|
2026 |
|
2027 static void |
|
2028 store_op1 (op, loc, arg) |
|
2029 re_opcode_t op; |
|
2030 unsigned char *loc; |
|
2031 int arg; |
|
2032 { |
|
2033 *loc = (unsigned char) op; |
|
2034 STORE_NUMBER (loc + 1, arg); |
|
2035 } |
|
2036 |
|
2037 |
|
2038 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ |
|
2039 |
|
2040 static void |
|
2041 store_op2 (op, loc, arg1, arg2) |
|
2042 re_opcode_t op; |
|
2043 unsigned char *loc; |
|
2044 int arg1, arg2; |
|
2045 { |
|
2046 *loc = (unsigned char) op; |
|
2047 STORE_NUMBER (loc + 1, arg1); |
|
2048 STORE_NUMBER (loc + 3, arg2); |
|
2049 } |
|
2050 |
|
2051 |
|
2052 /* Copy the bytes from LOC to END to open up three bytes of space at LOC |
|
2053 for OP followed by two-byte integer parameter ARG. */ |
|
2054 |
|
2055 static void |
|
2056 insert_op1 (op, loc, arg, end) |
|
2057 re_opcode_t op; |
|
2058 unsigned char *loc; |
|
2059 int arg; |
|
2060 unsigned char *end; |
|
2061 { |
|
2062 register unsigned char *pfrom = end; |
|
2063 register unsigned char *pto = end + 3; |
|
2064 |
|
2065 while (pfrom != loc) |
|
2066 *--pto = *--pfrom; |
|
2067 |
|
2068 store_op1 (op, loc, arg); |
|
2069 } |
|
2070 |
|
2071 |
|
2072 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ |
|
2073 |
|
2074 static void |
|
2075 insert_op2 (op, loc, arg1, arg2, end) |
|
2076 re_opcode_t op; |
|
2077 unsigned char *loc; |
|
2078 int arg1, arg2; |
|
2079 unsigned char *end; |
|
2080 { |
|
2081 register unsigned char *pfrom = end; |
|
2082 register unsigned char *pto = end + 5; |
|
2083 |
|
2084 while (pfrom != loc) |
|
2085 *--pto = *--pfrom; |
|
2086 |
|
2087 store_op2 (op, loc, arg1, arg2); |
|
2088 } |
|
2089 |
|
2090 |
|
2091 /* P points to just after a ^ in PATTERN. Return true if that ^ comes |
|
2092 after an alternative or a begin-subexpression. We assume there is at |
|
2093 least one character before the ^. */ |
|
2094 |
|
2095 static boolean |
|
2096 at_begline_loc_p (pattern, p, syntax) |
|
2097 const char *pattern, *p; |
|
2098 reg_syntax_t syntax; |
|
2099 { |
|
2100 const char *prev = p - 2; |
|
2101 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
|
2102 |
|
2103 return |
|
2104 /* After a subexpression? */ |
|
2105 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) |
|
2106 /* After an alternative? */ |
|
2107 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); |
|
2108 } |
|
2109 |
|
2110 |
|
2111 /* The dual of at_begline_loc_p. This one is for $. We assume there is |
|
2112 at least one character after the $, i.e., `P < PEND'. */ |
|
2113 |
|
2114 static boolean |
|
2115 at_endline_loc_p (p, pend, syntax) |
|
2116 const char *p, *pend; |
|
2117 int syntax; |
|
2118 { |
|
2119 const char *next = p; |
|
2120 boolean next_backslash = *next == '\\'; |
|
2121 const char *next_next = p + 1 < pend ? p + 1 : NULL; |
|
2122 |
|
2123 return |
|
2124 /* Before a subexpression? */ |
|
2125 (syntax & RE_NO_BK_PARENS ? *next == ')' |
|
2126 : next_backslash && next_next && *next_next == ')') |
|
2127 /* Before an alternative? */ |
|
2128 || (syntax & RE_NO_BK_VBAR ? *next == '|' |
|
2129 : next_backslash && next_next && *next_next == '|'); |
|
2130 } |
|
2131 |
|
2132 |
|
2133 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and |
|
2134 false if it's not. */ |
|
2135 |
|
2136 static boolean |
|
2137 group_in_compile_stack (compile_stack, regnum) |
|
2138 compile_stack_type compile_stack; |
|
2139 regnum_t regnum; |
|
2140 { |
|
2141 int this_element; |
|
2142 |
|
2143 for (this_element = compile_stack.avail - 1; |
|
2144 this_element >= 0; |
|
2145 this_element--) |
|
2146 if (compile_stack.stack[this_element].regnum == regnum) |
|
2147 return true; |
|
2148 |
|
2149 return false; |
|
2150 } |
|
2151 |
|
2152 |
|
2153 /* Read the ending character of a range (in a bracket expression) from the |
|
2154 uncompiled pattern *P_PTR (which ends at PEND). We assume the |
|
2155 starting character is in `P[-2]'. (`P[-1]' is the character `-'.) |
|
2156 Then we set the translation of all bits between the starting and |
|
2157 ending characters (inclusive) in the compiled pattern B. |
|
2158 |
|
2159 Return an error code. |
|
2160 |
|
2161 We use these short variable names so we can use the same macros as |
|
2162 `regex_compile' itself. */ |
|
2163 |
|
2164 static reg_errcode_t |
|
2165 compile_range (p_ptr, pend, translate, syntax, b) |
|
2166 const char **p_ptr, *pend; |
|
2167 char *translate; |
|
2168 reg_syntax_t syntax; |
|
2169 unsigned char *b; |
|
2170 { |
|
2171 unsigned this_char; |
|
2172 |
|
2173 const char *p = *p_ptr; |
|
2174 |
|
2175 /* Even though the pattern is a signed `char *', we need to fetch into |
|
2176 `unsigned char's. Reason: if the high bit of the pattern character |
|
2177 is set, the range endpoints will be negative if we fetch into a |
|
2178 signed `char *'. */ |
|
2179 unsigned char range_end; |
|
2180 unsigned char range_start = p[-2]; |
|
2181 |
|
2182 if (p == pend) |
|
2183 return REG_ERANGE; |
|
2184 |
|
2185 PATFETCH (range_end); |
|
2186 |
|
2187 /* Have to increment the pointer into the pattern string, so the |
|
2188 caller isn't still at the ending character. */ |
|
2189 (*p_ptr)++; |
|
2190 |
|
2191 /* If the start is after the end, the range is empty. */ |
|
2192 if (range_start > range_end) |
|
2193 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; |
|
2194 |
|
2195 /* Here we see why `this_char' has to be larger than an `unsigned |
|
2196 char' -- the range is inclusive, so if `range_end' == 0xff |
|
2197 (assuming 8-bit characters), we would otherwise go into an infinite |
|
2198 loop, since all characters <= 0xff. */ |
|
2199 for (this_char = range_start; this_char <= range_end; this_char++) |
|
2200 { |
|
2201 SET_LIST_BIT (TRANSLATE (this_char)); |
|
2202 } |
|
2203 |
|
2204 return REG_NOERROR; |
|
2205 } |
|
2206 |
|
2207 /* Failure stack declarations and macros; both re_compile_fastmap and |
|
2208 re_match_2 use a failure stack. These have to be macros because of |
|
2209 REGEX_ALLOCATE. */ |
|
2210 |
|
2211 |
|
2212 /* Number of failure points for which to initially allocate space |
|
2213 when matching. If this number is exceeded, we allocate more |
|
2214 space, so it is not a hard limit. */ |
|
2215 #ifndef INIT_FAILURE_ALLOC |
|
2216 #define INIT_FAILURE_ALLOC 5 |
|
2217 #endif |
|
2218 |
|
2219 /* Roughly the maximum number of failure points on the stack. Would be |
|
2220 exactly that if always used MAX_FAILURE_SPACE each time we failed. |
|
2221 This is a variable only so users of regex can assign to it; we never |
|
2222 change it ourselves. */ |
|
2223 int re_max_failures = 2000; |
|
2224 |
|
2225 typedef const unsigned char *fail_stack_elt_t; |
|
2226 |
|
2227 typedef struct |
|
2228 { |
|
2229 fail_stack_elt_t *stack; |
|
2230 unsigned size; |
|
2231 unsigned avail; /* Offset of next open position. */ |
|
2232 } fail_stack_type; |
|
2233 |
|
2234 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0) |
|
2235 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) |
|
2236 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) |
|
2237 #define FAIL_STACK_TOP() (fail_stack.stack[fail_stack.avail]) |
|
2238 |
|
2239 |
|
2240 /* Initialize `fail_stack'. Do `return -2' if the alloc fails. */ |
|
2241 |
|
2242 #define INIT_FAIL_STACK() \ |
|
2243 do { \ |
|
2244 fail_stack.stack = (fail_stack_elt_t *) \ |
|
2245 REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \ |
|
2246 \ |
|
2247 if (fail_stack.stack == NULL) \ |
|
2248 return -2; \ |
|
2249 \ |
|
2250 fail_stack.size = INIT_FAILURE_ALLOC; \ |
|
2251 fail_stack.avail = 0; \ |
|
2252 } while (0) |
|
2253 |
|
2254 |
|
2255 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. |
|
2256 |
|
2257 Return 1 if succeeds, and 0 if either ran out of memory |
|
2258 allocating space for it or it was already too large. |
|
2259 |
|
2260 REGEX_REALLOCATE requires `destination' be declared. */ |
|
2261 |
|
2262 #define DOUBLE_FAIL_STACK(fail_stack) \ |
|
2263 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ |
|
2264 ? 0 \ |
|
2265 : ((fail_stack).stack = (fail_stack_elt_t *) \ |
|
2266 REGEX_REALLOCATE ((fail_stack).stack, \ |
|
2267 (fail_stack).size * sizeof (fail_stack_elt_t), \ |
|
2268 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ |
|
2269 \ |
|
2270 (fail_stack).stack == NULL \ |
|
2271 ? 0 \ |
|
2272 : ((fail_stack).size <<= 1, \ |
|
2273 1))) |
|
2274 |
|
2275 |
|
2276 /* Push PATTERN_OP on FAIL_STACK. |
|
2277 |
|
2278 Return 1 if was able to do so and 0 if ran out of memory allocating |
|
2279 space to do so. */ |
|
2280 #define PUSH_PATTERN_OP(pattern_op, fail_stack) \ |
|
2281 ((FAIL_STACK_FULL () \ |
|
2282 && !DOUBLE_FAIL_STACK (fail_stack)) \ |
|
2283 ? 0 \ |
|
2284 : ((fail_stack).stack[(fail_stack).avail++] = pattern_op, \ |
|
2285 1)) |
|
2286 |
|
2287 /* This pushes an item onto the failure stack. Must be a four-byte |
|
2288 value. Assumes the variable `fail_stack'. Probably should only |
|
2289 be called from within `PUSH_FAILURE_POINT'. */ |
|
2290 #define PUSH_FAILURE_ITEM(item) \ |
|
2291 fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item |
|
2292 |
|
2293 /* The complement operation. Assumes `fail_stack' is nonempty. */ |
|
2294 #define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail] |
|
2295 |
|
2296 /* Used to omit pushing failure point id's when we're not debugging. */ |
|
2297 #ifdef DEBUG |
|
2298 #define DEBUG_PUSH PUSH_FAILURE_ITEM |
|
2299 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM () |
|
2300 #else |
|
2301 #define DEBUG_PUSH(item) |
|
2302 #define DEBUG_POP(item_addr) |
|
2303 #endif |
|
2304 |
|
2305 |
|
2306 /* Push the information about the state we will need |
|
2307 if we ever fail back to it. |
|
2308 |
|
2309 Requires variables fail_stack, regstart, regend, reg_info, and |
|
2310 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be |
|
2311 declared. |
|
2312 |
|
2313 Does `return FAILURE_CODE' if runs out of memory. */ |
|
2314 |
|
2315 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ |
|
2316 do { \ |
|
2317 char *destination; \ |
|
2318 /* Must be int, so when we don't save any registers, the arithmetic \ |
|
2319 of 0 + -1 isn't done as unsigned. */ \ |
|
2320 int this_reg; \ |
|
2321 \ |
|
2322 DEBUG_STATEMENT (failure_id++); \ |
|
2323 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ |
|
2324 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ |
|
2325 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ |
|
2326 \ |
|
2327 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ |
|
2328 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ |
|
2329 \ |
|
2330 /* Ensure we have enough space allocated for what we will push. */ \ |
|
2331 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ |
|
2332 { \ |
|
2333 if (!DOUBLE_FAIL_STACK (fail_stack)) \ |
|
2334 return failure_code; \ |
|
2335 \ |
|
2336 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ |
|
2337 (fail_stack).size); \ |
|
2338 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ |
|
2339 } \ |
|
2340 \ |
|
2341 /* Push the info, starting with the registers. */ \ |
|
2342 DEBUG_PRINT1 ("\n"); \ |
|
2343 \ |
|
2344 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ |
|
2345 this_reg++) \ |
|
2346 { \ |
|
2347 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \ |
|
2348 DEBUG_STATEMENT (num_regs_pushed++); \ |
|
2349 \ |
|
2350 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ |
|
2351 PUSH_FAILURE_ITEM (regstart[this_reg]); \ |
|
2352 \ |
|
2353 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ |
|
2354 PUSH_FAILURE_ITEM (regend[this_reg]); \ |
|
2355 \ |
|
2356 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \ |
|
2357 DEBUG_PRINT2 (" match_null=%d", \ |
|
2358 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ |
|
2359 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ |
|
2360 DEBUG_PRINT2 (" matched_something=%d", \ |
|
2361 MATCHED_SOMETHING (reg_info[this_reg])); \ |
|
2362 DEBUG_PRINT2 (" ever_matched=%d", \ |
|
2363 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ |
|
2364 DEBUG_PRINT1 ("\n"); \ |
|
2365 PUSH_FAILURE_ITEM (reg_info[this_reg].word); \ |
|
2366 } \ |
|
2367 \ |
|
2368 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\ |
|
2369 PUSH_FAILURE_ITEM (lowest_active_reg); \ |
|
2370 \ |
|
2371 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\ |
|
2372 PUSH_FAILURE_ITEM (highest_active_reg); \ |
|
2373 \ |
|
2374 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \ |
|
2375 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ |
|
2376 PUSH_FAILURE_ITEM (pattern_place); \ |
|
2377 \ |
|
2378 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \ |
|
2379 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ |
|
2380 size2); \ |
|
2381 DEBUG_PRINT1 ("'\n"); \ |
|
2382 PUSH_FAILURE_ITEM (string_place); \ |
|
2383 \ |
|
2384 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ |
|
2385 DEBUG_PUSH (failure_id); \ |
|
2386 } while (0) |
|
2387 |
|
2388 /* This is the number of items that are pushed and popped on the stack |
|
2389 for each register. */ |
|
2390 #define NUM_REG_ITEMS 3 |
|
2391 |
|
2392 /* Individual items aside from the registers. */ |
|
2393 #ifdef DEBUG |
|
2394 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ |
|
2395 #else |
|
2396 #define NUM_NONREG_ITEMS 4 |
|
2397 #endif |
|
2398 |
|
2399 /* We push at most this many items on the stack. */ |
|
2400 #define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS) |
|
2401 |
|
2402 /* We actually push this many items. */ |
|
2403 #define NUM_FAILURE_ITEMS \ |
|
2404 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \ |
|
2405 + NUM_NONREG_ITEMS) |
|
2406 |
|
2407 /* How many items can still be added to the stack without overflowing it. */ |
|
2408 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) |
|
2409 |
|
2410 |
|
2411 /* Pops what PUSH_FAIL_STACK pushes. |
|
2412 |
|
2413 We restore into the parameters, all of which should be lvalues: |
|
2414 STR -- the saved data position. |
|
2415 PAT -- the saved pattern position. |
|
2416 LOW_REG, HIGH_REG -- the highest and lowest active registers. |
|
2417 REGSTART, REGEND -- arrays of string positions. |
|
2418 REG_INFO -- array of information about each subexpression. |
|
2419 |
|
2420 Also assumes the variables `fail_stack' and (if debugging), `bufp', |
|
2421 `pend', `string1', `size1', `string2', and `size2'. */ |
|
2422 |
|
2423 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ |
|
2424 { \ |
|
2425 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \ |
|
2426 int this_reg; \ |
|
2427 const unsigned char *string_temp; \ |
|
2428 \ |
|
2429 assert (!FAIL_STACK_EMPTY ()); \ |
|
2430 \ |
|
2431 /* Remove failure points and point to how many regs pushed. */ \ |
|
2432 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ |
|
2433 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ |
|
2434 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ |
|
2435 \ |
|
2436 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ |
|
2437 \ |
|
2438 DEBUG_POP (&failure_id); \ |
|
2439 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ |
|
2440 \ |
|
2441 /* If the saved string location is NULL, it came from an \ |
|
2442 on_failure_keep_string_jump opcode, and we want to throw away the \ |
|
2443 saved NULL, thus retaining our current position in the string. */ \ |
|
2444 string_temp = POP_FAILURE_ITEM (); \ |
|
2445 if (string_temp != NULL) \ |
|
2446 str = (const char *) string_temp; \ |
|
2447 \ |
|
2448 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \ |
|
2449 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ |
|
2450 DEBUG_PRINT1 ("'\n"); \ |
|
2451 \ |
|
2452 pat = (unsigned char *) POP_FAILURE_ITEM (); \ |
|
2453 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \ |
|
2454 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ |
|
2455 \ |
|
2456 /* Restore register info. */ \ |
|
2457 high_reg = (unsigned) POP_FAILURE_ITEM (); \ |
|
2458 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \ |
|
2459 \ |
|
2460 low_reg = (unsigned) POP_FAILURE_ITEM (); \ |
|
2461 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ |
|
2462 \ |
|
2463 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ |
|
2464 { \ |
|
2465 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ |
|
2466 \ |
|
2467 reg_info[this_reg].word = POP_FAILURE_ITEM (); \ |
|
2468 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \ |
|
2469 \ |
|
2470 regend[this_reg] = (const char *) POP_FAILURE_ITEM (); \ |
|
2471 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ |
|
2472 \ |
|
2473 regstart[this_reg] = (const char *) POP_FAILURE_ITEM (); \ |
|
2474 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ |
|
2475 } \ |
|
2476 } /* POP_FAILURE_POINT */ |
|
2477 |
|
2478 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in |
|
2479 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible |
|
2480 characters can start a string that matches the pattern. This fastmap |
|
2481 is used by re_search to skip quickly over impossible starting points. |
|
2482 |
|
2483 The caller must supply the address of a (1 << BYTEWIDTH)-byte data |
|
2484 area as BUFP->fastmap. |
|
2485 |
|
2486 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in |
|
2487 the pattern buffer. |
|
2488 |
|
2489 Returns 0 if we succeed, -2 if an internal error. */ |
|
2490 |
|
2491 int |
|
2492 re_compile_fastmap (bufp) |
|
2493 struct re_pattern_buffer *bufp; |
|
2494 { |
|
2495 int j, k; |
|
2496 fail_stack_type fail_stack; |
|
2497 #ifndef REGEX_MALLOC |
|
2498 char *destination; |
|
2499 #endif |
|
2500 /* We don't push any register information onto the failure stack. */ |
|
2501 unsigned num_regs = 0; |
|
2502 |
|
2503 register char *fastmap = bufp->fastmap; |
|
2504 unsigned char *pattern = bufp->buffer; |
|
2505 unsigned long size = bufp->used; |
|
2506 const unsigned char *p = pattern; |
|
2507 register unsigned char *pend = pattern + size; |
|
2508 |
|
2509 /* Assume that each path through the pattern can be null until |
|
2510 proven otherwise. We set this false at the bottom of switch |
|
2511 statement, to which we get only if a particular path doesn't |
|
2512 match the empty string. */ |
|
2513 boolean path_can_be_null = true; |
|
2514 |
|
2515 /* We aren't doing a `succeed_n' to begin with. */ |
|
2516 boolean succeed_n_p = false; |
|
2517 |
|
2518 assert (fastmap != NULL && p != NULL); |
|
2519 |
|
2520 INIT_FAIL_STACK (); |
|
2521 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ |
|
2522 bufp->fastmap_accurate = 1; /* It will be when we're done. */ |
|
2523 bufp->can_be_null = 0; |
|
2524 |
|
2525 while (p != pend || !FAIL_STACK_EMPTY ()) |
|
2526 { |
|
2527 if (p == pend) |
|
2528 { |
|
2529 bufp->can_be_null |= path_can_be_null; |
|
2530 |
|
2531 /* Reset for next path. */ |
|
2532 path_can_be_null = true; |
|
2533 |
|
2534 p = fail_stack.stack[--fail_stack.avail]; |
|
2535 } |
|
2536 |
|
2537 /* We should never be about to go beyond the end of the pattern. */ |
|
2538 assert (p < pend); |
|
2539 |
|
2540 #ifdef SWITCH_ENUM_BUG |
|
2541 switch ((int) ((re_opcode_t) *p++)) |
|
2542 #else |
|
2543 switch ((re_opcode_t) *p++) |
|
2544 #endif |
|
2545 { |
|
2546 |
|
2547 /* I guess the idea here is to simply not bother with a fastmap |
|
2548 if a backreference is used, since it's too hard to figure out |
|
2549 the fastmap for the corresponding group. Setting |
|
2550 `can_be_null' stops `re_search_2' from using the fastmap, so |
|
2551 that is all we do. */ |
|
2552 case duplicate: |
|
2553 bufp->can_be_null = 1; |
|
2554 return 0; |
|
2555 |
|
2556 |
|
2557 /* Following are the cases which match a character. These end |
|
2558 with `break'. */ |
|
2559 |
|
2560 case exactn: |
|
2561 fastmap[p[1]] = 1; |
|
2562 break; |
|
2563 |
|
2564 |
|
2565 case charset: |
|
2566 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
|
2567 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) |
|
2568 fastmap[j] = 1; |
|
2569 break; |
|
2570 |
|
2571 |
|
2572 case charset_not: |
|
2573 /* Chars beyond end of map must be allowed. */ |
|
2574 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) |
|
2575 fastmap[j] = 1; |
|
2576 |
|
2577 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
|
2578 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) |
|
2579 fastmap[j] = 1; |
|
2580 break; |
|
2581 |
|
2582 |
|
2583 case wordchar: |
|
2584 for (j = 0; j < (1 << BYTEWIDTH); j++) |
|
2585 if (SYNTAX (j) == Sword) |
|
2586 fastmap[j] = 1; |
|
2587 break; |
|
2588 |
|
2589 |
|
2590 case notwordchar: |
|
2591 for (j = 0; j < (1 << BYTEWIDTH); j++) |
|
2592 if (SYNTAX (j) != Sword) |
|
2593 fastmap[j] = 1; |
|
2594 break; |
|
2595 |
|
2596 |
|
2597 case anychar: |
|
2598 /* `.' matches anything ... */ |
|
2599 for (j = 0; j < (1 << BYTEWIDTH); j++) |
|
2600 fastmap[j] = 1; |
|
2601 |
|
2602 /* ... except perhaps newline. */ |
|
2603 if (!(bufp->syntax & RE_DOT_NEWLINE)) |
|
2604 fastmap['\n'] = 0; |
|
2605 |
|
2606 /* Return if we have already set `can_be_null'; if we have, |
|
2607 then the fastmap is irrelevant. Something's wrong here. */ |
|
2608 else if (bufp->can_be_null) |
|
2609 return 0; |
|
2610 |
|
2611 /* Otherwise, have to check alternative paths. */ |
|
2612 break; |
|
2613 |
|
2614 |
|
2615 #ifdef emacs |
|
2616 case syntaxspec: |
|
2617 k = *p++; |
|
2618 for (j = 0; j < (1 << BYTEWIDTH); j++) |
|
2619 if (SYNTAX (j) == (enum syntaxcode) k) |
|
2620 fastmap[j] = 1; |
|
2621 break; |
|
2622 |
|
2623 |
|
2624 case notsyntaxspec: |
|
2625 k = *p++; |
|
2626 for (j = 0; j < (1 << BYTEWIDTH); j++) |
|
2627 if (SYNTAX (j) != (enum syntaxcode) k) |
|
2628 fastmap[j] = 1; |
|
2629 break; |
|
2630 |
|
2631 |
|
2632 /* All cases after this match the empty string. These end with |
|
2633 `continue'. */ |
|
2634 |
|
2635 |
|
2636 case before_dot: |
|
2637 case at_dot: |
|
2638 case after_dot: |
|
2639 continue; |
|
2640 #endif /* not emacs */ |
|
2641 |
|
2642 |
|
2643 case no_op: |
|
2644 case begline: |
|
2645 case endline: |
|
2646 case begbuf: |
|
2647 case endbuf: |
|
2648 case wordbound: |
|
2649 case notwordbound: |
|
2650 case wordbeg: |
|
2651 case wordend: |
|
2652 case push_dummy_failure: |
|
2653 continue; |
|
2654 |
|
2655 |
|
2656 case jump_n: |
|
2657 case pop_failure_jump: |
|
2658 case maybe_pop_jump: |
|
2659 case jump: |
|
2660 case jump_past_alt: |
|
2661 case dummy_failure_jump: |
|
2662 EXTRACT_NUMBER_AND_INCR (j, p); |
|
2663 p += j; |
|
2664 if (j > 0) |
|
2665 continue; |
|
2666 |
|
2667 /* Jump backward implies we just went through the body of a |
|
2668 loop and matched nothing. Opcode jumped to should be |
|
2669 `on_failure_jump' or `succeed_n'. Just treat it like an |
|
2670 ordinary jump. For a * loop, it has pushed its failure |
|
2671 point already; if so, discard that as redundant. */ |
|
2672 if ((re_opcode_t) *p != on_failure_jump |
|
2673 && (re_opcode_t) *p != succeed_n) |
|
2674 continue; |
|
2675 |
|
2676 p++; |
|
2677 EXTRACT_NUMBER_AND_INCR (j, p); |
|
2678 p += j; |
|
2679 |
|
2680 /* If what's on the stack is where we are now, pop it. */ |
|
2681 if (!FAIL_STACK_EMPTY () |
|
2682 && fail_stack.stack[fail_stack.avail - 1] == p) |
|
2683 fail_stack.avail--; |
|
2684 |
|
2685 continue; |
|
2686 |
|
2687 |
|
2688 case on_failure_jump: |
|
2689 case on_failure_keep_string_jump: |
|
2690 handle_on_failure_jump: |
|
2691 EXTRACT_NUMBER_AND_INCR (j, p); |
|
2692 |
|
2693 /* For some patterns, e.g., `(a?)?', `p+j' here points to the |
|
2694 end of the pattern. We don't want to push such a point, |
|
2695 since when we restore it above, entering the switch will |
|
2696 increment `p' past the end of the pattern. We don't need |
|
2697 to push such a point since we obviously won't find any more |
|
2698 fastmap entries beyond `pend'. Such a pattern can match |
|
2699 the null string, though. */ |
|
2700 if (p + j < pend) |
|
2701 { |
|
2702 if (!PUSH_PATTERN_OP (p + j, fail_stack)) |
|
2703 return -2; |
|
2704 } |
|
2705 else |
|
2706 bufp->can_be_null = 1; |
|
2707 |
|
2708 if (succeed_n_p) |
|
2709 { |
|
2710 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ |
|
2711 succeed_n_p = false; |
|
2712 } |
|
2713 |
|
2714 continue; |
|
2715 |
|
2716 |
|
2717 case succeed_n: |
|
2718 /* Get to the number of times to succeed. */ |
|
2719 p += 2; |
|
2720 |
|
2721 /* Increment p past the n for when k != 0. */ |
|
2722 EXTRACT_NUMBER_AND_INCR (k, p); |
|
2723 if (k == 0) |
|
2724 { |
|
2725 p -= 4; |
|
2726 succeed_n_p = true; /* Spaghetti code alert. */ |
|
2727 goto handle_on_failure_jump; |
|
2728 } |
|
2729 continue; |
|
2730 |
|
2731 |
|
2732 case set_number_at: |
|
2733 p += 4; |
|
2734 continue; |
|
2735 |
|
2736 |
|
2737 case start_memory: |
|
2738 case stop_memory: |
|
2739 p += 2; |
|
2740 continue; |
|
2741 |
|
2742 |
|
2743 default: |
|
2744 abort (); /* We have listed all the cases. */ |
|
2745 } /* switch *p++ */ |
|
2746 |
|
2747 /* Getting here means we have found the possible starting |
|
2748 characters for one path of the pattern -- and that the empty |
|
2749 string does not match. We need not follow this path further. |
|
2750 Instead, look at the next alternative (remembered on the |
|
2751 stack), or quit if no more. The test at the top of the loop |
|
2752 does these things. */ |
|
2753 path_can_be_null = false; |
|
2754 p = pend; |
|
2755 } /* while p */ |
|
2756 |
|
2757 /* Set `can_be_null' for the last path (also the first path, if the |
|
2758 pattern is empty). */ |
|
2759 bufp->can_be_null |= path_can_be_null; |
|
2760 return 0; |
|
2761 } /* re_compile_fastmap */ |
|
2762 |
|
2763 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and |
|
2764 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use |
|
2765 this memory for recording register information. STARTS and ENDS |
|
2766 must be allocated using the malloc library routine, and must each |
|
2767 be at least NUM_REGS * sizeof (regoff_t) bytes long. |
|
2768 |
|
2769 If NUM_REGS == 0, then subsequent matches should allocate their own |
|
2770 register data. |
|
2771 |
|
2772 Unless this function is called, the first search or match using |
|
2773 PATTERN_BUFFER will allocate its own register data, without |
|
2774 freeing the old data. */ |
|
2775 |
|
2776 void |
|
2777 re_set_registers (bufp, regs, num_regs, starts, ends) |
|
2778 struct re_pattern_buffer *bufp; |
|
2779 struct re_registers *regs; |
|
2780 unsigned num_regs; |
|
2781 regoff_t *starts, *ends; |
|
2782 { |
|
2783 if (num_regs) |
|
2784 { |
|
2785 bufp->regs_allocated = REGS_REALLOCATE; |
|
2786 regs->num_regs = num_regs; |
|
2787 regs->start = starts; |
|
2788 regs->end = ends; |
|
2789 } |
|
2790 else |
|
2791 { |
|
2792 bufp->regs_allocated = REGS_UNALLOCATED; |
|
2793 regs->num_regs = 0; |
|
2794 regs->start = regs->end = (regoff_t) 0; |
|
2795 } |
|
2796 } |
|
2797 |
|
2798 /* Searching routines. */ |
|
2799 |
|
2800 /* Like re_search_2, below, but only one string is specified, and |
|
2801 doesn't let you say where to stop matching. */ |
|
2802 |
|
2803 int |
|
2804 re_search (bufp, string, size, startpos, range, regs) |
|
2805 struct re_pattern_buffer *bufp; |
|
2806 const char *string; |
|
2807 int size, startpos, range; |
|
2808 struct re_registers *regs; |
|
2809 { |
|
2810 return re_search_2 (bufp, NULL, 0, string, size, startpos, range, |
|
2811 regs, size); |
|
2812 } |
|
2813 |
|
2814 |
|
2815 /* Using the compiled pattern in BUFP->buffer, first tries to match the |
|
2816 virtual concatenation of STRING1 and STRING2, starting first at index |
|
2817 STARTPOS, then at STARTPOS + 1, and so on. |
|
2818 |
|
2819 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. |
|
2820 |
|
2821 RANGE is how far to scan while trying to match. RANGE = 0 means try |
|
2822 only at STARTPOS; in general, the last start tried is STARTPOS + |
|
2823 RANGE. |
|
2824 |
|
2825 In REGS, return the indices of the virtual concatenation of STRING1 |
|
2826 and STRING2 that matched the entire BUFP->buffer and its contained |
|
2827 subexpressions. |
|
2828 |
|
2829 Do not consider matching one past the index STOP in the virtual |
|
2830 concatenation of STRING1 and STRING2. |
|
2831 |
|
2832 We return either the position in the strings at which the match was |
|
2833 found, -1 if no match, or -2 if error (such as failure |
|
2834 stack overflow). */ |
|
2835 |
|
2836 int |
|
2837 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop) |
|
2838 struct re_pattern_buffer *bufp; |
|
2839 const char *string1, *string2; |
|
2840 int size1, size2; |
|
2841 int startpos; |
|
2842 int range; |
|
2843 struct re_registers *regs; |
|
2844 int stop; |
|
2845 { |
|
2846 int val; |
|
2847 register char *fastmap = bufp->fastmap; |
|
2848 register char *translate = bufp->translate; |
|
2849 int total_size = size1 + size2; |
|
2850 int endpos = startpos + range; |
|
2851 |
|
2852 /* Check for out-of-range STARTPOS. */ |
|
2853 if (startpos < 0 || startpos > total_size) |
|
2854 return -1; |
|
2855 |
|
2856 /* Fix up RANGE if it might eventually take us outside |
|
2857 the virtual concatenation of STRING1 and STRING2. */ |
|
2858 if (endpos < -1) |
|
2859 range = -1 - startpos; |
|
2860 else if (endpos > total_size) |
|
2861 range = total_size - startpos; |
|
2862 |
|
2863 /* Update the fastmap now if not correct already. */ |
|
2864 if (fastmap && !bufp->fastmap_accurate) |
|
2865 if (re_compile_fastmap (bufp) == -2) |
|
2866 return -2; |
|
2867 |
|
2868 /* If the search isn't to be a backwards one, don't waste time in a |
|
2869 long search for a pattern that says it is anchored. */ |
|
2870 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf |
|
2871 && range > 0) |
|
2872 { |
|
2873 if (startpos > 0) |
|
2874 return -1; |
|
2875 else |
|
2876 range = 1; |
|
2877 } |
|
2878 |
|
2879 for (;;) |
|
2880 { |
|
2881 /* If a fastmap is supplied, skip quickly over characters that |
|
2882 cannot be the start of a match. If the pattern can match the |
|
2883 null string, however, we don't need to skip characters; we want |
|
2884 the first null string. */ |
|
2885 if (fastmap && startpos < total_size && !bufp->can_be_null) |
|
2886 { |
|
2887 if (range > 0) /* Searching forwards. */ |
|
2888 { |
|
2889 register const char *d; |
|
2890 register int lim = 0; |
|
2891 int irange = range; |
|
2892 |
|
2893 if (startpos < size1 && startpos + range >= size1) |
|
2894 lim = range - (size1 - startpos); |
|
2895 |
|
2896 d = (startpos >= size1 ? string2 - size1 : string1) + startpos; |
|
2897 |
|
2898 /* Written out as an if-else to avoid testing `translate' |
|
2899 inside the loop. */ |
|
2900 if (translate) |
|
2901 while (range > lim |
|
2902 && !fastmap[(unsigned char) translate[*d++]]) |
|
2903 range--; |
|
2904 else |
|
2905 while (range > lim && !fastmap[(unsigned char) *d++]) |
|
2906 range--; |
|
2907 |
|
2908 startpos += irange - range; |
|
2909 } |
|
2910 else /* Searching backwards. */ |
|
2911 { |
|
2912 register char c = (size1 == 0 || startpos >= size1 |
|
2913 ? string2[startpos - size1] |
|
2914 : string1[startpos]); |
|
2915 |
|
2916 if (!fastmap[TRANSLATE (c)]) |
|
2917 goto advance; |
|
2918 } |
|
2919 } |
|
2920 |
|
2921 /* If can't match the null string, and that's all we have left, fail. */ |
|
2922 if (range >= 0 && startpos == total_size && fastmap |
|
2923 && !bufp->can_be_null) |
|
2924 return -1; |
|
2925 |
|
2926 val = re_match_2 (bufp, string1, size1, string2, size2, |
|
2927 startpos, regs, stop); |
|
2928 if (val >= 0) |
|
2929 return startpos; |
|
2930 |
|
2931 if (val == -2) |
|
2932 return -2; |
|
2933 |
|
2934 advance: |
|
2935 if (!range) |
|
2936 break; |
|
2937 else if (range > 0) |
|
2938 { |
|
2939 range--; |
|
2940 startpos++; |
|
2941 } |
|
2942 else |
|
2943 { |
|
2944 range++; |
|
2945 startpos--; |
|
2946 } |
|
2947 } |
|
2948 return -1; |
|
2949 } /* re_search_2 */ |
|
2950 |
|
2951 /* Declarations and macros for re_match_2. */ |
|
2952 |
|
2953 static int bcmp_translate (); |
|
2954 static boolean alt_match_null_string_p (), |
|
2955 common_op_match_null_string_p (), |
|
2956 group_match_null_string_p (); |
|
2957 |
|
2958 /* Structure for per-register (a.k.a. per-group) information. |
|
2959 This must not be longer than one word, because we push this value |
|
2960 onto the failure stack. Other register information, such as the |
|
2961 starting and ending positions (which are addresses), and the list of |
|
2962 inner groups (which is a bits list) are maintained in separate |
|
2963 variables. |
|
2964 |
|
2965 We are making a (strictly speaking) nonportable assumption here: that |
|
2966 the compiler will pack our bit fields into something that fits into |
|
2967 the type of `word', i.e., is something that fits into one item on the |
|
2968 failure stack. */ |
|
2969 typedef union |
|
2970 { |
|
2971 fail_stack_elt_t word; |
|
2972 struct |
|
2973 { |
|
2974 /* This field is one if this group can match the empty string, |
|
2975 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ |
|
2976 #define MATCH_NULL_UNSET_VALUE 3 |
|
2977 unsigned match_null_string_p : 2; |
|
2978 unsigned is_active : 1; |
|
2979 unsigned matched_something : 1; |
|
2980 unsigned ever_matched_something : 1; |
|
2981 } bits; |
|
2982 } register_info_type; |
|
2983 |
|
2984 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) |
|
2985 #define IS_ACTIVE(R) ((R).bits.is_active) |
|
2986 #define MATCHED_SOMETHING(R) ((R).bits.matched_something) |
|
2987 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) |
|
2988 |
|
2989 |
|
2990 /* Call this when have matched something; it sets `matched' flags for the |
|
2991 registers corresponding to the group of which we currently are inside. |
|
2992 Also records whether this group ever matched something. We only care |
|
2993 about this information at `stop_memory', and then only about the |
|
2994 previous time through the loop (if the group is starred or whatever). |
|
2995 So it is ok to clear all the nonactive registers here. */ |
|
2996 #define SET_REGS_MATCHED() \ |
|
2997 do \ |
|
2998 { \ |
|
2999 unsigned r; \ |
|
3000 for (r = lowest_active_reg; r <= highest_active_reg; r++) \ |
|
3001 { \ |
|
3002 MATCHED_SOMETHING (reg_info[r]) \ |
|
3003 = EVER_MATCHED_SOMETHING (reg_info[r]) \ |
|
3004 = 1; \ |
|
3005 } \ |
|
3006 } \ |
|
3007 while (0) |
|
3008 |
|
3009 |
|
3010 /* This converts PTR, a pointer into one of the search strings `string1' |
|
3011 and `string2' into an offset from the beginning of that string. */ |
|
3012 #define POINTER_TO_OFFSET(ptr) \ |
|
3013 (FIRST_STRING_P (ptr) ? (ptr) - string1 : (ptr) - string2 + size1) |
|
3014 |
|
3015 /* Registers are set to a sentinel when they haven't yet matched. */ |
|
3016 #define REG_UNSET_VALUE ((char *) -1) |
|
3017 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE) |
|
3018 |
|
3019 |
|
3020 /* Macros for dealing with the split strings in re_match_2. */ |
|
3021 |
|
3022 #define MATCHING_IN_FIRST_STRING (dend == end_match_1) |
|
3023 |
|
3024 /* Call before fetching a character with *d. This switches over to |
|
3025 string2 if necessary. */ |
|
3026 #define PREFETCH() \ |
|
3027 while (d == dend) \ |
|
3028 { \ |
|
3029 /* End of string2 => fail. */ \ |
|
3030 if (dend == end_match_2) \ |
|
3031 goto fail; \ |
|
3032 /* End of string1 => advance to string2. */ \ |
|
3033 d = string2; \ |
|
3034 dend = end_match_2; \ |
|
3035 } |
|
3036 |
|
3037 |
|
3038 /* Test if at very beginning or at very end of the virtual concatenation |
|
3039 of `string1' and `string2'. If only one string, it's `string2'. */ |
|
3040 #define AT_STRINGS_BEG() (d == (size1 ? string1 : string2) || !size2) |
|
3041 #define AT_STRINGS_END() (d == end2) |
|
3042 |
|
3043 |
|
3044 /* Test if D points to a character which is word-constituent. We have |
|
3045 two special cases to check for: if past the end of string1, look at |
|
3046 the first character in string2; and if before the beginning of |
|
3047 string2, look at the last character in string1. |
|
3048 |
|
3049 Assumes `string1' exists, so use in conjunction with AT_STRINGS_BEG (). */ |
|
3050 #define LETTER_P(d) \ |
|
3051 (SYNTAX ((d) == end1 ? *string2 \ |
|
3052 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == Sword) |
|
3053 |
|
3054 /* Test if the character before D and the one at D differ with respect |
|
3055 to being word-constituent. */ |
|
3056 #define AT_WORD_BOUNDARY(d) \ |
|
3057 (AT_STRINGS_BEG () || AT_STRINGS_END () || LETTER_P (d - 1) != LETTER_P (d)) |
|
3058 |
|
3059 |
|
3060 /* Free everything we malloc. */ |
|
3061 #ifdef REGEX_MALLOC |
|
3062 #define FREE_VAR(var) if (var) free (var); var = NULL |
|
3063 #define FREE_VARIABLES() \ |
|
3064 do { \ |
|
3065 FREE_VAR (fail_stack.stack); \ |
|
3066 FREE_VAR (regstart); \ |
|
3067 FREE_VAR (regend); \ |
|
3068 FREE_VAR (old_regstart); \ |
|
3069 FREE_VAR (old_regend); \ |
|
3070 FREE_VAR (best_regstart); \ |
|
3071 FREE_VAR (best_regend); \ |
|
3072 FREE_VAR (reg_info); \ |
|
3073 FREE_VAR (reg_dummy); \ |
|
3074 FREE_VAR (reg_info_dummy); \ |
|
3075 } while (0) |
|
3076 #else /* not REGEX_MALLOC */ |
|
3077 /* Some MIPS systems (at least) want this to free alloca'd storage. */ |
|
3078 #define FREE_VARIABLES() alloca (0) |
|
3079 #endif /* not REGEX_MALLOC */ |
|
3080 |
|
3081 |
|
3082 /* These values must meet several constraints. They must not be valid |
|
3083 register values; since we have a limit of 255 registers (because |
|
3084 we use only one byte in the pattern for the register number), we can |
|
3085 use numbers larger than 255. They must differ by 1, because of |
|
3086 NUM_FAILURE_ITEMS above. And the value for the lowest register must |
|
3087 be larger than the value for the highest register, so we do not try |
|
3088 to actually save any registers when none are active. */ |
|
3089 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) |
|
3090 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) |
|
3091 |
|
3092 /* Matching routines. */ |
|
3093 |
|
3094 #ifndef emacs /* Emacs never uses this. */ |
|
3095 /* re_match is like re_match_2 except it takes only a single string. */ |
|
3096 |
|
3097 int |
|
3098 re_match (bufp, string, size, pos, regs) |
|
3099 struct re_pattern_buffer *bufp; |
|
3100 const char *string; |
|
3101 int size, pos; |
|
3102 struct re_registers *regs; |
|
3103 { |
|
3104 return re_match_2 (bufp, NULL, 0, string, size, pos, regs, size); |
|
3105 } |
|
3106 #endif /* not emacs */ |
|
3107 |
|
3108 |
|
3109 /* re_match_2 matches the compiled pattern in BUFP against the |
|
3110 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 |
|
3111 and SIZE2, respectively). We start matching at POS, and stop |
|
3112 matching at STOP. |
|
3113 |
|
3114 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we |
|
3115 store offsets for the substring each group matched in REGS. See the |
|
3116 documentation for exactly how many groups we fill. |
|
3117 |
|
3118 We return -1 if no match, -2 if an internal error (such as the |
|
3119 failure stack overflowing). Otherwise, we return the length of the |
|
3120 matched substring. */ |
|
3121 |
|
3122 int |
|
3123 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) |
|
3124 struct re_pattern_buffer *bufp; |
|
3125 const char *string1, *string2; |
|
3126 int size1, size2; |
|
3127 int pos; |
|
3128 struct re_registers *regs; |
|
3129 int stop; |
|
3130 { |
|
3131 /* General temporaries. */ |
|
3132 int mcnt; |
|
3133 unsigned char *p1; |
|
3134 |
|
3135 /* Just past the end of the corresponding string. */ |
|
3136 const char *end1, *end2; |
|
3137 |
|
3138 /* Pointers into string1 and string2, just past the last characters in |
|
3139 each to consider matching. */ |
|
3140 const char *end_match_1, *end_match_2; |
|
3141 |
|
3142 /* Where we are in the data, and the end of the current string. */ |
|
3143 const char *d, *dend; |
|
3144 |
|
3145 /* Where we are in the pattern, and the end of the pattern. */ |
|
3146 unsigned char *p = bufp->buffer; |
|
3147 register unsigned char *pend = p + bufp->used; |
|
3148 |
|
3149 /* We use this to map every character in the string. */ |
|
3150 char *translate = bufp->translate; |
|
3151 |
|
3152 /* Failure point stack. Each place that can handle a failure further |
|
3153 down the line pushes a failure point on this stack. It consists of |
|
3154 restart, regend, and reg_info for all registers corresponding to |
|
3155 the subexpressions we're currently inside, plus the number of such |
|
3156 registers, and, finally, two char *'s. The first char * is where |
|
3157 to resume scanning the pattern; the second one is where to resume |
|
3158 scanning the strings. If the latter is zero, the failure point is |
|
3159 a ``dummy''; if a failure happens and the failure point is a dummy, |
|
3160 it gets discarded and the next next one is tried. */ |
|
3161 fail_stack_type fail_stack; |
|
3162 #ifdef DEBUG |
|
3163 static unsigned failure_id = 0; |
|
3164 #endif |
|
3165 |
|
3166 /* We fill all the registers internally, independent of what we |
|
3167 return, for use in backreferences. The number here includes |
|
3168 an element for register zero. */ |
|
3169 unsigned num_regs = bufp->re_nsub + 1; |
|
3170 |
|
3171 /* The currently active registers. */ |
|
3172 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
|
3173 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
|
3174 |
|
3175 /* Information on the contents of registers. These are pointers into |
|
3176 the input strings; they record just what was matched (on this |
|
3177 attempt) by a subexpression part of the pattern, that is, the |
|
3178 regnum-th regstart pointer points to where in the pattern we began |
|
3179 matching and the regnum-th regend points to right after where we |
|
3180 stopped matching the regnum-th subexpression. (The zeroth register |
|
3181 keeps track of what the whole pattern matches.) */ |
|
3182 const char **regstart, **regend; |
|
3183 |
|
3184 /* If a group that's operated upon by a repetition operator fails to |
|
3185 match anything, then the register for its start will need to be |
|
3186 restored because it will have been set to wherever in the string we |
|
3187 are when we last see its open-group operator. Similarly for a |
|
3188 register's end. */ |
|
3189 const char **old_regstart, **old_regend; |
|
3190 |
|
3191 /* The is_active field of reg_info helps us keep track of which (possibly |
|
3192 nested) subexpressions we are currently in. The matched_something |
|
3193 field of reg_info[reg_num] helps us tell whether or not we have |
|
3194 matched any of the pattern so far this time through the reg_num-th |
|
3195 subexpression. These two fields get reset each time through any |
|
3196 loop their register is in. */ |
|
3197 register_info_type *reg_info; |
|
3198 |
|
3199 /* The following record the register info as found in the above |
|
3200 variables when we find a match better than any we've seen before. |
|
3201 This happens as we backtrack through the failure points, which in |
|
3202 turn happens only if we have not yet matched the entire string. */ |
|
3203 unsigned best_regs_set = false; |
|
3204 const char **best_regstart, **best_regend; |
|
3205 |
|
3206 /* Logically, this is `best_regend[0]'. But we don't want to have to |
|
3207 allocate space for that if we're not allocating space for anything |
|
3208 else (see below). Also, we never need info about register 0 for |
|
3209 any of the other register vectors, and it seems rather a kludge to |
|
3210 treat `best_regend' differently than the rest. So we keep track of |
|
3211 the end of the best match so far in a separate variable. We |
|
3212 initialize this to NULL so that when we backtrack the first time |
|
3213 and need to test it, it's not garbage. */ |
|
3214 const char *match_end = NULL; |
|
3215 |
|
3216 /* Used when we pop values we don't care about. */ |
|
3217 const char **reg_dummy; |
|
3218 register_info_type *reg_info_dummy; |
|
3219 |
|
3220 #ifdef DEBUG |
|
3221 /* Counts the total number of registers pushed. */ |
|
3222 unsigned num_regs_pushed = 0; |
|
3223 #endif |
|
3224 |
|
3225 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); |
|
3226 |
|
3227 INIT_FAIL_STACK (); |
|
3228 |
|
3229 /* Do not bother to initialize all the register variables if there are |
|
3230 no groups in the pattern, as it takes a fair amount of time. If |
|
3231 there are groups, we include space for register 0 (the whole |
|
3232 pattern), even though we never use it, since it simplifies the |
|
3233 array indexing. We should fix this. */ |
|
3234 if (bufp->re_nsub) |
|
3235 { |
|
3236 regstart = REGEX_TALLOC (num_regs, const char *); |
|
3237 regend = REGEX_TALLOC (num_regs, const char *); |
|
3238 old_regstart = REGEX_TALLOC (num_regs, const char *); |
|
3239 old_regend = REGEX_TALLOC (num_regs, const char *); |
|
3240 best_regstart = REGEX_TALLOC (num_regs, const char *); |
|
3241 best_regend = REGEX_TALLOC (num_regs, const char *); |
|
3242 reg_info = REGEX_TALLOC (num_regs, register_info_type); |
|
3243 reg_dummy = REGEX_TALLOC (num_regs, const char *); |
|
3244 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
|
3245 |
|
3246 if (!(regstart && regend && old_regstart && old_regend && reg_info |
|
3247 && best_regstart && best_regend && reg_dummy && reg_info_dummy)) |
|
3248 { |
|
3249 FREE_VARIABLES (); |
|
3250 return -2; |
|
3251 } |
|
3252 } |
|
3253 #ifdef REGEX_MALLOC |
|
3254 else |
|
3255 { |
|
3256 /* We must initialize all our variables to NULL, so that |
|
3257 `FREE_VARIABLES' doesn't try to free them. Too bad this isn't |
|
3258 Lisp, so we could have a list of variables. As it is, */ |
|
3259 regstart = regend = old_regstart = old_regend = best_regstart |
|
3260 = best_regend = reg_dummy = NULL; |
|
3261 reg_info = reg_info_dummy = (register_info_type *) NULL; |
|
3262 } |
|
3263 #endif /* REGEX_MALLOC */ |
|
3264 |
|
3265 /* The starting position is bogus. */ |
|
3266 if (pos < 0 || pos > size1 + size2) |
|
3267 { |
|
3268 FREE_VARIABLES (); |
|
3269 return -1; |
|
3270 } |
|
3271 |
|
3272 /* Initialize subexpression text positions to -1 to mark ones that no |
|
3273 start_memory/stop_memory has been seen for. Also initialize the |
|
3274 register information struct. */ |
|
3275 for (mcnt = 1; mcnt < num_regs; mcnt++) |
|
3276 { |
|
3277 regstart[mcnt] = regend[mcnt] |
|
3278 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; |
|
3279 |
|
3280 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; |
|
3281 IS_ACTIVE (reg_info[mcnt]) = 0; |
|
3282 MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
|
3283 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
|
3284 } |
|
3285 |
|
3286 /* We move `string1' into `string2' if the latter's empty -- but not if |
|
3287 `string1' is null. */ |
|
3288 if (size2 == 0 && string1 != NULL) |
|
3289 { |
|
3290 string2 = string1; |
|
3291 size2 = size1; |
|
3292 string1 = 0; |
|
3293 size1 = 0; |
|
3294 } |
|
3295 end1 = string1 + size1; |
|
3296 end2 = string2 + size2; |
|
3297 |
|
3298 /* Compute where to stop matching, within the two strings. */ |
|
3299 if (stop <= size1) |
|
3300 { |
|
3301 end_match_1 = string1 + stop; |
|
3302 end_match_2 = string2; |
|
3303 } |
|
3304 else |
|
3305 { |
|
3306 end_match_1 = end1; |
|
3307 end_match_2 = string2 + stop - size1; |
|
3308 } |
|
3309 |
|
3310 /* `p' scans through the pattern as `d' scans through the data. |
|
3311 `dend' is the end of the input string that `d' points within. `d' |
|
3312 is advanced into the following input string whenever necessary, but |
|
3313 this happens before fetching; therefore, at the beginning of the |
|
3314 loop, `d' can be pointing at the end of a string, but it cannot |
|
3315 equal `string2'. */ |
|
3316 if (size1 > 0 && pos <= size1) |
|
3317 { |
|
3318 d = string1 + pos; |
|
3319 dend = end_match_1; |
|
3320 } |
|
3321 else |
|
3322 { |
|
3323 d = string2 + pos - size1; |
|
3324 dend = end_match_2; |
|
3325 } |
|
3326 |
|
3327 DEBUG_PRINT1 ("The compiled pattern is: "); |
|
3328 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); |
|
3329 DEBUG_PRINT1 ("The string to match is: `"); |
|
3330 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); |
|
3331 DEBUG_PRINT1 ("'\n"); |
|
3332 |
|
3333 /* This loops over pattern commands. It exits by returning from the |
|
3334 function if the match is complete, or it drops through if the match |
|
3335 fails at this starting point in the input data. */ |
|
3336 for (;;) |
|
3337 { |
|
3338 DEBUG_PRINT2 ("\n0x%x: ", p); |
|
3339 |
|
3340 if (p == pend) |
|
3341 { /* End of pattern means we might have succeeded. */ |
|
3342 DEBUG_PRINT1 ("End of pattern: "); |
|
3343 /* If not end of string, try backtracking. Otherwise done. */ |
|
3344 if (d != end_match_2) |
|
3345 { |
|
3346 DEBUG_PRINT1 ("backtracking.\n"); |
|
3347 |
|
3348 if (!FAIL_STACK_EMPTY ()) |
|
3349 { /* More failure points to try. */ |
|
3350 boolean same_str_p = (FIRST_STRING_P (match_end) |
|
3351 == MATCHING_IN_FIRST_STRING); |
|
3352 |
|
3353 /* If exceeds best match so far, save it. */ |
|
3354 if (!best_regs_set |
|
3355 || (same_str_p && d > match_end) |
|
3356 || (!same_str_p && !MATCHING_IN_FIRST_STRING)) |
|
3357 { |
|
3358 best_regs_set = true; |
|
3359 match_end = d; |
|
3360 |
|
3361 DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); |
|
3362 |
|
3363 for (mcnt = 1; mcnt < num_regs; mcnt++) |
|
3364 { |
|
3365 best_regstart[mcnt] = regstart[mcnt]; |
|
3366 best_regend[mcnt] = regend[mcnt]; |
|
3367 } |
|
3368 } |
|
3369 goto fail; |
|
3370 } |
|
3371 |
|
3372 /* If no failure points, don't restore garbage. */ |
|
3373 else if (best_regs_set) |
|
3374 { |
|
3375 restore_best_regs: |
|
3376 /* Restore best match. It may happen that `dend == |
|
3377 end_match_1' while the restored d is in string2. |
|
3378 For example, the pattern `x.*y.*z' against the |
|
3379 strings `x-' and `y-z-', if the two strings are |
|
3380 not consecutive in memory. */ |
|
3381 d = match_end; |
|
3382 dend = ((d >= string1 && d <= end1) |
|
3383 ? end_match_1 : end_match_2); |
|
3384 |
|
3385 for (mcnt = 1; mcnt < num_regs; mcnt++) |
|
3386 { |
|
3387 regstart[mcnt] = best_regstart[mcnt]; |
|
3388 regend[mcnt] = best_regend[mcnt]; |
|
3389 } |
|
3390 } |
|
3391 } /* d != end_match_2 */ |
|
3392 |
|
3393 DEBUG_PRINT1 ("\nAccepting match.\n"); |
|
3394 |
|
3395 /* If caller wants register contents data back, do it. */ |
|
3396 if (regs && !bufp->no_sub) |
|
3397 { |
|
3398 /* Have the register data arrays been allocated? */ |
|
3399 if (bufp->regs_allocated == REGS_UNALLOCATED) |
|
3400 { /* No. So allocate them with malloc. We need one |
|
3401 extra element beyond `num_regs' for the `-1' marker |
|
3402 GNU code uses. */ |
|
3403 regs->num_regs = MAX (RE_NREGS, num_regs + 1); |
|
3404 regs->start = TALLOC (regs->num_regs, regoff_t); |
|
3405 regs->end = TALLOC (regs->num_regs, regoff_t); |
|
3406 if (regs->start == NULL || regs->end == NULL) |
|
3407 return -2; |
|
3408 bufp->regs_allocated = REGS_REALLOCATE; |
|
3409 } |
|
3410 else if (bufp->regs_allocated == REGS_REALLOCATE) |
|
3411 { /* Yes. If we need more elements than were already |
|
3412 allocated, reallocate them. If we need fewer, just |
|
3413 leave it alone. */ |
|
3414 if (regs->num_regs < num_regs + 1) |
|
3415 { |
|
3416 regs->num_regs = num_regs + 1; |
|
3417 RETALLOC (regs->start, regs->num_regs, regoff_t); |
|
3418 RETALLOC (regs->end, regs->num_regs, regoff_t); |
|
3419 if (regs->start == NULL || regs->end == NULL) |
|
3420 return -2; |
|
3421 } |
|
3422 } |
|
3423 else |
|
3424 assert (bufp->regs_allocated == REGS_FIXED); |
|
3425 |
|
3426 /* Convert the pointer data in `regstart' and `regend' to |
|
3427 indices. Register zero has to be set differently, |
|
3428 since we haven't kept track of any info for it. */ |
|
3429 if (regs->num_regs > 0) |
|
3430 { |
|
3431 regs->start[0] = pos; |
|
3432 regs->end[0] = (MATCHING_IN_FIRST_STRING ? d - string1 |
|
3433 : d - string2 + size1); |
|
3434 } |
|
3435 |
|
3436 /* Go through the first `min (num_regs, regs->num_regs)' |
|
3437 registers, since that is all we initialized. */ |
|
3438 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) |
|
3439 { |
|
3440 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) |
|
3441 regs->start[mcnt] = regs->end[mcnt] = -1; |
|
3442 else |
|
3443 { |
|
3444 regs->start[mcnt] = POINTER_TO_OFFSET (regstart[mcnt]); |
|
3445 regs->end[mcnt] = POINTER_TO_OFFSET (regend[mcnt]); |
|
3446 } |
|
3447 } |
|
3448 |
|
3449 /* If the regs structure we return has more elements than |
|
3450 were in the pattern, set the extra elements to -1. If |
|
3451 we (re)allocated the registers, this is the case, |
|
3452 because we always allocate enough to have at least one |
|
3453 -1 at the end. */ |
|
3454 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++) |
|
3455 regs->start[mcnt] = regs->end[mcnt] = -1; |
|
3456 } /* regs && !bufp->no_sub */ |
|
3457 |
|
3458 FREE_VARIABLES (); |
|
3459 DEBUG_PRINT2 ("%d registers pushed.\n", num_regs_pushed); |
|
3460 |
|
3461 mcnt = d - pos - (MATCHING_IN_FIRST_STRING |
|
3462 ? string1 |
|
3463 : string2 - size1); |
|
3464 |
|
3465 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
|
3466 |
|
3467 return mcnt; |
|
3468 } |
|
3469 |
|
3470 /* Otherwise match next pattern command. */ |
|
3471 #ifdef SWITCH_ENUM_BUG |
|
3472 switch ((int) ((re_opcode_t) *p++)) |
|
3473 #else |
|
3474 switch ((re_opcode_t) *p++) |
|
3475 #endif |
|
3476 { |
|
3477 /* Ignore these. Used to ignore the n of succeed_n's which |
|
3478 currently have n == 0. */ |
|
3479 case no_op: |
|
3480 DEBUG_PRINT1 ("EXECUTING no_op.\n"); |
|
3481 break; |
|
3482 |
|
3483 |
|
3484 /* Match the next n pattern characters exactly. The following |
|
3485 byte in the pattern defines n, and the n bytes after that |
|
3486 are the characters to match. */ |
|
3487 case exactn: |
|
3488 mcnt = *p++; |
|
3489 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); |
|
3490 |
|
3491 /* This is written out as an if-else so we don't waste time |
|
3492 testing `translate' inside the loop. */ |
|
3493 if (translate) |
|
3494 { |
|
3495 do |
|
3496 { |
|
3497 PREFETCH (); |
|
3498 if (translate[(unsigned char) *d++] != (char) *p++) |
|
3499 goto fail; |
|
3500 } |
|
3501 while (--mcnt); |
|
3502 } |
|
3503 else |
|
3504 { |
|
3505 do |
|
3506 { |
|
3507 PREFETCH (); |
|
3508 if (*d++ != (char) *p++) goto fail; |
|
3509 } |
|
3510 while (--mcnt); |
|
3511 } |
|
3512 SET_REGS_MATCHED (); |
|
3513 break; |
|
3514 |
|
3515 |
|
3516 /* Match any character except possibly a newline or a null. */ |
|
3517 case anychar: |
|
3518 DEBUG_PRINT1 ("EXECUTING anychar.\n"); |
|
3519 |
|
3520 PREFETCH (); |
|
3521 |
|
3522 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') |
|
3523 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) |
|
3524 goto fail; |
|
3525 |
|
3526 SET_REGS_MATCHED (); |
|
3527 DEBUG_PRINT2 (" Matched `%d'.\n", *d); |
|
3528 d++; |
|
3529 break; |
|
3530 |
|
3531 |
|
3532 case charset: |
|
3533 case charset_not: |
|
3534 { |
|
3535 register unsigned char c; |
|
3536 boolean not = (re_opcode_t) *(p - 1) == charset_not; |
|
3537 |
|
3538 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); |
|
3539 |
|
3540 PREFETCH (); |
|
3541 c = TRANSLATE (*d); /* The character to match. */ |
|
3542 |
|
3543 /* Cast to `unsigned' instead of `unsigned char' in case the |
|
3544 bit list is a full 32 bytes long. */ |
|
3545 if (c < (unsigned) (*p * BYTEWIDTH) |
|
3546 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
|
3547 not = !not; |
|
3548 |
|
3549 p += 1 + *p; |
|
3550 |
|
3551 if (!not) goto fail; |
|
3552 |
|
3553 SET_REGS_MATCHED (); |
|
3554 d++; |
|
3555 break; |
|
3556 } |
|
3557 |
|
3558 |
|
3559 /* The beginning of a group is represented by start_memory. |
|
3560 The arguments are the register number in the next byte, and the |
|
3561 number of groups inner to this one in the next. The text |
|
3562 matched within the group is recorded (in the internal |
|
3563 registers data structure) under the register number. */ |
|
3564 case start_memory: |
|
3565 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); |
|
3566 |
|
3567 /* Find out if this group can match the empty string. */ |
|
3568 p1 = p; /* To send to group_match_null_string_p. */ |
|
3569 |
|
3570 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) |
|
3571 REG_MATCH_NULL_STRING_P (reg_info[*p]) |
|
3572 = group_match_null_string_p (&p1, pend, reg_info); |
|
3573 |
|
3574 /* Save the position in the string where we were the last time |
|
3575 we were at this open-group operator in case the group is |
|
3576 operated upon by a repetition operator, e.g., with `(a*)*b' |
|
3577 against `ab'; then we want to ignore where we are now in |
|
3578 the string in case this attempt to match fails. */ |
|
3579 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
|
3580 ? REG_UNSET (regstart[*p]) ? d : regstart[*p] |
|
3581 : regstart[*p]; |
|
3582 DEBUG_PRINT2 (" old_regstart: %d\n", |
|
3583 POINTER_TO_OFFSET (old_regstart[*p])); |
|
3584 |
|
3585 regstart[*p] = d; |
|
3586 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); |
|
3587 |
|
3588 IS_ACTIVE (reg_info[*p]) = 1; |
|
3589 MATCHED_SOMETHING (reg_info[*p]) = 0; |
|
3590 |
|
3591 /* This is the new highest active register. */ |
|
3592 highest_active_reg = *p; |
|
3593 |
|
3594 /* If nothing was active before, this is the new lowest active |
|
3595 register. */ |
|
3596 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
|
3597 lowest_active_reg = *p; |
|
3598 |
|
3599 /* Move past the register number and inner group count. */ |
|
3600 p += 2; |
|
3601 break; |
|
3602 |
|
3603 |
|
3604 /* The stop_memory opcode represents the end of a group. Its |
|
3605 arguments are the same as start_memory's: the register |
|
3606 number, and the number of inner groups. */ |
|
3607 case stop_memory: |
|
3608 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
|
3609 |
|
3610 /* We need to save the string position the last time we were at |
|
3611 this close-group operator in case the group is operated |
|
3612 upon by a repetition operator, e.g., with `((a*)*(b*)*)*' |
|
3613 against `aba'; then we want to ignore where we are now in |
|
3614 the string in case this attempt to match fails. */ |
|
3615 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
|
3616 ? REG_UNSET (regend[*p]) ? d : regend[*p] |
|
3617 : regend[*p]; |
|
3618 DEBUG_PRINT2 (" old_regend: %d\n", |
|
3619 POINTER_TO_OFFSET (old_regend[*p])); |
|
3620 |
|
3621 regend[*p] = d; |
|
3622 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); |
|
3623 |
|
3624 /* This register isn't active anymore. */ |
|
3625 IS_ACTIVE (reg_info[*p]) = 0; |
|
3626 |
|
3627 /* If this was the only register active, nothing is active |
|
3628 anymore. */ |
|
3629 if (lowest_active_reg == highest_active_reg) |
|
3630 { |
|
3631 lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
|
3632 highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
|
3633 } |
|
3634 else |
|
3635 { /* We must scan for the new highest active register, since |
|
3636 it isn't necessarily one less than now: consider |
|
3637 (a(b)c(d(e)f)g). When group 3 ends, after the f), the |
|
3638 new highest active register is 1. */ |
|
3639 unsigned char r = *p - 1; |
|
3640 while (r > 0 && !IS_ACTIVE (reg_info[r])) |
|
3641 r--; |
|
3642 |
|
3643 /* If we end up at register zero, that means that we saved |
|
3644 the registers as the result of an `on_failure_jump', not |
|
3645 a `start_memory', and we jumped to past the innermost |
|
3646 `stop_memory'. For example, in ((.)*) we save |
|
3647 registers 1 and 2 as a result of the *, but when we pop |
|
3648 back to the second ), we are at the stop_memory 1. |
|
3649 Thus, nothing is active. */ |
|
3650 if (r == 0) |
|
3651 { |
|
3652 lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
|
3653 highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
|
3654 } |
|
3655 else |
|
3656 highest_active_reg = r; |
|
3657 } |
|
3658 |
|
3659 /* If just failed to match something this time around with a |
|
3660 group that's operated on by a repetition operator, try to |
|
3661 force exit from the ``loop,'' and restore the register |
|
3662 information for this group that we had before trying this |
|
3663 last match. */ |
|
3664 if ((!MATCHED_SOMETHING (reg_info[*p]) |
|
3665 || (re_opcode_t) p[-3] == start_memory) |
|
3666 && (p + 2) < pend) |
|
3667 { |
|
3668 boolean is_a_jump_n = false; |
|
3669 |
|
3670 p1 = p + 2; |
|
3671 mcnt = 0; |
|
3672 switch ((re_opcode_t) *p1++) |
|
3673 { |
|
3674 case jump_n: |
|
3675 is_a_jump_n = true; |
|
3676 case pop_failure_jump: |
|
3677 case maybe_pop_jump: |
|
3678 case jump: |
|
3679 case dummy_failure_jump: |
|
3680 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
3681 if (is_a_jump_n) |
|
3682 p1 += 2; |
|
3683 break; |
|
3684 |
|
3685 default: |
|
3686 /* do nothing */ ; |
|
3687 } |
|
3688 p1 += mcnt; |
|
3689 |
|
3690 /* If the next operation is a jump backwards in the pattern |
|
3691 to an on_failure_jump right before the start_memory |
|
3692 corresponding to this stop_memory, exit from the loop |
|
3693 by forcing a failure after pushing on the stack the |
|
3694 on_failure_jump's jump in the pattern, and d. */ |
|
3695 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump |
|
3696 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) |
|
3697 { |
|
3698 /* If this group ever matched anything, then restore |
|
3699 what its registers were before trying this last |
|
3700 failed match, e.g., with `(a*)*b' against `ab' for |
|
3701 regstart[1], and, e.g., with `((a*)*(b*)*)*' |
|
3702 against `aba' for regend[3]. |
|
3703 |
|
3704 Also restore the registers for inner groups for, |
|
3705 e.g., `((a*)(b*))*' against `aba' (register 3 would |
|
3706 otherwise get trashed). */ |
|
3707 |
|
3708 if (EVER_MATCHED_SOMETHING (reg_info[*p])) |
|
3709 { |
|
3710 unsigned r; |
|
3711 |
|
3712 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; |
|
3713 |
|
3714 /* Restore this and inner groups' (if any) registers. */ |
|
3715 for (r = *p; r < *p + *(p + 1); r++) |
|
3716 { |
|
3717 regstart[r] = old_regstart[r]; |
|
3718 |
|
3719 /* xx why this test? */ |
|
3720 if ((int) old_regend[r] >= (int) regstart[r]) |
|
3721 regend[r] = old_regend[r]; |
|
3722 } |
|
3723 } |
|
3724 p1++; |
|
3725 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
3726 PUSH_FAILURE_POINT (p1 + mcnt, d, -2); |
|
3727 |
|
3728 goto fail; |
|
3729 } |
|
3730 } |
|
3731 |
|
3732 /* Move past the register number and the inner group count. */ |
|
3733 p += 2; |
|
3734 break; |
|
3735 |
|
3736 |
|
3737 /* \<digit> has been turned into a `duplicate' command which is |
|
3738 followed by the numeric value of <digit> as the register number. */ |
|
3739 case duplicate: |
|
3740 { |
|
3741 register const char *d2, *dend2; |
|
3742 int regno = *p++; /* Get which register to match against. */ |
|
3743 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); |
|
3744 |
|
3745 /* Can't back reference a group which we've never matched. */ |
|
3746 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) |
|
3747 goto fail; |
|
3748 |
|
3749 /* Where in input to try to start matching. */ |
|
3750 d2 = regstart[regno]; |
|
3751 |
|
3752 /* Where to stop matching; if both the place to start and |
|
3753 the place to stop matching are in the same string, then |
|
3754 set to the place to stop, otherwise, for now have to use |
|
3755 the end of the first string. */ |
|
3756 |
|
3757 dend2 = ((FIRST_STRING_P (regstart[regno]) |
|
3758 == FIRST_STRING_P (regend[regno])) |
|
3759 ? regend[regno] : end_match_1); |
|
3760 for (;;) |
|
3761 { |
|
3762 /* If necessary, advance to next segment in register |
|
3763 contents. */ |
|
3764 while (d2 == dend2) |
|
3765 { |
|
3766 if (dend2 == end_match_2) break; |
|
3767 if (dend2 == regend[regno]) break; |
|
3768 |
|
3769 /* End of string1 => advance to string2. */ |
|
3770 d2 = string2; |
|
3771 dend2 = regend[regno]; |
|
3772 } |
|
3773 /* At end of register contents => success */ |
|
3774 if (d2 == dend2) break; |
|
3775 |
|
3776 /* If necessary, advance to next segment in data. */ |
|
3777 PREFETCH (); |
|
3778 |
|
3779 /* How many characters left in this segment to match. */ |
|
3780 mcnt = dend - d; |
|
3781 |
|
3782 /* Want how many consecutive characters we can match in |
|
3783 one shot, so, if necessary, adjust the count. */ |
|
3784 if (mcnt > dend2 - d2) |
|
3785 mcnt = dend2 - d2; |
|
3786 |
|
3787 /* Compare that many; failure if mismatch, else move |
|
3788 past them. */ |
|
3789 if (translate |
|
3790 ? bcmp_translate (d, d2, mcnt, translate) |
|
3791 : bcmp (d, d2, mcnt)) |
|
3792 goto fail; |
|
3793 d += mcnt, d2 += mcnt; |
|
3794 } |
|
3795 } |
|
3796 break; |
|
3797 |
|
3798 |
|
3799 /* begline matches the empty string at the beginning of the string |
|
3800 (unless `not_bol' is set in `bufp'), and, if |
|
3801 `newline_anchor' is set, after newlines. */ |
|
3802 case begline: |
|
3803 DEBUG_PRINT1 ("EXECUTING begline.\n"); |
|
3804 |
|
3805 if (AT_STRINGS_BEG ()) |
|
3806 { |
|
3807 if (!bufp->not_bol) break; |
|
3808 } |
|
3809 else if (d[-1] == '\n' && bufp->newline_anchor) |
|
3810 { |
|
3811 break; |
|
3812 } |
|
3813 /* In all other cases, we fail. */ |
|
3814 goto fail; |
|
3815 |
|
3816 |
|
3817 /* endline is the dual of begline. */ |
|
3818 case endline: |
|
3819 DEBUG_PRINT1 ("EXECUTING endline.\n"); |
|
3820 |
|
3821 if (AT_STRINGS_END ()) |
|
3822 { |
|
3823 if (!bufp->not_eol) break; |
|
3824 } |
|
3825 |
|
3826 /* We have to ``prefetch'' the next character. */ |
|
3827 else if ((d == end1 ? *string2 : *d) == '\n' |
|
3828 && bufp->newline_anchor) |
|
3829 { |
|
3830 break; |
|
3831 } |
|
3832 goto fail; |
|
3833 |
|
3834 |
|
3835 /* Match at the very beginning of the data. */ |
|
3836 case begbuf: |
|
3837 DEBUG_PRINT1 ("EXECUTING begbuf.\n"); |
|
3838 if (AT_STRINGS_BEG ()) |
|
3839 break; |
|
3840 goto fail; |
|
3841 |
|
3842 |
|
3843 /* Match at the very end of the data. */ |
|
3844 case endbuf: |
|
3845 DEBUG_PRINT1 ("EXECUTING endbuf.\n"); |
|
3846 if (AT_STRINGS_END ()) |
|
3847 break; |
|
3848 goto fail; |
|
3849 |
|
3850 |
|
3851 /* on_failure_keep_string_jump is used to optimize `.*\n'. It |
|
3852 pushes NULL as the value for the string on the stack. Then |
|
3853 `pop_failure_point' will keep the current value for the |
|
3854 string, instead of restoring it. To see why, consider |
|
3855 matching `foo\nbar' against `.*\n'. The .* matches the foo; |
|
3856 then the . fails against the \n. But the next thing we want |
|
3857 to do is match the \n against the \n; if we restored the |
|
3858 string value, we would be back at the foo. |
|
3859 |
|
3860 Because this is used only in specific cases, we don't need to |
|
3861 check all the things that `on_failure_jump' does, to make |
|
3862 sure the right things get saved on the stack. Hence we don't |
|
3863 share its code. The only reason to push anything on the |
|
3864 stack at all is that otherwise we would have to change |
|
3865 `anychar's code to do something besides goto fail in this |
|
3866 case; that seems worse than this. */ |
|
3867 case on_failure_keep_string_jump: |
|
3868 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
|
3869 |
|
3870 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
|
3871 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); |
|
3872 |
|
3873 PUSH_FAILURE_POINT (p + mcnt, NULL, -2); |
|
3874 break; |
|
3875 |
|
3876 |
|
3877 /* Uses of on_failure_jump: |
|
3878 |
|
3879 Each alternative starts with an on_failure_jump that points |
|
3880 to the beginning of the next alternative. Each alternative |
|
3881 except the last ends with a jump that in effect jumps past |
|
3882 the rest of the alternatives. (They really jump to the |
|
3883 ending jump of the following alternative, because tensioning |
|
3884 these jumps is a hassle.) |
|
3885 |
|
3886 Repeats start with an on_failure_jump that points past both |
|
3887 the repetition text and either the following jump or |
|
3888 pop_failure_jump back to this on_failure_jump. */ |
|
3889 case on_failure_jump: |
|
3890 on_failure: |
|
3891 DEBUG_PRINT1 ("EXECUTING on_failure_jump"); |
|
3892 |
|
3893 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
|
3894 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); |
|
3895 |
|
3896 /* If this on_failure_jump comes right before a group (i.e., |
|
3897 the original * applied to a group), save the information |
|
3898 for that group and all inner ones, so that if we fail back |
|
3899 to this point, the group's information will be correct. |
|
3900 For example, in \(a*\)*\1, we only need the preceding group, |
|
3901 and in \(\(a*\)b*\)\2, we need the inner group. */ |
|
3902 |
|
3903 /* We can't use `p' to check ahead because we push |
|
3904 a failure point to `p + mcnt' after we do this. */ |
|
3905 p1 = p; |
|
3906 |
|
3907 /* We need to skip no_op's before we look for the |
|
3908 start_memory in case this on_failure_jump is happening as |
|
3909 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 |
|
3910 against aba. */ |
|
3911 while (p1 < pend && (re_opcode_t) *p1 == no_op) |
|
3912 p1++; |
|
3913 |
|
3914 if (p1 < pend && (re_opcode_t) *p1 == start_memory) |
|
3915 { |
|
3916 /* We have a new highest active register now. This will |
|
3917 get reset at the start_memory we are about to get to, |
|
3918 but we will have saved all the registers relevant to |
|
3919 this repetition op, as described above. */ |
|
3920 highest_active_reg = *(p1 + 1) + *(p1 + 2); |
|
3921 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
|
3922 lowest_active_reg = *(p1 + 1); |
|
3923 } |
|
3924 |
|
3925 DEBUG_PRINT1 (":\n"); |
|
3926 PUSH_FAILURE_POINT (p + mcnt, d, -2); |
|
3927 break; |
|
3928 |
|
3929 |
|
3930 /* A smart repeat ends with a maybe_pop_jump. |
|
3931 We change it either to a pop_failure_jump or a jump. */ |
|
3932 case maybe_pop_jump: |
|
3933 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
|
3934 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); |
|
3935 { |
|
3936 register unsigned char *p2 = p; |
|
3937 |
|
3938 /* Compare the beginning of the repeat with what in the |
|
3939 pattern follows its end. If we can establish that there |
|
3940 is nothing that they would both match, i.e., that we |
|
3941 would have to backtrack because of (as in, e.g., `a*a') |
|
3942 then we can change to pop_failure_jump, because we'll |
|
3943 never have to backtrack. |
|
3944 |
|
3945 This is not true in the case of alternatives: in |
|
3946 `(a|ab)*' we do need to backtrack to the `ab' alternative |
|
3947 (e.g., if the string was `ab'). But instead of trying to |
|
3948 detect that here, the alternative has put on a dummy |
|
3949 failure point which is what we will end up popping. */ |
|
3950 |
|
3951 /* Skip over open/close-group commands. */ |
|
3952 while (p2 + 2 < pend |
|
3953 && ((re_opcode_t) *p2 == stop_memory |
|
3954 || (re_opcode_t) *p2 == start_memory)) |
|
3955 p2 += 3; /* Skip over args, too. */ |
|
3956 |
|
3957 /* If we're at the end of the pattern, we can change. */ |
|
3958 if (p2 == pend) |
|
3959 { |
|
3960 p[-3] = (unsigned char) pop_failure_jump; |
|
3961 DEBUG_PRINT1 |
|
3962 (" End of pattern: change to `pop_failure_jump'.\n"); |
|
3963 } |
|
3964 |
|
3965 else if ((re_opcode_t) *p2 == exactn |
|
3966 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) |
|
3967 { |
|
3968 register unsigned char c |
|
3969 = *p2 == (unsigned char) endline ? '\n' : p2[2]; |
|
3970 p1 = p + mcnt; |
|
3971 |
|
3972 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding |
|
3973 to the `maybe_finalize_jump' of this case. Examine what |
|
3974 follows. */ |
|
3975 if ((re_opcode_t) p1[3] == exactn && p1[5] != c) |
|
3976 p[-3] = (unsigned char) pop_failure_jump; |
|
3977 else if ((re_opcode_t) p1[3] == charset |
|
3978 || (re_opcode_t) p1[3] == charset_not) |
|
3979 { |
|
3980 int not = (re_opcode_t) p1[3] == charset_not; |
|
3981 |
|
3982 if (c < (unsigned char) (p1[4] * BYTEWIDTH) |
|
3983 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
|
3984 not = !not; |
|
3985 |
|
3986 /* `not' is equal to 1 if c would match, which means |
|
3987 that we can't change to pop_failure_jump. */ |
|
3988 if (!not) |
|
3989 { |
|
3990 p[-3] = (unsigned char) pop_failure_jump; |
|
3991 DEBUG_PRINT1 |
|
3992 (" No match: change to `pop_failure_jump'.\n"); |
|
3993 |
|
3994 } |
|
3995 } |
|
3996 } |
|
3997 } |
|
3998 p -= 2; /* Point at relative address again. */ |
|
3999 if ((re_opcode_t) p[-1] != pop_failure_jump) |
|
4000 { |
|
4001 p[-1] = (unsigned char) jump; |
|
4002 goto unconditional_jump; |
|
4003 } |
|
4004 /* Note fall through. */ |
|
4005 |
|
4006 |
|
4007 /* The end of a simple repeat has a pop_failure_jump back to |
|
4008 its matching on_failure_jump, where the latter will push a |
|
4009 failure point. The pop_failure_jump takes off failure |
|
4010 points put on by this pop_failure_jump's matching |
|
4011 on_failure_jump; we got through the pattern to here from the |
|
4012 matching on_failure_jump, so didn't fail. */ |
|
4013 case pop_failure_jump: |
|
4014 { |
|
4015 /* We need to pass separate storage for the lowest and |
|
4016 highest registers, even though we don't care about the |
|
4017 actual values. Otherwise, we will restore only one |
|
4018 register from the stack, since lowest will == highest in |
|
4019 `pop_failure_point'. */ |
|
4020 unsigned dummy_low_reg, dummy_high_reg; |
|
4021 unsigned char *pdummy; |
|
4022 const char *sdummy; |
|
4023 |
|
4024 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); |
|
4025 POP_FAILURE_POINT (sdummy, pdummy, |
|
4026 dummy_low_reg, dummy_high_reg, |
|
4027 reg_dummy, reg_dummy, reg_info_dummy); |
|
4028 } |
|
4029 /* Note fall through. */ |
|
4030 |
|
4031 |
|
4032 /* Unconditionally jump (without popping any failure points). */ |
|
4033 case jump: |
|
4034 unconditional_jump: |
|
4035 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ |
|
4036 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); |
|
4037 p += mcnt; /* Do the jump. */ |
|
4038 DEBUG_PRINT2 ("(to 0x%x).\n", p); |
|
4039 break; |
|
4040 |
|
4041 |
|
4042 /* We need this opcode so we can detect where alternatives end |
|
4043 in `group_match_null_string_p' et al. */ |
|
4044 case jump_past_alt: |
|
4045 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); |
|
4046 goto unconditional_jump; |
|
4047 |
|
4048 |
|
4049 /* Normally, the on_failure_jump pushes a failure point, which |
|
4050 then gets popped at pop_failure_jump. We will end up at |
|
4051 pop_failure_jump, also, and with a pattern of, say, `a+', we |
|
4052 are skipping over the on_failure_jump, so we have to push |
|
4053 something meaningless for pop_failure_jump to pop. */ |
|
4054 case dummy_failure_jump: |
|
4055 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); |
|
4056 /* It doesn't matter what we push for the string here. What |
|
4057 the code at `fail' tests is the value for the pattern. */ |
|
4058 PUSH_FAILURE_POINT (0, 0, -2); |
|
4059 goto unconditional_jump; |
|
4060 |
|
4061 |
|
4062 /* At the end of an alternative, we need to push a dummy failure |
|
4063 point in case we are followed by a pop_failure_jump', because |
|
4064 we don't want the failure point for the alternative to be |
|
4065 popped. For example, matching `(a|ab)*' against `aab' |
|
4066 requires that we match the `ab' alternative. */ |
|
4067 case push_dummy_failure: |
|
4068 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); |
|
4069 /* See comments just above at `dummy_failure_jump' about the |
|
4070 two zeroes. */ |
|
4071 PUSH_FAILURE_POINT (0, 0, -2); |
|
4072 break; |
|
4073 |
|
4074 /* Have to succeed matching what follows at least n times. |
|
4075 After that, handle like `on_failure_jump'. */ |
|
4076 case succeed_n: |
|
4077 EXTRACT_NUMBER (mcnt, p + 2); |
|
4078 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
|
4079 |
|
4080 assert (mcnt >= 0); |
|
4081 /* Originally, this is how many times we HAVE to succeed. */ |
|
4082 if (mcnt > 0) |
|
4083 { |
|
4084 mcnt--; |
|
4085 p += 2; |
|
4086 STORE_NUMBER_AND_INCR (p, mcnt); |
|
4087 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt); |
|
4088 } |
|
4089 else if (mcnt == 0) |
|
4090 { |
|
4091 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); |
|
4092 p[2] = (unsigned char) no_op; |
|
4093 p[3] = (unsigned char) no_op; |
|
4094 goto on_failure; |
|
4095 } |
|
4096 break; |
|
4097 |
|
4098 case jump_n: |
|
4099 EXTRACT_NUMBER (mcnt, p + 2); |
|
4100 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
|
4101 |
|
4102 /* Originally, this is how many times we CAN jump. */ |
|
4103 if (mcnt) |
|
4104 { |
|
4105 mcnt--; |
|
4106 STORE_NUMBER (p + 2, mcnt); |
|
4107 goto unconditional_jump; |
|
4108 } |
|
4109 /* If don't have to jump any more, skip over the rest of command. */ |
|
4110 else |
|
4111 p += 4; |
|
4112 break; |
|
4113 |
|
4114 case set_number_at: |
|
4115 { |
|
4116 DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); |
|
4117 |
|
4118 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
|
4119 p1 = p + mcnt; |
|
4120 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
|
4121 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); |
|
4122 STORE_NUMBER (p1, mcnt); |
|
4123 break; |
|
4124 } |
|
4125 |
|
4126 case wordbound: |
|
4127 DEBUG_PRINT1 ("EXECUTING wordbound.\n"); |
|
4128 if (AT_WORD_BOUNDARY (d)) |
|
4129 break; |
|
4130 goto fail; |
|
4131 |
|
4132 case notwordbound: |
|
4133 DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); |
|
4134 if (AT_WORD_BOUNDARY (d)) |
|
4135 goto fail; |
|
4136 break; |
|
4137 |
|
4138 case wordbeg: |
|
4139 DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); |
|
4140 if (LETTER_P (d) && (AT_STRINGS_BEG () || !LETTER_P (d - 1))) |
|
4141 break; |
|
4142 goto fail; |
|
4143 |
|
4144 case wordend: |
|
4145 DEBUG_PRINT1 ("EXECUTING wordend.\n"); |
|
4146 if (!AT_STRINGS_BEG () && LETTER_P (d - 1) |
|
4147 && (!LETTER_P (d) || AT_STRINGS_END ())) |
|
4148 break; |
|
4149 goto fail; |
|
4150 |
|
4151 #ifdef emacs |
|
4152 #ifdef emacs19 |
|
4153 case before_dot: |
|
4154 DEBUG_PRINT1 ("EXECUTING before_dot.\n"); |
|
4155 if (PTR_CHAR_POS ((unsigned char *) d) >= point) |
|
4156 goto fail; |
|
4157 break; |
|
4158 |
|
4159 case at_dot: |
|
4160 DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
|
4161 if (PTR_CHAR_POS ((unsigned char *) d) != point) |
|
4162 goto fail; |
|
4163 break; |
|
4164 |
|
4165 case after_dot: |
|
4166 DEBUG_PRINT1 ("EXECUTING after_dot.\n"); |
|
4167 if (PTR_CHAR_POS ((unsigned char *) d) <= point) |
|
4168 goto fail; |
|
4169 break; |
|
4170 #else /* not emacs19 */ |
|
4171 case at_dot: |
|
4172 DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
|
4173 if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point) |
|
4174 goto fail; |
|
4175 break; |
|
4176 #endif /* not emacs19 */ |
|
4177 |
|
4178 case syntaxspec: |
|
4179 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); |
|
4180 mcnt = *p++; |
|
4181 goto matchsyntax; |
|
4182 |
|
4183 case wordchar: |
|
4184 DEBUG_PRINT1 ("EXECUTING wordchar.\n"); |
|
4185 mcnt = (int) Sword; |
|
4186 matchsyntax: |
|
4187 PREFETCH (); |
|
4188 if (SYNTAX (*d++) != (enum syntaxcode) mcnt) goto fail; |
|
4189 SET_REGS_MATCHED (); |
|
4190 break; |
|
4191 |
|
4192 case notsyntaxspec: |
|
4193 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); |
|
4194 mcnt = *p++; |
|
4195 goto matchnotsyntax; |
|
4196 |
|
4197 case notwordchar: |
|
4198 DEBUG_PRINT1 ("EXECUTING notwordchar.\n"); |
|
4199 mcnt = (int) Sword; |
|
4200 matchnotsyntax: /* We goto here from notsyntaxspec. */ |
|
4201 PREFETCH (); |
|
4202 if (SYNTAX (*d++) == (enum syntaxcode) mcnt) goto fail; |
|
4203 SET_REGS_MATCHED (); |
|
4204 break; |
|
4205 |
|
4206 #else /* not emacs */ |
|
4207 case wordchar: |
|
4208 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); |
|
4209 PREFETCH (); |
|
4210 if (!LETTER_P (d)) |
|
4211 goto fail; |
|
4212 SET_REGS_MATCHED (); |
|
4213 break; |
|
4214 |
|
4215 case notwordchar: |
|
4216 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); |
|
4217 PREFETCH (); |
|
4218 if (LETTER_P (d)) |
|
4219 goto fail; |
|
4220 SET_REGS_MATCHED (); |
|
4221 break; |
|
4222 #endif /* not emacs */ |
|
4223 |
|
4224 default: |
|
4225 abort (); |
|
4226 } |
|
4227 continue; /* Successfully executed one pattern command; keep going. */ |
|
4228 |
|
4229 |
|
4230 /* We goto here if a matching operation fails. */ |
|
4231 fail: |
|
4232 if (!FAIL_STACK_EMPTY ()) |
|
4233 { /* A restart point is known. Restore to that state. */ |
|
4234 DEBUG_PRINT1 ("\nFAIL:\n"); |
|
4235 POP_FAILURE_POINT (d, p, |
|
4236 lowest_active_reg, highest_active_reg, |
|
4237 regstart, regend, reg_info); |
|
4238 |
|
4239 /* If this failure point is a dummy, try the next one. */ |
|
4240 if (!p) |
|
4241 goto fail; |
|
4242 |
|
4243 /* If we failed to the end of the pattern, don't examine *p. */ |
|
4244 assert (p <= pend); |
|
4245 if (p < pend) |
|
4246 { |
|
4247 boolean is_a_jump_n = false; |
|
4248 |
|
4249 /* If failed to a backwards jump that's part of a repetition |
|
4250 loop, need to pop this failure point and use the next one. */ |
|
4251 switch ((re_opcode_t) *p) |
|
4252 { |
|
4253 case jump_n: |
|
4254 is_a_jump_n = true; |
|
4255 case maybe_pop_jump: |
|
4256 case pop_failure_jump: |
|
4257 case jump: |
|
4258 p1 = p + 1; |
|
4259 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4260 p1 += mcnt; |
|
4261 |
|
4262 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) |
|
4263 || (!is_a_jump_n |
|
4264 && (re_opcode_t) *p1 == on_failure_jump)) |
|
4265 goto fail; |
|
4266 break; |
|
4267 default: |
|
4268 /* do nothing */ ; |
|
4269 } |
|
4270 } |
|
4271 |
|
4272 if (d >= string1 && d <= end1) |
|
4273 dend = end_match_1; |
|
4274 } |
|
4275 else |
|
4276 break; /* Matching at this starting point really fails. */ |
|
4277 } /* for (;;) */ |
|
4278 |
|
4279 if (best_regs_set) |
|
4280 goto restore_best_regs; |
|
4281 |
|
4282 FREE_VARIABLES (); |
|
4283 |
|
4284 return -1; /* Failure to match. */ |
|
4285 } /* re_match_2 */ |
|
4286 |
|
4287 /* Subroutine definitions for re_match_2. */ |
|
4288 |
|
4289 |
|
4290 /* We are passed P pointing to a register number after a start_memory. |
|
4291 |
|
4292 Return true if the pattern up to the corresponding stop_memory can |
|
4293 match the empty string, and false otherwise. |
|
4294 |
|
4295 If we find the matching stop_memory, sets P to point to one past its number. |
|
4296 Otherwise, sets P to an undefined byte less than or equal to END. |
|
4297 |
|
4298 We don't handle duplicates properly (yet). */ |
|
4299 |
|
4300 static boolean |
|
4301 group_match_null_string_p (p, end, reg_info) |
|
4302 unsigned char **p, *end; |
|
4303 register_info_type *reg_info; |
|
4304 { |
|
4305 int mcnt; |
|
4306 /* Point to after the args to the start_memory. */ |
|
4307 unsigned char *p1 = *p + 2; |
|
4308 |
|
4309 while (p1 < end) |
|
4310 { |
|
4311 /* Skip over opcodes that can match nothing, and return true or |
|
4312 false, as appropriate, when we get to one that can't, or to the |
|
4313 matching stop_memory. */ |
|
4314 |
|
4315 switch ((re_opcode_t) *p1) |
|
4316 { |
|
4317 /* Could be either a loop or a series of alternatives. */ |
|
4318 case on_failure_jump: |
|
4319 p1++; |
|
4320 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4321 |
|
4322 /* If the next operation is not a jump backwards in the |
|
4323 pattern. */ |
|
4324 |
|
4325 if (mcnt >= 0) |
|
4326 { |
|
4327 /* Go through the on_failure_jumps of the alternatives, |
|
4328 seeing if any of the alternatives cannot match nothing. |
|
4329 The last alternative starts with only a jump, |
|
4330 whereas the rest start with on_failure_jump and end |
|
4331 with a jump, e.g., here is the pattern for `a|b|c': |
|
4332 |
|
4333 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 |
|
4334 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 |
|
4335 /exactn/1/c |
|
4336 |
|
4337 So, we have to first go through the first (n-1) |
|
4338 alternatives and then deal with the last one separately. */ |
|
4339 |
|
4340 |
|
4341 /* Deal with the first (n-1) alternatives, which start |
|
4342 with an on_failure_jump (see above) that jumps to right |
|
4343 past a jump_past_alt. */ |
|
4344 |
|
4345 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) |
|
4346 { |
|
4347 /* `mcnt' holds how many bytes long the alternative |
|
4348 is, including the ending `jump_past_alt' and |
|
4349 its number. */ |
|
4350 |
|
4351 if (!alt_match_null_string_p (p1, p1 + mcnt - 3, |
|
4352 reg_info)) |
|
4353 return false; |
|
4354 |
|
4355 /* Move to right after this alternative, including the |
|
4356 jump_past_alt. */ |
|
4357 p1 += mcnt; |
|
4358 |
|
4359 /* Break if it's the beginning of an n-th alternative |
|
4360 that doesn't begin with an on_failure_jump. */ |
|
4361 if ((re_opcode_t) *p1 != on_failure_jump) |
|
4362 break; |
|
4363 |
|
4364 /* Still have to check that it's not an n-th |
|
4365 alternative that starts with an on_failure_jump. */ |
|
4366 p1++; |
|
4367 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4368 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) |
|
4369 { |
|
4370 /* Get to the beginning of the n-th alternative. */ |
|
4371 p1 -= 3; |
|
4372 break; |
|
4373 } |
|
4374 } |
|
4375 |
|
4376 /* Deal with the last alternative: go back and get number |
|
4377 of the `jump_past_alt' just before it. `mcnt' contains |
|
4378 the length of the alternative. */ |
|
4379 EXTRACT_NUMBER (mcnt, p1 - 2); |
|
4380 |
|
4381 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) |
|
4382 return false; |
|
4383 |
|
4384 p1 += mcnt; /* Get past the n-th alternative. */ |
|
4385 } /* if mcnt > 0 */ |
|
4386 break; |
|
4387 |
|
4388 |
|
4389 case stop_memory: |
|
4390 assert (p1[1] == **p); |
|
4391 *p = p1 + 2; |
|
4392 return true; |
|
4393 |
|
4394 |
|
4395 default: |
|
4396 if (!common_op_match_null_string_p (&p1, end, reg_info)) |
|
4397 return false; |
|
4398 } |
|
4399 } /* while p1 < end */ |
|
4400 |
|
4401 return false; |
|
4402 } /* group_match_null_string_p */ |
|
4403 |
|
4404 |
|
4405 /* Similar to group_match_null_string_p, but doesn't deal with alternatives: |
|
4406 It expects P to be the first byte of a single alternative and END one |
|
4407 byte past the last. The alternative can contain groups. */ |
|
4408 |
|
4409 static boolean |
|
4410 alt_match_null_string_p (p, end, reg_info) |
|
4411 unsigned char *p, *end; |
|
4412 register_info_type *reg_info; |
|
4413 { |
|
4414 int mcnt; |
|
4415 unsigned char *p1 = p; |
|
4416 |
|
4417 while (p1 < end) |
|
4418 { |
|
4419 /* Skip over opcodes that can match nothing, and break when we get |
|
4420 to one that can't. */ |
|
4421 |
|
4422 switch ((re_opcode_t) *p1) |
|
4423 { |
|
4424 /* It's a loop. */ |
|
4425 case on_failure_jump: |
|
4426 p1++; |
|
4427 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4428 p1 += mcnt; |
|
4429 break; |
|
4430 |
|
4431 default: |
|
4432 if (!common_op_match_null_string_p (&p1, end, reg_info)) |
|
4433 return false; |
|
4434 } |
|
4435 } /* while p1 < end */ |
|
4436 |
|
4437 return true; |
|
4438 } /* alt_match_null_string_p */ |
|
4439 |
|
4440 |
|
4441 /* Deals with the ops common to group_match_null_string_p and |
|
4442 alt_match_null_string_p. |
|
4443 |
|
4444 Sets P to one after the op and its arguments, if any. */ |
|
4445 |
|
4446 static boolean |
|
4447 common_op_match_null_string_p (p, end, reg_info) |
|
4448 unsigned char **p, *end; |
|
4449 register_info_type *reg_info; |
|
4450 { |
|
4451 int mcnt; |
|
4452 boolean ret; |
|
4453 int reg_no; |
|
4454 unsigned char *p1 = *p; |
|
4455 |
|
4456 switch ((re_opcode_t) *p1++) |
|
4457 { |
|
4458 case no_op: |
|
4459 case begline: |
|
4460 case endline: |
|
4461 case begbuf: |
|
4462 case endbuf: |
|
4463 case wordbeg: |
|
4464 case wordend: |
|
4465 case wordbound: |
|
4466 case notwordbound: |
|
4467 #ifdef emacs |
|
4468 case before_dot: |
|
4469 case at_dot: |
|
4470 case after_dot: |
|
4471 #endif |
|
4472 break; |
|
4473 |
|
4474 case start_memory: |
|
4475 reg_no = *p1; |
|
4476 assert (reg_no > 0 && reg_no <= MAX_REGNUM); |
|
4477 ret = group_match_null_string_p (&p1, end, reg_info); |
|
4478 |
|
4479 /* Have to set this here in case we're checking a group which |
|
4480 contains a group and a back reference to it. */ |
|
4481 |
|
4482 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) |
|
4483 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; |
|
4484 |
|
4485 if (!ret) |
|
4486 return false; |
|
4487 break; |
|
4488 |
|
4489 /* If this is an optimized succeed_n for zero times, make the jump. */ |
|
4490 case jump: |
|
4491 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4492 if (mcnt >= 0) |
|
4493 p1 += mcnt; |
|
4494 else |
|
4495 return false; |
|
4496 break; |
|
4497 |
|
4498 case succeed_n: |
|
4499 /* Get to the number of times to succeed. */ |
|
4500 p1 += 2; |
|
4501 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4502 |
|
4503 if (mcnt == 0) |
|
4504 { |
|
4505 p1 -= 4; |
|
4506 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
|
4507 p1 += mcnt; |
|
4508 } |
|
4509 else |
|
4510 return false; |
|
4511 break; |
|
4512 |
|
4513 case duplicate: |
|
4514 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) |
|
4515 return false; |
|
4516 break; |
|
4517 |
|
4518 case set_number_at: |
|
4519 p1 += 4; |
|
4520 |
|
4521 default: |
|
4522 /* All other opcodes mean we cannot match the empty string. */ |
|
4523 return false; |
|
4524 } |
|
4525 |
|
4526 *p = p1; |
|
4527 return true; |
|
4528 } /* common_op_match_null_string_p */ |
|
4529 |
|
4530 |
|
4531 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN |
|
4532 bytes; nonzero otherwise. */ |
|
4533 |
|
4534 static int |
|
4535 bcmp_translate (s1, s2, len, translate) |
|
4536 unsigned char *s1, *s2; |
|
4537 register int len; |
|
4538 char *translate; |
|
4539 { |
|
4540 register unsigned char *p1 = s1, *p2 = s2; |
|
4541 while (len) |
|
4542 { |
|
4543 if (translate[*p1++] != translate[*p2++]) return 1; |
|
4544 len--; |
|
4545 } |
|
4546 return 0; |
|
4547 } |
|
4548 |
|
4549 /* Entry points for GNU code. */ |
|
4550 |
|
4551 /* re_compile_pattern is the GNU regular expression compiler: it |
|
4552 compiles PATTERN (of length SIZE) and puts the result in BUFP. |
|
4553 Returns 0 if the pattern was valid, otherwise an error string. |
|
4554 |
|
4555 Assumes the `allocated' (and perhaps `buffer') and `translate' fields |
|
4556 are set in BUFP on entry. |
|
4557 |
|
4558 We call regex_compile to do the actual compilation. */ |
|
4559 |
|
4560 const char * |
|
4561 re_compile_pattern (pattern, length, bufp) |
|
4562 const char *pattern; |
|
4563 int length; |
|
4564 struct re_pattern_buffer *bufp; |
|
4565 { |
|
4566 reg_errcode_t ret; |
|
4567 |
|
4568 /* GNU code is written to assume at least RE_NREGS registers will be set |
|
4569 (and at least one extra will be -1). */ |
|
4570 bufp->regs_allocated = REGS_UNALLOCATED; |
|
4571 |
|
4572 /* And GNU code determines whether or not to get register information |
|
4573 by passing null for the REGS argument to re_match, etc., not by |
|
4574 setting no_sub. */ |
|
4575 bufp->no_sub = 0; |
|
4576 |
|
4577 /* Match anchors at newline. */ |
|
4578 bufp->newline_anchor = 1; |
|
4579 |
|
4580 ret = regex_compile (pattern, length, re_syntax_options, bufp); |
|
4581 |
|
4582 return re_error_msg[(int) ret]; |
|
4583 } |
|
4584 |
|
4585 /* Entry points compatible with 4.2 BSD regex library. We don't define |
|
4586 them if this is an Emacs or POSIX compilation. */ |
|
4587 |
|
4588 #if !defined (emacs) && !defined (_POSIX_SOURCE) |
|
4589 |
|
4590 /* BSD has one and only one pattern buffer. */ |
|
4591 static struct re_pattern_buffer re_comp_buf; |
|
4592 |
|
4593 char * |
|
4594 re_comp (s) |
|
4595 const char *s; |
|
4596 { |
|
4597 reg_errcode_t ret; |
|
4598 |
|
4599 if (!s) |
|
4600 { |
|
4601 if (!re_comp_buf.buffer) |
|
4602 return "No previous regular expression"; |
|
4603 return 0; |
|
4604 } |
|
4605 |
|
4606 if (!re_comp_buf.buffer) |
|
4607 { |
|
4608 re_comp_buf.buffer = (unsigned char *) malloc (200); |
|
4609 if (re_comp_buf.buffer == NULL) |
|
4610 return "Memory exhausted"; |
|
4611 re_comp_buf.allocated = 200; |
|
4612 |
|
4613 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH); |
|
4614 if (re_comp_buf.fastmap == NULL) |
|
4615 return "Memory exhausted"; |
|
4616 } |
|
4617 |
|
4618 /* Since `re_exec' always passes NULL for the `regs' argument, we |
|
4619 don't need to initialize the pattern buffer fields which affect it. */ |
|
4620 |
|
4621 /* Match anchors at newlines. */ |
|
4622 re_comp_buf.newline_anchor = 1; |
|
4623 |
|
4624 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); |
|
4625 |
|
4626 /* Yes, we're discarding `const' here. */ |
|
4627 return (char *) re_error_msg[(int) ret]; |
|
4628 } |
|
4629 |
|
4630 |
|
4631 int |
|
4632 re_exec (s) |
|
4633 const char *s; |
|
4634 { |
|
4635 const int len = strlen (s); |
|
4636 return |
|
4637 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); |
|
4638 } |
|
4639 #endif /* not emacs and not _POSIX_SOURCE */ |
|
4640 |
|
4641 /* POSIX.2 functions. Don't define these for Emacs. */ |
|
4642 |
|
4643 #ifndef emacs |
|
4644 |
|
4645 /* regcomp takes a regular expression as a string and compiles it. |
|
4646 |
|
4647 PREG is a regex_t *. We do not expect any fields to be initialized, |
|
4648 since POSIX says we shouldn't. Thus, we set |
|
4649 |
|
4650 `buffer' to the compiled pattern; |
|
4651 `used' to the length of the compiled pattern; |
|
4652 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the |
|
4653 REG_EXTENDED bit in CFLAGS is set; otherwise, to |
|
4654 RE_SYNTAX_POSIX_BASIC; |
|
4655 `newline_anchor' to REG_NEWLINE being set in CFLAGS; |
|
4656 `fastmap' and `fastmap_accurate' to zero; |
|
4657 `re_nsub' to the number of subexpressions in PATTERN. |
|
4658 |
|
4659 PATTERN is the address of the pattern string. |
|
4660 |
|
4661 CFLAGS is a series of bits which affect compilation. |
|
4662 |
|
4663 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we |
|
4664 use POSIX basic syntax. |
|
4665 |
|
4666 If REG_NEWLINE is set, then . and [^...] don't match newline. |
|
4667 Also, regexec will try a match beginning after every newline. |
|
4668 |
|
4669 If REG_ICASE is set, then we considers upper- and lowercase |
|
4670 versions of letters to be equivalent when matching. |
|
4671 |
|
4672 If REG_NOSUB is set, then when PREG is passed to regexec, that |
|
4673 routine will report only success or failure, and nothing about the |
|
4674 registers. |
|
4675 |
|
4676 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for |
|
4677 the return codes and their meanings.) */ |
|
4678 |
|
4679 int |
|
4680 regcomp (preg, pattern, cflags) |
|
4681 regex_t *preg; |
|
4682 const char *pattern; |
|
4683 int cflags; |
|
4684 { |
|
4685 reg_errcode_t ret; |
|
4686 unsigned syntax |
|
4687 = cflags & REG_EXTENDED ? RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; |
|
4688 |
|
4689 /* regex_compile will allocate the space for the compiled pattern. */ |
|
4690 preg->buffer = 0; |
|
4691 |
|
4692 /* Don't bother to use a fastmap when searching. This simplifies the |
|
4693 REG_NEWLINE case: if we used a fastmap, we'd have to put all the |
|
4694 characters after newlines into the fastmap. This way, we just try |
|
4695 every character. */ |
|
4696 preg->fastmap = 0; |
|
4697 |
|
4698 if (cflags & REG_ICASE) |
|
4699 { |
|
4700 unsigned i; |
|
4701 |
|
4702 preg->translate = (char *) malloc (CHAR_SET_SIZE); |
|
4703 if (preg->translate == NULL) |
|
4704 return (int) REG_ESPACE; |
|
4705 |
|
4706 /* Map uppercase characters to corresponding lowercase ones. */ |
|
4707 for (i = 0; i < CHAR_SET_SIZE; i++) |
|
4708 preg->translate[i] = isupper (i) ? tolower (i) : i; |
|
4709 } |
|
4710 else |
|
4711 preg->translate = NULL; |
|
4712 |
|
4713 /* If REG_NEWLINE is set, newlines are treated differently. */ |
|
4714 if (cflags & REG_NEWLINE) |
|
4715 { /* REG_NEWLINE implies neither . nor [^...] match newline. */ |
|
4716 syntax &= ~RE_DOT_NEWLINE; |
|
4717 syntax |= RE_HAT_LISTS_NOT_NEWLINE; |
|
4718 /* It also changes the matching behavior. */ |
|
4719 preg->newline_anchor = 1; |
|
4720 } |
|
4721 else |
|
4722 preg->newline_anchor = 0; |
|
4723 |
|
4724 preg->no_sub = !!(cflags & REG_NOSUB); |
|
4725 |
|
4726 /* POSIX says a null character in the pattern terminates it, so we |
|
4727 can use strlen here in compiling the pattern. */ |
|
4728 ret = regex_compile (pattern, strlen (pattern), syntax, preg); |
|
4729 |
|
4730 /* POSIX doesn't distinguish between an unmatched open-group and an |
|
4731 unmatched close-group: both are REG_EPAREN. */ |
|
4732 if (ret == REG_ERPAREN) ret = REG_EPAREN; |
|
4733 |
|
4734 return (int) ret; |
|
4735 } |
|
4736 |
|
4737 |
|
4738 /* regexec searches for a given pattern, specified by PREG, in the |
|
4739 string STRING. |
|
4740 |
|
4741 If NMATCH is zero or REG_NOSUB was set in the cflags argument to |
|
4742 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at |
|
4743 least NMATCH elements, and we set them to the offsets of the |
|
4744 corresponding matched substrings. |
|
4745 |
|
4746 EFLAGS specifies `execution flags' which affect matching: if |
|
4747 REG_NOTBOL is set, then ^ does not match at the beginning of the |
|
4748 string; if REG_NOTEOL is set, then $ does not match at the end. |
|
4749 |
|
4750 We return 0 if we find a match and REG_NOMATCH if not. */ |
|
4751 |
|
4752 int |
|
4753 regexec (preg, string, nmatch, pmatch, eflags) |
|
4754 const regex_t *preg; |
|
4755 const char *string; |
|
4756 size_t nmatch; |
|
4757 regmatch_t pmatch[]; |
|
4758 int eflags; |
|
4759 { |
|
4760 int ret; |
|
4761 struct re_registers regs; |
|
4762 regex_t private_preg; |
|
4763 int len = strlen (string); |
|
4764 boolean want_reg_info = !preg->no_sub && nmatch > 0; |
|
4765 |
|
4766 private_preg = *preg; |
|
4767 |
|
4768 private_preg.not_bol = !!(eflags & REG_NOTBOL); |
|
4769 private_preg.not_eol = !!(eflags & REG_NOTEOL); |
|
4770 |
|
4771 /* The user has told us exactly how many registers to return |
|
4772 information about, via `nmatch'. We have to pass that on to the |
|
4773 matching routines. */ |
|
4774 private_preg.regs_allocated = REGS_FIXED; |
|
4775 |
|
4776 if (want_reg_info) |
|
4777 { |
|
4778 regs.num_regs = nmatch; |
|
4779 regs.start = TALLOC (nmatch, regoff_t); |
|
4780 regs.end = TALLOC (nmatch, regoff_t); |
|
4781 if (regs.start == NULL || regs.end == NULL) |
|
4782 return (int) REG_NOMATCH; |
|
4783 } |
|
4784 |
|
4785 /* Perform the searching operation. */ |
|
4786 ret = re_search (&private_preg, string, len, |
|
4787 /* start: */ 0, /* range: */ len, |
|
4788 want_reg_info ? ®s : (struct re_registers *) 0); |
|
4789 |
|
4790 /* Copy the register information to the POSIX structure. */ |
|
4791 if (want_reg_info) |
|
4792 { |
|
4793 if (ret >= 0) |
|
4794 { |
|
4795 unsigned r; |
|
4796 |
|
4797 for (r = 0; r < nmatch; r++) |
|
4798 { |
|
4799 pmatch[r].rm_so = regs.start[r]; |
|
4800 pmatch[r].rm_eo = regs.end[r]; |
|
4801 } |
|
4802 } |
|
4803 |
|
4804 /* If we needed the temporary register info, free the space now. */ |
|
4805 free (regs.start); |
|
4806 free (regs.end); |
|
4807 } |
|
4808 |
|
4809 /* We want zero return to mean success, unlike `re_search'. */ |
|
4810 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; |
|
4811 } |
|
4812 |
|
4813 |
|
4814 /* Returns a message corresponding to an error code, ERRCODE, returned |
|
4815 from either regcomp or regexec. */ |
|
4816 |
|
4817 size_t |
|
4818 regerror (errcode, preg, errbuf, errbuf_size) |
|
4819 int errcode; |
|
4820 const regex_t *preg; |
|
4821 char *errbuf; |
|
4822 size_t errbuf_size; |
|
4823 { |
|
4824 const char *msg |
|
4825 = re_error_msg[errcode] == NULL ? "Success" : re_error_msg[errcode]; |
|
4826 size_t msg_size = strlen (msg) + 1; /* Includes the null. */ |
|
4827 |
|
4828 if (errbuf_size != 0) |
|
4829 { |
|
4830 if (msg_size > errbuf_size) |
|
4831 { |
|
4832 strncpy (errbuf, msg, errbuf_size - 1); |
|
4833 errbuf[errbuf_size - 1] = 0; |
|
4834 } |
|
4835 else |
|
4836 strcpy (errbuf, msg); |
|
4837 } |
|
4838 |
|
4839 return msg_size; |
|
4840 } |
|
4841 |
|
4842 |
|
4843 /* Free dynamically allocated space used by PREG. */ |
|
4844 |
|
4845 void |
|
4846 regfree (preg) |
|
4847 regex_t *preg; |
|
4848 { |
|
4849 if (preg->buffer != NULL) |
|
4850 free (preg->buffer); |
|
4851 preg->buffer = NULL; |
|
4852 |
|
4853 preg->allocated = 0; |
|
4854 preg->used = 0; |
|
4855 |
|
4856 if (preg->fastmap != NULL) |
|
4857 free (preg->fastmap); |
|
4858 preg->fastmap = NULL; |
|
4859 preg->fastmap_accurate = 0; |
|
4860 |
|
4861 if (preg->translate != NULL) |
|
4862 free (preg->translate); |
|
4863 preg->translate = NULL; |
|
4864 } |
|
4865 |
|
4866 #endif /* not emacs */ |
|
4867 |
|
4868 /* |
|
4869 Local variables: |
|
4870 make-backup-files: t |
|
4871 version-control: t |
|
4872 trim-versions-without-asking: nil |
|
4873 End: |
|
4874 */ |