annotate lib/readtokens.c @ 375:4d5f9b2487d2

.
author Jim Meyering <jim@meyering.net>
date Fri, 27 Jan 1995 05:37:03 +0000
parents
children c30155ecefad
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
375
Jim Meyering <jim@meyering.net>
parents:
diff changeset
1 /* readtokens.c -- Functions for reading tokens from an input stream.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
2 Copyright (C) 1990-1991 Jim Meyering.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
3
Jim Meyering <jim@meyering.net>
parents:
diff changeset
4 This program is free software; you can redistribute it and/or modify
Jim Meyering <jim@meyering.net>
parents:
diff changeset
5 it under the terms of the GNU General Public License as published by
Jim Meyering <jim@meyering.net>
parents:
diff changeset
6 the Free Software Foundation; either version 2, or (at your option)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
7 any later version.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
8
Jim Meyering <jim@meyering.net>
parents:
diff changeset
9 This program is distributed in the hope that it will be useful,
Jim Meyering <jim@meyering.net>
parents:
diff changeset
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
Jim Meyering <jim@meyering.net>
parents:
diff changeset
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
Jim Meyering <jim@meyering.net>
parents:
diff changeset
12 GNU General Public License for more details.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
13
Jim Meyering <jim@meyering.net>
parents:
diff changeset
14 You should have received a copy of the GNU General Public License
Jim Meyering <jim@meyering.net>
parents:
diff changeset
15 along with this program; if not, write to the Free Software
Jim Meyering <jim@meyering.net>
parents:
diff changeset
16 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
17
Jim Meyering <jim@meyering.net>
parents:
diff changeset
18 Written by Jim Meyering. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
19
Jim Meyering <jim@meyering.net>
parents:
diff changeset
20 /* This almost supercedes xreadline stuff -- using delim="\n"
Jim Meyering <jim@meyering.net>
parents:
diff changeset
21 gives the same functionality, except that these functions
Jim Meyering <jim@meyering.net>
parents:
diff changeset
22 would never return empty lines.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
23
Jim Meyering <jim@meyering.net>
parents:
diff changeset
24 To Do:
Jim Meyering <jim@meyering.net>
parents:
diff changeset
25 - To allow '\0' as a delimiter, I will have to change
Jim Meyering <jim@meyering.net>
parents:
diff changeset
26 interfaces to permit specification of delimiter-string
Jim Meyering <jim@meyering.net>
parents:
diff changeset
27 length.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
28 */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
29
Jim Meyering <jim@meyering.net>
parents:
diff changeset
30 #ifdef HAVE_CONFIG_H
Jim Meyering <jim@meyering.net>
parents:
diff changeset
31 #include <config.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
32 #endif
Jim Meyering <jim@meyering.net>
parents:
diff changeset
33
Jim Meyering <jim@meyering.net>
parents:
diff changeset
34 #include <stdio.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
35
Jim Meyering <jim@meyering.net>
parents:
diff changeset
36 #ifdef STDC_HEADERS
Jim Meyering <jim@meyering.net>
parents:
diff changeset
37 #include <stdlib.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
38 #endif
Jim Meyering <jim@meyering.net>
parents:
diff changeset
39
Jim Meyering <jim@meyering.net>
parents:
diff changeset
40 #if defined (STDC_HEADERS) || defined(HAVE_STRING_H)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
41 #include <string.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
42 /* An ANSI string.h and pre-ANSI memory.h might conflict. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
43 #if !defined (STDC_HEADERS) && defined (HAVE_MEMORY_H)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
44 #include <memory.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
45 #endif /* not STDC_HEADERS and HAVE_MEMORY_H */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
46 #else /* not STDC_HEADERS and not HAVE_STRING_H */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
47 #include <strings.h>
Jim Meyering <jim@meyering.net>
parents:
diff changeset
48 /* memory.h and strings.h conflict on some systems. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
49 #endif /* not STDC_HEADERS and not HAVE_STRING_H */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
50
Jim Meyering <jim@meyering.net>
parents:
diff changeset
51 #include "readtokens.h"
Jim Meyering <jim@meyering.net>
parents:
diff changeset
52 #include "xalloc.h"
Jim Meyering <jim@meyering.net>
parents:
diff changeset
53
Jim Meyering <jim@meyering.net>
parents:
diff changeset
54 #define STREQ(a,b) ((a) == (b) || ((a) && (b) && *(a) == *(b) \
Jim Meyering <jim@meyering.net>
parents:
diff changeset
55 && strcmp(a, b) == 0))
Jim Meyering <jim@meyering.net>
parents:
diff changeset
56
Jim Meyering <jim@meyering.net>
parents:
diff changeset
57 /* Initialize a tokenbuffer. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
58
Jim Meyering <jim@meyering.net>
parents:
diff changeset
59 void
Jim Meyering <jim@meyering.net>
parents:
diff changeset
60 init_tokenbuffer (token_buffer *tokenbuffer)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
61 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
62 tokenbuffer->size = INITIAL_TOKEN_LENGTH;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
63 tokenbuffer->buffer = ((char *) xmalloc (INITIAL_TOKEN_LENGTH));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
64 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
65
Jim Meyering <jim@meyering.net>
parents:
diff changeset
66 /* Read a token from `stream' into `tokenbuffer'.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
67 Upon return, the token is in tokenbuffer->buffer and
Jim Meyering <jim@meyering.net>
parents:
diff changeset
68 has a trailing '\0' instead of the original delimiter.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
69 The function value is the length of the token not including
Jim Meyering <jim@meyering.net>
parents:
diff changeset
70 the final '\0'. When EOF is reached (i.e. on the call
Jim Meyering <jim@meyering.net>
parents:
diff changeset
71 after the last token is read), -1 is returned and tokenbuffer
Jim Meyering <jim@meyering.net>
parents:
diff changeset
72 isn't modified.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
73
Jim Meyering <jim@meyering.net>
parents:
diff changeset
74 This function will work properly on lines containing NUL bytes
Jim Meyering <jim@meyering.net>
parents:
diff changeset
75 and on files that aren't newline-terminated. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
76
Jim Meyering <jim@meyering.net>
parents:
diff changeset
77 long
Jim Meyering <jim@meyering.net>
parents:
diff changeset
78 readtoken (FILE *stream, const char *delim, int n_delim,
Jim Meyering <jim@meyering.net>
parents:
diff changeset
79 token_buffer *tokenbuffer)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
80 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
81 char *p;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
82 int c, i, n;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
83 static const char *saved_delim = NULL;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
84 static char isdelim[256];
Jim Meyering <jim@meyering.net>
parents:
diff changeset
85 int same_delimiters;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
86
Jim Meyering <jim@meyering.net>
parents:
diff changeset
87 if (delim == NULL && saved_delim == NULL)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
88 abort ();
Jim Meyering <jim@meyering.net>
parents:
diff changeset
89
Jim Meyering <jim@meyering.net>
parents:
diff changeset
90 same_delimiters = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
91 if (delim != saved_delim && saved_delim != NULL)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
92 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
93 same_delimiters = 1;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
94 for (i = 0; i < n_delim; i++)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
95 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
96 if (delim[i] != saved_delim[i])
Jim Meyering <jim@meyering.net>
parents:
diff changeset
97 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
98 same_delimiters = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
99 break;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
100 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
101 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
102 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
103
Jim Meyering <jim@meyering.net>
parents:
diff changeset
104 if (!same_delimiters)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
105 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
106 const char *t;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
107 saved_delim = delim;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
108 for (i = 0; i < sizeof (isdelim); i++)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
109 isdelim[i] = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
110 for (t = delim; *t; t++)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
111 isdelim[(unsigned int) *t] = 1;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
112 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
113
Jim Meyering <jim@meyering.net>
parents:
diff changeset
114 p = tokenbuffer->buffer;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
115 n = tokenbuffer->size;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
116 i = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
117
Jim Meyering <jim@meyering.net>
parents:
diff changeset
118 /* FIXME: don't fool with this caching BS. Use strchr instead. */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
119 /* skip over any leading delimiters */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
120 for (c = getc (stream); c >= 0 && isdelim[c]; c = getc (stream))
Jim Meyering <jim@meyering.net>
parents:
diff changeset
121 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
122 /* empty */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
123 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
124
Jim Meyering <jim@meyering.net>
parents:
diff changeset
125 for (;;)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
126 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
127 if (i >= n)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
128 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
129 n = 3 * (n / 2 + 1);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
130 p = xrealloc (p, (unsigned int) n);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
131 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
132 if (c < 0)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
133 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
134 if (i == 0)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
135 return (-1);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
136 p[i] = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
137 break;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
138 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
139 if (isdelim[c])
Jim Meyering <jim@meyering.net>
parents:
diff changeset
140 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
141 p[i] = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
142 break;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
143 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
144 p[i++] = c;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
145 c = getc (stream);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
146 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
147
Jim Meyering <jim@meyering.net>
parents:
diff changeset
148 tokenbuffer->buffer = p;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
149 tokenbuffer->size = n;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
150 return (i);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
151 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
152
Jim Meyering <jim@meyering.net>
parents:
diff changeset
153 /* Return a NULL-terminated array of pointers to tokens
Jim Meyering <jim@meyering.net>
parents:
diff changeset
154 read from `stream.' The number of tokens is returned
Jim Meyering <jim@meyering.net>
parents:
diff changeset
155 as the value of the function.
Jim Meyering <jim@meyering.net>
parents:
diff changeset
156 All storage is obtained through calls to malloc();
Jim Meyering <jim@meyering.net>
parents:
diff changeset
157
Jim Meyering <jim@meyering.net>
parents:
diff changeset
158 %%% Question: is it worth it to do a single
Jim Meyering <jim@meyering.net>
parents:
diff changeset
159 %%% realloc() of `tokens' just before returning? */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
160
Jim Meyering <jim@meyering.net>
parents:
diff changeset
161 int
Jim Meyering <jim@meyering.net>
parents:
diff changeset
162 readtokens (FILE *stream, int projected_n_tokens,
Jim Meyering <jim@meyering.net>
parents:
diff changeset
163 const char *delim, int n_delim,
Jim Meyering <jim@meyering.net>
parents:
diff changeset
164 char ***tokens_out, long **token_lengths)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
165 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
166 token_buffer tb, *token = &tb;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
167 int token_length;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
168 char **tokens;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
169 long *lengths;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
170 int sz;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
171 int n_tokens;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
172
Jim Meyering <jim@meyering.net>
parents:
diff changeset
173 n_tokens = 0;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
174 if (projected_n_tokens > 0)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
175 projected_n_tokens++; /* add one for trailing NULL pointer */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
176 else
Jim Meyering <jim@meyering.net>
parents:
diff changeset
177 projected_n_tokens = 64;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
178 sz = projected_n_tokens;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
179 tokens = (char **) xmalloc (sz * sizeof (char *));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
180 lengths = (long *) xmalloc (sz * sizeof (long));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
181
Jim Meyering <jim@meyering.net>
parents:
diff changeset
182 init_tokenbuffer (token);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
183 for (;;)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
184 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
185 char *tmp;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
186 token_length = readtoken (stream, delim, n_delim, token);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
187 if (n_tokens >= sz)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
188 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
189 sz *= 2;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
190 tokens = (char **) xrealloc (tokens, sz * sizeof (char *));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
191 lengths = (long *) xrealloc (lengths, sz * sizeof (long));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
192 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
193
Jim Meyering <jim@meyering.net>
parents:
diff changeset
194 if (token_length < 0)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
195 {
Jim Meyering <jim@meyering.net>
parents:
diff changeset
196 /* don't increment n_tokens for NULL entry */
Jim Meyering <jim@meyering.net>
parents:
diff changeset
197 tokens[n_tokens] = NULL;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
198 lengths[n_tokens] = -1;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
199 break;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
200 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
201 tmp = (char *) xmalloc ((token_length + 1) * sizeof (char));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
202 lengths[n_tokens] = token_length;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
203 tokens[n_tokens] = strncpy (tmp, token->buffer,
Jim Meyering <jim@meyering.net>
parents:
diff changeset
204 (unsigned) (token_length + 1));
Jim Meyering <jim@meyering.net>
parents:
diff changeset
205 n_tokens++;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
206 }
Jim Meyering <jim@meyering.net>
parents:
diff changeset
207
Jim Meyering <jim@meyering.net>
parents:
diff changeset
208 free (token->buffer);
Jim Meyering <jim@meyering.net>
parents:
diff changeset
209 *tokens_out = tokens;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
210 if (token_lengths != NULL)
Jim Meyering <jim@meyering.net>
parents:
diff changeset
211 *token_lengths = lengths;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
212 return n_tokens;
Jim Meyering <jim@meyering.net>
parents:
diff changeset
213 }