2 * UCW Library -- Interface to Regular Expression Libraries
4 * (c) 1997--2004 Martin Mares <mj@ucw.cz>
5 * (c) 2001 Robert Spalek <robert@ucw.cz>
7 * This software may be freely distributed and used according to the terms
8 * of the GNU Lesser General Public License.
12 #include "lib/chartype.h"
13 #include "lib/hashfunc.h"
18 #ifdef CONFIG_POSIX_REGEX
20 /* POSIX regular expression library */
26 regmatch_t matches[10];
30 rx_compile(const char *p, int icase)
32 regex *r = xmalloc_zero(sizeof(regex));
34 int err = regcomp(&r->rx, p, REG_EXTENDED | (icase ? REG_ICASE : 0));
38 regerror(err, &r->rx, msg, sizeof(msg)-1);
39 /* regfree(&r->rx) not needed */
40 die("Error parsing regular expression `%s': %s", p, msg);
53 rx_match(regex *r, const char *s)
55 int err = regexec(&r->rx, s, 10, r->matches, 0);
58 /* regexec doesn't support anchored expressions, so we have to check ourselves that the full string is matched */
59 return !(r->matches[0].rm_so || s[r->matches[0].rm_eo]);
61 else if (err == REG_NOMATCH)
63 else if (err == REG_ESPACE)
64 die("Regex matching ran out of memory");
66 die("Regex matching failed with unknown error %d", err);
70 rx_subst(regex *r, const char *by, const char *src, char *dest, uns destlen)
72 char *end = dest + destlen - 1;
74 if (!rx_match(r, src))
82 if (*by >= '0' && *by <= '9') /* \0 gets replaced by entire pattern */
85 if (j <= r->rx.re_nsub && r->matches[j].rm_so >= 0)
87 const char *s = src + r->matches[j].rm_so;
88 uns i = r->matches[j].rm_eo - r->matches[j].rm_so;
106 #elif defined(CONFIG_PCRE)
115 uns match_array_size;
117 int matches[0]; /* (max_matches+1) pairs (pos,len) plus some workspace */
121 rx_compile(const char *p, int icase)
124 int errpos, match_array_size, eno;
126 pcre *rx = pcre_compile(p, PCRE_ANCHORED | PCRE_EXTRA | (icase ? PCRE_CASELESS : 0), &err, &errpos, NULL);
128 die("Error parsing regular expression `%s': %s at position %d", p, err, errpos);
129 eno = pcre_fullinfo(rx, NULL, PCRE_INFO_CAPTURECOUNT, &match_array_size);
131 die("Internal error: pcre_fullinfo() failed with error %d", eno);
132 match_array_size = 3*(match_array_size+1);
133 regex *r = xmalloc_zero(sizeof(regex) + match_array_size * sizeof(int));
135 r->match_array_size = match_array_size;
136 r->extra = pcre_study(r->rx, 0, &err);
138 die("Error studying regular expression `%s': %s", p, err);
151 rx_match(regex *r, const char *s)
153 int len = str_len(s);
154 int err = pcre_exec(r->rx, r->extra, s, len, 0, 0, r->matches, r->match_array_size);
157 r->real_matches = err;
158 /* need to check that the full string matches */
159 return !(r->matches[0] || s[r->matches[1]]);
161 else if (err == PCRE_ERROR_NOMATCH)
163 else if (err == PCRE_ERROR_NOMEMORY)
164 die("Regex matching ran out of memory");
166 die("Regex matching failed with unknown error %d", err);
170 rx_subst(regex *r, const char *by, const char *src, char *dest, uns destlen)
172 char *end = dest + destlen - 1;
174 if (!rx_match(r, src))
182 if (*by >= '0' && *by <= '9') /* \0 gets replaced by entire pattern */
185 if (j < r->real_matches && r->matches[2*j] >= 0)
187 char *s = src + r->matches[2*j];
188 uns i = r->matches[2*j+1] - r->matches[2*j];
208 /* BSD regular expression library */
210 #ifdef CONFIG_OWN_BSD_REGEX
211 #include "lib/regex/regex-sh.h"
216 #define INITIAL_MEM 1024 /* Initial space allocated for each pattern */
217 #define CHAR_SET_SIZE 256 /* How many characters in the character set. */
220 struct re_pattern_buffer buf;
221 struct re_registers regs; /* Must not change between re_match() calls */
226 rx_compile(const char *p, int icase)
228 regex *r = xmalloc_zero(sizeof(regex));
231 r->buf.buffer = xmalloc(INITIAL_MEM);
232 r->buf.allocated = INITIAL_MEM;
236 r->buf.translate = xmalloc (CHAR_SET_SIZE);
237 /* Map uppercase characters to corresponding lowercase ones. */
238 for (i = 0; i < CHAR_SET_SIZE; i++)
239 r->buf.translate[i] = Cupcase(i);
242 r->buf.translate = NULL;
243 re_set_syntax(RE_SYNTAX_POSIX_EXTENDED);
244 msg = re_compile_pattern(p, strlen(p), &r->buf);
247 die("Error parsing pattern `%s': %s", p, msg);
253 xfree(r->buf.buffer);
254 if (r->buf.translate)
255 xfree(r->buf.translate);
260 rx_match(regex *r, const char *s)
265 if (re_match(&r->buf, s, len, 0, &r->regs) < 0)
267 if (r->regs.start[0] || r->regs.end[0] != len) /* XXX: Why regex doesn't enforce implicit "^...$" ? */
273 rx_subst(regex *r, const char *by, const char *src, char *dest, uns destlen)
275 char *end = dest + destlen - 1;
277 if (!rx_match(r, src))
285 if (*by >= '0' && *by <= '9') /* \0 gets replaced by entire pattern */
288 if (j < r->regs.num_regs)
290 const char *s = src + r->regs.start[j];
291 uns i = r->regs.end[j] - r->regs.start[j];
292 if (r->regs.start[j] > r->len_cache || r->regs.end[j] > r->len_cache)
315 int main(int argc, char **argv)
318 char buf1[4096], buf2[4096];
321 if (!strcmp(argv[1], "-i"))
327 r = rx_compile(argv[1], opt_i);
328 while (fgets(buf1, sizeof(buf1), stdin))
330 char *p = strchr(buf1, '\n');
335 if (rx_match(r, buf1))
342 int i = rx_subst(r, argv[2], buf1, buf2, sizeof(buf2));