--- code/trunk/pcreposix.c 2007/02/24 21:38:53 29 +++ code/tags/pcre-3.0/pcreposix.c 2007/02/24 21:39:23 44 @@ -12,7 +12,7 @@ Written by: Philip Hazel - Copyright (c) 1997-1999 University of Cambridge + Copyright (c) 1997-2000 University of Cambridge ----------------------------------------------------------------------------- Permission is granted to anyone to use this software for any purpose on any @@ -46,7 +46,8 @@ static const char *estring[] = { ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9, ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19, ERR20, - ERR21, ERR22, ERR23, ERR24, ERR25 }; + ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR29, ERR29, ERR30, + ERR31 }; static int eint[] = { REG_EESCAPE, /* "\\ at end of pattern" */ @@ -76,7 +77,10 @@ REG_BADPAT, /* "lookbehind assertion is not fixed length" */ REG_BADPAT, /* "malformed number after (?(" */ REG_BADPAT, /* "conditional group containe more than two branches" */ - REG_BADPAT /* "assertion expected after (?(" */ + REG_BADPAT, /* "assertion expected after (?(" */ + REG_BADPAT, /* "(?p must be followed by )" */ + REG_ECTYPE, /* "unknown POSIX class name" */ + REG_BADPAT /* "POSIX collating elements are not supported" */ }; /* Table of texts corresponding to POSIX error codes */ @@ -211,39 +215,61 @@ * Match a regular expression * *************************************************/ +/* Unfortunately, PCRE requires 3 ints of working space for each captured +substring, so we have to get and release working store instead of just using +the POSIX structures as was done in earlier releases when PCRE needed only 2 +ints. */ + int regexec(regex_t *preg, const char *string, size_t nmatch, regmatch_t pmatch[], int eflags) { int rc; int options = 0; +int *ovector = NULL; if ((eflags & REG_NOTBOL) != 0) options |= PCRE_NOTBOL; if ((eflags & REG_NOTEOL) != 0) options |= PCRE_NOTEOL; preg->re_erroffset = (size_t)(-1); /* Only has meaning after compile */ -rc = pcre_exec(preg->re_pcre, NULL, string, (int)strlen(string), options, - (int *)pmatch, nmatch * 2); +if (nmatch > 0) + { + ovector = (int *)malloc(sizeof(int) * nmatch * 3); + if (ovector == NULL) return REG_ESPACE; + } + +rc = pcre_exec(preg->re_pcre, NULL, string, (int)strlen(string), 0, options, + ovector, nmatch * 3); -if (rc == 0) return 0; /* All pmatch were filled in */ +if (rc == 0) rc = nmatch; /* All captured slots were filled in */ -if (rc > 0) +if (rc >= 0) { size_t i; - for (i = rc; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1; + for (i = 0; i < rc; i++) + { + pmatch[i].rm_so = ovector[i*2]; + pmatch[i].rm_eo = ovector[i*2+1]; + } + if (ovector != NULL) free(ovector); + for (; i < nmatch; i++) pmatch[i].rm_so = pmatch[i].rm_eo = -1; return 0; } -else switch(rc) +else { - case PCRE_ERROR_NOMATCH: return REG_NOMATCH; - case PCRE_ERROR_NULL: return REG_INVARG; - case PCRE_ERROR_BADOPTION: return REG_INVARG; - case PCRE_ERROR_BADMAGIC: return REG_INVARG; - case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT; - case PCRE_ERROR_NOMEMORY: return REG_ESPACE; - default: return REG_ASSERT; + if (ovector != NULL) free(ovector); + switch(rc) + { + case PCRE_ERROR_NOMATCH: return REG_NOMATCH; + case PCRE_ERROR_NULL: return REG_INVARG; + case PCRE_ERROR_BADOPTION: return REG_INVARG; + case PCRE_ERROR_BADMAGIC: return REG_INVARG; + case PCRE_ERROR_UNKNOWN_NODE: return REG_ASSERT; + case PCRE_ERROR_NOMEMORY: return REG_ESPACE; + default: return REG_ASSERT; + } } }