/* * cfindloop - the heart of cfind */ static int cfindloop(struct vars * v, struct cnfa * cnfa, struct colormap * cm, struct dfa * d, struct dfa * s, chr **coldp) /* where to put coldstart pointer */ { chr *begin; chr *end; chr *cold; chr *open; /* open and close of range of possible starts */ chr *close; chr *estart; chr *estop; int er; int shorter = v->g->tree->flags & SHORTER; int hitend; assert(d != NULL && s != NULL); cold = NULL; close = v->search_start; do { MDEBUG(("\ncsearch at %ld\n", LOFF(close))); close = shortest(v, s, close, close, v->stop, &cold, (int *) NULL); if (close == NULL) break; /* NOTE BREAK */ assert(cold != NULL); open = cold; cold = NULL; MDEBUG(("cbetween %ld and %ld\n", LOFF(open), LOFF(close))); for (begin = open; begin <= close; begin++) { MDEBUG(("\ncfind trying at %ld\n", LOFF(begin))); estart = begin; estop = v->stop; for (;;) { if (shorter) end = shortest(v, d, begin, estart, estop, (chr **) NULL, &hitend); else end = longest(v, d, begin, estop, &hitend); if (hitend && cold == NULL) cold = begin; if (end == NULL) break; /* NOTE BREAK OUT */ MDEBUG(("tentative end %ld\n", LOFF(end))); zapsubs(v->pmatch, v->nmatch); zapmem(v, v->g->tree); er = cdissect(v, v->g->tree, begin, end); if (er == REG_OKAY) { if (v->nmatch > 0) { v->pmatch[0].rm_so = OFF(begin); v->pmatch[0].rm_eo = OFF(end); } *coldp = cold; return REG_OKAY; } if (er != REG_NOMATCH) { ERR(er); *coldp = cold; return er; } if ((shorter) ? end == estop : end == begin) { /* no point in trying again */ *coldp = cold; return REG_NOMATCH; } /* go around and try again */ if (shorter) estart = end + 1; else estop = end - 1; } } } while (close < v->stop); *coldp = cold; return REG_NOMATCH; }
/* * pg_regexec - match regular expression */ int pg_regexec(regex_t *re, const chr *string, size_t len, size_t search_start, rm_detail_t *details, size_t nmatch, regmatch_t pmatch[], int flags) { struct vars var; register struct vars *v = &var; int st; size_t n; int backref; #define LOCALMAT 20 regmatch_t mat[LOCALMAT]; #define LOCALMEM 40 regoff_t mem[LOCALMEM]; /* sanity checks */ if (re == NULL || string == NULL || re->re_magic != REMAGIC) return REG_INVARG; if (re->re_csize != sizeof(chr)) return REG_MIXED; /* setup */ v->re = re; v->g = (struct guts *) re->re_guts; if ((v->g->cflags & REG_EXPECT) && details == NULL) return REG_INVARG; if (v->g->info & REG_UIMPOSSIBLE) return REG_NOMATCH; backref = (v->g->info & REG_UBACKREF) ? 1 : 0; v->eflags = flags; if (v->g->cflags & REG_NOSUB) nmatch = 0; /* override client */ v->nmatch = nmatch; if (backref) { /* need work area */ if (v->g->nsub + 1 <= LOCALMAT) v->pmatch = mat; else v->pmatch = (regmatch_t *) MALLOC((v->g->nsub + 1) * sizeof(regmatch_t)); if (v->pmatch == NULL) return REG_ESPACE; v->nmatch = v->g->nsub + 1; } else v->pmatch = pmatch; v->details = details; v->start = (chr *) string; v->search_start = (chr *) string + search_start; v->stop = (chr *) string + len; v->err = 0; if (backref) { /* need retry memory */ assert(v->g->ntree >= 0); n = (size_t) v->g->ntree; if (n <= LOCALMEM) v->mem = mem; else v->mem = (regoff_t *) MALLOC(n * sizeof(regoff_t)); if (v->mem == NULL) { if (v->pmatch != pmatch && v->pmatch != mat) FREE(v->pmatch); return REG_ESPACE; } } else v->mem = NULL; /* do it */ assert(v->g->tree != NULL); if (backref) st = cfind(v, &v->g->tree->cnfa, &v->g->cmap); else st = find(v, &v->g->tree->cnfa, &v->g->cmap); /* copy (portion of) match vector over if necessary */ if (st == REG_OKAY && v->pmatch != pmatch && nmatch > 0) { zapsubs(pmatch, nmatch); n = (nmatch < v->nmatch) ? nmatch : v->nmatch; memcpy(VS(pmatch), VS(v->pmatch), n * sizeof(regmatch_t)); } /* clean up */ if (v->pmatch != pmatch && v->pmatch != mat) FREE(v->pmatch); if (v->mem != NULL && v->mem != mem) FREE(v->mem); return st; }
/* * find - find a match for the main NFA (no-complications case) */ static int find(struct vars * v, struct cnfa * cnfa, struct colormap * cm) { struct dfa *s; struct dfa *d; chr *begin; chr *end = NULL; chr *cold; chr *open; /* open and close of range of possible starts */ chr *close; int hitend; int shorter = (v->g->tree->flags & SHORTER) ? 1 : 0; /* first, a shot with the search RE */ s = newdfa(v, &v->g->search, cm, &v->dfa1); assert(!(ISERR() && s != NULL)); NOERR(); MDEBUG(("\nsearch at %ld\n", LOFF(v->start))); cold = NULL; close = shortest(v, s, v->search_start, v->search_start, v->stop, &cold, (int *) NULL); freedfa(s); NOERR(); if (v->g->cflags & REG_EXPECT) { assert(v->details != NULL); if (cold != NULL) v->details->rm_extend.rm_so = OFF(cold); else v->details->rm_extend.rm_so = OFF(v->stop); v->details->rm_extend.rm_eo = OFF(v->stop); /* unknown */ } if (close == NULL) /* not found */ return REG_NOMATCH; if (v->nmatch == 0) /* found, don't need exact location */ return REG_OKAY; /* find starting point and match */ assert(cold != NULL); open = cold; cold = NULL; MDEBUG(("between %ld and %ld\n", LOFF(open), LOFF(close))); d = newdfa(v, cnfa, cm, &v->dfa1); assert(!(ISERR() && d != NULL)); NOERR(); for (begin = open; begin <= close; begin++) { MDEBUG(("\nfind trying at %ld\n", LOFF(begin))); if (shorter) end = shortest(v, d, begin, begin, v->stop, (chr **) NULL, &hitend); else end = longest(v, d, begin, v->stop, &hitend); NOERR(); if (hitend && cold == NULL) cold = begin; if (end != NULL) break; /* NOTE BREAK OUT */ } assert(end != NULL); /* search RE succeeded so loop should */ freedfa(d); /* and pin down details */ assert(v->nmatch > 0); v->pmatch[0].rm_so = OFF(begin); v->pmatch[0].rm_eo = OFF(end); if (v->g->cflags & REG_EXPECT) { if (cold != NULL) v->details->rm_extend.rm_so = OFF(cold); else v->details->rm_extend.rm_so = OFF(v->stop); v->details->rm_extend.rm_eo = OFF(v->stop); /* unknown */ } if (v->nmatch == 1) /* no need for submatches */ return REG_OKAY; /* submatches */ zapsubs(v->pmatch, v->nmatch); return dissect(v, v->g->tree, begin, end); }
/* - exec - match regular expression ^ int exec(regex_t *, const chr *, size_t, rm_detail_t *, ^ size_t, regmatch_t [], int); */ int exec( regex_t *re, const chr *string, size_t len, rm_detail_t *details, size_t nmatch, regmatch_t pmatch[], int flags) { AllocVars(v); int st; size_t n; int backref; #define LOCALMAT 20 regmatch_t mat[LOCALMAT]; #define LOCALMEM 40 regoff_t mem[LOCALMEM]; /* * Sanity checks. */ if (re == NULL || string == NULL || re->re_magic != REMAGIC) { FreeVars(v); return REG_INVARG; } if (re->re_csize != sizeof(chr)) { FreeVars(v); return REG_MIXED; } /* * Setup. */ v->re = re; v->g = (struct guts *)re->re_guts; if ((v->g->cflags®_EXPECT) && details == NULL) { FreeVars(v); return REG_INVARG; } if (v->g->info®_UIMPOSSIBLE) { FreeVars(v); return REG_NOMATCH; } backref = (v->g->info®_UBACKREF) ? 1 : 0; v->eflags = flags; if (v->g->cflags®_NOSUB) { nmatch = 0; /* override client */ } v->nmatch = nmatch; if (backref) { /* * Need work area. */ if (v->g->nsub + 1 <= LOCALMAT) { v->pmatch = mat; } else { v->pmatch = (regmatch_t *) MALLOC((v->g->nsub + 1) * sizeof(regmatch_t)); } if (v->pmatch == NULL) { FreeVars(v); return REG_ESPACE; } v->nmatch = v->g->nsub + 1; } else { v->pmatch = pmatch; } v->details = details; v->start = (chr *)string; v->stop = (chr *)string + len; v->err = 0; if (backref) { /* * Need retry memory. */ assert(v->g->ntree >= 0); n = (size_t)v->g->ntree; if (n <= LOCALMEM) { v->mem = mem; } else { v->mem = (regoff_t *) MALLOC(n*sizeof(regoff_t)); } if (v->mem == NULL) { if (v->pmatch != pmatch && v->pmatch != mat) { FREE(v->pmatch); } FreeVars(v); return REG_ESPACE; } } else { v->mem = NULL; } /* * Do it. */ assert(v->g->tree != NULL); if (backref) { st = cfind(v, &v->g->tree->cnfa, &v->g->cmap); } else { st = find(v, &v->g->tree->cnfa, &v->g->cmap); } /* * Copy (portion of) match vector over if necessary. */ if (st == REG_OKAY && v->pmatch != pmatch && nmatch > 0) { zapsubs(pmatch, nmatch); n = (nmatch < v->nmatch) ? nmatch : v->nmatch; memcpy(VS(pmatch), VS(v->pmatch), n*sizeof(regmatch_t)); } /* * Clean up. */ if (v->pmatch != pmatch && v->pmatch != mat) { FREE(v->pmatch); } if (v->mem != NULL && v->mem != mem) { FREE(v->mem); } FreeVars(v); return st; }