diff options
Diffstat (limited to 'src/backend/regex/regexec.c')
| -rw-r--r-- | src/backend/regex/regexec.c | 48 |
1 files changed, 36 insertions, 12 deletions
diff --git a/src/backend/regex/regexec.c b/src/backend/regex/regexec.c index aa5ba855149..2411e6561d7 100644 --- a/src/backend/regex/regexec.c +++ b/src/backend/regex/regexec.c @@ -216,14 +216,14 @@ pg_regexec(regex_t *re, if (backref && nmatch <= v->g->nsub) { /* need larger work area */ - if (v->g->nsub + 1 <= LOCALMAT) + v->nmatch = v->g->nsub + 1; + if (v->nmatch <= LOCALMAT) v->pmatch = mat; else - v->pmatch = (regmatch_t *) MALLOC((v->g->nsub + 1) * - sizeof(regmatch_t)); + v->pmatch = (regmatch_t *) MALLOC(v->nmatch * sizeof(regmatch_t)); if (v->pmatch == NULL) return REG_ESPACE; - v->nmatch = v->g->nsub + 1; + zapallsubs(v->pmatch, v->nmatch); } else { @@ -488,7 +488,6 @@ find(struct vars *v, return REG_OKAY; /* find submatches */ - zapallsubs(v->pmatch, v->nmatch); return cdissect(v, v->g->tree, begin, end); } @@ -599,7 +598,6 @@ cfindloop(struct vars *v, break; /* no match with this begin point, try next */ MDEBUG(("tentative end %ld\n", LOFF(end))); /* Dissect the potential match to see if it really matches */ - zapallsubs(v->pmatch, v->nmatch); er = cdissect(v, v->g->tree, begin, end); if (er == REG_OKAY) { @@ -647,6 +645,8 @@ cfindloop(struct vars *v, /* * zapallsubs - initialize all subexpression matches to "no match" + * + * Note that p[0], the overall-match location, is not touched. */ static void zapallsubs(regmatch_t *p, @@ -716,8 +716,30 @@ subset(struct vars *v, * DFA and found that the proposed substring satisfies the DFA. (We make * the caller do that because in concatenation and iteration nodes, it's * much faster to check all the substrings against the child DFAs before we - * recurse.) Also, caller must have cleared subexpression match data via - * zaptreesubs (or zapallsubs at the top level). + * recurse.) + * + * A side-effect of a successful match is to save match locations for + * capturing subexpressions in v->pmatch[]. This is a little bit tricky, + * so we make the following rules: + * 1. Before initial entry to cdissect, all match data must have been + * cleared (this is seen to by zapallsubs). + * 2. Before any recursive entry to cdissect, the match data for that + * subexpression tree must be guaranteed clear (see zaptreesubs). + * 3. When returning REG_OKAY, each level of cdissect will have saved + * any relevant match locations. + * 4. When returning REG_NOMATCH, each level of cdissect will guarantee + * that its subexpression match locations are again clear. + * 5. No guarantees are made for error cases (i.e., other result codes). + * 6. When a level of cdissect abandons a successful sub-match, it will + * clear that subtree's match locations with zaptreesubs before trying + * any new DFA match or cdissect call for that subtree or any subtree + * to its right (that is, any subtree that could have a backref into the + * abandoned match). + * This may seem overly complicated, but it's difficult to simplify it + * because of the provision that match locations must be reset before + * any fresh DFA match (a rule that is needed to make dfa_backref safe). + * That means it won't work to just reset relevant match locations at the + * start of each cdissect level. */ static int /* regexec return code */ cdissect(struct vars *v, @@ -841,6 +863,8 @@ ccondissect(struct vars *v, MDEBUG(("%d: successful\n", t->id)); return REG_OKAY; } + /* Reset left's matches (right should have done so itself) */ + zaptreesubs(v, left); } if (er != REG_NOMATCH) return er; @@ -863,8 +887,6 @@ ccondissect(struct vars *v, return REG_NOMATCH; } MDEBUG(("%d: new midpoint %ld\n", t->id, LOFF(mid))); - zaptreesubs(v, left); - zaptreesubs(v, right); } /* can't get here */ @@ -922,6 +944,8 @@ crevcondissect(struct vars *v, MDEBUG(("%d: successful\n", t->id)); return REG_OKAY; } + /* Reset left's matches (right should have done so itself) */ + zaptreesubs(v, left); } if (er != REG_NOMATCH) return er; @@ -944,8 +968,6 @@ crevcondissect(struct vars *v, return REG_NOMATCH; } MDEBUG(("%d: new midpoint %ld\n", t->id, LOFF(mid))); - zaptreesubs(v, left); - zaptreesubs(v, right); } /* can't get here */ @@ -1214,6 +1236,7 @@ citerdissect(struct vars *v, for (i = nverified + 1; i <= k; i++) { + /* zap any match data from a non-last iteration */ zaptreesubs(v, t->child); er = cdissect(v, t->child, endpts[i - 1], endpts[i]); if (er == REG_OKAY) @@ -1426,6 +1449,7 @@ creviterdissect(struct vars *v, for (i = nverified + 1; i <= k; i++) { + /* zap any match data from a non-last iteration */ zaptreesubs(v, t->child); er = cdissect(v, t->child, endpts[i - 1], endpts[i]); if (er == REG_OKAY) |