Chicken out: the hash randomisation is not on by default.
[p5sagit/p5-mst-13.2.git] / regexec.c
index 4cf8069..db3cecb 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -67,7 +67,8 @@
  *
  ****    Alterations to Henry's code are...
  ****
- ****    Copyright (c) 1991-2002, Larry Wall
+ ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+ ****    2000, 2001, 2002, 2003, by Larry Wall and others
  ****
  ****    You may distribute under the terms of either the GNU General Public
  ****    License or the Artistic License, as specified in the README file.
@@ -171,7 +172,7 @@ S_regcppush(pTHX_ I32 parenfloor)
        Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
 
 #define REGCP_OTHER_ELEMS 6
-    SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS);
+    SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
     for (p = PL_regsize; p > parenfloor; p--) {
 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
        SSPUSHINT(PL_regendp[p]);
@@ -400,6 +401,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
     char *i_strpos = strpos;
     SV *dsv = PERL_DEBUG_PAD_ZERO(0);
 #endif
+    RX_MATCH_UTF8_set(prog,do_utf8);
 
     if (prog->reganch & ROPT_UTF8) {
        DEBUG_r(PerlIO_printf(Perl_debug_log,
@@ -541,6 +543,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
                            start_shift + (s - strbeg), end_shift, pp, 0);
        else
            goto fail_finish;
+       /* we may be pointing at the wrong string */
+       if (s && RX_MATCH_COPIED(prog))
+           s = strbeg + (s - SvPVX(sv));
        if (data)
            *data->scream_olds = s;
     }
@@ -848,10 +853,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
        char *startpos = strbeg;
 
        t = s;
-       if (prog->reganch & ROPT_UTF8) {        
-           PL_regdata = prog->data;
-           PL_bostr = startpos;
-       }
+       cache_re(prog);
        s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1);
        if (!s) {
 #ifdef DEBUGGING
@@ -959,25 +961,40 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
        /* We know what class it must start with. */
        switch (OP(c)) {
        case ANYOF:
-           while (s < strend) {
-               STRLEN skip = do_utf8 ? UTF8SKIP(s) : 1;
-                 
-               if (do_utf8 ?
-                   reginclass(c, (U8*)s, 0, do_utf8) :
-                   REGINCLASS(c, (U8*)s) ||
-                   (ANYOF_FOLD_SHARP_S(c, s, strend) &&
-                    /* The assignment of 2 is intentional:
-                     * for the sharp s, the skip is 2. */
-                    (skip = SHARP_S_SKIP)
-                    )) {
-                   if (tmp && (norun || regtry(prog, s)))
-                       goto got_it;
-                   else
-                       tmp = doevery;
-               }
-               else 
-                   tmp = 1;
-               s += skip;
+           if (do_utf8) {
+                while (s < strend) {
+                     if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
+                         !UTF8_IS_INVARIANT((U8)s[0]) ?
+                         reginclass(c, (U8*)s, 0, do_utf8) :
+                         REGINCLASS(c, (U8*)s)) {
+                          if (tmp && (norun || regtry(prog, s)))
+                               goto got_it;
+                          else
+                               tmp = doevery;
+                     }
+                     else 
+                          tmp = 1;
+                     s += UTF8SKIP(s);
+                }
+           }
+           else {
+                while (s < strend) {
+                     STRLEN skip = 1;
+
+                     if (REGINCLASS(c, (U8*)s) ||
+                         (ANYOF_FOLD_SHARP_S(c, s, strend) &&
+                          /* The assignment of 2 is intentional:
+                           * for the folded sharp s, the skip is 2. */
+                          (skip = SHARP_S_SKIP))) {
+                          if (tmp && (norun || regtry(prog, s)))
+                               goto got_it;
+                          else
+                               tmp = doevery;
+                     }
+                     else 
+                          tmp = 1;
+                     s += skip;
+                }
            }
            break;
        case CANY:
@@ -1597,6 +1614,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
     SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
     SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
 #endif
+    RX_MATCH_UTF8_set(prog,do_utf8);
 
     PL_regcc = 0;
 
@@ -1843,6 +1861,9 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
                 : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend),
                                  (unsigned char*)strend, must,
                                  PL_multiline ? FBMrf_MULTILINE : 0))) ) {
+           /* we may be pointing at the wrong string */
+           if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
+               s = strbeg + (s - SvPVX(sv));
            DEBUG_r( did_match = 1 );
            if (HOPc(s, -back_max) > last1) {
                last1 = HOPc(s, -back_min);
@@ -1929,6 +1950,9 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
                                   end_shift, &scream_pos, 1); /* last one */
                if (!last)
                    last = scream_olds; /* Only one occurrence. */
+               /* we may be pointing at the wrong string */
+               else if (RX_MATCH_COPIED(prog))
+                   s = strbeg + (s - SvPVX(sv));
            }
            else {
                STRLEN len;
@@ -1996,17 +2020,28 @@ got_it:
 
     /* make sure $`, $&, $', and $digit will work later */
     if ( !(flags & REXEC_NOT_FIRST) ) {
-       if (RX_MATCH_COPIED(prog)) {
-           Safefree(prog->subbeg);
-           RX_MATCH_COPIED_off(prog);
-       }
+       RX_MATCH_COPY_FREE(prog);
        if (flags & REXEC_COPY_STR) {
            I32 i = PL_regeol - startpos + (stringarg - strbeg);
-
-           s = savepvn(strbeg, i);
-           prog->subbeg = s;
+#ifdef PERL_COPY_ON_WRITE
+           if ((SvIsCOW(sv)
+                || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
+               if (DEBUG_C_TEST) {
+                   PerlIO_printf(Perl_debug_log,
+                                 "Copy on write: regexp capture, type %d\n",
+                                 (int) SvTYPE(sv));
+               }
+               prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
+               prog->subbeg = SvPVX(prog->saved_copy);
+               assert (SvPOKp(prog->saved_copy));
+           } else
+#endif
+           {
+               RX_MATCH_COPIED_on(prog);
+               s = savepvn(strbeg, i);
+               prog->subbeg = s;
+           }
            prog->sublen = i;
-           RX_MATCH_COPIED_on(prog);
        }
        else {
            prog->subbeg = strbeg;
@@ -2096,6 +2131,9 @@ S_regtry(pTHX_ regexp *prog, char *startpos)
                $` inside (?{}) could fail... */
            PL_reg_oldsaved = prog->subbeg;
            PL_reg_oldsavedlen = prog->sublen;
+#ifdef PERL_COPY_ON_WRITE
+           PL_nrs = prog->saved_copy;
+#endif
            RX_MATCH_COPIED_off(prog);
        }
        else
@@ -2797,6 +2835,7 @@ S_regmatch(pTHX_ regnode *prog)
            COP *ocurcop = PL_curcop;
            PAD *old_comppad;
            SV *ret;
+           struct regexp *oreg = PL_reg_re;
        
            n = ARG(scan);
            PL_op = (OP_4tree*)PL_regdata->data[n];
@@ -2826,13 +2865,17 @@ S_regmatch(pTHX_ regnode *prog)
                    re_cc_state state;
                    CHECKPOINT cp, lastcp;
                     int toggleutf;
+                   register SV *sv;
 
-                   if(SvROK(ret) || SvRMAGICAL(ret)) {
-                       SV *sv = SvROK(ret) ? SvRV(ret) : ret;
-
-                       if(SvMAGICAL(sv))
-                           mg = mg_find(sv, PERL_MAGIC_qr);
+                   if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
+                       mg = mg_find(sv, PERL_MAGIC_qr);
+                   else if (SvSMAGICAL(ret)) {
+                       if (SvGMAGICAL(ret))
+                           sv_unmagic(ret, PERL_MAGIC_qr);
+                       else
+                           mg = mg_find(ret, PERL_MAGIC_qr);
                    }
+
                    if (mg) {
                        re = (regexp *)mg->mg_obj;
                        (void)ReREFCNT_inc(re);
@@ -2849,7 +2892,8 @@ S_regmatch(pTHX_ regnode *prog)
                         if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
                        re = CALLREGCOMP(aTHX_ t, t + len, &pm);
                        if (!(SvFLAGS(ret)
-                             & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
+                             & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
+                               | SVs_GMG)))
                            sv_magic(ret,(SV*)ReREFCNT_inc(re),
                                        PERL_MAGIC_qr,0,0);
                        PL_regprecomp = oprecomp;
@@ -2924,8 +2968,10 @@ S_regmatch(pTHX_ regnode *prog)
                sw = SvTRUE(ret);
                logical = 0;
            }
-           else
+           else {
                sv_setsv(save_scalar(PL_replgv), ret);
+               cache_re(oreg);
+           }
            break;
        }
        case OPEN:
@@ -4528,6 +4574,9 @@ restore_pos(pTHX_ void *arg)
        if (PL_reg_oldsaved) {
            PL_reg_re->subbeg = PL_reg_oldsaved;
            PL_reg_re->sublen = PL_reg_oldsavedlen;
+#ifdef PERL_COPY_ON_WRITE
+           PL_reg_re->saved_copy = PL_nrs;
+#endif
            RX_MATCH_COPIED_on(PL_reg_re);
        }
        PL_reg_magic->mg_len = PL_reg_oldpos;