This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
threads-shared 1.03 - Changes, Makefile.PL
[perl5.git] / regexec.c
index 96b84eb..f7f0d2b 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -1637,7 +1637,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
     I32 scream_pos = -1;               /* Internal iterator of scream. */
     char *scream_olds = NULL;
     SV* const oreplsv = GvSV(PL_replgv);
-    const bool do_utf8 = DO_UTF8(sv);
+    const bool do_utf8 = (bool)DO_UTF8(sv);
     I32 multiline;
 
     regmatch_info reginfo;  /* create some info to pass to regtry etc */
@@ -1773,7 +1773,11 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
            }
        }
        goto phooey;
-    } else if (prog->reganch & ROPT_ANCH_GPOS) {
+    } else if (ROPT_GPOS_CHECK == (prog->reganch & ROPT_GPOS_CHECK)) 
+    {
+        /* the warning about reginfo.ganch being used without intialization
+           is bogus -- we set it above, when prog->reganch & ROPT_GPOS_SEEN 
+           and we only enter this block when the same bit is set. */
        if (regtry(&reginfo, reginfo.ganch))
            goto got_it;
        goto phooey;
@@ -2197,19 +2201,16 @@ S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos)
 
 #define sayYES goto yes
 #define sayNO goto no
-#define sayNO_ANYOF goto no_anyof
-#define sayYES_FINAL goto yes_final
-#define sayNO_FINAL  goto no_final
-#define sayNO_SILENT goto do_no
+#define sayNO_SILENT goto no_silent
 #define saySAME(x) if (x) goto yes; else goto no
 
-#define CACHEsayNO STMT_START { \
+/* we dont use STMT_START/END here because it leads to 
+   "unreachable code" warnings, which are bogus, but distracting. */
+#define CACHEsayNO \
     if (st->u.whilem.cache_offset | st->u.whilem.cache_bit) \
        PL_reg_poscache[st->u.whilem.cache_offset] |= \
            (1<<st->u.whilem.cache_bit); \
-    sayNO; \
-} STMT_END
-
+    sayNO
 
 /* this is used to determine how far from the left messages like
    'failed...' are printed. It should be set such that messages 
@@ -2472,7 +2473,7 @@ S_dump_exec_pos(pTHX_ const char *locinput,
                    len1, s1,
                    (docolor ? "" : "> <"),
                    len2, s2,
-                   tlen > 19 ? 0 :  19 - tlen,
+                   (int)(tlen > 19 ? 0 :  19 - tlen),
                    "");
     }
 }
@@ -2715,7 +2716,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 
                while ( state && uc <= (U8*)PL_regeol ) {
                     U32 base = trie->states[ state ].trans.base;
-                    UV uvc;
+                    UV uvc = 0;
                     U16 charid;
                     /* We use charid to hold the wordnum as we don't use it
                        for charid until after we have done the wordnum logic. 
@@ -3019,7 +3020,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                STRLEN inclasslen = PL_regeol - locinput;
 
                if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
-                   sayNO_ANYOF;
+                   goto anyof_fail;
                if (locinput >= PL_regeol)
                    sayNO;
                locinput += inclasslen ? inclasslen : UTF8SKIP(locinput);
@@ -3030,13 +3031,13 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                if (nextchr < 0)
                    nextchr = UCHARAT(locinput);
                if (!REGINCLASS(rex, scan, (U8*)locinput))
-                   sayNO_ANYOF;
+                   goto anyof_fail;
                if (!nextchr && locinput >= PL_regeol)
                    sayNO;
                nextchr = UCHARAT(++locinput);
                break;
            }
-       no_anyof:
+       anyof_fail:
            /* If we might have the case of the German sharp s
             * in a casefolding Unicode character class. */
 
@@ -3389,7 +3390,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 
                        Zero(&pm, 1, PMOP);
                        if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
-                       re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm);
+                       re = CALLREGCOMP((char*)t, (char*)t + len, &pm);
                        if (!(SvFLAGS(ret)
                              & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
                                | SVs_GMG)))
@@ -3434,7 +3435,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                /* NOTREACHED */
            }
            /* /(?(?{...})X|Y)/ */
-           st->sw = SvTRUE(ret);
+           st->sw = (bool)SvTRUE(ret);
            st->logical = 0;
            break;
        }
@@ -3449,7 +3450,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
            cur_curlyx = ST.prev_curlyx;
            /* XXXX This is too dramatic a measure... */
            PL_reg_maxiter = 0;
-           sayYES_FINAL;
+           sayYES;
 
 
        case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */
@@ -3484,7 +3485,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
            break;
        case GROUPP:
            n = ARG(scan);  /* which paren pair */
-           st->sw = ((I32)*PL_reglastparen >= n && PL_regendp[n] != -1);
+           st->sw = (bool)((I32)*PL_reglastparen >= n && PL_regendp[n] != -1);
            break;
        case IFTHEN:
            PL_reg_leftiter = PL_reg_maxiter;           /* Void cache */
@@ -3664,8 +3665,6 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                {
                    st->u.whilem.savecc = cur_curlyx;
                    cur_curlyx = cur_curlyx->u.curlyx.outercc;
-                   if (cur_curlyx)
-                       st->ln = cur_curlyx->u.curlyx.cur;
                    DEBUG_EXECUTE_r(
                        PerlIO_printf(Perl_debug_log,
                           "%*s  empty match detected, try continuation...\n",
@@ -3676,8 +3675,6 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                    cur_curlyx = st->u.whilem.savecc;
                    if (result)
                        sayYES;
-                   if (cur_curlyx->u.curlyx.outercc)
-                       cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln;
                    sayNO;
                }
 
@@ -3748,8 +3745,6 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                if (cur_curlyx->minmod) {
                    st->u.whilem.savecc = cur_curlyx;
                    cur_curlyx = cur_curlyx->u.curlyx.outercc;
-                   if (cur_curlyx)
-                       st->ln = cur_curlyx->u.curlyx.cur;
                    st->u.whilem.cp = regcppush(st->u.whilem.savecc->u.curlyx.parenfloor);
                    REGCP_SET(st->u.whilem.lastcp);
                    REGMATCH(st->u.whilem.savecc->next, WHILEM3);
@@ -3761,8 +3756,6 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                    }
                    REGCP_UNWIND(st->u.whilem.lastcp);
                    regcppop(rex);
-                   if (cur_curlyx->u.curlyx.outercc)
-                       cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln;
 
                    if (n >= cur_curlyx->u.curlyx.max) { /* Maximum greed exceeded? */
                        if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
@@ -3832,15 +3825,11 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                /* Failed deeper matches of scan, so see if this one works. */
                st->u.whilem.savecc = cur_curlyx;
                cur_curlyx = cur_curlyx->u.curlyx.outercc;
-               if (cur_curlyx)
-                   st->ln = cur_curlyx->u.curlyx.cur;
                REGMATCH(st->u.whilem.savecc->next, WHILEM6);
                /*** all unsaved local vars undefined at this point */
                cur_curlyx = st->u.whilem.savecc;
                if (result)
                    sayYES;
-               if (cur_curlyx->u.curlyx.outercc)
-                   cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln;
                cur_curlyx->u.curlyx.cur = n - 1;
                cur_curlyx->u.curlyx.lastloc = st->u.whilem.lastloc;
                CACHEsayNO;
@@ -4369,10 +4358,10 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                                      (long)(locinput - PL_reg_starttry),
                                      (long)(reginfo->till - PL_reg_starttry),
                                      PL_colors[5]));
-               sayNO_FINAL;            /* Cannot match: too short. */
+               sayNO_SILENT;           /* Cannot match: too short. */
            }
            PL_reginput = locinput;     /* put where regtry can find it */
-           sayYES_FINAL;               /* Success! */
+           sayYES;                     /* Success! */
 
        case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */
            DEBUG_EXECUTE_r(
@@ -4380,7 +4369,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
                "%*s  %ssubpattern success...%s\n",
                REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]));
            PL_reginput = locinput;     /* put where regtry can find it */
-           sayYES_FINAL;               /* Success! */
+           sayYES;                     /* Success! */
 
 #undef  ST
 #define ST st->u.ifmatch
@@ -4505,6 +4494,8 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 
            DEBUG_STATE_pp("push");
            depth++;
+           st->u.yes.prev_yes_state = yes_state;
+           yes_state = st;
 
            /* grab the next free state slot */
            st++;
@@ -4525,8 +4516,6 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
        }
     }
 
-
-
     /*
     * We get here only if there's trouble -- normally "case END" is
     * the terminating point.
@@ -4535,8 +4524,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
     /*NOTREACHED*/
     sayNO;
 
-yes_final:
-
+yes:
     if (yes_state) {
        /* we have successfully completed a subexpression, but we must now
         * pop to the state marked by yes_state and continue from there */
@@ -4567,6 +4555,38 @@ yes_final:
        PL_regmatch_state = st;
 
        switch (st->resume_state) {
+       case resume_CURLYX:
+       case resume_WHILEM1:
+       case resume_WHILEM2:
+       case resume_WHILEM3:
+       case resume_WHILEM4:
+       case resume_WHILEM5:
+       case resume_WHILEM6:
+           result = 1;
+           /* restore previous state and re-enter */
+           scan        = st->scan;
+           next        = st->next;
+           n   = st->n;
+           locinput= st->locinput;
+           nextchr = UCHARAT(locinput);
+           switch (st->resume_state) {
+           case resume_CURLYX:
+               goto resume_point_CURLYX;
+           case resume_WHILEM1:
+               goto resume_point_WHILEM1;
+           case resume_WHILEM2:
+               goto resume_point_WHILEM2;
+           case resume_WHILEM3:
+               goto resume_point_WHILEM3;
+           case resume_WHILEM4:
+               goto resume_point_WHILEM4;
+           case resume_WHILEM5:
+               goto resume_point_WHILEM5;
+           case resume_WHILEM6:
+               goto resume_point_WHILEM6;
+           }
+           Perl_croak(aTHX_ "unexpected whilem resume state");
+
        case EVAL_AB: 
        case IFMATCH_A:
        case CURLYM_A:
@@ -4584,59 +4604,8 @@ yes_final:
 
     DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
                          PL_colors[4], PL_colors[5]));
-yes:
 
     result = 1;
-    /* XXX this is duplicate(ish) code to that in the do_no section.
-     * will disappear when REGFMATCH goes */
-    if (depth) {
-       /* restore previous state and re-enter */
-       st--;
-       if (st < SLAB_FIRST(PL_regmatch_slab)) {
-           PL_regmatch_slab = PL_regmatch_slab->prev;
-           st = SLAB_LAST(PL_regmatch_slab);
-       }
-       PL_regmatch_state = st;
-       scan    = st->scan;
-       next    = st->next;
-       n       = st->n;
-       locinput= st->locinput;
-       nextchr = UCHARAT(locinput);
-
-       DEBUG_STATE_pp("pop");
-       depth--;
-
-       switch (st->resume_state) {
-       case resume_CURLYX:
-           goto resume_point_CURLYX;
-       case resume_WHILEM1:
-           goto resume_point_WHILEM1;
-       case resume_WHILEM2:
-           goto resume_point_WHILEM2;
-       case resume_WHILEM3:
-           goto resume_point_WHILEM3;
-       case resume_WHILEM4:
-           goto resume_point_WHILEM4;
-       case resume_WHILEM5:
-           goto resume_point_WHILEM5;
-       case resume_WHILEM6:
-           goto resume_point_WHILEM6;
-
-       case TRIE_next:
-       case CURLYM_A:
-       case CURLYM_B:
-       case EVAL_AB:
-       case IFMATCH_A:
-       case BRANCH_next:
-       case CURLY_B_max:
-       case CURLY_B_min:
-       case CURLY_B_min_known:
-           break;
-
-       default:
-           Perl_croak(aTHX_ "regexp resume memory corruption");
-       }
-    }
     goto final_exit;
 
 no:
@@ -4646,9 +4615,8 @@ no:
             REPORT_CODE_OFF+depth*2, "", 
             PL_colors[4], PL_colors[5])
        );
-no_final:
-do_no:
 
+no_silent:
     result = 0;
 
     if (depth) {
@@ -4667,6 +4635,8 @@ do_no:
 
        DEBUG_STATE_pp("pop");
        depth--;
+       if (yes_state == st)
+           yes_state = st->u.yes.prev_yes_state;
 
        switch (st->resume_state) {
        case resume_CURLYX:
@@ -4693,8 +4663,6 @@ do_no:
        case CURLY_B_max:
        case CURLY_B_min:
        case CURLY_B_min_known:
-           if (yes_state == st)
-               yes_state = st->u.yes.prev_yes_state;
            state_num = st->resume_state + 1; /* failure = success + 1 */
            goto reenter_switch;
 
@@ -4703,7 +4671,7 @@ do_no:
        }
     }
 
-final_exit:
+  final_exit:
 
     /* restore original high-water mark */
     PL_regmatch_slab  = orig_slab;
@@ -4721,7 +4689,6 @@ final_exit:
     }
 
     return result;
-
 }
 
 /*
@@ -5178,6 +5145,11 @@ S_reghop3(U8 *s, I32 off, const U8* lim)
     return s;
 }
 
+#ifdef XXX_dmq
+/* there are a bunch of places where we use two reghop3's that should
+   be replaced with this routine. but since thats not done yet 
+   we ifdef it out - dmq
+*/
 STATIC U8 *
 S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
 {
@@ -5200,7 +5172,7 @@ S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim)
     }
     return s;
 }
-
+#endif
 
 STATIC U8 *
 S_reghopmaybe3(U8* s, I32 off, const U8* lim)