Re: [ID 20001023.003] PATCH perlfaq5 [perl-current]

[p5sagit/p5-mst-13.2.git] / regexec.c

diff --git a/regexec.c b/regexec.c
index 6401710..350f432 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -325,6 +325,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
     register I32 end_shift;
     register char *s;
     register SV *check;
+    char *strbeg;
     char *t;
     I32 ml_anch;
     char *tmp;
@@ -351,23 +352,25 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
        DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n"));
        goto fail;
     }
+    strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
     check = prog->check_substr;
     if (prog->reganch & ROPT_ANCH) {   /* Match at beg-of-str or after \n */
        ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
                     || ( (prog->reganch & ROPT_ANCH_BOL)
                          && !PL_multiline ) ); /* Check after \n? */
 
-       if ((prog->check_offset_min == prog->check_offset_max) && !ml_anch) {
+       if (!ml_anch) {
+         if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */
+              /* SvCUR is not set on references: SvRV and SvPVX overlap */
+              && sv && !SvROK(sv)
+              && (strpos != strbeg)) {
+             DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
+             goto fail;
+         }
+         if (prog->check_offset_min == prog->check_offset_max) {
            /* Substring at constant offset from beg-of-str... */
            I32 slen;
 
-           if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */
-                /* SvCUR is not set on references: SvRV and SvPVX overlap */
-                && sv && !SvROK(sv)
-                && (strpos + SvCUR(sv) != strend)) {
-               DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
-               goto fail;
-           }
            PL_regeol = strend;                 /* Used in HOP() */
            s = HOPc(strpos, prog->check_offset_min);
            if (SvTAIL(check)) {
@@ -393,6 +396,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
                         && memNE(SvPVX(check), s, slen)))
                goto report_neq;
            goto success_at_start;
+         }
        }
        /* Match is anchored, but substr is not anchored wrt beg-of-str. */
        s = strpos;
@@ -426,7 +430,6 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
     /* Find a possible match in the region s..strend by looking for
        the "check" substring in the region corrected by start/end_shift. */
     if (flags & REXEC_SCREAM) {
-       char *strbeg = SvPVX(sv);       /* XXXX Assume PV_force() on SCREAM! */
        I32 p = -1;                     /* Internal iterator of scream. */
        I32 *pp = data ? data->scream_pos : &p;
 
@@ -668,7 +671,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
        /* Even in this situation we may use MBOL flag if strpos is offset
           wrt the start of the string. */
        if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
-           && (strpos + SvCUR(sv) != strend) && strpos[-1] != '\n'
+           && (strpos != strbeg) && strpos[-1] != '\n'
            /* May be due to an implicit anchor of m{.*foo}  */
            && !(prog->reganch & ROPT_IMPLICIT))
        {
@@ -719,7 +722,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
                ? s + (prog->minlen? cl_l : 0)
                : (prog->float_substr ? check_at - start_shift + cl_l
                                      : strend) ;
-       char *startpos = sv && SvPOK(sv) ? strend - SvCUR(sv) : s;
+       char *startpos = strbeg;
 
        t = s;
        if (prog->reganch & ROPT_UTF8) {        
@@ -914,7 +917,9 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
            PL_reg_flags |= RF_tainted;
            /* FALL THROUGH */
        case BOUNDUTF8:
-           tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : '\n';
+           tmp = (I32)(s != startpos) ? utf8_to_uv_chk(reghop((U8*)s, -1),
+                                                       strend - s,
+                                                       0, 0) : '\n';
            tmp = ((OP(c) == BOUNDUTF8 ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
            while (s < strend) {
                if (tmp == !(OP(c) == BOUNDUTF8 ?
@@ -950,7 +955,9 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
            PL_reg_flags |= RF_tainted;
            /* FALL THROUGH */
        case NBOUNDUTF8:
-           tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : '\n';
+           tmp = (I32)(s != startpos) ? utf8_to_uv_chk(reghop((U8*)s, -1),
+                                                       strend - s,
+                                                       0, 0) : '\n';
            tmp = ((OP(c) == NBOUNDUTF8 ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
            while (s < strend) {
                if (tmp == !(OP(c) == NBOUNDUTF8 ?
@@ -1995,7 +2002,7 @@ S_regmatch(pTHX_ regnode *prog)
                while (s < e) {
                    if (l >= PL_regeol)
                        sayNO;
-                   if (utf8_to_uv((U8*)s, 0) != (c1 ?
+                   if (utf8_to_uv_chk((U8*)s, e - s, 0, 0) != (c1 ?
                                                  toLOWER_utf8((U8*)l) :
                                                  toLOWER_LC_utf8((U8*)l)))
                    {
@@ -2133,7 +2140,8 @@ S_regmatch(pTHX_ regnode *prog)
        case NBOUNDUTF8:
            /* was last char in word? */
            ln = (locinput != PL_regbol)
-               ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
+               ? utf8_to_uv_chk(reghop((U8*)locinput, -1),
+                                PL_regeol - locinput, 0, 0) : PL_regprev;
            if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
                ln = isALNUM_uni(ln);
                n = swash_fetch(PL_utf8_alnum, (U8*)locinput);