latest switch/say/~~

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

diff --git a/regexec.c b/regexec.c
index 6a644d9..13cc68b 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -518,6 +518,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
                     || ((slen = SvCUR(check)) > 1
                         && memNE(SvPVX_const(check), s, slen)))
                goto report_neq;
+           check_at = s;
            goto success_at_start;
          }
        }
@@ -1039,14 +1040,15 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32
                U8 *sm = (U8 *) m;
                U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
                U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
+               const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
 
                to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
                to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
 
                c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, 
-                                   0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+                                   0, uniflags);
                c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
-                                   0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+                                   0, uniflags);
                lnc = 0;
                while (sm < ((U8 *) m + ln)) {
                    lnc++;
@@ -1085,14 +1087,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32
                UV c, f;
                U8 tmpbuf [UTF8_MAXBYTES+1];
                STRLEN len, foldlen;
-               
+               const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
                if (c1 == c2) {
                    /* Upper and lower of 1st char are equal -
                     * probably not a "letter". */
                    while (s <= e) {
                        c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
-                                          ckWARN(WARN_UTF8) ?
-                                          0 : UTF8_ALLOW_ANY);
+                                          uniflags);
                        if ( c == c1
                             && (ln == len ||
                                 ibcmp_utf8(s, (char **)0, 0,  do_utf8,
@@ -1119,8 +1120,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32
                else {
                    while (s <= e) {
                      c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
-                                          ckWARN(WARN_UTF8) ?
-                                          0 : UTF8_ALLOW_ANY);
+                                          uniflags);
 
                        /* Handle some of the three Greek sigmas cases.
                         * Note that not all the possible combinations
@@ -2408,7 +2408,7 @@ S_regmatch(pTHX_ regnode *prog)
 #if 0
     I32 firstcp = PL_savestack_ix;
 #endif
-    const register bool do_utf8 = PL_reg_match_utf8;
+    register const bool do_utf8 = PL_reg_match_utf8;
 #ifdef DEBUGGING
     SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
     SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
@@ -2416,6 +2416,7 @@ S_regmatch(pTHX_ regnode *prog)
 
     SV *re_debug_flags = NULL;
 #endif
+    U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
 
     GET_RE_DEBUG_FLAGS;
 
@@ -2583,8 +2584,6 @@ S_regmatch(pTHX_ regnode *prog)
        case TRIEF:
        case TRIEFL:
            {
-
-               const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
                U8 *uc = ( U8* )locinput;
                U32 state = 1;
                U16 charid = 0;
@@ -2652,7 +2651,6 @@ S_regmatch(pTHX_ regnode *prog)
               from previous if blocks */
        case TRIE:
            {
-               const U32 uniflags = ckWARN( WARN_UTF8 ) ? 0 : UTF8_ALLOW_ANY;
                U8 *uc = (U8*)locinput;
                U32 state = 1;
                U16 charid = 0;
@@ -2808,8 +2806,7 @@ S_regmatch(pTHX_ regnode *prog)
                             sayNO;
                        if (NATIVE_TO_UNI(*(U8*)s) !=
                            utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
-                                          ckWARN(WARN_UTF8) ?
-                                          0 : UTF8_ALLOW_ANY))
+                                           uniflags))
                             sayNO;
                        l += ulen;
                        s ++;
@@ -2823,8 +2820,7 @@ S_regmatch(pTHX_ regnode *prog)
                            sayNO;
                        if (NATIVE_TO_UNI(*((U8*)l)) !=
                            utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
-                                          ckWARN(WARN_UTF8) ?
-                                          0 : UTF8_ALLOW_ANY))
+                                          uniflags))
                            sayNO;
                        s += ulen;
                        l ++;
@@ -3941,16 +3937,13 @@ S_regmatch(pTHX_ regnode *prog)
                             to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
 
                             c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
-                                                ckWARN(WARN_UTF8) ?
-                                                0 : UTF8_ALLOW_ANY);
+                                                uniflags);
                             c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
-                                                ckWARN(WARN_UTF8) ?
-                                                0 : UTF8_ALLOW_ANY);
+                                                uniflags);
                        }
                        else {
                            c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
-                                                    ckWARN(WARN_UTF8) ?
-                                                    0 : UTF8_ALLOW_ANY);
+                                                    uniflags);
                        }
                    }
                }
@@ -4011,8 +4004,7 @@ S_regmatch(pTHX_ regnode *prog)
                                while (locinput <= e &&
                                       utf8n_to_uvchr((U8*)locinput,
                                                      UTF8_MAXBYTES, &len,
-                                                     ckWARN(WARN_UTF8) ?
-                                                     0 : UTF8_ALLOW_ANY) != (UV)c1) {
+                                                     uniflags) != (UV)c1) {
                                    locinput += len;
                                    count++;
                                }
@@ -4023,8 +4015,7 @@ S_regmatch(pTHX_ regnode *prog)
                                while (locinput <= e) {
                                    UV c = utf8n_to_uvchr((U8*)locinput,
                                                          UTF8_MAXBYTES, &len,
-                                                         ckWARN(WARN_UTF8) ?
-                                                         0 : UTF8_ALLOW_ANY);
+                                                         uniflags);
                                    if (c == (UV)c1 || c == (UV)c2)
                                        break;
                                    locinput += len;
@@ -4060,8 +4051,7 @@ S_regmatch(pTHX_ regnode *prog)
                        if (do_utf8)
                            c = utf8n_to_uvchr((U8*)PL_reginput,
                                               UTF8_MAXBYTES, 0,
-                                              ckWARN(WARN_UTF8) ?
-                                              0 : UTF8_ALLOW_ANY);
+                                              uniflags);
                        else
                            c = UCHARAT(PL_reginput);
                        /* If it could work, try it. */
@@ -4110,8 +4100,7 @@ S_regmatch(pTHX_ regnode *prog)
                            if (do_utf8)
                                c = utf8n_to_uvchr((U8*)PL_reginput,
                                                   UTF8_MAXBYTES, 0,
-                                                  ckWARN(WARN_UTF8) ?
-                                                  0 : UTF8_ALLOW_ANY);
+                                                  uniflags);
                            else
                                c = UCHARAT(PL_reginput);
                        }
@@ -4133,8 +4122,7 @@ S_regmatch(pTHX_ regnode *prog)
                            if (do_utf8)
                                c = utf8n_to_uvchr((U8*)PL_reginput,
                                                   UTF8_MAXBYTES, 0,
-                                                  ckWARN(WARN_UTF8) ?
-                                                  0 : UTF8_ALLOW_ANY);
+                                                  uniflags);
                            else
                                c = UCHARAT(PL_reginput);
                        }
@@ -4673,14 +4661,14 @@ Perl_regclass_swash(pTHX_ register const regnode* node, bool doinit, SV** listsv
        if (PL_regdata->what[n] == 's') {
            SV * const rv = (SV*)PL_regdata->data[n];
            AV * const av = (AV*)SvRV((SV*)rv);
-           SV **ary = AvARRAY(av);
+           SV **const ary = AvARRAY(av);
            SV **a, **b;
        
-           /* See the end of regcomp.c:S_reglass() for
+           /* See the end of regcomp.c:S_regclass() for
             * documentation of these array elements. */
 
            si = *ary;
-           a  = SvTYPE(ary[1]) == SVt_RV   ? &ary[1] : 0;
+           a  = SvROK(ary[1]) ? &ary[1] : 0;
            b  = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0;
 
            if (a)
@@ -4722,9 +4710,13 @@ S_reginclass(pTHX_ register const regnode *n, register const U8* p, STRLEN* lenp
     STRLEN len = 0;
     STRLEN plen;
 
-    if (do_utf8 && !UTF8_IS_INVARIANT(c))
-        c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
-                           ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+    if (do_utf8 && !UTF8_IS_INVARIANT(c)) {
+       c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
+                           ckWARN(WARN_UTF8) ? UTF8_CHECK_ONLY :
+                                       UTF8_ALLOW_ANYUV|UTF8_CHECK_ONLY);
+       if (len == (STRLEN)-1)
+           Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
+    }
 
     plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c));
     if (do_utf8 || (flags & ANYOF_UNICODE)) {