locale.c: Remove vars unused on some platforms

[perl5.git] / regcomp.c

diff --git a/regcomp.c b/regcomp.c
index 1761811..a82171a 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -4388,13 +4388,13 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    scan = NEXTOPER(scan);
                    goto do_curly;
                }
-               is_inf = is_inf_internal = 1;
-               scan = regnext(scan);
                if (flags & SCF_DO_SUBSTR) {
                     scan_commit(pRExC_state, data, minlenp, is_inf);
                     /* Cannot extend fixed substrings */
                    data->longest = &(data->longest_float);
                }
+                is_inf = is_inf_internal = 1;
+                scan = regnext(scan);
                goto optimize_curly_tail;
            case CURLY:
                if (stopparen>0 && (OP(scan)==CURLYN || OP(scan)==CURLYM)
@@ -4630,6 +4630,7 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                    pars++;
                if (flags & SCF_DO_SUBSTR) {
                    SV *last_str = NULL;
+                    STRLEN last_chrs = 0;
                    int counted = mincount != 0;
 
                     if (data->last_end > 0 && mincount != 0) { /* Ends with a
@@ -4645,9 +4646,12 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                        l -= old;
                        /* Get the added string: */
                        last_str = newSVpvn_utf8(s  + old, l, UTF);
+                        last_chrs = UTF ? utf8_length((U8*)(s + old),
+                                            (U8*)(s + old + l)) : l;
                        if (deltanext == 0 && pos_before == b) {
                            /* What was added is a constant string */
                            if (mincount > 1) {
+
                                SvGROW(last_str, (mincount * l) + 1);
                                repeatcpy(SvPVX(last_str) + l,
                                          SvPVX_const(last_str), l,
@@ -4663,8 +4667,9 @@ S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
                                        SvUTF8(sv) && SvMAGICAL(sv) ?
                                        mg_find(sv, PERL_MAGIC_utf8) : NULL;
                                    if (mg && mg->mg_len >= 0)
-                                       mg->mg_len += CHR_SVLEN(last_str) - l;
+                                       mg->mg_len += last_chrs * (mincount-1);
                                }
+                                last_chrs *= mincount;
                                data->last_end += l * (mincount - 1);
                            }
                        } else {
@@ -4706,12 +4711,10 @@ PerlIO_printf(Perl_debug_log, "LHS=%"UVdf" RHS=%"UVdf"\n",
                                mg->mg_len = -1;
                            sv_setsv(sv, last_str);
                            data->last_end = data->pos_min;
-                           data->last_start_min =
-                               data->pos_min - CHR_SVLEN(last_str);
+                           data->last_start_min = data->pos_min - last_chrs;
                            data->last_start_max = is_inf
                                ? SSize_t_MAX
-                               : data->pos_min + data->pos_delta
-                               - CHR_SVLEN(last_str);
+                               : data->pos_min + data->pos_delta - last_chrs;
                        }
                        data->longest = &(data->longest_float);
                    }
@@ -6950,14 +6953,16 @@ reStudy:
         /* A temporary algorithm prefers floated substr to fixed one to dig
          * more info. */
        if (longest_fixed_length > longest_float_length) {
+           r->substrs->check_ix = 0;
            r->check_end_shift = r->anchored_end_shift;
            r->check_substr = r->anchored_substr;
            r->check_utf8 = r->anchored_utf8;
            r->check_offset_min = r->check_offset_max = r->anchored_offset;
-            if (r->intflags & PREGf_ANCH_SINGLE)
+            if (r->intflags & (PREGf_ANCH_SBOL|PREGf_ANCH_GPOS))
                 r->intflags |= PREGf_NOSCAN;
        }
        else {
+           r->substrs->check_ix = 1;
            r->check_end_shift = r->float_end_shift;
            r->check_substr = r->float_substr;
            r->check_utf8 = r->float_utf8;
@@ -6969,6 +6974,8 @@ reStudy:
            if (SvTAIL(r->check_substr ? r->check_substr : r->check_utf8))
                r->extflags |= RXf_INTUIT_TAIL;
        }
+        r->substrs->data[0].max_offset = r->substrs->data[0].min_offset;
+
        /* XXX Unneeded? dmq (shouldn't as this is handled elsewhere)
        if ( (STRLEN)minlen < longest_float_length )
             minlen= longest_float_length;
@@ -11277,7 +11284,8 @@ tryagain:
            FLAGS(ret) = get_regex_charset(RExC_flags);
            *flagp |= SIMPLE;
            if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
-               ckWARNdep(RExC_parse, "\"\\b{\" is deprecated; use \"\\b\\{\" or \"\\b[{]\" instead");
+                /* diag_listed_as: Use "%s" instead of "%s" */
+               vFAIL("Use \"\\b\\{\" instead of \"\\b{\"");
            }
            goto finish_meta_pat;
        case 'B':
@@ -11291,7 +11299,8 @@ tryagain:
            FLAGS(ret) = get_regex_charset(RExC_flags);
            *flagp |= SIMPLE;
            if (! SIZE_ONLY && (U8) *(RExC_parse + 1) == '{') {
-               ckWARNdep(RExC_parse, "\"\\B{\" is deprecated; use \"\\B\\{\" or \"\\B[{]\" instead");
+                /* diag_listed_as: Use "%s" instead of "%s" */
+               vFAIL("Use \"\\B\\{\" instead of \"\\B{\"");
            }
            goto finish_meta_pat;
 
@@ -11785,7 +11794,7 @@ tryagain:
                        }
                    case 'c':
                        p++;
-                       ender = grok_bslash_c(*p++, UTF, SIZE_ONLY);
+                       ender = grok_bslash_c(*p++, SIZE_ONLY);
                        break;
                     case '8': case '9': /* must be a backreference */
                         --p;
@@ -13528,7 +13537,7 @@ parseit:
                    goto recode_encoding;
                break;
            case 'c':
-               value = grok_bslash_c(*RExC_parse++, UTF, SIZE_ONLY);
+               value = grok_bslash_c(*RExC_parse++, SIZE_ONLY);
                break;
            case '0': case '1': case '2': case '3': case '4':
            case '5': case '6': case '7':