int num_code_blocks; /* size of code_blocks[] */
int code_index; /* next code_blocks[] slot */
SSize_t maxlen; /* mininum possible number of chars in string to match */
-#if ADD_TO_REGEXEC
+#ifdef ADD_TO_REGEXEC
char *starttry; /* -Dr: where regtry was called. */
#define RExC_starttry (pRExC_state->starttry)
#endif
} scan_frame;
-#define SCAN_COMMIT(s, data, m) scan_commit(s, data, m, is_inf)
-
STATIC SSize_t
S_study_chunk(pTHX_ RExC_state_t *pRExC_state, regnode **scanp,
SSize_t *minlenp, SSize_t *deltap,
regnode_ssc accum;
regnode * const startbranch=scan;
- if (flags & SCF_DO_SUBSTR)
- SCAN_COMMIT(pRExC_state, data, minlenp); /* Cannot merge
- strings after
- this. */
+ if (flags & SCF_DO_SUBSTR) {
+ /* Cannot merge strings after this. */
+ scan_commit(pRExC_state, data, minlenp, is_inf);
+ }
+
if (flags & SCF_DO_STCLASS)
ssc_init_zero(pRExC_state, &accum);
/* some form of infinite recursion, assume infinite length
* */
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->longest = &(data->longest_float);
}
is_inf = is_inf_internal = 1;
/* Search for fixed substrings supports EXACT only. */
if (flags & SCF_DO_SUBSTR) {
assert(data);
- SCAN_COMMIT(pRExC_state, data, minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
}
if (UTF) {
const U8 * const s = (U8 *)STRING(scan);
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);
+ 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)
next_is_eval = (OP(scan) == EVAL);
do_curly:
if (flags & SCF_DO_SUBSTR) {
- if (mincount == 0) SCAN_COMMIT(pRExC_state,data,minlenp);
+ if (mincount == 0)
+ scan_commit(pRExC_state, data, minlenp, is_inf);
/* Cannot extend fixed substrings */
pos_before = data->pos_min;
}
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
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,
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 {
if (mincount != maxcount) {
/* Cannot extend fixed substrings found inside
the group. */
- SCAN_COMMIT(pRExC_state,data,minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
if (mincount && last_str) {
SV * const sv = data->last_found;
MAGIC * const mg = SvUTF8(sv) && SvMAGICAL(sv) ?
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);
}
case REF:
case CLUMP:
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect
- anything... */
+ /* Cannot expect anything... */
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->longest = &(data->longest_float);
}
is_inf = is_inf_internal = 1;
min++;
delta++; /* Because of the 2 char string cr-lf */
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect
- anything... */
+ /* Cannot expect anything... */
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->pos_min += 1;
data->pos_delta += 1;
data->longest = &(data->longest_float);
else if (REGNODE_SIMPLE(OP(scan))) {
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->pos_min++;
}
min++;
data->flags |= (OP(scan) == MEOL
? SF_BEFORE_MEOL
: SF_BEFORE_SEOL);
- SCAN_COMMIT(pRExC_state, data, minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
}
else if ( PL_regkind[OP(scan)] == BRANCHJ
if ((flags & SCF_DO_SUBSTR) && data->last_found) {
f |= SCF_DO_SUBSTR;
if (scan->flags)
- SCAN_COMMIT(pRExC_state, &data_fake,minlenp);
+ scan_commit(pRExC_state, &data_fake, minlenp, is_inf);
data_fake.last_found=newSVsv(data->last_found);
}
}
if ((flags & SCF_DO_SUBSTR) && data_fake.last_found) {
if (RExC_rx->minlen<*minnextp)
RExC_rx->minlen=*minnextp;
- SCAN_COMMIT(pRExC_state, &data_fake, minnextp);
+ scan_commit(pRExC_state, &data_fake, minnextp, is_inf);
SvREFCNT_dec_NN(data_fake.last_found);
if ( data_fake.minlen_fixed != minlenp )
}
else if ( PL_regkind[OP(scan)] == ENDLIKE ) {
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
flags &= ~SCF_DO_SUBSTR;
}
if (data && OP(scan)==ACCEPT) {
else if (OP(scan) == LOGICAL && scan->flags == 2) /* Embedded follows */
{
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp);
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->longest = &(data->longest_float);
}
is_inf = is_inf_internal = 1;
SSize_t max1 = 0, min1 = SSize_t_MAX;
regnode_ssc accum;
- if (flags & SCF_DO_SUBSTR) /* XXXX Add !SUSPEND? */
- SCAN_COMMIT(pRExC_state, data,minlenp); /* Cannot merge strings
- after this. */
+ if (flags & SCF_DO_SUBSTR) { /* XXXX Add !SUSPEND? */
+ /* Cannot merge strings after this. */
+ scan_commit(pRExC_state, data, minlenp, is_inf);
+ }
if (flags & SCF_DO_STCLASS)
ssc_init_zero(pRExC_state, &accum);
delta += (trie->maxlen - trie->minlen);
flags &= ~SCF_DO_STCLASS; /* xxx */
if (flags & SCF_DO_SUBSTR) {
- SCAN_COMMIT(pRExC_state,data,minlenp); /* Cannot expect
- anything... */
+ /* Cannot expect anything... */
+ scan_commit(pRExC_state, data, minlenp, is_inf);
data->pos_min += trie->minlen;
data->pos_delta += (trie->maxlen - trie->minlen);
if (trie->maxlen != trie->minlen)
/* 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;
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;
assert(s >= rx->subbeg);
assert((STRLEN)rx->sublen >= (STRLEN)((s - rx->subbeg) + i) );
if (i >= 0) {
-#if NO_TAINT_SUPPORT
+#ifdef NO_TAINT_SUPPORT
sv_setpvn(sv, s, i);
#else
const int oldtainted = TAINT_get;
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':
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 '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;
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':
https://perl5.git.perl.org / perl5.git / blobdiff