RExC_naughty += RExC_naughty / (exp) + (add)
#define ISMULT1(c) ((c) == '*' || (c) == '+' || (c) == '?')
-#define ISMULT2(s) ((*s) == '*' || (*s) == '+' || (*s) == '?' || \
- ((*s) == '{' && regcurly(s)))
+#define ISMULT2(s) (ISMULT1(*s) || ((*s) == '{' && regcurly(s)))
/*
* Flags to be passed up and down.
*/
-#define WORST 0 /* Worst case. */
#define HASWIDTH 0x01 /* Known to not match null strings, could match
non-null ones. */
-
-/* Simple enough to be STAR/PLUS operand; in an EXACTish node must be a single
- * character. (There needs to be a case: in the switch statement in regexec.c
- * for any node marked SIMPLE.) Note that this is not the same thing as
- * REGNODE_SIMPLE */
-#define SIMPLE 0x02
-#define SPSTART 0x04 /* Starts with * or + */
+#define SIMPLE 0x02 /* Exactly one character wide */
+ /* (or LNBREAK as a special case) */
#define POSTPONED 0x08 /* (?1),(?&name), (??{...}) or similar */
#define TRYAGAIN 0x10 /* Weeded out a declaration. */
#define RESTART_PARSE 0x20 /* Need to redo the parse */
* returned list must, and will, contain every code point that is a
* possibility. */
- dVAR;
SV* invlist = NULL;
SV* only_utf8_locale_invlist = NULL;
unsigned int i;
STATIC void
S_rck_elide_nothing(pTHX_ regnode *node)
{
- dVAR;
-
PERL_ARGS_ASSERT_RCK_ELIDE_NOTHING;
if (OP(node) != CURLYX) {
/* recursed: which subroutines have we recursed into */
/* and_withp: Valid if flags & SCF_DO_STCLASS_OR */
{
- dVAR;
SSize_t final_minlen;
/* There must be at least this number of characters to match */
SSize_t min = 0;
* might result in a minlen of 1 and not of 4,
* but this doesn't make us mismatch, just try a bit
* harder than we should.
- * */
+ *
+ * However we must assume this GOSUB is infinite, to
+ * avoid wrongly applying other optimizations in the
+ * enclosing scope - see GH 18096, for example.
+ */
+ is_inf = is_inf_internal = 1;
scan= regnext(scan);
continue;
}
}
if (flags & SCF_DO_SUBSTR)
data->pos_min++;
+ /* This will bypass the formal 'min += minnext * mincount'
+ * calculation in the do_curly path, so assumes min width
+ * of the PLUS payload is exactly one. */
min++;
/* FALLTHROUGH */
case STAR:
PERL_ARGS_ASSERT_SET_REGEX_PV;
/* make sure PL_bitcount bounds not exceeded */
- assert(sizeof(STD_PAT_MODS) <= 8);
+ STATIC_ASSERT_STMT(sizeof(STD_PAT_MODS) <= 8);
p = sv_grow(MUTABLE_SV(Rx), wraplen + 1); /* +1 for the ending NUL */
SvPOK_on(Rx);
OP *expr, const regexp_engine* eng, REGEXP *old_re,
bool *is_bare_re, const U32 orig_rx_flags, const U32 pm_flags)
{
- dVAR;
REGEXP *Rx; /* Capital 'R' means points to a REGEXP */
STRLEN plen;
char *exp;
/* We have that number in RExC_npar */
RExC_total_parens = RExC_npar;
-
- /* XXX For backporting, use long jumps if there is any possibility of
- * overflow */
- if (RExC_size > U16_MAX && ! RExC_use_BRANCHJ) {
- RExC_use_BRANCHJ = TRUE;
- flags |= RESTART_PARSE;
- }
}
else if (! MUST_RESTART(flags)) {
ReREFCNT_dec(Rx);
} else if (flags & RXapif_ONE) {
ret = CALLREG_NAMED_BUFF_ALL(r, (flags | RXapif_REGNAMES));
av = MUTABLE_AV(SvRV(ret));
- length = av_tindex(av);
+ length = av_count(av);
SvREFCNT_dec_NN(ret);
- return newSViv(length + 1);
+ return newSViv(length);
} else {
Perl_croak(aTHX_ "panic: Unknown flags %d in named_buff_scalar",
(int)flags);
STATIC SV*
S_make_exactf_invlist(pTHX_ RExC_state_t *pRExC_state, regnode *node)
{
- dVAR;
const U8 * s = (U8*)STRING(node);
SSize_t bytelen = STR_LEN(node);
UV uc;
vFAIL("Too many nested open parens");
}
- *flagp = 0; /* Tentatively. */
+ *flagp = 0; /* Initialize. */
if (RExC_in_lookbehind) {
RExC_in_lookbehind++;
}
else if (paren != '?') /* Not Conditional */
ret = br;
- *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
+ *flagp |= flags & (HASWIDTH | POSTPONED);
lastbr = br;
while (*RExC_parse == '|') {
if (RExC_use_BRANCHJ) {
REQUIRE_BRANCHJ(flagp, 0);
}
lastbr = br;
- *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
+ *flagp |= flags & (HASWIDTH | POSTPONED);
}
if (have_branch || paren != ':') {
}
}
- *flagp = WORST; /* Tentatively. */
+ *flagp = 0; /* Initialize. */
skip_to_be_ignored_text(pRExC_state, &RExC_parse,
FALSE /* Don't force to /x */ );
else if (ret == 0)
ret = latest;
*flagp |= flags&(HASWIDTH|POSTPONED);
- if (chain == 0) /* First piece. */
- *flagp |= flags&SPSTART;
- else {
+ if (chain != 0) {
/* FIXME adding one for every branch after the first is probably
* excessive now we have TRIE support. (hv) */
MARK_NAUGHTY(1);
FAIL2("panic: regatom returned failure, flags=%#" UVxf, (UV) flags);
}
+ if (! ISMULT2(RExC_parse)) {
+ *flagp = flags;
+ return(ret);
+ }
+
+ /* Here we know the input is a legal quantifier, including {m,n} */
+
op = *RExC_parse;
- if (op == '{' && regcurly(RExC_parse)) {
- maxpos = NULL;
#ifdef RE_TRACK_PATTERN_OFFSETS
- parse_start = RExC_parse; /* MJD */
+ parse_start = RExC_parse;
#endif
- next = RExC_parse + 1;
- while (isDIGIT(*next) || *next == ',') {
- if (*next == ',') {
- if (maxpos)
- break;
- else
- maxpos = next;
- }
- next++;
- }
- if (*next == '}') { /* got one */
- const char* endptr;
- if (!maxpos)
- maxpos = next;
- RExC_parse++;
- if (isDIGIT(*RExC_parse)) {
- endptr = RExC_end;
- if (!grok_atoUV(RExC_parse, &uv, &endptr))
- vFAIL("Invalid quantifier in {,}");
- if (uv >= REG_INFTY)
- vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
- min = (I32)uv;
- } else {
- min = 0;
- }
- if (*maxpos == ',')
- maxpos++;
- else
- maxpos = RExC_parse;
- if (isDIGIT(*maxpos)) {
- endptr = RExC_end;
- if (!grok_atoUV(maxpos, &uv, &endptr))
- vFAIL("Invalid quantifier in {,}");
- if (uv >= REG_INFTY)
- vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
- max = (I32)uv;
- } else {
- max = REG_INFTY; /* meaning "infinity" */
- }
- RExC_parse = next;
- nextchar(pRExC_state);
- if (max < min) { /* If can't match, warn and optimize to fail
- unconditionally */
- reginsert(pRExC_state, OPFAIL, orig_emit, depth+1);
- ckWARNreg(RExC_parse, "Quantifier {n,m} with n > m can't match");
- NEXT_OFF(REGNODE_p(orig_emit)) =
- regarglen[OPFAIL] + NODE_STEP_REGNODE;
- return ret;
- }
- else if (min == max && *RExC_parse == '?')
- {
- ckWARN2reg(RExC_parse + 1,
- "Useless use of greediness modifier '%c'",
- *RExC_parse);
- }
- do_curly:
- if ((flags&SIMPLE)) {
- if (min == 0 && max == REG_INFTY) {
-
- /* Going from 0..inf is currently forbidden in wildcard
- * subpatterns. The only reason is to make it harder to
- * write patterns that take a long long time to halt, and
- * because the use of this construct isn't necessary in
- * matching Unicode property values */
- if (RExC_pm_flags & PMf_WILDCARD) {
- RExC_parse++;
- /* diag_listed_as: Use of %s is not allowed in Unicode
- property wildcard subpatterns in regex; marked by
- <-- HERE in m/%s/ */
- vFAIL("Use of quantifier '*' is not allowed in"
- " Unicode property wildcard subpatterns");
- /* Note, don't need to worry about {0,}, as a '}' isn't
- * legal at all in wildcards, so wouldn't get this far
- * */
- }
- reginsert(pRExC_state, STAR, ret, depth+1);
- MARK_NAUGHTY(4);
- RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
- goto nest_check;
- }
- if (min == 1 && max == REG_INFTY) {
- reginsert(pRExC_state, PLUS, ret, depth+1);
- MARK_NAUGHTY(3);
- RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
- goto nest_check;
- }
- MARK_NAUGHTY_EXP(2, 2);
- reginsert(pRExC_state, CURLY, ret, depth+1);
- Set_Node_Offset(REGNODE_p(ret), parse_start+1); /* MJD */
- Set_Node_Cur_Length(REGNODE_p(ret), parse_start);
- }
- else {
- const regnode_offset w = reg_node(pRExC_state, WHILEM);
+ if (op != '{') {
+ nextchar(pRExC_state);
- FLAGS(REGNODE_p(w)) = 0;
- if (! REGTAIL(pRExC_state, ret, w)) {
- REQUIRE_BRANCHJ(flagp, 0);
- }
- if (RExC_use_BRANCHJ) {
- reginsert(pRExC_state, LONGJMP, ret, depth+1);
- reginsert(pRExC_state, NOTHING, ret, depth+1);
- NEXT_OFF(REGNODE_p(ret)) = 3; /* Go over LONGJMP. */
- }
- reginsert(pRExC_state, CURLYX, ret, depth+1);
- /* MJD hk */
- Set_Node_Offset(REGNODE_p(ret), parse_start+1);
- Set_Node_Length(REGNODE_p(ret),
- op == '{' ? (RExC_parse - parse_start) : 1);
+ *flagp = HASWIDTH;
- if (RExC_use_BRANCHJ)
- NEXT_OFF(REGNODE_p(ret)) = 3; /* Go over NOTHING to
- LONGJMP. */
- if (! REGTAIL(pRExC_state, ret, reg_node(pRExC_state,
- NOTHING)))
- {
- REQUIRE_BRANCHJ(flagp, 0);
- }
- RExC_whilem_seen++;
- MARK_NAUGHTY_EXP(1, 4); /* compound interest */
- }
- FLAGS(REGNODE_p(ret)) = 0;
-
- if (min > 0)
- *flagp = WORST;
- if (max > 0)
- *flagp |= HASWIDTH;
- ARG1_SET(REGNODE_p(ret), (U16)min);
- ARG2_SET(REGNODE_p(ret), (U16)max);
- if (max == REG_INFTY)
- RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
-
- goto nest_check;
- }
+ if (op == '*') {
+ min = 0;
+ }
+ else if (op == '+') {
+ min = 1;
+ }
+ else if (op == '?') {
+ min = 0; max = 1;
+ }
}
+ else { /* is '{' */
+ const char* endptr;
- if (!ISMULT1(op)) {
- *flagp = flags;
- return(ret);
- }
+ maxpos = NULL;
+ next = RExC_parse + 1;
+ while (isDIGIT(*next) || *next == ',') {
+ if (*next == ',') {
+ if (maxpos)
+ break;
+ else
+ maxpos = next;
+ }
+ next++;
+ }
-#if 0 /* Now runtime fix should be reliable. */
+ assert(*next == '}');
- /* if this is reinstated, don't forget to put this back into perldiag:
+ if (!maxpos)
+ maxpos = next;
+ RExC_parse++;
+ if (isDIGIT(*RExC_parse)) {
+ endptr = RExC_end;
+ if (!grok_atoUV(RExC_parse, &uv, &endptr))
+ vFAIL("Invalid quantifier in {,}");
+ if (uv >= REG_INFTY)
+ vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
+ min = (I32)uv;
+ } else {
+ min = 0;
+ }
+ if (*maxpos == ',')
+ maxpos++;
+ else
+ maxpos = RExC_parse;
+ if (isDIGIT(*maxpos)) {
+ endptr = RExC_end;
+ if (!grok_atoUV(maxpos, &uv, &endptr))
+ vFAIL("Invalid quantifier in {,}");
+ if (uv >= REG_INFTY)
+ vFAIL2("Quantifier in {,} bigger than %d", REG_INFTY - 1);
+ max = (I32)uv;
+ } else {
+ max = REG_INFTY; /* meaning "infinity" */
+ }
+ RExC_parse = next;
+ nextchar(pRExC_state);
+ if (max < min) { /* If can't match, warn and optimize to fail
+ unconditionally */
+ reginsert(pRExC_state, OPFAIL, orig_emit, depth+1);
+ ckWARNreg(RExC_parse, "Quantifier {n,m} with n > m can't match");
+ NEXT_OFF(REGNODE_p(orig_emit)) =
+ regarglen[OPFAIL] + NODE_STEP_REGNODE;
+ return ret;
+ }
+ else if (min == max && *RExC_parse == '?')
+ {
+ ckWARN2reg(RExC_parse + 1,
+ "Useless use of greediness modifier '%c'",
+ *RExC_parse);
+ }
+ }
- =item Regexp *+ operand could be empty at {#} in regex m/%s/
+ if (! (flags & (HASWIDTH|POSTPONED))) {
+ if (max > REG_INFTY/3) {
+ if (origparse[0] == '\\' && origparse[1] == 'K') {
+ vFAIL2utf8f(
+ "%" UTF8f " is forbidden - matches null string"
+ " many times",
+ UTF8fARG(UTF, (RExC_parse >= origparse
+ ? RExC_parse - origparse
+ : 0),
+ origparse));
+ } else {
+ ckWARN2reg(RExC_parse,
+ "%" UTF8f " matches null string many times",
+ UTF8fARG(UTF, (RExC_parse >= origparse
+ ? RExC_parse - origparse
+ : 0),
+ origparse));
+ }
+ }
+ }
- (F) The part of the regexp subject to either the * or + quantifier
- could match an empty string. The {#} shows in the regular
- expression about where the problem was discovered.
+ if ((flags&SIMPLE)) {
+ if (min == 0 && max == REG_INFTY) {
- */
+ /* Going from 0..inf is currently forbidden in wildcard
+ * subpatterns. The only reason is to make it harder to
+ * write patterns that take a long long time to halt, and
+ * because the use of this construct isn't necessary in
+ * matching Unicode property values */
+ if (RExC_pm_flags & PMf_WILDCARD) {
+ RExC_parse++;
+ /* diag_listed_as: Use of %s is not allowed in Unicode
+ property wildcard subpatterns in regex; marked by
+ <-- HERE in m/%s/ */
+ vFAIL("Use of quantifier '*' is not allowed in"
+ " Unicode property wildcard subpatterns");
+ /* Note, don't need to worry about {0,}, as a '}' isn't
+ * legal at all in wildcards, so wouldn't get this far
+ * */
+ }
+ reginsert(pRExC_state, STAR, ret, depth+1);
+ MARK_NAUGHTY(4);
+ RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
+ goto nest_check;
+ }
+ if (min == 1 && max == REG_INFTY) {
+ reginsert(pRExC_state, PLUS, ret, depth+1);
+ MARK_NAUGHTY(3);
+ RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
+ goto nest_check;
+ }
+ MARK_NAUGHTY_EXP(2, 2);
+ reginsert(pRExC_state, CURLY, ret, depth+1);
+ Set_Node_Offset(REGNODE_p(ret), parse_start+1); /* MJD */
+ Set_Node_Cur_Length(REGNODE_p(ret), parse_start);
+ }
+ else {
+ const regnode_offset w = reg_node(pRExC_state, WHILEM);
- if (!(flags&HASWIDTH) && op != '?')
- vFAIL("Regexp *+ operand could be empty");
-#endif
+ FLAGS(REGNODE_p(w)) = 0;
+ if (! REGTAIL(pRExC_state, ret, w)) {
+ REQUIRE_BRANCHJ(flagp, 0);
+ }
+ if (RExC_use_BRANCHJ) {
+ reginsert(pRExC_state, LONGJMP, ret, depth+1);
+ reginsert(pRExC_state, NOTHING, ret, depth+1);
+ NEXT_OFF(REGNODE_p(ret)) = 3; /* Go over LONGJMP. */
+ }
+ reginsert(pRExC_state, CURLYX, ret, depth+1);
+ /* MJD hk */
+ Set_Node_Offset(REGNODE_p(ret), parse_start+1);
+ Set_Node_Length(REGNODE_p(ret),
+ op == '{' ? (RExC_parse - parse_start) : 1);
+
+ if (RExC_use_BRANCHJ)
+ NEXT_OFF(REGNODE_p(ret)) = 3; /* Go over NOTHING to
+ LONGJMP. */
+ if (! REGTAIL(pRExC_state, ret, reg_node(pRExC_state,
+ NOTHING)))
+ {
+ REQUIRE_BRANCHJ(flagp, 0);
+ }
+ RExC_whilem_seen++;
+ MARK_NAUGHTY_EXP(1, 4); /* compound interest */
+ }
+ FLAGS(REGNODE_p(ret)) = 0;
-#ifdef RE_TRACK_PATTERN_OFFSETS
- parse_start = RExC_parse;
-#endif
- nextchar(pRExC_state);
+ if (min > 0)
+ *flagp = 0;
+ if (max > 0)
+ *flagp |= HASWIDTH;
+ ARG1_SET(REGNODE_p(ret), (U16)min);
+ ARG2_SET(REGNODE_p(ret), (U16)max);
+ if (max == REG_INFTY)
+ RExC_seen |= REG_UNBOUNDED_QUANTIFIER_SEEN;
- *flagp = (op != '+') ? (WORST|SPSTART|HASWIDTH) : (WORST|HASWIDTH);
+ goto nest_check;
- if (op == '*') {
- min = 0;
- goto do_curly;
- }
- else if (op == '+') {
- min = 1;
- goto do_curly;
- }
- else if (op == '?') {
- min = 0; max = 1;
- goto do_curly;
- }
nest_check:
- if (!(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3) {
- if (origparse[0] == '\\' && origparse[1] == 'K') {
- vFAIL2utf8f(
- "%" UTF8f " is forbidden - matches null string many times",
- UTF8fARG(UTF, (RExC_parse >= origparse
- ? RExC_parse - origparse
- : 0),
- origparse));
- /* NOT-REACHED */
- } else {
- ckWARN2reg(RExC_parse,
- "%" UTF8f " matches null string many times",
- UTF8fARG(UTF, (RExC_parse >= origparse
- ? RExC_parse - origparse
- : 0),
- origparse));
- }
- }
if (*RExC_parse == '?') {
- nextchar(pRExC_state);
- reginsert(pRExC_state, MINMOD, ret, depth+1);
+ nextchar(pRExC_state);
+ reginsert(pRExC_state, MINMOD, ret, depth+1);
if (! REGTAIL(pRExC_state, ret, ret + NODE_STEP_REGNODE)) {
REQUIRE_BRANCHJ(flagp, 0);
}
}
if (ISMULT2(RExC_parse)) {
- RExC_parse++;
- vFAIL("Nested quantifiers");
+ RExC_parse++;
+ vFAIL("Nested quantifiers");
}
return(ret);
FAIL2("panic: reg returned failure to grok_bslash_N, flags=%#" UVxf,
(UV) flags);
}
- *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
+ *flagp |= flags&(HASWIDTH|SIMPLE|POSTPONED);
nextchar(pRExC_state);
STATIC regnode_offset
S_regatom(pTHX_ RExC_state_t *pRExC_state, I32 *flagp, U32 depth)
{
- dVAR;
regnode_offset ret = 0;
I32 flags = 0;
char *parse_start;
DECLARE_AND_GET_RE_DEBUG_FLAGS;
- *flagp = WORST; /* Tentatively. */
+ *flagp = 0; /* Initialize. */
DEBUG_PARSE("atom");
FAIL2("panic: reg returned failure to regatom, flags=%#" UVxf,
(UV) flags);
}
- *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
+ *flagp |= flags&(HASWIDTH|SIMPLE|POSTPONED);
break;
case '|':
case ')':
}
ret = reg_node(pRExC_state, GPOS);
RExC_seen |= REG_GPOS_SEEN;
- *flagp |= SIMPLE;
goto finish_meta_pat;
case 'K':
if (!RExC_in_lookbehind && !RExC_in_lookahead) {
RExC_seen_zerolen++;
ret = reg_node(pRExC_state, KEEPS);
- *flagp |= SIMPLE;
/* XXX:dmq : disabling in-place substitution seems to
* be necessary here to avoid cases of memory corruption, as
* with: C<$_="x" x 80; s/x\K/y/> -- rgs
ret = reg_node(pRExC_state, op);
FLAGS(REGNODE_p(ret)) = flags;
- *flagp |= SIMPLE;
-
goto finish_meta_pat;
}
num > 9
/* any numeric escape < RExC_npar is a backref */
&& num >= RExC_npar
- /* cannot be an octal escape if it starts with 8 */
- && *RExC_parse != '8'
- /* cannot be an octal escape if it starts with 9 */
- && *RExC_parse != '9'
+ /* cannot be an octal escape if it starts with [89] */
+ && ! inRANGE(*RExC_parse, '8', '9')
) {
/* Probably not meant to be a backref, instead likely
* to be an octal character escape, e.g. \35 or \777.
FAIL2("panic: loc_correspondence[%d] is 0",
(int) (s - s_start));
}
+ Safefree(locfold_buf);
+ Safefree(loc_correspondence);
}
else {
upper_fill = s - s0;
* sets up the bitmap and any flags, removing those code points from the
* inversion list, setting it to NULL should it become completely empty */
- dVAR;
PERL_ARGS_ASSERT_POPULATE_ANYOF_FROM_INVLIST;
assert(PL_regkind[OP(node)] == ANYOF);
? end
: NUM_ANYOF_CODE_POINTS - 1;
for (i = start; i <= (int) high; i++) {
- if (! ANYOF_BITMAP_TEST(node, i)) {
- ANYOF_BITMAP_SET(node, i);
- }
+ ANYOF_BITMAP_SET(node, i);
}
}
invlist_iterfinish(*invlist_ptr);
if ( posix_warnings
&& RExC_warn_text
- && av_top_index(RExC_warn_text) > -1)
+ && av_count(RExC_warn_text) > 0)
{
*posix_warnings = RExC_warn_text;
}
*
* There is a line below that uses the same white space criteria but is outside
* this macro. Both here and there must use the same definition */
-#define SKIP_BRACKETED_WHITE_SPACE(do_skip, p) \
+#define SKIP_BRACKETED_WHITE_SPACE(do_skip, p, stop_p) \
STMT_START { \
if (do_skip) { \
- while (isBLANK_A(UCHARAT(p))) \
+ while (p < stop_p && isBLANK_A(UCHARAT(p))) \
{ \
p++; \
} \
* UTF-8
*/
- dVAR;
UV prevvalue = OOB_UNICODE, save_prevvalue = OOB_UNICODE;
IV range = 0;
UV value = OOB_UNICODE, save_value = OOB_UNICODE;
initial_listsv_len = SvCUR(listsv);
SvTEMP_off(listsv); /* Grr, TEMPs and mortals are conflated. */
- SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
+ SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse, RExC_end);
assert(RExC_parse <= RExC_end);
invert = TRUE;
allow_mutiple_chars = FALSE;
MARK_NAUGHTY(1);
- SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
+ SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse, RExC_end);
}
/* Check that they didn't say [:posix:] instead of [[:posix:]] */
output_posix_warnings(pRExC_state, posix_warnings);
}
+ SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse, RExC_end);
+
if (RExC_parse >= stop_ptr) {
break;
}
- SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
-
if (UCHARAT(RExC_parse) == ']') {
break;
}
}
/* For each multi-character string ... */
- while (av_tindex(strings) >= 0) {
+ while (av_count(strings) > 0) {
/* ... Each entry is itself an array of code
* points. */
AV * this_string = (AV *) av_shift( strings);
- STRLEN cp_count = av_tindex(this_string) + 1;
+ STRLEN cp_count = av_count(this_string);
SV * final = newSV(cp_count * 4);
SvPVCLEAR(final);
/* Create another string of sequences of \x{...} */
- while (av_tindex(this_string) >= 0) {
+ while (av_count(this_string) > 0) {
SV * character = av_shift(this_string);
UV cp = SvUV(character);
}
} /* end of namedclass \blah */
- SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse);
+ SKIP_BRACKETED_WHITE_SPACE(skip_white, RExC_parse, RExC_end);
/* If 'range' is set, 'value' is the ending of a range--check its
* validity. (If value isn't a single code point in the case of a
char* next_char_ptr = RExC_parse + 1;
/* Get the next real char after the '-' */
- SKIP_BRACKETED_WHITE_SPACE(skip_white, next_char_ptr);
+ SKIP_BRACKETED_WHITE_SPACE(skip_white, next_char_ptr, RExC_end);
/* If the '-' is at the end of the class (just before the ']',
* it is a literal minus; otherwise it is a range */
ret = reg(pRExC_state, 1, ®_flags, depth+1);
- *flagp |= reg_flags & (HASWIDTH|SIMPLE|SPSTART|POSTPONED|RESTART_PARSE|NEED_UTF8);
+ *flagp |= reg_flags & (HASWIDTH|SIMPLE|POSTPONED|RESTART_PARSE|NEED_UTF8);
/* And restore so can parse the rest of the pattern */
RExC_parse = save_parse;
av_store(av, INVLIST_INDEX, SvREFCNT_inc_NN(cp_list));
}
+ /* (Note that if any of this changes, the size calculations in
+ * S_optimize_regclass() might need to be updated.) */
+
if (only_utf8_locale_list) {
av_store(av, ONLY_LOCALE_MATCHES_INDEX,
SvREFCNT_inc_NN(only_utf8_locale_list));
PERL_UNUSED_ARG(depth);
#endif
- /* Find last node. */
+ /* The final node in the chain is the first one with a nonzero next pointer
+ * */
scan = (regnode_offset) p;
for (;;) {
regnode * const temp = regnext(REGNODE_p(scan));
scan = REGNODE_OFFSET(temp);
}
+ /* Populate this node's next pointer */
assert(val >= scan);
if (reg_off_by_arg[OP(REGNODE_p(scan))]) {
assert((UV) (val - scan) <= U32_MAX);
Perl_regprop(pTHX_ const regexp *prog, SV *sv, const regnode *o, const regmatch_info *reginfo, const RExC_state_t *pRExC_state)
{
#ifdef DEBUGGING
- dVAR;
int k;
RXi_GET_DECL(prog, progi);
DECLARE_AND_GET_RE_DEBUG_FLAGS;
U32 refcount;
reg_ac_data *aho=(reg_ac_data*)ri->data->data[n];
#ifdef USE_ITHREADS
- dVAR;
#endif
OP_REFCNT_LOCK;
refcount = --aho->refcount;
U32 refcount;
reg_trie_data *trie=(reg_trie_data*)ri->data->data[n];
#ifdef USE_ITHREADS
- dVAR;
#endif
OP_REFCNT_LOCK;
refcount = --trie->refcount;
#define SAVEPVN(p, n) ((p) ? savepvn(p, n) : NULL)
/*
- re_dup_guts - duplicate a regexp.
+=for apidoc_section REGEXP Functions
+=for apidoc re_dup_guts
+Duplicate a regexp.
- This routine is expected to clone a given regexp structure. It is only
- compiled under USE_ITHREADS.
+This routine is expected to clone a given regexp structure. It is only
+compiled under USE_ITHREADS.
- After all of the core data stored in struct regexp is duplicated
- the regexp_engine.dupe method is used to copy any private data
- stored in the *pprivate pointer. This allows extensions to handle
- any duplication it needs to do.
+After all of the core data stored in struct regexp is duplicated
+the regexp_engine.dupe method is used to copy any private data
+stored in the *pprivate pointer. This allows extensions to handle
+any duplication they need to do.
+
+=cut
See pregfree() and regfree_internal() if you change anything here.
*/
void
Perl_re_dup_guts(pTHX_ const REGEXP *sstr, REGEXP *dstr, CLONE_PARAMS *param)
{
- dVAR;
I32 npar;
const struct regexp *r = ReANY(sstr);
struct regexp *ret = ReANY(dstr);
void *
Perl_regdupe_internal(pTHX_ REGEXP * const rx, CLONE_PARAMS *param)
{
- dVAR;
struct regexp *const r = ReANY(rx);
regexp_internal *reti;
int len;
UV this_end;
const char * format;
- if (end - start < min_range_count) {
-
- /* Output chars individually when they occur in short ranges */
+ if ( end - start < min_range_count
+ && (end - start <= 2 || (isPRINT_A(start) && isPRINT_A(end))))
+ {
+ /* Output a range of 1 or 2 chars individually, or longer ranges
+ * when printable */
for (; start <= end; start++) {
put_code_point(sv, start);
}
* output would have been only the inversion indicator '^', NULL is instead
* returned. */
- dVAR;
SV * output;
PERL_ARGS_ASSERT_PUT_CHARCLASS_BITMAP_INNARDS_COMMON;
* cases where it can't try inverting, as what actually matches isn't known
* until runtime, and hence the inversion isn't either. */
- dVAR;
bool inverting_allowed = ! force_as_is_display;
int i;
void
Perl_init_uniprops(pTHX)
{
- dVAR;
# ifdef DEBUGGING
char * dump_len_string;
* properties. This is a function so it can be set up to be called even if
* the program unexpectedly quits */
- dVAR;
SV ** current_entry;
const STRLEN key_len = strlen((const char *) key);
DECLARATION_FOR_GLOBAL_CONTEXT;
this */
const STRLEN level) /* Recursion level of this call */
{
- dVAR;
char* lookup_name; /* normalized name for lookup in our tables */
unsigned lookup_len; /* Its length */
enum { Not_Strict = 0, /* Some properties have stricter name */
goto append_name_to_msg;
}
- lookup_loose = get_cv("_charnames::_loose_regcomp_lookup", 0);
+ lookup_loose = get_cvs("_charnames::_loose_regcomp_lookup", 0);
if (! lookup_loose) {
Perl_croak(aTHX_
"panic: Can't find '_charnames::_loose_regcomp_lookup");
/* Try again stripping off any initial 'Is'. This is because we
* promise that an initial Is is optional. The same isn't true of
* names that start with 'In'. Those can match only blocks, and the
- * lookup table already has those accounted for. */
- if (starts_with_Is) {
+ * lookup table already has those accounted for. The lookup table also
+ * has already accounted for Perl extensions (without and = sign)
+ * starting with 'i's'. */
+ if (starts_with_Is && equals_pos >= 0) {
lookup_name += 2;
lookup_len -= 2;
equals_pos -= 2;