STATIC CHECKPOINT
S_regcppush(pTHX_ const regexp *rex, I32 parenfloor, U32 maxopenparen)
{
- dVAR;
const int retval = PL_savestack_ix;
const int paren_elems_to_push =
(maxopenparen - parenfloor) * REGCP_PAREN_ELEMS;
PERL_ARGS_ASSERT_REGCPPUSH;
if (paren_elems_to_push < 0)
- Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0, maxopenparen: %i parenfloor: %i REGCP_PAREN_ELEMS: %i",
- paren_elems_to_push, maxopenparen, parenfloor, REGCP_PAREN_ELEMS);
+ Perl_croak(aTHX_ "panic: paren_elems_to_push, %i < 0, maxopenparen: %i parenfloor: %i REGCP_PAREN_ELEMS: %u",
+ (int)paren_elems_to_push, (int)maxopenparen,
+ (int)parenfloor, (unsigned)REGCP_PAREN_ELEMS);
if ((elems_shifted >> SAVE_TIGHT_SHIFT) != total_elems)
Perl_croak(aTHX_ "panic: paren_elems_to_push offset %"UVuf
STATIC void
S_regcppop(pTHX_ regexp *rex, U32 *maxopenparen_p)
{
- dVAR;
UV i;
U32 paren;
GET_RE_DEBUG_FLAGS_DECL;
case _CC_ENUM_BLANK: return is_HORIZWS_high(character);
case _CC_ENUM_XDIGIT: return is_XDIGIT_high(character);
case _CC_ENUM_VERTSPACE: return is_VERTWS_high(character);
- default: return 0; /* Things like CNTRL are always
- below 256 */
+ default: break;
}
- assert(0); /* NOTREACHED */
- return FALSE;
+ return FALSE; /* Things like CNTRL are always below 256 */
}
/*
const U32 flags,
re_scream_pos_data *data)
{
- dVAR;
struct regexp *const prog = ReANY(rx);
SSize_t start_shift = prog->check_offset_min;
/* Should be nonnegative! */
});
if (prog->intflags & PREGf_ANCH) { /* Match at \G, beg-of-str or after \n */
- /* Check after \n? */
- ml_anch = (prog->intflags & PREGf_ANCH_MBOL);
- if (!ml_anch) {
+ /* ml_anch: check after \n?
+ *
+ * A note about IMPLICIT: on an un-anchored pattern beginning
+ * with /.*.../, these flags will have been added by the
+ * compiler:
+ * /.*abc/, /.*abc/m: PREGf_IMPLICIT | PREGf_ANCH_MBOL
+ * /.*abc/s: PREGf_IMPLICIT | PREGf_ANCH_SBOL
+ */
+ ml_anch = (prog->intflags & PREGf_ANCH_MBOL)
+ && !(prog->intflags & PREGf_IMPLICIT);
+
+ if (!ml_anch && !(prog->intflags & PREGf_IMPLICIT)) {
/* we are only allowed to match at BOS or \G */
/* trivially reject if there's a BOS anchor and we're not at BOS.
- * In the case of \G, we hope(!) that the caller has already
- * set strpos to pos()-gofs, and will already have checked
- * that this anchor position is legal. So we can skip it here.
+ *
+ * Note that we don't try to do a similar quick reject for
+ * \G, since generally the caller will have calculated strpos
+ * based on pos() and gofs, so the string is already correctly
+ * anchored by definition; and handling the exceptions would
+ * be too fiddly (e.g. REXEC_IGNOREPOS).
*/
- if ( !(prog->intflags & PREGf_ANCH_GPOS)
- && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */
- && (strpos != strbeg))
+ if ( strpos != strbeg
+ && (prog->intflags & PREGf_ANCH_SBOL))
{
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
" Not at start...\n"));
{
/* Substring at constant offset from beg-of-str... */
SSize_t slen = SvCUR(check);
- char *s;
-
- s = HOP3c(strpos, prog->check_offset_min, strend);
+ char *s = HOP3c(strpos, prog->check_offset_min, strend);
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
" Looking for check substr at fixed offset %"IVdf"...\n",
(IV)prog->check_offset_min));
- if (SvTAIL(check) && !multiline) {
- /* In this case, the regex is anchored at the end too,
- * so the lengths must match exactly, give or take a \n.
- * NB: slen >= 1 since the last char of check is \n */
- if ( strend - s > slen || strend - s < slen - 1
- || (strend - s == slen && strend[-1] != '\n'))
+ if (SvTAIL(check)) {
+ /* In this case, the regex is anchored at the end too.
+ * Unless it's a multiline match, the lengths must match
+ * exactly, give or take a \n. NB: slen >= 1 since
+ * the last char of check is \n */
+ if (!multiline
+ && ( strend - s > slen
+ || strend - s < slen - 1
+ || (strend - s == slen && strend[-1] != '\n')))
{
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
" String too long...\n"));
}
- /* if the regex is absolutely anchored to the start of the string,
- * then check_offset_max represents an upper bound on the string
- * where the substr could start */
+ /* If the regex is absolutely anchored to either the start of the
+ * string (SBOL) or to pos() (ANCH_GPOS), then
+ * check_offset_max represents an upper bound on the string where
+ * the substr could start. For the ANCH_GPOS case, we assume that
+ * the caller of intuit will have already set strpos to
+ * pos()-gofs, so in this case strpos + offset_max will still be
+ * an upper bound on the substr.
+ */
if (!ml_anch
&& prog->intflags & PREGf_ANCH
- && prog->check_offset_max != SSize_t_MAX
- && start_shift < prog->check_offset_max)
+ && prog->check_offset_max != SSize_t_MAX)
{
SSize_t len = SvCUR(check) - !!SvTAIL(check);
- end_point = HOP3lim(start_point,
- prog->check_offset_max - start_shift,
- end_point -len)
- + len;
+ const char * const anchor =
+ (prog->intflags & PREGf_ANCH_GPOS ? strpos : strbeg);
+
+ /* do a bytes rather than chars comparison. It's conservative;
+ * so it skips doing the HOP if the result can't possibly end
+ * up earlier than the old value of end_point.
+ */
+ if ((char*)end_point - anchor > prog->check_offset_max) {
+ end_point = HOP3lim((U8*)anchor,
+ prog->check_offset_max,
+ end_point -len)
+ + len;
+ }
}
DEBUG_OPTIMISE_MORE_r({
/* handle the extra constraint of /^.../m if present */
- if (ml_anch && rx_origin != strbeg && rx_origin[-1] != '\n'
- /* May be due to an implicit anchor of m{.*foo} */
- && !(prog->intflags & PREGf_IMPLICIT))
- {
+ if (ml_anch && rx_origin != strbeg && rx_origin[-1] != '\n') {
char *s;
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
}
DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
" This position contradicts STCLASS...\n") );
- if ((prog->intflags & PREGf_ANCH) && !ml_anch)
+ if ((prog->intflags & PREGf_ANCH) && !ml_anch
+ && !(prog->intflags & PREGf_IMPLICIT))
goto fail;
/* Contradict one of substrings */
prog->check_substr = prog->check_utf8 = NULL; /* disable */
prog->float_substr = prog->float_utf8 = NULL; /* clear */
check = NULL; /* abort */
- /* XXXX If the check string was an implicit check MBOL, then we need to unset the relevant flag
- see http://bugs.activestate.com/show_bug.cgi?id=87173 */
- if (prog->intflags & PREGf_IMPLICIT) {
- prog->intflags &= ~PREGf_ANCH_MBOL;
- /* maybe we have no anchors left after this... */
- if (!(prog->intflags & PREGf_ANCH))
- prog->extflags &= ~RXf_IS_ANCHORED;
- }
/* XXXX This is a remnant of the old implementation. It
looks wasteful, since now INTUIT can use many
other heuristics. */
prog->extflags &= ~RXf_USE_INTUIT;
- /* XXXX What other flags might need to be cleared in this branch? */
}
}
switch (trie_type) { \
case trie_utf8_exactfa_fold: \
flags |= FOLD_FLAGS_NOMIX_ASCII; \
- /* FALL THROUGH */ \
+ /* FALLTHROUGH */ \
case trie_utf8_fold: \
if ( foldlen>0 ) { \
uvc = utf8n_to_uvchr( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
break; \
case trie_latin_utf8_exactfa_fold: \
flags |= FOLD_FLAGS_NOMIX_ASCII; \
- /* FALL THROUGH */ \
+ /* FALLTHROUGH */ \
case trie_latin_utf8_fold: \
if ( foldlen>0 ) { \
uvc = utf8n_to_uvchr( (const U8*) uscan, UTF8_MAXLEN, &len, uniflags ); \
} \
} STMT_END
-#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
+#define DUMP_EXEC_POS(li,s,doutf8) \
+ dump_exec_pos(li,s,(reginfo->strend),(reginfo->strbeg), \
+ startpos, doutf8)
+
+#define REXEC_FBC_EXACTISH_SCAN(COND) \
STMT_START { \
while (s <= e) { \
- if ( (CoNd) \
+ if ( (COND) \
&& (ln == 1 || folder(s, pat_string, ln)) \
&& (reginfo->intuit || regtry(reginfo, &s)) )\
goto got_it; \
} \
} STMT_END
-#define REXEC_FBC_UTF8_SCAN(CoDe) \
+#define REXEC_FBC_UTF8_SCAN(CODE) \
STMT_START { \
while (s < strend) { \
- CoDe \
+ CODE \
s += UTF8SKIP(s); \
} \
} STMT_END
-#define REXEC_FBC_SCAN(CoDe) \
+#define REXEC_FBC_SCAN(CODE) \
STMT_START { \
while (s < strend) { \
- CoDe \
+ CODE \
s++; \
} \
} STMT_END
-#define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \
-REXEC_FBC_UTF8_SCAN( \
- if (CoNd) { \
- if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
- goto got_it; \
- else \
- tmp = doevery; \
- } \
- else \
- tmp = 1; \
+#define REXEC_FBC_UTF8_CLASS_SCAN(COND) \
+REXEC_FBC_UTF8_SCAN( /* Loops while (s < strend) */ \
+ if (COND) { \
+ if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
+ goto got_it; \
+ else \
+ tmp = doevery; \
+ } \
+ else \
+ tmp = 1; \
)
-#define REXEC_FBC_CLASS_SCAN(CoNd) \
-REXEC_FBC_SCAN( \
- if (CoNd) { \
- if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
- goto got_it; \
- else \
- tmp = doevery; \
- } \
- else \
- tmp = 1; \
+#define REXEC_FBC_CLASS_SCAN(COND) \
+REXEC_FBC_SCAN( /* Loops while (s < strend) */ \
+ if (COND) { \
+ if (tmp && (reginfo->intuit || regtry(reginfo, &s))) \
+ goto got_it; \
+ else \
+ tmp = doevery; \
+ } \
+ else \
+ tmp = 1; \
)
-#define REXEC_FBC_TRYIT \
-if ((reginfo->intuit || regtry(reginfo, &s))) \
- goto got_it
-
-#define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \
+#define REXEC_FBC_CSCAN(CONDUTF8,COND) \
if (utf8_target) { \
- REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \
+ REXEC_FBC_UTF8_CLASS_SCAN(CONDUTF8); \
} \
else { \
- REXEC_FBC_CLASS_SCAN(CoNd); \
+ REXEC_FBC_CLASS_SCAN(COND); \
}
-
-#define DUMP_EXEC_POS(li,s,doutf8) \
- dump_exec_pos(li,s,(reginfo->strend),(reginfo->strbeg), \
- startpos, doutf8)
-
-#define UTF8_NOLOAD(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
- tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
- tmp = TEST_NON_UTF8(tmp); \
- REXEC_FBC_UTF8_SCAN( \
- if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
- tmp = !tmp; \
- IF_SUCCESS; \
- } \
- else { \
- IF_FAIL; \
- } \
- ); \
-
-#define UTF8_LOAD(TeSt1_UtF8, TeSt2_UtF8, IF_SUCCESS, IF_FAIL) \
- if (s == reginfo->strbeg) { \
- tmp = '\n'; \
- } \
- else { \
- U8 * const r = reghop3((U8*)s, -1, (U8*)reginfo->strbeg); \
- tmp = utf8n_to_uvchr(r, (U8*) reginfo->strend - r, \
+/* The three macros below are slightly different versions of the same logic.
+ *
+ * The first is for /a and /aa when the target string is UTF-8. This can only
+ * match ascii, but it must advance based on UTF-8. The other two handle the
+ * non-UTF-8 and the more generic UTF-8 cases. In all three, we are looking
+ * for the boundary (or non-boundary) between a word and non-word character.
+ * The utf8 and non-utf8 cases have the same logic, but the details must be
+ * different. Find the "wordness" of the character just prior to this one, and
+ * compare it with the wordness of this one. If they differ, we have a
+ * boundary. At the beginning of the string, pretend that the previous
+ * character was a new-line.
+ *
+ * All these macros uncleanly have side-effects with each other and outside
+ * variables. So far it's been too much trouble to clean-up
+ *
+ * TEST_NON_UTF8 is the macro or function to call to test if its byte input is
+ * a word character or not.
+ * IF_SUCCESS is code to do if it finds that we are at a boundary between
+ * word/non-word
+ * IF_FAIL is code to do if we aren't at a boundary between word/non-word
+ *
+ * Exactly one of the two IF_FOO parameters is a no-op, depending on whether we
+ * are looking for a boundary or for a non-boundary. If we are looking for a
+ * boundary, we want IF_FAIL to be the no-op, and for IF_SUCCESS to go out and
+ * see if this tentative match actually works, and if so, to quit the loop
+ * here. And vice-versa if we are looking for a non-boundary.
+ *
+ * 'tmp' below in the next three macros in the REXEC_FBC_SCAN and
+ * REXEC_FBC_UTF8_SCAN loops is a loop invariant, a bool giving the return of
+ * TEST_NON_UTF8(s-1). To see this, note that that's what it is defined to be
+ * at entry to the loop, and to get to the IF_FAIL branch, tmp must equal
+ * TEST_NON_UTF8(s), and in the opposite branch, IF_SUCCESS, tmp is that
+ * complement. But in that branch we complement tmp, meaning that at the
+ * bottom of the loop tmp is always going to be equal to TEST_NON_UTF8(s),
+ * which means at the top of the loop in the next iteration, it is
+ * TEST_NON_UTF8(s-1) */
+#define FBC_UTF8_A(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
+ tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
+ tmp = TEST_NON_UTF8(tmp); \
+ REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */ \
+ if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
+ tmp = !tmp; \
+ IF_SUCCESS; /* Is a boundary if values for s-1 and s differ */ \
+ } \
+ else { \
+ IF_FAIL; \
+ } \
+ ); \
+
+/* Like FBC_UTF8_A, but TEST_UV is a macro which takes a UV as its input, and
+ * TEST_UTF8 is a macro that for the same input code points returns identically
+ * to TEST_UV, but takes a pointer to a UTF-8 encoded string instead */
+#define FBC_UTF8(TEST_UV, TEST_UTF8, IF_SUCCESS, IF_FAIL) \
+ if (s == reginfo->strbeg) { \
+ tmp = '\n'; \
+ } \
+ else { /* Back-up to the start of the previous character */ \
+ U8 * const r = reghop3((U8*)s, -1, (U8*)reginfo->strbeg); \
+ tmp = utf8n_to_uvchr(r, (U8*) reginfo->strend - r, \
0, UTF8_ALLOW_DEFAULT); \
- } \
- tmp = TeSt1_UtF8; \
- LOAD_UTF8_CHARCLASS_ALNUM(); \
- REXEC_FBC_UTF8_SCAN( \
- if (tmp == ! (TeSt2_UtF8)) { \
- tmp = !tmp; \
- IF_SUCCESS; \
- } \
- else { \
- IF_FAIL; \
- } \
- ); \
+ } \
+ tmp = TEST_UV(tmp); \
+ LOAD_UTF8_CHARCLASS_ALNUM(); \
+ REXEC_FBC_UTF8_SCAN( /* advances s while s < strend */ \
+ if (tmp == ! (TEST_UTF8((U8 *) s))) { \
+ tmp = !tmp; \
+ IF_SUCCESS; \
+ } \
+ else { \
+ IF_FAIL; \
+ } \
+ );
-/* The only difference between the BOUND and NBOUND cases is that
- * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
- * NBOUND. This is accomplished by passing it in either the if or else clause,
- * with the other one being empty */
-#define FBC_BOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
- FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
-
-#define FBC_BOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
- FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
-
-#define FBC_NBOUND(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
- FBC_BOUND_COMMON(UTF8_LOAD(TEST1_UTF8, TEST2_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
-
-#define FBC_NBOUND_NOLOAD(TEST_NON_UTF8, TEST1_UTF8, TEST2_UTF8) \
- FBC_BOUND_COMMON(UTF8_NOLOAD(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
-
-
-/* Common to the BOUND and NBOUND cases. Unfortunately the UTF8 tests need to
- * be passed in completely with the variable name being tested, which isn't
- * such a clean interface, but this is easier to read than it was before. We
- * are looking for the boundary (or non-boundary between a word and non-word
- * character. The utf8 and non-utf8 cases have the same logic, but the details
- * must be different. Find the "wordness" of the character just prior to this
- * one, and compare it with the wordness of this one. If they differ, we have
- * a boundary. At the beginning of the string, pretend that the previous
- * character was a new-line */
+/* Like the above two macros. UTF8_CODE is the complete code for handling
+ * UTF-8. Common to the BOUND and NBOUND cases, set-up by the FBC_BOUND, etc
+ * macros below */
#define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
if (utf8_target) { \
- UTF8_CODE \
+ UTF8_CODE \
} \
else { /* Not utf8 */ \
tmp = (s != reginfo->strbeg) ? UCHARAT(s - 1) : '\n'; \
tmp = TEST_NON_UTF8(tmp); \
- REXEC_FBC_SCAN( \
+ REXEC_FBC_SCAN( /* advances s while s < strend */ \
if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
- tmp = !tmp; \
IF_SUCCESS; \
+ tmp = !tmp; \
} \
else { \
IF_FAIL; \
} \
); \
} \
- if ((!prog->minlen && tmp) && (reginfo->intuit || regtry(reginfo, &s))) \
- goto got_it;
+ /* Here, things have been set up by the previous code so that tmp is the \
+ * return of TEST_NON_UTF(s-1) or TEST_UTF8(s-1) (depending on the \
+ * utf8ness of the target). We also have to check if this matches against \
+ * the EOS, which we treat as a \n (which is the same value in both UTF-8 \
+ * or non-UTF8, so can use the non-utf8 test condition even for a UTF-8 \
+ * string */ \
+ if (tmp == ! TEST_NON_UTF8('\n')) { \
+ IF_SUCCESS; \
+ } \
+ else { \
+ IF_FAIL; \
+ }
+
+/* This is the macro to use when we want to see if something that looks like it
+ * could match, actually does, and if so exits the loop */
+#define REXEC_FBC_TRYIT \
+ if ((reginfo->intuit || regtry(reginfo, &s))) \
+ goto got_it
+
+/* The only difference between the BOUND and NBOUND cases is that
+ * REXEC_FBC_TRYIT is called when matched in BOUND, and when non-matched in
+ * NBOUND. This is accomplished by passing it as either the if or else clause,
+ * with the other one being empty (PLACEHOLDER is defined as empty).
+ *
+ * The TEST_FOO parameters are for operating on different forms of input, but
+ * all should be ones that return identically for the same underlying code
+ * points */
+#define FBC_BOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
+ FBC_BOUND_COMMON( \
+ FBC_UTF8(TEST_UV, TEST_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
+ TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
+
+#define FBC_BOUND_A(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
+ FBC_BOUND_COMMON( \
+ FBC_UTF8_A(TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER), \
+ TEST_NON_UTF8, REXEC_FBC_TRYIT, PLACEHOLDER)
+
+#define FBC_NBOUND(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
+ FBC_BOUND_COMMON( \
+ FBC_UTF8(TEST_UV, TEST_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
+ TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
+
+#define FBC_NBOUND_A(TEST_NON_UTF8, TEST_UV, TEST_UTF8) \
+ FBC_BOUND_COMMON( \
+ FBC_UTF8_A(TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT), \
+ TEST_NON_UTF8, PLACEHOLDER, REXEC_FBC_TRYIT)
+
/* We know what class REx starts with. Try to find this position... */
/* if reginfo->intuit, its a dryrun */
/* annoyingly all the vars in this routine have different names from their counterparts
in regmatch. /grrr */
-
STATIC char *
S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s,
const char *strend, regmatch_info *reginfo)
case EXACTFA_NO_TRIE: /* This node only generated for non-utf8 patterns */
assert(! is_utf8_pat);
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case EXACTFA:
if (is_utf8_pat || utf8_target) {
utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
fold_array = PL_fold_latin1;
folder = foldEQ_latin1;
- /* FALL THROUGH */
+ /* FALLTHROUGH */
do_exactf_non_utf8: /* Neither pattern nor string are UTF8, and there
are no glitches with fold-length differences
}
break;
}
+
case BOUNDL:
- FBC_BOUND(isWORDCHAR_LC,
- isWORDCHAR_LC_uvchr(tmp),
- isWORDCHAR_LC_utf8((U8*)s));
+ FBC_BOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
break;
case NBOUNDL:
- FBC_NBOUND(isWORDCHAR_LC,
- isWORDCHAR_LC_uvchr(tmp),
- isWORDCHAR_LC_utf8((U8*)s));
+ FBC_NBOUND(isWORDCHAR_LC, isWORDCHAR_LC_uvchr, isWORDCHAR_LC_utf8);
break;
case BOUND:
- FBC_BOUND(isWORDCHAR,
- isWORDCHAR_uni(tmp),
- cBOOL(swash_fetch(PL_utf8_swash_ptrs[_CC_WORDCHAR], (U8*)s, utf8_target)));
+ FBC_BOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
break;
case BOUNDA:
- FBC_BOUND_NOLOAD(isWORDCHAR_A,
- isWORDCHAR_A(tmp),
- isWORDCHAR_A((U8*)s));
+ FBC_BOUND_A(isWORDCHAR_A, isWORDCHAR_A, isWORDCHAR_A);
break;
case NBOUND:
- FBC_NBOUND(isWORDCHAR,
- isWORDCHAR_uni(tmp),
- cBOOL(swash_fetch(PL_utf8_swash_ptrs[_CC_WORDCHAR], (U8*)s, utf8_target)));
+ FBC_NBOUND(isWORDCHAR, isWORDCHAR_uni, isWORDCHAR_utf8);
break;
case NBOUNDA:
- FBC_NBOUND_NOLOAD(isWORDCHAR_A,
- isWORDCHAR_A(tmp),
- isWORDCHAR_A((U8*)s));
+ FBC_NBOUND_A(isWORDCHAR_A, isWORDCHAR_A, isWORDCHAR_A);
break;
case BOUNDU:
- FBC_BOUND(isWORDCHAR_L1,
- isWORDCHAR_uni(tmp),
- cBOOL(swash_fetch(PL_utf8_swash_ptrs[_CC_WORDCHAR], (U8*)s, utf8_target)));
+ FBC_BOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
break;
case NBOUNDU:
- FBC_NBOUND(isWORDCHAR_L1,
- isWORDCHAR_uni(tmp),
- cBOOL(swash_fetch(PL_utf8_swash_ptrs[_CC_WORDCHAR], (U8*)s, utf8_target)));
+ FBC_NBOUND(isWORDCHAR_L1, isWORDCHAR_uni, isWORDCHAR_utf8);
break;
case LNBREAK:
REXEC_FBC_CSCAN(is_LNBREAK_utf8_safe(s, strend),
case NPOSIXA:
if (utf8_target) {
/* The complement of something that matches only ASCII matches all
- * UTF-8 variant code points, plus everything in ASCII that isn't
- * in the class */
- REXEC_FBC_UTF8_CLASS_SCAN(! UTF8_IS_INVARIANT(*s)
+ * non-ASCII, plus everything in ASCII that isn't in the class. */
+ REXEC_FBC_UTF8_CLASS_SCAN(! isASCII_utf8(s)
|| ! _generic_isCC_A(*s, FLAGS(c)));
break;
}
macros */
case _CC_ENUM_SPACE: /* XXX would require separate code if we
revert the change of \v matching this */
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case _CC_ENUM_PSXSPC:
REXEC_FBC_UTF8_CLASS_SCAN(
break;
default:
Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
- break;
}
return 0;
got_it:
/* flags: For optimizations. See REXEC_* in regexp.h */
{
- dVAR;
struct regexp *const prog = ReANY(rx);
char *s;
regnode *c;
/* Be paranoid... */
if (prog == NULL || stringarg == NULL) {
Perl_croak(aTHX_ "NULL regexp parameter");
- return 0;
}
DEBUG_EXECUTE_r(
Perl_croak(aTHX_ "corrupted regexp program");
}
+ RX_MATCH_TAINTED_off(rx);
+
reginfo->prog = rx; /* Yes, sorry that this is confusing. */
reginfo->intuit = 0;
reginfo->is_utf8_target = cBOOL(utf8_target);
magic belonging to this SV.
Not newSVsv, either, as it does not COW.
*/
- assert(!IS_PADGV(sv));
reginfo->sv = newSV(0);
SvSetSV_nosteal(reginfo->sv, sv);
SAVEFREESV(reginfo->sv);
}
/* Simplest case: anchored match need be tried only once. */
- /* [unless only anchor is BOL and multiline is set] */
+ /* [unless only anchor is MBOL - implying multiline is set] */
if (prog->intflags & (PREGf_ANCH & ~PREGf_ANCH_GPOS)) {
if (s == startpos && regtry(reginfo, &s))
goto got_it;
* and replaced it with this one. Yves */
DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
- "String does not contain required substring, cannot match.\n"
+ "%sString does not contain required substring, cannot match.%s\n",
+ PL_colors[4], PL_colors[5]
));
goto phooey;
}
STATIC I32 /* 0 failure, 1 success */
S_regtry(pTHX_ regmatch_info *reginfo, char **startposp)
{
- dVAR;
CHECKPOINT lastcp;
REGEXP *const rx = reginfo->prog;
regexp *const prog = ReANY(rx);
* or 0 if non of the buffers matched.
*/
STATIC I32
-S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan)
+S_reg_check_named_buff_matched(const regexp *rex, const regnode *scan)
{
I32 n;
RXi_GET_DECL(rex,rexi);
const bool utf8_target = reginfo->is_utf8_target;
- UV c1 = CHRTEST_NOT_A_CP_1;
- UV c2 = CHRTEST_NOT_A_CP_2;
+ UV c1 = (UV)CHRTEST_NOT_A_CP_1;
+ UV c2 = (UV)CHRTEST_NOT_A_CP_2;
bool use_chrtest_void = FALSE;
const bool is_utf8_pat = reginfo->is_utf8_pat;
}
else { /* an EXACTFish node which doesn't begin with a multi-char fold */
c1 = is_utf8_pat ? valid_utf8_to_uvchr(pat, NULL) : *pat;
- if (c1 > 256) {
+ if (c1 > 255) {
/* Load the folds hash, if not already done */
SV** listp;
if (! PL_utf8_foldclosures) {
- if (! PL_utf8_tofold) {
- U8 dummy[UTF8_MAXBYTES_CASE+1];
-
- /* Force loading this by folding an above-Latin1 char */
- to_utf8_fold((U8*) HYPHEN_UTF8, dummy, NULL);
- assert(PL_utf8_tofold); /* Verify that worked */
- }
- PL_utf8_foldclosures = _swash_inversion_hash(PL_utf8_tofold);
+ _load_PL_utf8_foldclosures();
}
/* The fold closures data structure is a hash with the keys
/* Folds that cross the 255/256 boundary are forbidden
* if EXACTFL (and isnt a UTF8 locale), or EXACTFA and
* one is ASCIII. Since the pattern character is above
- * 256, and its only other match is below 256, the only
+ * 255, and its only other match is below 256, the only
* legal match will be to itself. We have thrown away
* the original, so have to compute which is the one
- * above 255 */
+ * above 255. */
if ((c1 < 256) != (c2 < 256)) {
if ((OP(text_node) == EXACTFL
&& ! IN_UTF8_CTYPE_LOCALE)
}
}
}
- else /* Here, c1 is < 255 */
+ else /* Here, c1 is <= 255 */
if (utf8_target
&& HAS_NONLATIN1_FOLD_CLOSURE(c1)
&& ( ! (OP(text_node) == EXACTFL && ! IN_UTF8_CTYPE_LOCALE))
case EXACTFA_NO_TRIE: /* This node only generated for
non-utf8 patterns */
assert(! is_utf8_pat);
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case EXACTFA:
case EXACTFU_SS:
case EXACTFU:
assert(nextchr < 256 && (nextchr >= 0 || nextchr == NEXTCHR_EOS));
switch (state_num) {
- case BOL: /* /^../ */
- case SBOL: /* /^../s */
+ case SBOL: /* /^../ and /\A../ */
if (locinput == reginfo->strbeg)
break;
sayNO;
st->u.keeper.val = rex->offs[0].start;
rex->offs[0].start = locinput - reginfo->strbeg;
PUSH_STATE_GOTO(KEEPS_next, next, locinput);
- assert(0); /*NOTREACHED*/
+ /* NOTREACHED */
+ assert(0);
+
case KEEPS_next_fail:
/* rollback the start point change */
rex->offs[0].start = st->u.keeper.val;
sayNO_SILENT;
- assert(0); /*NOTREACHED*/
+ /* NOTREACHED */
+ assert(0);
case MEOL: /* /..$/m */
if (!NEXTCHR_IS_EOS && nextchr != '\n')
sayNO;
break;
- case EOL: /* /..$/ */
- /* FALL THROUGH */
- case SEOL: /* /..$/s */
+ case SEOL: /* /..$/ */
if (!NEXTCHR_IS_EOS && nextchr != '\n')
sayNO;
if (reginfo->strend - locinput > 1)
REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
);
sayNO_SILENT;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case TRIE: /* (ab|cd) */
/* the basic plan of execution of the trie is:
* At the beginning, run though all the states, and
);
goto trie_first_try; /* jump into the fail handler */
}}
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case TRIE_next_fail: /* we failed - try next alternative */
{
if (ST.accepted > 1 || has_cutgroup) {
PUSH_STATE_GOTO(TRIE_next, scan, (char*)uc);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
/* only one choice left - just continue */
DEBUG_EXECUTE_r({
locinput = (char*)uc;
continue; /* execute rest of RE */
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
#undef ST
case EXACTFA_NO_TRIE: /* This node only generated for non-utf8
patterns */
assert(! is_utf8_pat);
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case EXACTFA: /* /abc/iaa */
folder = foldEQ_latin1;
fold_array = PL_fold_latin1;
break;
default:
Perl_croak(aTHX_ "panic: Unexpected FLAGS %u in op %u", FLAGS(scan), OP(scan));
- break;
}
}
/* Note requires that all BOUNDs be lower than all NBOUNDs in
case TAIL: /* placeholder while compiling (A|B|C) */
break;
- case BACK: /* ??? doesn't appear to be used ??? */
- break;
-
#undef ST
#define ST st->u.eval
{
/* and then jump to the code we share with EVAL */
goto eval_recurse_doit;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */
if (cur_eval && cur_eval->locinput==locinput) {
assert(o->op_targ == OP_LEAVE);
o = cUNOPo->op_first;
assert(o->op_type == OP_ENTER);
- o = o->op_sibling;
+ o = OP_SIBLING(o);
}
if (o->op_type != OP_STUB) {
re_sv = rex->engine->op_comp(aTHX_ &ret, 1, NULL,
rex->engine, NULL, NULL,
/* copy /msix etc to inner pattern */
- scan->flags,
+ ARG2L(scan),
pm_flags);
if (!(SvFLAGS(ret)
cur_eval = st;
/* now continue from first node in postoned RE */
PUSH_YES_STATE_GOTO(EVAL_AB, startpoint, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
case EVAL_AB: /* cleanup after a successful (??{A})B */
}
}
goto fake_end;
- /*NOTREACHED*/
+ /* NOTREACHED */
case GROUPP: /* (?(1)) */
n = ARG(scan); /* which paren pair */
ST.lastloc = NULL; /* this will be updated by WHILEM */
PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next), locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
case CURLYX_end: /* just finished matching all of A*B */
cur_curlyx = ST.prev_curlyx;
sayYES;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case CURLYX_end_fail: /* just failed to match all of A*B */
regcpblow(ST.cp);
cur_curlyx = ST.prev_curlyx;
sayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
#undef ST
{
/* see the discussion above about CURLYX/WHILEM */
I32 n;
- int min = ARG1(cur_curlyx->u.curlyx.me);
- int max = ARG2(cur_curlyx->u.curlyx.me);
- regnode *A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
+ int min, max;
+ regnode *A;
assert(cur_curlyx); /* keep Coverity happy */
+
+ min = ARG1(cur_curlyx->u.curlyx.me);
+ max = ARG2(cur_curlyx->u.curlyx.me);
+ A = NEXTOPER(cur_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS;
n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */
ST.save_lastloc = cur_curlyx->u.curlyx.lastloc;
ST.cache_offset = 0;
REGCP_SET(ST.lastcp);
PUSH_STATE_GOTO(WHILEM_A_pre, A, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
/* If degenerate A matches "", assume A done. */
REGCP_SET(ST.lastcp);
PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B,
locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
/* Prefer A over B for maximal matching. */
cur_curlyx->u.curlyx.lastloc = locinput;
REGCP_SET(ST.lastcp);
PUSH_STATE_GOTO(WHILEM_A_max, A, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
goto do_whilem_B_max;
}
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case WHILEM_B_min: /* just matched B in a minimal match */
case WHILEM_B_max: /* just matched B in a maximal match */
cur_curlyx = ST.save_curlyx;
sayYES;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case WHILEM_B_max_fail: /* just failed to match B in a maximal match */
cur_curlyx = ST.save_curlyx;
cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
cur_curlyx->u.curlyx.count--;
CACHEsayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case WHILEM_A_pre_fail: /* just failed to match even minimal A */
REGCP_UNWIND(ST.lastcp);
regcppop(rex, &maxopenparen);
cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
cur_curlyx->u.curlyx.count--;
CACHEsayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case WHILEM_A_max_fail: /* just failed to match A in a maximal match */
REGCP_UNWIND(ST.lastcp);
cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx;
PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B,
locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case WHILEM_B_min_fail: /* just failed to match B in a minimal match */
cur_curlyx = ST.save_curlyx;
PUSH_STATE_GOTO(WHILEM_A_min,
/*A*/ NEXTOPER(ST.save_curlyx->u.curlyx.me) + EXTRA_STEP_2ARGS,
locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
#undef ST
#define ST st->u.branch
if (next == scan)
next = NULL;
scan = NEXTOPER(scan);
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case BRANCH: /* /(...|A|...)/ */
scan = NEXTOPER(scan); /* scan now points to inner node */
} else {
PUSH_STATE_GOTO(BRANCH_next, scan, locinput);
}
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case CUTGROUP: /* /(*THEN)/ */
sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL :
MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
PUSH_STATE_GOTO(CUTGROUP_next, next, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case CUTGROUP_next_fail:
do_cutgroup = 1;
if (st->u.mark.mark_name)
sv_commit = st->u.mark.mark_name;
sayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case BRANCH_next:
sayYES;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case BRANCH_next_fail: /* that branch failed; try the next, if any */
if (do_cutgroup) {
sayNO_SILENT;
}
continue; /* execute next BRANCH[J] op */
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case MINMOD: /* next op will be non-greedy, e.g. A*? */
minmod = 1;
curlym_do_A: /* execute the A in /A{m,n}B/ */
PUSH_YES_STATE_GOTO(CURLYM_A, ST.A, locinput); /* match A */
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case CURLYM_A: /* we've just matched an A */
ST.count++;
/* calculate c1 and c2 for possible match of 1st char
* following curly */
ST.c1 = ST.c2 = CHRTEST_VOID;
+ assert(ST.B);
if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) {
regnode *text_node = ST.B;
if (! HAS_TEXT(text_node))
}
PUSH_STATE_GOTO(CURLYM_B, ST.B, locinput); /* match B */
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case CURLYM_B_fail: /* just failed to match a B */
REGCP_UNWIND(ST.cp);
REGCP_SET(ST.cp);
goto curly_try_B_max;
}
- assert(0); /* NOTREACHED */
-
+ /* NOTREACHED */
+ assert(0);
case CURLY_B_min_known_fail:
/* failed to find B in a non-greedy match where c1,c2 valid */
}
PUSH_STATE_GOTO(CURLY_B_min_known, ST.B, locinput);
}
- assert(0); /* NOTREACHED */
-
+ /* NOTREACHED */
+ assert(0);
case CURLY_B_min_fail:
/* failed to find B in a non-greedy match where c1,c2 invalid */
}
}
sayNO;
- assert(0); /* NOTREACHED */
-
+ /* NOTREACHED */
+ assert(0);
curly_try_B_max:
/* a successful greedy match: now try to match B */
if (ST.c1 == CHRTEST_VOID || could_match) {
CURLY_SETPAREN(ST.paren, ST.count);
PUSH_STATE_GOTO(CURLY_B_max, ST.B, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
}
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case CURLY_B_max_fail:
/* failed to find B in a greedy match */
/* execute body of (?...A) */
PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan)), newstart);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
case IFMATCH_A_fail: /* body of (?...A) failed */
ST.wanted = !ST.wanted;
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case IFMATCH_A: /* body of (?...A) succeeded */
if (ST.logical) {
if (!scan->flags)
sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]);
PUSH_STATE_GOTO(COMMIT_next, next, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case COMMIT_next_fail:
no_final = 1;
case OPFAIL: /* (*FAIL) */
sayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
#define ST st->u.mark
case MARKPOINT: /* (*MARK:foo) */
mark_state = st;
ST.mark_loc = locinput;
PUSH_YES_STATE_GOTO(MARKPOINT_next, next, locinput);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case MARKPOINT_next:
mark_state = ST.prev_mark;
sayYES;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case MARKPOINT_next_fail:
if (popmark && sv_eq(ST.mark_name,popmark))
sv_yes_mark = mark_state ?
mark_state->u.mark.mark_name : NULL;
sayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
case SKIP: /* (*SKIP) */
if (scan->flags) {
}
no_final = 1;
sayNO;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
#undef ST
case LNBREAK: /* \R */
/* switch break jumps here */
scan = next; /* prepare to execute the next op and ... */
continue; /* ... jump back to the top, reusing st */
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
push_yes_state:
/* push a state that backtracks on success */
st->u.yes.prev_yes_state = yes_state;
yes_state = st;
- /* FALL THROUGH */
+ /* FALLTHROUGH */
push_state:
/* push a new regex state, then continue at scan */
{
locinput = pushinput;
st = newst;
continue;
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
}
* the terminating point.
*/
Perl_croak(aTHX_ "corrupted regexp pointers");
- /*NOTREACHED*/
+ /* NOTREACHED */
sayNO;
yes:
sv_commit = &PL_sv_yes;
sv_yes_mark = &PL_sv_no;
}
+ assert(sv_err);
+ assert(sv_mrk);
sv_setsv(sv_err, sv_commit);
sv_setsv(sv_mrk, sv_yes_mark);
}
S_regrepeat(pTHX_ regexp *prog, char **startposp, const regnode *p,
regmatch_info *const reginfo, I32 max, int depth)
{
- dVAR;
char *scan; /* Pointer to current position in target string */
I32 c;
char *loceol = reginfo->strend; /* local version */
case EXACTFA_NO_TRIE: /* This node only generated for non-utf8 patterns */
assert(! reginfo->is_utf8_pat);
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case EXACTFA:
utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
goto do_exactf;
to_complement = 1;
goto utf8_posix;
}
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case NPOSIXA:
if (! utf8_target) {
else {
/* The complement of something that matches only ASCII matches all
- * UTF-8 variant code points, plus everything in ASCII that isn't
- * in the class. */
+ * non-ASCII, plus everything in ASCII that isn't in the class. */
while (hardcount < max && scan < loceol
- && (! UTF8_IS_INVARIANT(*scan)
+ && (! isASCII_utf8(scan)
|| ! _generic_isCC_A((U8) *scan, FLAGS(p))))
{
scan += UTF8SKIP(scan);
case _CC_ENUM_SPACE: /* XXX would require separate code
if we revert the change of \v
matching this */
- /* FALL THROUGH */
+ /* FALLTHROUGH */
case _CC_ENUM_PSXSPC:
while (hardcount < max
&& scan < loceol
default:
Perl_croak(aTHX_ "panic: regrepeat() called with unrecognized node type %d='%s'", OP(p), PL_reg_name[OP(p)]);
- assert(0); /* NOTREACHED */
+ /* NOTREACHED */
+ assert(0);
}
*altsvp = NULL;
}
- return newSVsv(_get_regclass_nonbitmap_data(prog, node, doinit, listsvp, NULL));
+ return newSVsv(_get_regclass_nonbitmap_data(prog, node, doinit, listsvp, NULL, NULL));
}
-SV *
-Perl__get_regclass_nonbitmap_data(pTHX_ const regexp *prog,
- const regnode* node,
- bool doinit,
- SV** listsvp,
- SV** only_utf8_locale_ptr)
-{
- /* For internal core use only.
- * Returns the swash for the input 'node' in the regex 'prog'.
- * If <doinit> is 'true', will attempt to create the swash if not already
- * done.
- * If <listsvp> is non-null, will return the printable contents of the
- * swash. This can be used to get debugging information even before the
- * swash exists, by calling this function with 'doinit' set to false, in
- * which case the components that will be used to eventually create the
- * swash are returned (in a printable form).
- * Tied intimately to how regcomp.c sets up the data structure */
-
- dVAR;
- SV *sw = NULL;
- SV *si = NULL; /* Input swash initialization string */
- SV* invlist = NULL;
-
- RXi_GET_DECL(prog,progi);
- const struct reg_data * const data = prog ? progi->data : NULL;
-
- PERL_ARGS_ASSERT__GET_REGCLASS_NONBITMAP_DATA;
-
- assert(ANYOF_FLAGS(node)
- & (ANYOF_UTF8|ANYOF_NONBITMAP_NON_UTF8|ANYOF_LOC_FOLD));
-
- if (data && data->count) {
- const U32 n = ARG(node);
-
- if (data->what[n] == 's') {
- SV * const rv = MUTABLE_SV(data->data[n]);
- AV * const av = MUTABLE_AV(SvRV(rv));
- SV **const ary = AvARRAY(av);
- U8 swash_init_flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
-
- si = *ary; /* ary[0] = the string to initialize the swash with */
-
- /* Elements 3 and 4 are either both present or both absent. [3] is
- * any inversion list generated at compile time; [4] indicates if
- * that inversion list has any user-defined properties in it. */
- if (av_tindex(av) >= 2) {
- if (only_utf8_locale_ptr
- && ary[2]
- && ary[2] != &PL_sv_undef)
- {
- *only_utf8_locale_ptr = ary[2];
- }
- else {
- *only_utf8_locale_ptr = NULL;
- }
-
- if (av_tindex(av) >= 3) {
- invlist = ary[3];
- if (SvUV(ary[4])) {
- swash_init_flags |= _CORE_SWASH_INIT_USER_DEFINED_PROPERTY;
- }
- }
- else {
- invlist = NULL;
- }
- }
-
- /* Element [1] is reserved for the set-up swash. If already there,
- * return it; if not, create it and store it there */
- if (ary[1] && SvROK(ary[1])) {
- sw = ary[1];
- }
- else if (doinit && ((si && si != &PL_sv_undef)
- || (invlist && invlist != &PL_sv_undef))) {
-
- sw = _core_swash_init("utf8", /* the utf8 package */
- "", /* nameless */
- si,
- 1, /* binary */
- 0, /* not from tr/// */
- invlist,
- &swash_init_flags);
- (void)av_store(av, 1, sw);
- }
- }
- }
-
- /* If requested, return a printable version of what this swash matches */
- if (listsvp) {
- SV* matches_string = newSVpvn("", 0);
-
- /* The swash should be used, if possible, to get the data, as it
- * contains the resolved data. But this function can be called at
- * compile-time, before everything gets resolved, in which case we
- * return the currently best available information, which is the string
- * that will eventually be used to do that resolving, 'si' */
- if ((! sw || (invlist = _get_swash_invlist(sw)) == NULL)
- && (si && si != &PL_sv_undef))
- {
- sv_catsv(matches_string, si);
- }
-
- /* Add the inversion list to whatever we have. This may have come from
- * the swash, or from an input parameter */
- if (invlist) {
- sv_catsv(matches_string, _invlist_contents(invlist));
- }
- *listsvp = matches_string;
- }
-
- return sw;
-}
#endif /* !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) */
/*
}
/* If this character is potentially in the bitmap, check it */
- if (c < 256) {
+ if (c < NUM_ANYOF_CODE_POINTS) {
if (ANYOF_BITMAP_TEST(n, c))
match = TRUE;
- else if (flags & ANYOF_NON_UTF8_NON_ASCII_ALL
- && ! utf8_target
- && ! isASCII(c))
+ else if ((flags & ANYOF_MATCHES_ALL_NON_UTF8_NON_ASCII)
+ && ! utf8_target
+ && ! isASCII(c))
{
match = TRUE;
}
else if (flags & ANYOF_LOCALE_FLAGS) {
- if (flags & ANYOF_LOC_FOLD) {
- if (ANYOF_BITMAP_TEST(n, PL_fold_locale[c])) {
- match = TRUE;
- }
+ if ((flags & ANYOF_LOC_FOLD)
+ && c < 256
+ && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
+ {
+ match = TRUE;
}
- if (! match && ANYOF_POSIXL_TEST_ANY_SET(n)) {
+ else if (ANYOF_POSIXL_TEST_ANY_SET(n)
+ && c < 256
+ ) {
/* The data structure is arranged so bits 0, 2, 4, ... are set
* if the class includes the Posix character class given by
/* If the bitmap didn't (or couldn't) match, and something outside the
* bitmap could match, try that. */
if (!match) {
- if (c >= 256 && (flags & ANYOF_ABOVE_LATIN1_ALL)) {
- match = TRUE; /* Everything above 255 matches */
+ if (c >= NUM_ANYOF_CODE_POINTS
+ && (flags & ANYOF_MATCHES_ALL_ABOVE_BITMAP))
+ {
+ match = TRUE; /* Everything above the bitmap matches */
}
- else if ((flags & ANYOF_NONBITMAP_NON_UTF8)
- || (utf8_target && (flags & ANYOF_UTF8))
+ else if ((flags & ANYOF_HAS_NONBITMAP_NON_UTF8_MATCHES)
+ || (utf8_target && (flags & ANYOF_HAS_UTF8_NONBITMAP_MATCHES))
|| ((flags & ANYOF_LOC_FOLD)
&& IN_UTF8_CTYPE_LOCALE
- && ARG(n) != ANYOF_NONBITMAP_EMPTY))
+ && ARG(n) != ANYOF_ONLY_HAS_BITMAP))
{
SV* only_utf8_locale = NULL;
SV * const sw = _get_regclass_nonbitmap_data(prog, n, TRUE, 0,
- &only_utf8_locale);
+ &only_utf8_locale, NULL);
if (sw) {
+ U8 utf8_buffer[2];
U8 * utf8_p;
if (utf8_target) {
utf8_p = (U8 *) p;
} else { /* Convert to utf8 */
- STRLEN len = 1;
- utf8_p = bytes_to_utf8(p, &len);
+ utf8_p = utf8_buffer;
+ append_utf8_from_native_byte(*p, &utf8_p);
+ utf8_p = utf8_buffer;
}
if (swash_fetch(sw, utf8_p, TRUE)) {
match = TRUE;
}
-
- /* If we allocated a string above, free it */
- if (! utf8_target) Safefree(utf8_p);
}
if (! match && only_utf8_locale && IN_UTF8_CTYPE_LOCALE) {
match = _invlist_contains_cp(only_utf8_locale, c);
* 'off' >= 0, backwards if negative. But don't go outside of position
* 'lim', which better be < s if off < 0 */
- dVAR;
-
PERL_ARGS_ASSERT_REGHOP3;
if (off >= 0) {
STATIC U8 *
S_reghop4(U8 *s, SSize_t off, const U8* llim, const U8* rlim)
{
- dVAR;
-
PERL_ARGS_ASSERT_REGHOP4;
if (off >= 0) {
STATIC U8 *
S_reghopmaybe3(U8* s, SSize_t off, const U8* lim)
{
- dVAR;
-
PERL_ARGS_ASSERT_REGHOPMAYBE3;
if (off >= 0) {
static void
S_cleanup_regmatch_info_aux(pTHX_ void *arg)
{
- dVAR;
regmatch_info_aux *aux = (regmatch_info_aux *) arg;
regmatch_info_aux_eval *eval_state = aux->info_aux_eval;
regmatch_slab *s;
/* Converts substr fields in prog from UTF-8 to bytes, calling fbm_compile
* on the converted value; returns FALSE if can't be converted. */
- dVAR;
int i = 1;
PERL_ARGS_ASSERT_TO_BYTE_SUBSTR;
https://perl5.git.perl.org / perl5.git / blobdiff