# include "regcomp.h"
#endif
-#define RF_tainted 1 /* tainted information used? */
+#define RF_tainted 1 /* tainted information used? e.g. locale */
#define RF_warned 2 /* warned about big count? */
#define RF_utf8 8 /* Pattern contains multibyte chars? */
#define STATIC static
#endif
-/* Valid for non-utf8 strings only: avoids the reginclass call if there are no
- * complications: i.e., if everything matchable is straight forward in the
- * bitmap */
+/* Valid for non-utf8 strings, non-ANYOFV nodes only: avoids the reginclass
+ * call if there are no complications: i.e., if everything matchable is
+ * straight forward in the bitmap */
#define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) \
: ANYOF_BITMAP_TEST(p,*(c)))
/* these are unrolled below in the CCC_TRY_XXX defined */
#define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \
- if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END
+ if (!CAT2(PL_utf8_,class)) { \
+ bool ok; \
+ ENTER; save_re_context(); \
+ ok=CAT2(is_utf8_,class)((const U8*)str); \
+ assert(ok); LEAVE; } } STMT_END
/* Doesn't do an assert to verify that is correct */
#define LOAD_UTF8_CHARCLASS_NO_CHECK(class) STMT_START { \
- if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)" "); LEAVE; } } STMT_END
+ if (!CAT2(PL_utf8_,class)) { \
+ bool throw_away __attribute__unused__; \
+ ENTER; save_re_context(); \
+ throw_away = CAT2(is_utf8_,class)((const U8*)" "); \
+ LEAVE; } } STMT_END
#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
LOAD_UTF8_CHARCLASS_NO_CHECK(X_T); /* U+11A8 "\xe1\x86\xa8" */ \
LOAD_UTF8_CHARCLASS_NO_CHECK(X_V) /* U+1160 "\xe1\x85\xa0" */
-/*
- We dont use PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS as the direct test
- so that it is possible to override the option here without having to
- rebuild the entire core. as we are required to do if we change regcomp.h
- which is where PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS is defined.
-*/
-#if PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS
-#define BROKEN_UNICODE_CHARCLASS_MAPPINGS
-#endif
-
-#ifdef BROKEN_UNICODE_CHARCLASS_MAPPINGS
-#define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS_ALNUM()
-#define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS_SPACE()
-#define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS_DIGIT()
-#define RE_utf8_perl_word PL_utf8_alnum
-#define RE_utf8_perl_space PL_utf8_space
-#define RE_utf8_posix_digit PL_utf8_digit
-#define perl_word alnum
-#define perl_space space
-#define posix_digit digit
-#else
-#define LOAD_UTF8_CHARCLASS_PERL_WORD() LOAD_UTF8_CHARCLASS(perl_word,"a")
-#define LOAD_UTF8_CHARCLASS_PERL_SPACE() LOAD_UTF8_CHARCLASS(perl_space," ")
-#define LOAD_UTF8_CHARCLASS_POSIX_DIGIT() LOAD_UTF8_CHARCLASS(posix_digit,"0")
-#define RE_utf8_perl_word PL_utf8_perl_word
-#define RE_utf8_perl_space PL_utf8_perl_space
-#define RE_utf8_posix_digit PL_utf8_posix_digit
-#endif
-
-
-#define _CCC_TRY_AFF_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC) \
- case NAMEL: \
- PL_reg_flags |= RF_tainted; \
- /* FALL THROUGH */ \
- case NAME: \
- if (!nextchr) \
- sayNO; \
- if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
- if (!CAT2(PL_utf8_,CLASS)) { \
- bool ok; \
- ENTER; \
- save_re_context(); \
- ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
- assert(ok); \
- LEAVE; \
- } \
- if (!(OP(scan) == NAME \
- ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, utf8_target)) \
- : LCFUNC_utf8((U8*)locinput))) \
- { \
- sayNO; \
- } \
- locinput += PL_utf8skip[nextchr]; \
- nextchr = UCHARAT(locinput); \
- break; \
- } \
- /* Drops through to the macro that calls this one */
-
-#define CCC_TRY_AFF(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
- _CCC_TRY_AFF_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC) \
- if (!(OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
- sayNO; \
- nextchr = UCHARAT(++locinput); \
- break
-
-/* Almost identical to the above, but has a case for a node that matches chars
- * between 128 and 255 using Unicode (latin1) semantics. */
-#define CCC_TRY_AFF_U(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNCU,LCFUNC) \
- _CCC_TRY_AFF_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC) \
- if (!(OP(scan) == NAMEL ? LCFUNC(nextchr) : (FUNCU(nextchr) && (isASCII(nextchr) || (FLAGS(scan) & USE_UNI))))) \
- sayNO; \
- nextchr = UCHARAT(++locinput); \
- break
-
-#define _CCC_TRY_NEG_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC) \
- case NAMEL: \
- PL_reg_flags |= RF_tainted; \
- /* FALL THROUGH */ \
- case NAME : \
- if (!nextchr && locinput >= PL_regeol) \
- sayNO; \
- if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
- if (!CAT2(PL_utf8_,CLASS)) { \
- bool ok; \
- ENTER; \
- save_re_context(); \
- ok=CAT2(is_utf8_,CLASS)((const U8*)STR); \
- assert(ok); \
- LEAVE; \
- } \
- if ((OP(scan) == NAME \
- ? cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), (U8*)locinput, utf8_target)) \
- : LCFUNC_utf8((U8*)locinput))) \
- { \
- sayNO; \
- } \
- locinput += PL_utf8skip[nextchr]; \
- nextchr = UCHARAT(locinput); \
- break; \
- }
-
-#define CCC_TRY_NEG(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC,LCFUNC) \
- _CCC_TRY_NEG_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNC) \
- if ((OP(scan) == NAME ? FUNC(nextchr) : LCFUNC(nextchr))) \
- sayNO; \
- nextchr = UCHARAT(++locinput); \
- break
-
-
-#define CCC_TRY_NEG_U(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNCU,LCFUNC) \
- _CCC_TRY_NEG_COMMON(NAME,NAMEL,CLASS,STR,LCFUNC_utf8,FUNCU) \
- if ((OP(scan) == NAMEL ? LCFUNC(nextchr) : (FUNCU(nextchr) && (isASCII(nextchr) || (FLAGS(scan) & USE_UNI))))) \
- sayNO; \
- nextchr = UCHARAT(++locinput); \
- break
-
-
+#define PLACEHOLDER /* Something for the preprocessor to grab onto */
+
+/* The actual code for CCC_TRY, which uses several variables from the routine
+ * it's callable from. It is designed to be the bulk of a case statement.
+ * FUNC is the macro or function to call on non-utf8 targets that indicate if
+ * nextchr matches the class.
+ * UTF8_TEST is the whole test string to use for utf8 targets
+ * LOAD is what to use to test, and if not present to load in the swash for the
+ * class
+ * POS_OR_NEG is either empty or ! to complement the results of FUNC or
+ * UTF8_TEST test.
+ * The logic is: Fail if we're at the end-of-string; otherwise if the target is
+ * utf8 and a variant, load the swash if necessary and test using the utf8
+ * test. Advance to the next character if test is ok, otherwise fail; If not
+ * utf8 or an invariant under utf8, use the non-utf8 test, and fail if it
+ * fails, or advance to the next character */
+
+#define _CCC_TRY_CODE(POS_OR_NEG, FUNC, UTF8_TEST, CLASS, STR) \
+ if (locinput >= PL_regeol) { \
+ sayNO; \
+ } \
+ if (utf8_target && UTF8_IS_CONTINUED(nextchr)) { \
+ LOAD_UTF8_CHARCLASS(CLASS, STR); \
+ if (POS_OR_NEG (UTF8_TEST)) { \
+ sayNO; \
+ } \
+ locinput += PL_utf8skip[nextchr]; \
+ nextchr = UCHARAT(locinput); \
+ break; \
+ } \
+ if (POS_OR_NEG (FUNC(nextchr))) { \
+ sayNO; \
+ } \
+ nextchr = UCHARAT(++locinput); \
+ break;
+
+/* Handle the non-locale cases for a character class and its complement. It
+ * calls _CCC_TRY_CODE with a ! to complement the test for the character class.
+ * This is because that code fails when the test succeeds, so we want to have
+ * the test fail so that the code succeeds. The swash is stored in a
+ * predictable PL_ place */
+#define _CCC_TRY_NONLOCALE(NAME, NNAME, FUNC, \
+ CLASS, STR) \
+ case NAME: \
+ _CCC_TRY_CODE( !, FUNC, \
+ cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), \
+ (U8*)locinput, TRUE)), \
+ CLASS, STR) \
+ case NNAME: \
+ _CCC_TRY_CODE( PLACEHOLDER , FUNC, \
+ cBOOL(swash_fetch(CAT2(PL_utf8_,CLASS), \
+ (U8*)locinput, TRUE)), \
+ CLASS, STR) \
+
+/* Generate the case statements for both locale and non-locale character
+ * classes in regmatch for classes that don't have special unicode semantics.
+ * Locales don't use an immediate swash, but an intermediary special locale
+ * function that is called on the pointer to the current place in the input
+ * string. That function will resolve to needing the same swash. One might
+ * think that because we don't know what the locale will match, we shouldn't
+ * check with the swash loading function that it loaded properly; ie, that we
+ * should use LOAD_UTF8_CHARCLASS_NO_CHECK for those, but what is passed to the
+ * regular LOAD_UTF8_CHARCLASS is in non-locale terms, and so locale is
+ * irrelevant here */
+#define CCC_TRY(NAME, NNAME, FUNC, \
+ NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
+ NAMEA, NNAMEA, FUNCA, \
+ CLASS, STR) \
+ case NAMEL: \
+ PL_reg_flags |= RF_tainted; \
+ _CCC_TRY_CODE( !, LCFUNC, LCFUNC_utf8((U8*)locinput), CLASS, STR) \
+ case NNAMEL: \
+ PL_reg_flags |= RF_tainted; \
+ _CCC_TRY_CODE( PLACEHOLDER, LCFUNC, LCFUNC_utf8((U8*)locinput), \
+ CLASS, STR) \
+ case NAMEA: \
+ if (locinput >= PL_regeol || ! FUNCA(nextchr)) { \
+ sayNO; \
+ } \
+ /* Matched a utf8-invariant, so don't have to worry about utf8 */ \
+ nextchr = UCHARAT(++locinput); \
+ break; \
+ case NNAMEA: \
+ if (locinput >= PL_regeol || FUNCA(nextchr)) { \
+ sayNO; \
+ } \
+ if (utf8_target) { \
+ locinput += PL_utf8skip[nextchr]; \
+ nextchr = UCHARAT(locinput); \
+ } \
+ else { \
+ nextchr = UCHARAT(++locinput); \
+ } \
+ break; \
+ /* Generate the non-locale cases */ \
+ _CCC_TRY_NONLOCALE(NAME, NNAME, FUNC, CLASS, STR)
+
+/* This is like CCC_TRY, but has an extra set of parameters for generating case
+ * statements to handle separate Unicode semantics nodes */
+#define CCC_TRY_U(NAME, NNAME, FUNC, \
+ NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
+ NAMEU, NNAMEU, FUNCU, \
+ NAMEA, NNAMEA, FUNCA, \
+ CLASS, STR) \
+ CCC_TRY(NAME, NNAME, FUNC, \
+ NAMEL, NNAMEL, LCFUNC, LCFUNC_utf8, \
+ NAMEA, NNAMEA, FUNCA, \
+ CLASS, STR) \
+ _CCC_TRY_NONLOCALE(NAMEU, NNAMEU, FUNCU, CLASS, STR)
/* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */
/* Currently these are only used when PL_regkind[OP(rn)] == EXACT so
we don't need this definition. */
#define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF )
-#define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF )
+#define IS_TEXTF(rn) ( (OP(rn)==EXACTFU || OP(rn)==EXACTFA || OP(rn)==EXACTF) || OP(rn)==REFF || OP(rn)==NREFF )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL )
#else
/* ... so we use this as its faster. */
#define IS_TEXT(rn) ( OP(rn)==EXACT )
+#define IS_TEXTFU(rn) ( OP(rn)==EXACTFU || OP(rn) == EXACTFA)
#define IS_TEXTF(rn) ( OP(rn)==EXACTF )
#define IS_TEXTFL(rn) ( OP(rn)==EXACTFL )
a) Anchored substring;
b) Fixed substring;
c) Whether we are anchored (beginning-of-line or \G);
- d) First node (of those at offset 0) which may distingush positions;
+ d) First node (of those at offset 0) which may distinguish positions;
We use a)b)d) and multiline-part of c), and try to find a position in the
string which does not contradict any of them.
*/
prog->float_substr = prog->float_utf8 = NULL; /* clear */
check = NULL; /* abort */
s = strpos;
- /* XXXX If the check string was an implicit check MBOL, then we need to unset the relevent flag
+ /* 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->extflags &= ~RXf_ANCH_MBOL;
even for \b or \B. But (minlen? 1 : 0) below assumes that
regstclass does not come from lookahead... */
/* If regstclass takes bytelength more than 1: If charlength==1, OK.
- This leaves EXACTF only, which is dealt with in find_byclass(). */
+ This leaves EXACTF-ish only, which are dealt with in find_byclass(). */
const U8* const str = (U8*)STRING(progi->regstclass);
const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT
? CHR_DIST(str+STR_LEN(progi->regstclass), str)
}
if (!(utf8_target ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
goto fail;
- /* Check is floating subtring. */
+ /* Check is floating substring. */
retry_floating_check:
t = check_at - start_shift;
DEBUG_EXECUTE_r( what = "floating" );
} \
} STMT_END
-#define REXEC_FBC_EXACTISH_CHECK(CoNd) \
-{ \
- char *my_strend= (char *)strend; \
- if ( (CoNd) \
- && (ln == len || \
- foldEQ_utf8(s, &my_strend, 0, utf8_target, \
- m, NULL, ln, cBOOL(UTF_PATTERN))) \
- && (!reginfo || regtry(reginfo, &s)) ) \
- goto got_it; \
- else { \
- U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \
- uvchr_to_utf8(tmpbuf, c); \
- f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \
- if ( f != c \
- && (f == c1 || f == c2) \
- && (ln == len || \
- foldEQ_utf8(s, &my_strend, 0, utf8_target,\
- m, NULL, ln, cBOOL(UTF_PATTERN)))\
- && (!reginfo || regtry(reginfo, &s)) ) \
- goto got_it; \
- } \
-} \
-s += len
-
#define REXEC_FBC_EXACTISH_SCAN(CoNd) \
STMT_START { \
while (s <= e) { \
if ( (CoNd) \
- && (ln == 1 || (OP(c) == EXACTF \
- ? foldEQ(s, m, ln) \
- : foldEQ_locale(s, m, ln))) \
- && (!reginfo || regtry(reginfo, &s)) ) \
+ && (ln == 1 || folder(s, pat_string, ln)) \
+ && (!reginfo || regtry(reginfo, &s)) ) \
goto got_it; \
s++; \
} \
} \
else { \
REXEC_FBC_CLASS_SCAN(CoNd); \
- } \
- break
+ }
#define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \
if (utf8_target) { \
} \
else { \
REXEC_FBC_CLASS_SCAN(CoNd); \
- } \
- break
+ }
#define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \
PL_reg_flags |= RF_tainted; \
} \
else { \
REXEC_FBC_CLASS_SCAN(CoNd); \
- } \
- break
+ }
#define DUMP_EXEC_POS(li,s,doutf8) \
dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8)
+
+#define UTF8_NOLOAD(TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
+ tmp = (s != PL_bostr) ? 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 == PL_bostr) { \
+ tmp = '\n'; \
+ } \
+ else { \
+ U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); \
+ tmp = utf8n_to_uvchr(r, UTF8SKIP(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; \
+ } \
+ ); \
+
+/* 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 */
+#define FBC_BOUND_COMMON(UTF8_CODE, TEST_NON_UTF8, IF_SUCCESS, IF_FAIL) \
+ if (utf8_target) { \
+ UTF8_CODE \
+ } \
+ else { /* Not utf8 */ \
+ tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; \
+ tmp = TEST_NON_UTF8(tmp); \
+ REXEC_FBC_SCAN( \
+ if (tmp == ! TEST_NON_UTF8((U8) *s)) { \
+ tmp = !tmp; \
+ IF_SUCCESS; \
+ } \
+ else { \
+ IF_FAIL; \
+ } \
+ ); \
+ } \
+ if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) \
+ goto got_it;
+
/* We know what class REx starts with. Try to find this position... */
/* if reginfo is NULL, its a dryrun */
/* annoyingly all the vars in this routine have different names from their counterparts
{
dVAR;
const I32 doevery = (prog->intflags & PREGf_SKIP) == 0;
- char *m;
+ char *pat_string; /* The pattern's exactish string */
+ char *pat_end; /* ptr to end char of pat_string */
+ re_fold_t folder; /* Function for computing non-utf8 folds */
+ const U8 *fold_array; /* array for folding ords < 256 */
STRLEN ln;
STRLEN lnc;
register STRLEN uskip;
- unsigned int c1;
- unsigned int c2;
+ U8 c1;
+ U8 c2;
char *e;
register I32 tmp = 1; /* Scratch variable? */
register const bool utf8_target = PL_reg_match_utf8;
+ UV utf8_fold_flags = 0;
RXi_GET_DECL(prog,progi);
PERL_ARGS_ASSERT_FIND_BYCLASS;
/* We know what class it must start with. */
switch (OP(c)) {
+ case ANYOFV:
case ANYOF:
- if (utf8_target) {
- REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
- !UTF8_IS_INVARIANT((U8)s[0]) ?
- reginclass(prog, c, (U8*)s, 0, utf8_target) :
- REGINCLASS(prog, c, (U8*)s));
+ if (utf8_target || OP(c) == ANYOFV) {
+ STRLEN inclasslen = strend - s;
+ REXEC_FBC_UTF8_CLASS_SCAN(
+ reginclass(prog, c, (U8*)s, &inclasslen, utf8_target));
}
else {
- while (s < strend) {
- STRLEN skip = 1;
-
- if (REGINCLASS(prog, c, (U8*)s) ||
- (ANYOF_FOLD_SHARP_S(c, s, strend) &&
- /* The assignment of 2 is intentional:
- * for the folded sharp s, the skip is 2. */
- (skip = SHARP_S_SKIP))) {
- if (tmp && (!reginfo || regtry(reginfo, &s)))
- goto got_it;
- else
- tmp = doevery;
- }
- else
- tmp = 1;
- s += skip;
- }
+ REXEC_FBC_CLASS_SCAN(REGINCLASS(prog, c, (U8*)s));
}
break;
case CANY:
tmp = doevery;
);
break;
- case EXACTF:
- m = STRING(c);
- ln = STR_LEN(c); /* length to match in octets/bytes */
- lnc = (I32) ln; /* length to match in characters */
- if (UTF_PATTERN) {
- STRLEN ulen1, ulen2;
- U8 *sm = (U8 *) m;
- U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
- U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
- /* used by commented-out code below */
- /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/
-
- /* XXX: Since the node will be case folded at compile
- time this logic is a little odd, although im not
- sure that its actually wrong. --dmq */
-
- c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
- c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
-
- /* XXX: This is kinda strange. to_utf8_XYZ returns the
- codepoint of the first character in the converted
- form, yet originally we did the extra step.
- No tests fail by commenting this code out however
- so Ive left it out. -- dmq.
-
- c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE,
- 0, uniflags);
- c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE,
- 0, uniflags);
- */
-
- lnc = 0;
- while (sm < ((U8 *) m + ln)) {
- lnc++;
- sm += UTF8SKIP(sm);
- }
+
+ case EXACTFA:
+ if (UTF_PATTERN || utf8_target) {
+ utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
+ goto do_exactf_utf8;
}
- else {
- c1 = *(U8*)m;
- c2 = PL_fold[c1];
+ fold_array = PL_fold_latin1; /* Latin1 folds are not affected by */
+ folder = foldEQ_latin1; /* /a, except the sharp s one which */
+ goto do_exactf_non_utf8; /* isn't dealt with by these */
+
+ case EXACTFU:
+ if (UTF_PATTERN || utf8_target) {
+ utf8_fold_flags = 0;
+ goto do_exactf_utf8;
}
- goto do_exactf;
+ fold_array = PL_fold_latin1;
+ folder = foldEQ_latin1;
+ /* XXX This uses the full utf8 fold because if the pattern contains
+ * 'ss' it could match LATIN_SMALL_LETTER SHARP_S in the string.
+ * There could be a new node type, say EXACTFU_SS, which is
+ * generated by regcomp only if there is an 'ss', and then every
+ * other case could goto do_exactf_non_utf8;*/
+ goto do_exactf_utf8;
+
+ case EXACTF:
+ if (UTF_PATTERN || utf8_target) {
+ utf8_fold_flags = 0;
+ goto do_exactf_utf8;
+ }
+ fold_array = PL_fold;
+ folder = foldEQ;
+ goto do_exactf_non_utf8;
+
case EXACTFL:
- m = STRING(c);
- ln = STR_LEN(c);
- lnc = (I32) ln;
- c1 = *(U8*)m;
- c2 = PL_fold_locale[c1];
- do_exactf:
- e = HOP3c(strend, -((I32)lnc), s);
+ if (UTF_PATTERN || utf8_target) {
+ utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
+ goto do_exactf_utf8;
+ }
+ fold_array = PL_fold_locale;
+ folder = foldEQ_locale;
- if (!reginfo && e < s)
- e = s; /* Due to minlen logic of intuit() */
+ /* FALL THROUGH */
- /* The idea in the EXACTF* cases is to first find the
- * first character of the EXACTF* node and then, if
- * necessary, case-insensitively compare the full
- * text of the node. The c1 and c2 are the first
- * characters (though in Unicode it gets a bit
- * more complicated because there are more cases
- * than just upper and lower: one needs to use
- * the so-called folding case for case-insensitive
- * matching (called "loose matching" in Unicode).
- * foldEQ_utf8() will do just that. */
+ do_exactf_non_utf8: /* Neither pattern nor string are UTF8 */
- if (utf8_target || UTF_PATTERN) {
- UV c, f;
- U8 tmpbuf [UTF8_MAXBYTES+1];
- STRLEN len = 1;
- STRLEN foldlen;
- const U32 uniflags = UTF8_ALLOW_DEFAULT;
- if (c1 == c2) {
- /* Upper and lower of 1st char are equal -
- * probably not a "letter". */
- while (s <= e) {
- if (utf8_target) {
- c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
- uniflags);
- } else {
- c = *((U8*)s);
- }
- REXEC_FBC_EXACTISH_CHECK(c == c1);
- }
- }
- else {
- while (s <= e) {
- if (utf8_target) {
- c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
- uniflags);
- } else {
- c = *((U8*)s);
- }
+ /* The idea in the non-utf8 EXACTF* cases is to first find the
+ * first character of the EXACTF* node and then, if necessary,
+ * case-insensitively compare the full text of the node. c1 is the
+ * first character. c2 is its fold. This logic will not work for
+ * Unicode semantics and the german sharp ss, which hence should
+ * not be compiled into a node that gets here. */
+ pat_string = STRING(c);
+ ln = STR_LEN(c); /* length to match in octets/bytes */
- /* Handle some of the three Greek sigmas cases.
- * Note that not all the possible combinations
- * are handled here: some of them are handled
- * by the standard folding rules, and some of
- * them (the character class or ANYOF cases)
- * are handled during compiletime in
- * regexec.c:S_regclass(). */
- if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA ||
- c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA)
- c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA;
-
- REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2);
- }
- }
+ e = HOP3c(strend, -((I32)ln), s);
+
+ if (!reginfo && e < s) {
+ e = s; /* Due to minlen logic of intuit() */
+ }
+
+ c1 = *pat_string;
+ c2 = fold_array[c1];
+ if (c1 == c2) { /* If char and fold are the same */
+ REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
}
else {
- /* Neither pattern nor string are UTF8 */
- if (c1 == c2)
- REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1);
- else
- REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
+ REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2);
}
break;
- case BOUNDL:
- PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
- case BOUND:
- if (utf8_target) {
- if (s == PL_bostr)
- tmp = '\n';
- else {
- U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
- tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
- }
- tmp = ((OP(c) == BOUND ?
- isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
- LOAD_UTF8_CHARCLASS_ALNUM();
- REXEC_FBC_UTF8_SCAN(
- if (tmp == !(OP(c) == BOUND ?
- cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
- isALNUM_LC_utf8((U8*)s)))
- {
- tmp = !tmp;
- REXEC_FBC_TRYIT;
- }
- );
+
+ do_exactf_utf8:
+
+ /* If one of the operands is in utf8, we can't use the simpler
+ * folding above, due to the fact that many different characters
+ * can have the same fold, or portion of a fold, or different-
+ * length fold */
+ pat_string = STRING(c);
+ ln = STR_LEN(c); /* length to match in octets/bytes */
+ pat_end = pat_string + ln;
+ lnc = (UTF_PATTERN) /* length to match in characters */
+ ? utf8_length((U8 *) pat_string, (U8 *) pat_end)
+ : ln;
+
+ e = HOP3c(strend, -((I32)lnc), s);
+
+ if (!reginfo && e < s) {
+ e = s; /* Due to minlen logic of intuit() */
}
- else { /* Not utf8 */
- tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
- tmp = cBOOL((OP(c) == BOUNDL)
- ? isALNUM_LC(tmp)
- : (isWORDCHAR_L1(tmp)
- && (isASCII(tmp) || (FLAGS(c) & USE_UNI))));
- REXEC_FBC_SCAN(
- if (tmp ==
- !((OP(c) == BOUNDL)
- ? isALNUM_LC(*s)
- : (isWORDCHAR_L1((U8) *s)
- && (isASCII((U8) *s) || (FLAGS(c) & USE_UNI)))))
- {
- tmp = !tmp;
- REXEC_FBC_TRYIT;
+
+ while (s <= e) {
+ char *my_strend= (char *)strend;
+ if (foldEQ_utf8_flags(s, &my_strend, 0, utf8_target,
+ pat_string, NULL, ln, cBOOL(UTF_PATTERN), utf8_fold_flags)
+ && (!reginfo || regtry(reginfo, &s)) )
+ {
+ goto got_it;
}
- );
+ s += UTF8SKIP(s);
}
- if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s)))
- goto got_it;
+ break;
+ case BOUNDL:
+ PL_reg_flags |= RF_tainted;
+ FBC_BOUND(isALNUM_LC,
+ isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
+ isALNUM_LC_utf8((U8*)s));
break;
case NBOUNDL:
PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
+ FBC_NBOUND(isALNUM_LC,
+ isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp)),
+ isALNUM_LC_utf8((U8*)s));
+ break;
+ case BOUND:
+ FBC_BOUND(isWORDCHAR,
+ isALNUM_uni(tmp),
+ cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+ break;
+ case BOUNDA:
+ FBC_BOUND_NOLOAD(isWORDCHAR_A,
+ isWORDCHAR_A(tmp),
+ isWORDCHAR_A((U8*)s));
+ break;
case NBOUND:
- if (utf8_target) {
- if (s == PL_bostr)
- tmp = '\n';
- else {
- U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
- tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT);
- }
- tmp = ((OP(c) == NBOUND ?
- isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
- LOAD_UTF8_CHARCLASS_ALNUM();
- REXEC_FBC_UTF8_SCAN(
- if (tmp == !(OP(c) == NBOUND ?
- cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)) :
- isALNUM_LC_utf8((U8*)s)))
- tmp = !tmp;
- else REXEC_FBC_TRYIT;
- );
- }
- else {
- tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n';
- tmp = cBOOL((OP(c) == NBOUNDL)
- ? isALNUM_LC(tmp)
- : (isWORDCHAR_L1(tmp)
- && (isASCII(tmp) || (FLAGS(c) & USE_UNI))));
- REXEC_FBC_SCAN(
- if (tmp == ! cBOOL(
- (OP(c) == NBOUNDL)
- ? isALNUM_LC(*s)
- : (isWORDCHAR_L1((U8) *s)
- && (isASCII((U8) *s) || (FLAGS(c) & USE_UNI)))))
- {
- tmp = !tmp;
- }
- else REXEC_FBC_TRYIT;
- );
- }
- if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s)))
- goto got_it;
+ FBC_NBOUND(isWORDCHAR,
+ isALNUM_uni(tmp),
+ cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+ break;
+ case NBOUNDA:
+ FBC_NBOUND_NOLOAD(isWORDCHAR_A,
+ isWORDCHAR_A(tmp),
+ isWORDCHAR_A((U8*)s));
+ break;
+ case BOUNDU:
+ FBC_BOUND(isWORDCHAR_L1,
+ isALNUM_uni(tmp),
+ cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
+ break;
+ case NBOUNDU:
+ FBC_NBOUND(isWORDCHAR_L1,
+ isALNUM_uni(tmp),
+ cBOOL(swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target)));
break;
- case ALNUM:
- REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_PERL_WORD(),
- swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
- (FLAGS(c) & USE_UNI) ? isWORDCHAR_L1((U8) *s) : isALNUM(*s)
- );
case ALNUML:
REXEC_FBC_CSCAN_TAINT(
isALNUM_LC_utf8((U8*)s),
isALNUM_LC(*s)
);
+ break;
+ case ALNUMU:
+ REXEC_FBC_CSCAN_PRELOAD(
+ LOAD_UTF8_CHARCLASS_ALNUM(),
+ swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
+ isWORDCHAR_L1((U8) *s)
+ );
+ break;
+ case ALNUM:
+ REXEC_FBC_CSCAN_PRELOAD(
+ LOAD_UTF8_CHARCLASS_ALNUM(),
+ swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
+ isWORDCHAR((U8) *s)
+ );
+ break;
+ case ALNUMA:
+ /* Don't need to worry about utf8, as it can match only a single
+ * byte invariant character */
+ REXEC_FBC_CLASS_SCAN( isWORDCHAR_A(*s));
+ break;
+ case NALNUMU:
+ REXEC_FBC_CSCAN_PRELOAD(
+ LOAD_UTF8_CHARCLASS_ALNUM(),
+ !swash_fetch(PL_utf8_alnum,(U8*)s, utf8_target),
+ ! isWORDCHAR_L1((U8) *s)
+ );
+ break;
case NALNUM:
REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_PERL_WORD(),
- !swash_fetch(RE_utf8_perl_word, (U8*)s, utf8_target),
- ! ((FLAGS(c) & USE_UNI) ? isWORDCHAR_L1((U8) *s) : isALNUM(*s))
+ LOAD_UTF8_CHARCLASS_ALNUM(),
+ !swash_fetch(PL_utf8_alnum, (U8*)s, utf8_target),
+ ! isALNUM(*s)
);
+ break;
+ case NALNUMA:
+ REXEC_FBC_CSCAN(
+ !isWORDCHAR_A(*s),
+ !isWORDCHAR_A(*s)
+ );
+ break;
case NALNUML:
REXEC_FBC_CSCAN_TAINT(
!isALNUM_LC_utf8((U8*)s),
!isALNUM_LC(*s)
);
+ break;
+ case SPACEU:
+ REXEC_FBC_CSCAN_PRELOAD(
+ LOAD_UTF8_CHARCLASS_SPACE(),
+ *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
+ isSPACE_L1((U8) *s)
+ );
+ break;
case SPACE:
REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_PERL_SPACE(),
- *s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target),
- isSPACE_L1((U8) *s) && (isASCII((U8) *s) || (FLAGS(c) & USE_UNI))
+ LOAD_UTF8_CHARCLASS_SPACE(),
+ *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target),
+ isSPACE((U8) *s)
);
+ break;
+ case SPACEA:
+ /* Don't need to worry about utf8, as it can match only a single
+ * byte invariant character */
+ REXEC_FBC_CLASS_SCAN( isSPACE_A(*s));
+ break;
case SPACEL:
REXEC_FBC_CSCAN_TAINT(
- *s == ' ' || isSPACE_LC_utf8((U8*)s),
+ isSPACE_LC_utf8((U8*)s),
isSPACE_LC(*s)
);
+ break;
+ case NSPACEU:
+ REXEC_FBC_CSCAN_PRELOAD(
+ LOAD_UTF8_CHARCLASS_SPACE(),
+ !( *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
+ ! isSPACE_L1((U8) *s)
+ );
+ break;
case NSPACE:
REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_PERL_SPACE(),
- !(*s == ' ' || swash_fetch(RE_utf8_perl_space,(U8*)s, utf8_target)),
- !(isSPACE_L1((U8) *s) && (isASCII((U8) *s) || (FLAGS(c) & USE_UNI)))
+ LOAD_UTF8_CHARCLASS_SPACE(),
+ !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, utf8_target)),
+ ! isSPACE((U8) *s)
);
+ break;
+ case NSPACEA:
+ REXEC_FBC_CSCAN(
+ !isSPACE_A(*s),
+ !isSPACE_A(*s)
+ );
+ break;
case NSPACEL:
REXEC_FBC_CSCAN_TAINT(
- !(*s == ' ' || isSPACE_LC_utf8((U8*)s)),
+ !isSPACE_LC_utf8((U8*)s),
!isSPACE_LC(*s)
);
+ break;
case DIGIT:
REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
- swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
+ LOAD_UTF8_CHARCLASS_DIGIT(),
+ swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
isDIGIT(*s)
);
+ break;
+ case DIGITA:
+ /* Don't need to worry about utf8, as it can match only a single
+ * byte invariant character */
+ REXEC_FBC_CLASS_SCAN( isDIGIT_A(*s));
+ break;
case DIGITL:
REXEC_FBC_CSCAN_TAINT(
isDIGIT_LC_utf8((U8*)s),
isDIGIT_LC(*s)
);
+ break;
case NDIGIT:
REXEC_FBC_CSCAN_PRELOAD(
- LOAD_UTF8_CHARCLASS_POSIX_DIGIT(),
- !swash_fetch(RE_utf8_posix_digit,(U8*)s, utf8_target),
+ LOAD_UTF8_CHARCLASS_DIGIT(),
+ !swash_fetch(PL_utf8_digit,(U8*)s, utf8_target),
!isDIGIT(*s)
);
+ break;
+ case NDIGITA:
+ REXEC_FBC_CSCAN(
+ !isDIGIT_A(*s),
+ !isDIGIT_A(*s)
+ );
+ break;
case NDIGITL:
REXEC_FBC_CSCAN_TAINT(
!isDIGIT_LC_utf8((U8*)s),
!isDIGIT_LC(*s)
);
+ break;
case LNBREAK:
REXEC_FBC_CSCAN(
is_LNBREAK_utf8(s),
is_LNBREAK_latin1(s)
);
+ break;
case VERTWS:
REXEC_FBC_CSCAN(
is_VERTWS_utf8(s),
is_VERTWS_latin1(s)
);
+ break;
case NVERTWS:
REXEC_FBC_CSCAN(
!is_VERTWS_utf8(s),
!is_VERTWS_latin1(s)
);
+ break;
case HORIZWS:
REXEC_FBC_CSCAN(
is_HORIZWS_utf8(s),
is_HORIZWS_latin1(s)
);
+ break;
case NHORIZWS:
REXEC_FBC_CSCAN(
!is_HORIZWS_utf8(s),
!is_HORIZWS_latin1(s)
);
+ break;
case AHOCORASICKC:
case AHOCORASICK:
{
PerlIO_printf( Perl_debug_log,
" Scanning for legal start char...\n");
}
- );
- while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
- uc++;
- }
+ );
+ if (utf8_target) {
+ while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
+ uc += UTF8SKIP(uc);
+ }
+ } else {
+ while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) {
+ uc++;
+ }
+ }
s= (char *)uc;
}
if (uc >(U8*)last_start) break;
goto phooey;
} else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK))
{
- /* the warning about reginfo.ganch being used without intialization
+ /* the warning about reginfo.ganch being used without initialization
is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN
and we only enter this block when the same bit is set. */
char *tmp_s = reginfo.ganch - prog->gofs;
/* This is safe against NULLs: */
ReREFCNT_dec(PM_GETRE(PL_reg_curpm));
/* PM_reg_curpm owns a reference to this regexp. */
- ReREFCNT_inc(rx);
+ (void)ReREFCNT_inc(rx);
#endif
PM_SETRE(PL_reg_curpm, rx);
PL_reg_oldcurpm = PL_curpm;
(?>A), which pop all the backtrack states associated with A before
continuing.
-Bascktrack state structs are allocated in slabs of about 4K in size.
+Backtrack state structs are allocated in slabs of about 4K in size.
PL_regmatch_state and st always point to the currently active state,
and PL_regmatch_slab points to the slab currently containing
PL_regmatch_state. The first time regmatch() is called, the first slab is
SV *popmark = NULL; /* are we looking for a mark? */
SV *sv_commit = NULL; /* last mark name seen in failure */
SV *sv_yes_mark = NULL; /* last mark name we have seen
- during a successfull match */
+ during a successful match */
U32 lastopen = 0; /* last open we saw */
bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0;
SV* const oreplsv = GvSV(PL_replgv);
"%*s %smatched empty string...%s\n",
REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])
);
- break;
+ if (!trie->jump)
+ break;
} else {
DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
ST.nextword,
tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0,
PL_colors[0], PL_colors[1],
- (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)
+ (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0)|PERL_PV_ESCAPE_NONASCII
)
: "not compiled under -Dr",
PL_colors[5] );
nextchr = UCHARAT(locinput);
break;
}
- case EXACTFL:
+ case EXACTFL: {
+ re_fold_t folder;
+ const U8 * fold_array;
+ const char * s;
+ U32 fold_utf8_flags;
+
PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
- case EXACTF: {
- char * const s = STRING(scan);
+ folder = foldEQ_locale;
+ fold_array = PL_fold_locale;
+ fold_utf8_flags = FOLDEQ_UTF8_LOCALE;
+ goto do_exactf;
+
+ case EXACTFU:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ fold_utf8_flags = 0;
+ goto do_exactf;
+
+ case EXACTFA:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ fold_utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
+ goto do_exactf;
+
+ case EXACTF:
+ folder = foldEQ;
+ fold_array = PL_fold;
+ fold_utf8_flags = 0;
+
+ do_exactf:
+ s = STRING(scan);
ln = STR_LEN(scan);
if (utf8_target || UTF_PATTERN) {
const char * const l = locinput;
char *e = PL_regeol;
- if (! foldEQ_utf8(s, 0, ln, cBOOL(UTF_PATTERN),
- l, &e, 0, utf8_target)) {
- /* One more case for the sharp s:
- * pack("U0U*", 0xDF) =~ /ss/i,
- * the 0xC3 0x9F are the UTF-8
- * byte sequence for the U+00DF. */
-
- if (!(utf8_target &&
- toLOWER(s[0]) == 's' &&
- ln >= 2 &&
- toLOWER(s[1]) == 's' &&
- (U8)l[0] == 0xC3 &&
- e - l >= 2 &&
- (U8)l[1] == 0x9F))
- sayNO;
+ if (! foldEQ_utf8_flags(s, 0, ln, cBOOL(UTF_PATTERN),
+ l, &e, 0, utf8_target, fold_utf8_flags))
+ {
+ sayNO;
}
locinput = e;
nextchr = UCHARAT(locinput);
break;
}
- /* Neither the target and the pattern are utf8. */
-
- /* Inline the first character, for speed. */
+ /* Neither the target nor the pattern are utf8 */
if (UCHARAT(s) != nextchr &&
- UCHARAT(s) != ((OP(scan) == EXACTF)
- ? PL_fold : PL_fold_locale)[nextchr])
+ UCHARAT(s) != fold_array[nextchr])
+ {
sayNO;
+ }
if (PL_regeol - locinput < ln)
sayNO;
- if (ln > 1 && (OP(scan) == EXACTF
- ? ! foldEQ(s, locinput, ln)
- : ! foldEQ_locale(s, locinput, ln)))
+ if (ln > 1 && ! folder(s, locinput, ln))
sayNO;
locinput += ln;
nextchr = UCHARAT(locinput);
break;
- }
+ }
+
+ /* XXX Could improve efficiency by separating these all out using a
+ * macro or in-line function. At that point regcomp.c would no longer
+ * have to set the FLAGS fields of these */
case BOUNDL:
case NBOUNDL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case BOUND:
+ case BOUNDU:
+ case BOUNDA:
case NBOUND:
+ case NBOUNDU:
+ case NBOUNDA:
/* was last char in word? */
- if (utf8_target) {
+ if (utf8_target && FLAGS(scan) != REGEX_ASCII_RESTRICTED_CHARSET) {
if (locinput == PL_bostr)
ln = '\n';
else {
ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags);
}
- if (OP(scan) == BOUND || OP(scan) == NBOUND) {
+ if (FLAGS(scan) != REGEX_LOCALE_CHARSET) {
ln = isALNUM_uni(ln);
LOAD_UTF8_CHARCLASS_ALNUM();
n = swash_fetch(PL_utf8_alnum, (U8*)locinput, utf8_target);
}
}
else {
+
+ /* Here the string isn't utf8, or is utf8 and only ascii
+ * characters are to match \w. In the latter case looking at
+ * the byte just prior to the current one may be just the final
+ * byte of a multi-byte character. This is ok. There are two
+ * cases:
+ * 1) it is a single byte character, and then the test is doing
+ * just what it's supposed to.
+ * 2) it is a multi-byte character, in which case the final
+ * byte is never mistakable for ASCII, and so the test
+ * will say it is not a word character, which is the
+ * correct answer. */
ln = (locinput != PL_bostr) ?
UCHARAT(locinput - 1) : '\n';
- if (FLAGS(scan) & USE_UNI) {
-
- /* Here, can't be BOUNDL or NBOUNDL because they never set
- * the flags to USE_UNI */
- ln = isWORDCHAR_L1(ln);
- n = isWORDCHAR_L1(nextchr);
- }
- else if (OP(scan) == BOUND || OP(scan) == NBOUND) {
- ln = isALNUM(ln);
- n = isALNUM(nextchr);
- }
- else {
- ln = isALNUM_LC(ln);
- n = isALNUM_LC(nextchr);
+ switch (FLAGS(scan)) {
+ case REGEX_UNICODE_CHARSET:
+ ln = isWORDCHAR_L1(ln);
+ n = isWORDCHAR_L1(nextchr);
+ break;
+ case REGEX_LOCALE_CHARSET:
+ ln = isALNUM_LC(ln);
+ n = isALNUM_LC(nextchr);
+ break;
+ case REGEX_DEPENDS_CHARSET:
+ ln = isALNUM(ln);
+ n = isALNUM(nextchr);
+ break;
+ case REGEX_ASCII_RESTRICTED_CHARSET:
+ ln = isWORDCHAR_A(ln);
+ n = isWORDCHAR_A(nextchr);
+ break;
+ default:
+ Perl_croak(aTHX_ "panic: Unexpected FLAGS %u in op %u", FLAGS(scan), OP(scan));
+ break;
}
}
- if (((!ln) == (!n)) == (OP(scan) == BOUND ||
- OP(scan) == BOUNDL))
+ /* Note requires that all BOUNDs be lower than all NBOUNDs in
+ * regcomp.sym */
+ if (((!ln) == (!n)) == (OP(scan) < NBOUND))
sayNO;
break;
+ case ANYOFV:
case ANYOF:
- if (utf8_target) {
+ if (utf8_target || state_num == ANYOFV) {
STRLEN inclasslen = PL_regeol - locinput;
if (locinput >= PL_regeol)
sayNO;
if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, utf8_target))
- goto anyof_fail;
+ sayNO;
locinput += inclasslen;
nextchr = UCHARAT(locinput);
break;
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (!REGINCLASS(rex, scan, (U8*)locinput))
- goto anyof_fail;
+ sayNO;
nextchr = UCHARAT(++locinput);
break;
}
- anyof_fail:
- /* If we might have the case of the German sharp s
- * in a casefolding Unicode character class. */
-
- if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) {
- locinput += SHARP_S_SKIP;
- nextchr = UCHARAT(locinput);
- }
- else
- sayNO;
break;
/* Special char classes - The defines start on line 129 or so */
- CCC_TRY_AFF_U( ALNUM, ALNUML, perl_word, "a", isALNUM_LC_utf8, isWORDCHAR_L1, isALNUM_LC);
- CCC_TRY_NEG_U(NALNUM, NALNUML, perl_word, "a", isALNUM_LC_utf8, isWORDCHAR_L1, isALNUM_LC);
-
- CCC_TRY_AFF_U( SPACE, SPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE_L1, isSPACE_LC);
- CCC_TRY_NEG_U(NSPACE, NSPACEL, perl_space, " ", isSPACE_LC_utf8, isSPACE_L1, isSPACE_LC);
-
- CCC_TRY_AFF( DIGIT, DIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
- CCC_TRY_NEG(NDIGIT, NDIGITL, posix_digit, "0", isDIGIT_LC_utf8, isDIGIT, isDIGIT_LC);
+ CCC_TRY_U(ALNUM, NALNUM, isWORDCHAR,
+ ALNUML, NALNUML, isALNUM_LC, isALNUM_LC_utf8,
+ ALNUMU, NALNUMU, isWORDCHAR_L1,
+ ALNUMA, NALNUMA, isWORDCHAR_A,
+ alnum, "a");
+
+ CCC_TRY_U(SPACE, NSPACE, isSPACE,
+ SPACEL, NSPACEL, isSPACE_LC, isSPACE_LC_utf8,
+ SPACEU, NSPACEU, isSPACE_L1,
+ SPACEA, NSPACEA, isSPACE_A,
+ space, " ");
+
+ CCC_TRY(DIGIT, NDIGIT, isDIGIT,
+ DIGITL, NDIGITL, isDIGIT_LC, isDIGIT_LC_utf8,
+ DIGITA, NDIGITA, isDIGIT_A,
+ digit, "0");
case CLUMP: /* Match \X: logical Unicode character. This is defined as
a Unicode extended Grapheme Cluster */
break;
case NREFFL:
- {
+ { /* The capture buffer cases. The ones beginning with N for the
+ named buffers just convert to the equivalent numbered and
+ pretend they were called as the corresponding numbered buffer
+ op. */
+ /* don't initialize these in the declaration, it makes C++
+ unhappy */
char *s;
char type;
+ re_fold_t folder;
+ const U8 *fold_array;
+ UV utf8_fold_flags;
+
PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
- case NREF:
+ folder = foldEQ_locale;
+ fold_array = PL_fold_locale;
+ type = REFFL;
+ utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
+ goto do_nref;
+
+ case NREFFA:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ type = REFFA;
+ utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
+ goto do_nref;
+
+ case NREFFU:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ type = REFFU;
+ utf8_fold_flags = 0;
+ goto do_nref;
+
case NREFF:
- type = OP(scan);
+ folder = foldEQ;
+ fold_array = PL_fold;
+ type = REFF;
+ utf8_fold_flags = 0;
+ goto do_nref;
+
+ case NREF:
+ type = REF;
+ folder = NULL;
+ fold_array = NULL;
+ utf8_fold_flags = 0;
+ do_nref:
+
+ /* For the named back references, find the corresponding buffer
+ * number */
n = reg_check_named_buff_matched(rex,scan);
- if ( n ) {
- type = REF + ( type - NREF );
- goto do_ref;
- } else {
+ if ( ! n ) {
sayNO;
- }
- /* unreached */
+ }
+ goto do_nref_ref_common;
+
case REFFL:
PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
+ folder = foldEQ_locale;
+ fold_array = PL_fold_locale;
+ utf8_fold_flags = FOLDEQ_UTF8_LOCALE;
+ goto do_ref;
+
+ case REFFA:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ utf8_fold_flags = FOLDEQ_UTF8_NOMIX_ASCII;
+ goto do_ref;
+
+ case REFFU:
+ folder = foldEQ_latin1;
+ fold_array = PL_fold_latin1;
+ utf8_fold_flags = 0;
+ goto do_ref;
+
+ case REFF:
+ folder = foldEQ;
+ fold_array = PL_fold;
+ utf8_fold_flags = 0;
+ goto do_ref;
+
case REF:
- case REFF:
- n = ARG(scan); /* which paren pair */
+ folder = NULL;
+ fold_array = NULL;
+ utf8_fold_flags = 0;
+
+ do_ref:
type = OP(scan);
- do_ref:
+ n = ARG(scan); /* which paren pair */
+
+ do_nref_ref_common:
ln = PL_regoffs[n].start;
PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
if (*PL_reglastparen < n || ln == -1)
break;
s = PL_bostr + ln;
- if (utf8_target && type != REF) { /* REF can do byte comparison */
- char *l = locinput;
- const char *e = PL_bostr + PL_regoffs[n].end;
- /*
- * Note that we can't do the "other character" lookup trick as
- * in the 8-bit case (no pun intended) because in Unicode we
- * have to map both upper and title case to lower case.
- */
- if (type == REFF) {
- while (s < e) {
- STRLEN ulen1, ulen2;
- U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
- U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
-
- if (l >= PL_regeol)
- sayNO;
- toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
- toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
- if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1))
- sayNO;
- s += ulen1;
- l += ulen2;
- }
+ if (type != REF /* REF can do byte comparison */
+ && (utf8_target || type == REFFU))
+ { /* XXX handle REFFL better */
+ char * limit = PL_regeol;
+
+ /* This call case insensitively compares the entire buffer
+ * at s, with the current input starting at locinput, but
+ * not going off the end given by PL_regeol, and returns in
+ * limit upon success, how much of the current input was
+ * matched */
+ if (! foldEQ_utf8_flags(s, NULL, PL_regoffs[n].end - ln, utf8_target,
+ locinput, &limit, 0, utf8_target, utf8_fold_flags))
+ {
+ sayNO;
}
- locinput = l;
+ locinput = limit;
nextchr = UCHARAT(locinput);
break;
}
- /* Inline the first character, for speed. */
+ /* Not utf8: Inline the first character, for speed. */
if (UCHARAT(s) != nextchr &&
(type == REF ||
- (UCHARAT(s) != (type == REFF
- ? PL_fold : PL_fold_locale)[nextchr])))
+ UCHARAT(s) != fold_array[nextchr]))
sayNO;
ln = PL_regoffs[n].end - ln;
if (locinput + ln > PL_regeol)
sayNO;
if (ln > 1 && (type == REF
? memNE(s, locinput, ln)
- : (type == REFF
- ? ! foldEQ(s, locinput, ln)
- : ! foldEQ_locale(s, locinput, ln))))
+ : ! folder(s, locinput, ln)))
sayNO;
locinput += ln;
nextchr = UCHARAT(locinput);
/* First just match a string of min A's. */
if (n < min) {
+ ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor);
cur_curlyx->u.curlyx.lastloc = locinput;
+ REGCP_SET(ST.lastcp);
+
PUSH_STATE_GOTO(WHILEM_A_pre, A);
/* NOTREACHED */
}
/* NOTREACHED */
case WHILEM_A_min_fail: /* just failed to match A in a minimal match */
- REGCP_UNWIND(ST.lastcp);
- regcppop(rex);
/* FALL THROUGH */
case WHILEM_A_pre_fail: /* just failed to match even minimal A */
+ REGCP_UNWIND(ST.lastcp);
+ regcppop(rex);
cur_curlyx->u.curlyx.lastloc = ST.save_lastloc;
cur_curlyx->u.curlyx.count--;
CACHEsayNO;
{
ST.c1 = (U8)*STRING(text_node);
- ST.c2 =
- (IS_TEXTF(text_node))
- ? PL_fold[ST.c1]
- : (IS_TEXTFL(text_node))
- ? PL_fold_locale[ST.c1]
- : ST.c1;
+ switch (OP(text_node)) {
+ case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
+ case EXACTFA:
+ case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
+ case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
+ default: ST.c2 = ST.c1;
+ }
}
}
}
if this changes back then the macro for IS_TEXT and
friends need to change. */
if (!UTF_PATTERN) {
- ST.c2 = ST.c1 = *s;
- if (IS_TEXTF(text_node))
- ST.c2 = PL_fold[ST.c1];
- else if (IS_TEXTFL(text_node))
- ST.c2 = PL_fold_locale[ST.c1];
+ ST.c1 = *s;
+ switch (OP(text_node)) {
+ case EXACTF: ST.c2 = PL_fold[ST.c1]; break;
+ case EXACTFA:
+ case EXACTFU: ST.c2 = PL_fold_latin1[ST.c1]; break;
+ case EXACTFL: ST.c2 = PL_fold_locale[ST.c1]; break;
+ default: ST.c2 = ST.c1; break;
+ }
}
else { /* UTF_PATTERN */
- if (IS_TEXTF(text_node)) {
+ if (IS_TEXTFU(text_node) || IS_TEXTF(text_node)) {
STRLEN ulen1, ulen2;
U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
rex = (struct regexp *)SvANY(rex_sv);
rexi = RXi_GET(rex);
cur_curlyx = cur_eval->u.eval.prev_curlyx;
- ReREFCNT_inc(rex_sv);
+ (void)ReREFCNT_inc(rex_sv);
st->u.eval.cp = regcppush(0); /* Save *all* the positions. */
/* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */
n = ARG(scan);
if ( n == (U32)what_len_TRICKYFOLD(locinput,utf8_target,ln) ) {
locinput += ln;
- } else if ( 0xDF == n && !utf8_target && !UTF_PATTERN ) {
+ } else if ( LATIN_SMALL_LETTER_SHARP_S == n && !utf8_target && !UTF_PATTERN ) {
sayNO;
} else {
U8 folded[UTF8_MAXBYTES_CASE+1];
register char *loceol = PL_regeol;
register I32 hardcount = 0;
register bool utf8_target = PL_reg_match_utf8;
+ UV utf8_flags;
#ifndef DEBUGGING
PERL_UNUSED_ARG(depth);
#endif
case CANY:
scan = loceol;
break;
- case EXACTFL:
- PL_reg_flags |= RF_tainted;
- /* FALL THROUGH */
case EXACT:
- case EXACTF:
- /* To get here, EXACTish nodes must have *byte* length == 1. That means
- * they match only characters in the string that can be expressed as a
- * single byte. For non-utf8 strings, that means a simple match. For
- * utf8 strings, the character matched must be an invariant, or
+ /* To get here, EXACTish nodes must have *byte* length == 1. That
+ * means they match only characters in the string that can be expressed
+ * as a single byte. For non-utf8 strings, that means a simple match.
+ * For utf8 strings, the character matched must be an invariant, or
* downgradable to a single byte. The pattern's utf8ness is
- * irrelevant, as it must be a single byte, so either it isn't utf8, or
- * if it is it's an invariant */
+ * irrelevant, as since it's a single byte, it either isn't utf8, or if
+ * it is, it's an invariant */
c = (U8)*STRING(p);
assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));
- if ((! utf8_target) || UNI_IS_INVARIANT(c)) {
-
- /* Here, the string isn't utf8, or the character in the EXACT
- * node is the same in utf8 as not, so can just do equality.
- * Each matching char must be 1 byte long */
- switch (OP(p)) {
- case EXACT:
- while (scan < loceol && UCHARAT(scan) == c) {
- scan++;
- }
- break;
- case EXACTF:
- while (scan < loceol &&
- (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
- {
- scan++;
- }
- break;
- case EXACTFL:
- while (scan < loceol &&
- (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
- {
- scan++;
- }
- break;
- default:
- Perl_croak(aTHX_ "panic: Unexpected op %u", OP(p));
+ if (! utf8_target || UNI_IS_INVARIANT(c)) {
+ while (scan < loceol && UCHARAT(scan) == c) {
+ scan++;
}
}
else {
/* Here, the string is utf8, and the pattern char is different
- * in utf8 than not. */
+ * in utf8 than not, so can't compare them directly. Outside the
+ * loop, find find the two utf8 bytes that represent c, and then
+ * look for those in sequence in the utf8 string */
+ U8 high = UTF8_TWO_BYTE_HI(c);
+ U8 low = UTF8_TWO_BYTE_LO(c);
+ loceol = PL_regeol;
- switch (OP(p)) {
- case EXACT:
- {
- /* Fastest to find the two utf8 bytes that represent c, and
- * then look for those in sequence in the utf8 string */
- U8 high = UTF8_TWO_BYTE_HI(c);
- U8 low = UTF8_TWO_BYTE_LO(c);
- loceol = PL_regeol;
-
- while (hardcount < max
- && scan + 1 < loceol
- && UCHARAT(scan) == high
- && UCHARAT(scan + 1) == low)
- {
- scan += 2;
- hardcount++;
- }
- }
- break;
- case EXACTFL: /* Doesn't really make sense, but is best we can
- do. The documents warn against mixing locale
- and utf8 */
- case EXACTF:
- { /* utf8 string, so use utf8 foldEQ */
- char *tmpeol = loceol;
- while (hardcount < max
- && foldEQ_utf8(scan, &tmpeol, 0, utf8_target,
- STRING(p), NULL, 1, UTF_PATTERN))
- {
- scan = tmpeol;
- tmpeol = loceol;
- hardcount++;
- }
+ while (hardcount < max
+ && scan + 1 < loceol
+ && UCHARAT(scan) == high
+ && UCHARAT(scan + 1) == low)
+ {
+ scan += 2;
+ hardcount++;
+ }
+ }
+ break;
+ case EXACTFA:
+ utf8_flags = FOLDEQ_UTF8_NOMIX_ASCII;
+ goto do_exactf;
- /* XXX Note that the above handles properly the German
- * sharp ss in the pattern matching ss in the string. But
- * it doesn't handle properly cases where the string
- * contains say 'LIGATURE ff' and the pattern is 'f+'.
- * This would require, say, a new function or revised
- * interface to foldEQ_utf8(), in which the maximum number
- * of characters to match could be passed and it would
- * return how many actually did. This is just one of many
- * cases where multi-char folds don't work properly, and so
- * the fix is being deferred */
- }
- break;
- default:
- Perl_croak(aTHX_ "panic: Unexpected op %u", OP(p));
+ case EXACTFL:
+ PL_reg_flags |= RF_tainted;
+ utf8_flags = FOLDEQ_UTF8_LOCALE;
+ goto do_exactf;
+
+ case EXACTF:
+ case EXACTFU:
+ utf8_flags = 0;
+
+ /* The comments for the EXACT case above apply as well to these fold
+ * ones */
+
+ do_exactf:
+ c = (U8)*STRING(p);
+ assert(! UTF_PATTERN || UNI_IS_INVARIANT(c));
+
+ if (utf8_target) { /* Use full Unicode fold matching */
+ char *tmpeol = loceol;
+ while (hardcount < max
+ && foldEQ_utf8_flags(scan, &tmpeol, 0, utf8_target,
+ STRING(p), NULL, 1, cBOOL(UTF_PATTERN), utf8_flags))
+ {
+ scan = tmpeol;
+ tmpeol = loceol;
+ hardcount++;
+ }
+
+ /* XXX Note that the above handles properly the German sharp s in
+ * the pattern matching ss in the string. But it doesn't handle
+ * properly cases where the string contains say 'LIGATURE ff' and
+ * the pattern is 'f+'. This would require, say, a new function or
+ * revised interface to foldEQ_utf8(), in which the maximum number
+ * of characters to match could be passed and it would return how
+ * many actually did. This is just one of many cases where
+ * multi-char folds don't work properly, and so the fix is being
+ * deferred */
+ }
+ else {
+ U8 folded;
+
+ /* Here, the string isn't utf8 and c is a single byte; and either
+ * the pattern isn't utf8 or c is an invariant, so its utf8ness
+ * doesn't affect c. Can just do simple comparisons for exact or
+ * fold matching. */
+ switch (OP(p)) {
+ case EXACTF: folded = PL_fold[c]; break;
+ case EXACTFA:
+ case EXACTFU: folded = PL_fold_latin1[c]; break;
+ case EXACTFL: folded = PL_fold_locale[c]; break;
+ default: Perl_croak(aTHX_ "panic: Unexpected op %u", OP(p));
+ }
+ while (scan < loceol &&
+ (UCHARAT(scan) == c || UCHARAT(scan) == folded))
+ {
+ scan++;
}
}
break;
+ case ANYOFV:
case ANYOF:
- if (utf8_target) {
+ if (utf8_target || OP(p) == ANYOFV) {
+ STRLEN inclasslen;
loceol = PL_regeol;
- while (hardcount < max && scan < loceol &&
- reginclass(prog, p, (U8*)scan, 0, utf8_target)) {
- scan += UTF8SKIP(scan);
+ inclasslen = loceol - scan;
+ while (hardcount < max
+ && ((inclasslen = loceol - scan) > 0)
+ && reginclass(prog, p, (U8*)scan, &inclasslen, utf8_target))
+ {
+ scan += inclasslen;
hardcount++;
}
} else {
scan++;
}
break;
- case ALNUM:
+ case ALNUMU:
if (utf8_target) {
+ utf8_wordchar:
loceol = PL_regeol;
LOAD_UTF8_CHARCLASS_ALNUM();
while (hardcount < max && scan < loceol &&
scan += UTF8SKIP(scan);
hardcount++;
}
- } else if (FLAGS(p) & USE_UNI) {
+ } else {
while (scan < loceol && isWORDCHAR_L1((U8) *scan)) {
scan++;
}
- } else {
- while (scan < loceol && isALNUM((U8) *scan)) {
- scan++;
- }
+ }
+ break;
+ case ALNUM:
+ if (utf8_target)
+ goto utf8_wordchar;
+ while (scan < loceol && isALNUM((U8) *scan)) {
+ scan++;
+ }
+ break;
+ case ALNUMA:
+ while (scan < loceol && isWORDCHAR_A((U8) *scan)) {
+ scan++;
}
break;
case ALNUML:
scan++;
}
break;
- case NALNUM:
+ case NALNUMU:
if (utf8_target) {
+
+ utf8_Nwordchar:
+
loceol = PL_regeol;
LOAD_UTF8_CHARCLASS_ALNUM();
while (hardcount < max && scan < loceol &&
- !swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
+ ! swash_fetch(PL_utf8_alnum, (U8*)scan, utf8_target))
{
scan += UTF8SKIP(scan);
hardcount++;
}
- } else if (FLAGS(p) & USE_UNI) {
+ } else {
while (scan < loceol && ! isWORDCHAR_L1((U8) *scan)) {
scan++;
}
- } else {
- while (scan < loceol && ! isALNUM((U8) *scan)) {
- scan++;
- }
+ }
+ break;
+ case NALNUM:
+ if (utf8_target)
+ goto utf8_Nwordchar;
+ while (scan < loceol && ! isALNUM((U8) *scan)) {
+ scan++;
+ }
+ break;
+ case NALNUMA:
+ if (utf8_target) {
+ while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
+ scan += UTF8SKIP(scan);
+ }
+ }
+ else {
+ while (scan < loceol && ! isWORDCHAR_A((U8) *scan)) {
+ scan++;
+ }
}
break;
case NALNUML:
scan++;
}
break;
- case SPACE:
+ case SPACEU:
if (utf8_target) {
+
+ utf8_space:
+
loceol = PL_regeol;
LOAD_UTF8_CHARCLASS_SPACE();
while (hardcount < max && scan < loceol &&
scan += UTF8SKIP(scan);
hardcount++;
}
- } else if (FLAGS(p) & USE_UNI) {
+ break;
+ }
+ else {
while (scan < loceol && isSPACE_L1((U8) *scan)) {
scan++;
}
- } else {
- while (scan < loceol && isSPACE((U8) *scan))
- scan++;
+ break;
+ }
+ case SPACE:
+ if (utf8_target)
+ goto utf8_space;
+
+ while (scan < loceol && isSPACE((U8) *scan)) {
+ scan++;
+ }
+ break;
+ case SPACEA:
+ while (scan < loceol && isSPACE_A((U8) *scan)) {
+ scan++;
}
break;
case SPACEL:
if (utf8_target) {
loceol = PL_regeol;
while (hardcount < max && scan < loceol &&
- (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
+ isSPACE_LC_utf8((U8*)scan)) {
scan += UTF8SKIP(scan);
hardcount++;
}
scan++;
}
break;
- case NSPACE:
+ case NSPACEU:
if (utf8_target) {
+
+ utf8_Nspace:
+
loceol = PL_regeol;
LOAD_UTF8_CHARCLASS_SPACE();
while (hardcount < max && scan < loceol &&
- !(*scan == ' ' ||
- swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
+ ! (*scan == ' ' ||
+ swash_fetch(PL_utf8_space,(U8*)scan, utf8_target)))
{
scan += UTF8SKIP(scan);
hardcount++;
}
- } else if (FLAGS(p) & USE_UNI) {
+ break;
+ }
+ else {
while (scan < loceol && ! isSPACE_L1((U8) *scan)) {
scan++;
}
- } else {
- while (scan < loceol && ! isSPACE((U8) *scan)) {
- scan++;
- }
+ }
+ break;
+ case NSPACE:
+ if (utf8_target)
+ goto utf8_Nspace;
+
+ while (scan < loceol && ! isSPACE((U8) *scan)) {
+ scan++;
+ }
+ break;
+ case NSPACEA:
+ if (utf8_target) {
+ while (scan < loceol && ! isSPACE_A((U8) *scan)) {
+ scan += UTF8SKIP(scan);
+ }
+ }
+ else {
+ while (scan < loceol && ! isSPACE_A((U8) *scan)) {
+ scan++;
+ }
}
break;
case NSPACEL:
if (utf8_target) {
loceol = PL_regeol;
while (hardcount < max && scan < loceol &&
- !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
+ !isSPACE_LC_utf8((U8*)scan)) {
scan += UTF8SKIP(scan);
hardcount++;
}
scan++;
}
break;
+ case DIGITA:
+ while (scan < loceol && isDIGIT_A((U8) *scan)) {
+ scan++;
+ }
+ break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ if (utf8_target) {
+ loceol = PL_regeol;
+ while (hardcount < max && scan < loceol &&
+ isDIGIT_LC_utf8((U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ } else {
+ while (scan < loceol && isDIGIT_LC(*scan))
+ scan++;
+ }
+ break;
case NDIGIT:
if (utf8_target) {
loceol = PL_regeol;
while (scan < loceol && !isDIGIT(*scan))
scan++;
}
+ break;
+ case NDIGITA:
+ if (utf8_target) {
+ while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
+ scan += UTF8SKIP(scan);
+ }
+ }
+ else {
+ while (scan < loceol && ! isDIGIT_A((U8) *scan)) {
+ scan++;
+ }
+ }
+ break;
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ if (utf8_target) {
+ loceol = PL_regeol;
+ while (hardcount < max && scan < loceol &&
+ !isDIGIT_LC_utf8((U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ } else {
+ while (scan < loceol && !isDIGIT_LC(*scan))
+ scan++;
+ }
+ break;
case LNBREAK:
if (utf8_target) {
loceol = PL_regeol;
PERL_ARGS_ASSERT_REGCLASS_SWASH;
+ assert(ANYOF_NONBITMAP(node));
+
if (data && data->count) {
const U32 n = ARG(node);
bytes in p were matched. If there was no match, the value is undefined,
possibly changed from the input.
+ Note that this can be a synthetic start class, a combination of various
+ nodes, so things you think might be mutually exclusive, such as locale,
+ aren't. It can match both locale and non-locale
+
*/
STATIC bool
maxlen = c_len;
}
- if (utf8_target || (flags & ANYOF_UNICODE)) {
- if (utf8_target && !ANYOF_RUNTIME(n)) {
- if (c < 256 && ANYOF_BITMAP_TEST(n, c))
+ /* If this character is potentially in the bitmap, check it */
+ if (c < 256) {
+ if (ANYOF_BITMAP_TEST(n, c))
+ match = TRUE;
+ else if (flags & ANYOF_NON_UTF8_LATIN1_ALL
+ && ! utf8_target
+ && ! isASCII(c))
+ {
+ match = TRUE;
+ }
+
+ else if (flags & ANYOF_LOCALE) {
+ PL_reg_flags |= RF_tainted;
+
+ if ((flags & ANYOF_LOC_NONBITMAP_FOLD)
+ && ANYOF_BITMAP_TEST(n, PL_fold_locale[c]))
+ {
+ match = TRUE;
+ }
+ else if (ANYOF_CLASS_TEST_ANY_SET(n) &&
+ ((ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) ||
+ (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c))
+ ) /* How's that for a conditional? */
+ ) {
match = TRUE;
+ }
}
- if (!match && utf8_target && (flags & ANYOF_UNICODE_ALL) && c >= 256)
- match = TRUE;
- if (!match) {
+ }
+
+ /* If the bitmap didn't (or couldn't) match, and something outside the
+ * bitmap could match, try that. Locale nodes specifiy completely the
+ * behavior of code points in the bit map (otherwise, a utf8 target would
+ * cause them to be treated as Unicode and not locale), except in
+ * the very unlikely event when this node is a synthetic start class, which
+ * could be a combination of locale and non-locale nodes. So allow locale
+ * to match for the synthetic start class, which will give a false
+ * positive that will be resolved when the match is done again as not part
+ * of the synthetic start class */
+ if (!match) {
+ if (utf8_target && (flags & ANYOF_UNICODE_ALL) && c >= 256) {
+ match = TRUE; /* Everything above 255 matches */
+ }
+ else if (ANYOF_NONBITMAP(n)
+ && ((flags & ANYOF_NONBITMAP_NON_UTF8)
+ || (utf8_target
+ && (c >=256
+ || (! (flags & ANYOF_LOCALE))
+ || (flags & ANYOF_IS_SYNTHETIC)))))
+ {
AV *av;
SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av);
-
+
if (sw) {
U8 * utf8_p;
if (utf8_target) {
utf8_p = (U8 *) p;
} else {
- STRLEN len = 1;
+
+ /* Not utf8. Convert as much of the string as available up
+ * to the limit of how far the (single) character in the
+ * pattern can possibly match (no need to go further). If
+ * the node is a straight ANYOF or not folding, it can't
+ * match more than one. Otherwise, It can match up to how
+ * far a single char can fold to. Since not utf8, each
+ * character is a single byte, so the max it can be in
+ * bytes is the same as the max it can be in characters */
+ STRLEN len = (OP(n) == ANYOF
+ || ! (flags & ANYOF_LOC_NONBITMAP_FOLD))
+ ? 1
+ : (maxlen < UTF8_MAX_FOLD_CHAR_EXPAND)
+ ? maxlen
+ : UTF8_MAX_FOLD_CHAR_EXPAND;
utf8_p = bytes_to_utf8(p, &len);
}
- if (swash_fetch(sw, utf8_p, 1))
+
+ if (swash_fetch(sw, utf8_p, TRUE))
match = TRUE;
- else if (flags & ANYOF_FOLD) {
- if (!match && lenp && av) {
+ else if (flags & ANYOF_LOC_NONBITMAP_FOLD) {
+
+ /* Here, we need to test if the fold of the target string
+ * matches. The non-multi char folds have all been moved to
+ * the compilation phase, and the multi-char folds have
+ * been stored by regcomp into 'av'; we linearly check to
+ * see if any match the target string (folded). We know
+ * that the originals were each one character, but we don't
+ * currently know how many characters/bytes each folded to,
+ * except we do know that there are small limits imposed by
+ * Unicode. XXX A performance enhancement would be to have
+ * regcomp.c store the max number of chars/bytes that are
+ * in an av entry, as, say the 0th element. Even better
+ * would be to have a hash of the few characters that can
+ * start a multi-char fold to the max number of chars of
+ * those folds.
+ *
+ * If there is a match, we will need to advance (if lenp is
+ * specified) the match pointer in the target string. But
+ * what we are comparing here isn't that string directly,
+ * but its fold, whose length may differ from the original.
+ * As we go along in constructing the fold, therefore, we
+ * create a map so that we know how many bytes in the
+ * source to advance given that we have matched a certain
+ * number of bytes in the fold. This map is stored in
+ * 'map_fold_len_back'. Let n mean the number of bytes in
+ * the fold of the first character that we are folding.
+ * Then map_fold_len_back[n] is set to the number of bytes
+ * in that first character. Similarly let m be the
+ * corresponding number for the second character to be
+ * folded. Then map_fold_len_back[n+m] is set to the
+ * number of bytes occupied by the first two source
+ * characters. ... */
+ U8 map_fold_len_back[UTF8_MAXBYTES_CASE+1] = { 0 };
+ U8 folded[UTF8_MAXBYTES_CASE+1];
+ STRLEN foldlen = 0; /* num bytes in fold of 1st char */
+ STRLEN total_foldlen = 0; /* num bytes in fold of all
+ chars */
+
+ if (OP(n) == ANYOF || maxlen == 1 || ! lenp || ! av) {
+
+ /* Here, only need to fold the first char of the target
+ * string. It the source wasn't utf8, is 1 byte long */
+ to_utf8_fold(utf8_p, folded, &foldlen);
+ total_foldlen = foldlen;
+ map_fold_len_back[foldlen] = (utf8_target)
+ ? UTF8SKIP(utf8_p)
+ : 1;
+ }
+ else {
+
+ /* Here, need to fold more than the first char. Do so
+ * up to the limits */
+ U8* source_ptr = utf8_p; /* The source for the fold
+ is the regex target
+ string */
+ U8* folded_ptr = folded;
+ U8* e = utf8_p + maxlen; /* Can't go beyond last
+ available byte in the
+ target string */
+ U8 i;
+ for (i = 0;
+ i < UTF8_MAX_FOLD_CHAR_EXPAND && source_ptr < e;
+ i++)
+ {
+
+ /* Fold the next character */
+ U8 this_char_folded[UTF8_MAXBYTES_CASE+1];
+ STRLEN this_char_foldlen;
+ to_utf8_fold(source_ptr,
+ this_char_folded,
+ &this_char_foldlen);
+
+ /* Bail if it would exceed the byte limit for
+ * folding a single char. */
+ if (this_char_foldlen + folded_ptr - folded >
+ UTF8_MAXBYTES_CASE)
+ {
+ break;
+ }
+
+ /* Add the fold of this character */
+ Copy(this_char_folded,
+ folded_ptr,
+ this_char_foldlen,
+ U8);
+ source_ptr += UTF8SKIP(source_ptr);
+ folded_ptr += this_char_foldlen;
+ total_foldlen = folded_ptr - folded;
+
+ /* Create map from the number of bytes in the fold
+ * back to the number of bytes in the source. If
+ * the source isn't utf8, the byte count is just
+ * the number of characters so far */
+ map_fold_len_back[total_foldlen]
+ = (utf8_target)
+ ? source_ptr - utf8_p
+ : i + 1;
+ }
+ *folded_ptr = '\0';
+ }
+
+
+ /* Do the linear search to see if the fold is in the list
+ * of multi-char folds. */
+ if (av) {
I32 i;
for (i = 0; i <= av_len(av); i++) {
SV* const sv = *av_fetch(av, i, FALSE);
STRLEN len;
const char * const s = SvPV_const(sv, len);
- if (len <= maxlen && memEQ(s, (char*)utf8_p, len)) {
- *lenp = len;
+
+ if (len <= total_foldlen
+ && memEQ(s, (char*)folded, len)
+
+ /* If 0, means matched a partial char. See
+ * [perl #90536] */
+ && map_fold_len_back[len])
+ {
+
+ /* Advance the target string ptr to account for
+ * this fold, but have to translate from the
+ * folded length to the corresponding source
+ * length. */
+ if (lenp) {
+ *lenp = map_fold_len_back[len];
+ }
match = TRUE;
break;
}
}
}
- if (!match) {
- U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
-
- STRLEN tmplen;
- to_utf8_fold(utf8_p, tmpbuf, &tmplen);
- if (swash_fetch(sw, tmpbuf, 1))
- match = TRUE;
- }
}
/* If we allocated a string above, free it */
}
}
}
- if (!match && c < 256) {
- if (ANYOF_BITMAP_TEST(n, c))
- match = TRUE;
- else if (flags & ANYOF_FOLD) {
- U8 f;
-
- if (flags & ANYOF_LOCALE) {
- PL_reg_flags |= RF_tainted;
- f = PL_fold_locale[c];
- }
- else
- f = PL_fold[c];
- if (f != c && ANYOF_BITMAP_TEST(n, f))
- match = TRUE;
- }
-
- if (!match && (flags & ANYOF_CLASS)) {
- PL_reg_flags |= RF_tainted;
- if (
- (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) ||
- (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c))
- ) /* How's that for a conditional? */
- {
- match = TRUE;
- }
- }
- }
return (flags & ANYOF_INVERT) ? !match : match;
}
prog->substrs->data[i].utf8_substr = sv;
sv_utf8_upgrade(sv);
if (SvVALID(prog->substrs->data[i].substr)) {
- const U8 flags = BmFLAGS(prog->substrs->data[i].substr);
- if (flags & FBMcf_TAIL) {
+ if (SvTAIL(prog->substrs->data[i].substr)) {
/* Trim the trailing \n that fbm_compile added last
time. */
SvCUR_set(sv, SvCUR(sv) - 1);
/* Whilst this makes the SV technically "invalid" (as its
buffer is no longer followed by "\0") when fbm_compile()
adds the "\n" back, a "\0" is restored. */
- }
- fbm_compile(sv, flags);
+ fbm_compile(sv, FBMcf_TAIL);
+ } else
+ fbm_compile(sv, 0);
}
if (prog->substrs->data[i].substr == prog->check_substr)
prog->check_utf8 = sv;
SV* sv = newSVsv(prog->substrs->data[i].utf8_substr);
if (sv_utf8_downgrade(sv, TRUE)) {
if (SvVALID(prog->substrs->data[i].utf8_substr)) {
- const U8 flags
- = BmFLAGS(prog->substrs->data[i].utf8_substr);
- if (flags & FBMcf_TAIL) {
+ if (SvTAIL(prog->substrs->data[i].utf8_substr)) {
/* Trim the trailing \n that fbm_compile added last
time. */
SvCUR_set(sv, SvCUR(sv) - 1);
- }
- fbm_compile(sv, flags);
- }
+ fbm_compile(sv, FBMcf_TAIL);
+ } else
+ fbm_compile(sv, 0);
+ }
} else {
SvREFCNT_dec(sv);
sv = &PL_sv_undef;
https://perl5.git.perl.org / perl5.git / blobdiff