# define PERL_IN_XSUB_RE
# endif
/* need access to debugger hooks */
-# ifndef DEBUGGING
+# if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
# define DEBUGGING
# endif
#endif
# define Perl_regexec_flags my_regexec
# define Perl_regdump my_regdump
# define Perl_regprop my_regprop
+# define Perl_re_intuit_start my_re_intuit_start
/* *These* symbols are masked to allow static link. */
# define Perl_pregexec my_pregexec
+# define Perl_reginitcolors my_reginitcolors
+
+# define PERL_NO_GET_CONTEXT
#endif
/*SUPPRESS 112*/
*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-1998, Larry Wall
+ **** Copyright (c) 1991-1999, Larry Wall
****
**** You may distribute under the terms of either the GNU General Public
**** License or the Artistic License, as specified in the README file.
* regular-expression syntax might require a total rethink.
*/
#include "EXTERN.h"
+#define PERL_IN_REGEXEC_C
#include "perl.h"
+#ifdef PERL_IN_XSUB_RE
+# if defined(PERL_CAPI) || defined(PERL_OBJECT)
+# include "XSUB.h"
+# endif
+#endif
+
#include "regcomp.h"
#define RF_tainted 1 /* tainted information used? */
#define STATIC static
#endif
-#ifndef PERL_OBJECT
-typedef I32 CHECKPOINT;
-
/*
* Forwards.
*/
-static I32 regmatch _((regnode *prog));
-static I32 regrepeat _((regnode *p, I32 max));
-static I32 regrepeat_hard _((regnode *p, I32 max, I32 *lp));
-static I32 regtry _((regexp *prog, char *startpos));
-
-static bool reginclass _((char *p, I32 c));
-static bool reginclassutf8 _((regnode *f, U8* p));
-static CHECKPOINT regcppush _((I32 parenfloor));
-static char * regcppop _((void));
-static char * regcp_set_to _((I32 ss));
-static void cache_re _((regexp *prog));
-#endif
-
-#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c))
+#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c) : ANYOF_BITMAP_TEST(p,c))
#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p))
#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
-#ifndef PERL_OBJECT
-static U8 * reghop _((U8 *pos, I32 off));
-static U8 * reghopmaybe _((U8 *pos, I32 off));
-#endif
#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off))
#define HOPc(pos,off) ((char*)HOP(pos,off))
#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
+static void restore_pos(pTHXo_ void *arg);
+
+
STATIC CHECKPOINT
-regcppush(I32 parenfloor)
+S_regcppush(pTHX_ I32 parenfloor)
{
dTHR;
int retval = PL_savestack_ix;
SSCHECK(i + 5);
for (p = PL_regsize; p > parenfloor; p--) {
- SSPUSHPTR(PL_regendp[p]);
- SSPUSHPTR(PL_regstartp[p]);
+ SSPUSHINT(PL_regendp[p]);
+ SSPUSHINT(PL_regstartp[p]);
SSPUSHPTR(PL_reg_start_tmp[p]);
SSPUSHINT(p);
}
lastcp, PL_savestack_ix) : 0); regcpblow(lastcp)
STATIC char *
-regcppop(void)
+S_regcppop(pTHX)
{
dTHR;
I32 i = SSPOPINT;
U32 paren = 0;
char *input;
- char *tmps;
+ I32 tmps;
assert(i == SAVEt_REGCONTEXT);
i = SSPOPINT;
input = (char *) SSPOPPTR;
for (i -= 3; i > 0; i -= 4) {
paren = (U32)SSPOPINT;
PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
- PL_regstartp[paren] = (char *) SSPOPPTR;
- tmps = (char*)SSPOPPTR;
+ PL_regstartp[paren] = SSPOPINT;
+ tmps = SSPOPINT;
if (paren <= *PL_reglastparen)
PL_regendp[paren] = tmps;
DEBUG_r(
PerlIO_printf(Perl_debug_log,
" restoring \\%d to %d(%d)..%d%s\n",
- paren, PL_regstartp[paren] - PL_regbol,
- PL_reg_start_tmp[paren] - PL_regbol,
- PL_regendp[paren] - PL_regbol,
+ paren, PL_regstartp[paren],
+ PL_reg_start_tmp[paren] - PL_bostr,
+ PL_regendp[paren],
(paren > *PL_reglastparen ? "(no)" : ""));
);
}
);
for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) {
if (paren > PL_regsize)
- PL_regstartp[paren] = Nullch;
- PL_regendp[paren] = Nullch;
+ PL_regstartp[paren] = -1;
+ PL_regendp[paren] = -1;
}
return input;
}
STATIC char *
-regcp_set_to(I32 ss)
+S_regcp_set_to(pTHX_ I32 ss)
{
dTHR;
I32 tmp = PL_savestack_ix;
- pregexec - match a regexp against a string
*/
I32
-pregexec(register regexp *prog, char *stringarg, register char *strend,
+Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend,
char *strbeg, I32 minend, SV *screamer, U32 nosave)
/* strend: pointer to null at end of string */
/* strbeg: real beginning of string */
}
STATIC void
-cache_re(regexp *prog)
+S_cache_re(pTHX_ regexp *prog)
{
dTHR;
PL_regprecomp = prog->precomp; /* Needed for FAIL. */
PL_regdata = prog->data;
PL_reg_re = prog;
}
-
+
+/*
+ * Need to implement the following flags for reg_anch:
+ *
+ * USE_INTUIT_NOML - Useful to call re_intuit_start() first
+ * USE_INTUIT_ML
+ * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer
+ * INTUIT_AUTORITATIVE_ML
+ * INTUIT_ONCE_NOML - Intuit can match in one location only.
+ * INTUIT_ONCE_ML
+ *
+ * Another flag for this function: SECOND_TIME (so that float substrs
+ * with giant delta may be not rechecked).
+ */
+
+/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
+
+/* If SCREAM, then sv should be compatible with strpos and strend.
+ Otherwise, only SvCUR(sv) is used to get strbeg. */
+
+/* XXXX We assume that strpos is strbeg unless sv. */
+
+char *
+Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
+ char *strend, U32 flags, re_scream_pos_data *data)
+{
+ I32 start_shift;
+ /* Should be nonnegative! */
+ I32 end_shift;
+ char *s;
+ char *t;
+ I32 ml_anch;
+
+ DEBUG_r( if (!PL_colorset) reginitcolors() );
+ DEBUG_r(PerlIO_printf(Perl_debug_log,
+ "%sGuessing start of match:%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
+ PL_colors[4],PL_colors[5],PL_colors[0],
+ prog->precomp,
+ PL_colors[1],
+ (strlen(prog->precomp) > 60 ? "..." : ""),
+ PL_colors[0],
+ (strend - strpos > 60 ? 60 : strend - strpos),
+ strpos, PL_colors[1],
+ (strend - strpos > 60 ? "..." : ""))
+ );
+
+ if (prog->minlen > strend - strpos)
+ goto fail;
+
+ /* XXXX Move further down? */
+ start_shift = prog->check_offset_min; /* okay to underestimate on CC */
+ /* Should be nonnegative! */
+ end_shift = prog->minlen - start_shift -
+ CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0);
+
+ if (prog->reganch & ROPT_ANCH) {
+ ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
+ || ( (prog->reganch & ROPT_ANCH_BOL)
+ && !PL_multiline ) );
+
+ if ((prog->check_offset_min == prog->check_offset_max) && !ml_anch) {
+ /* Anchored... */
+ I32 slen;
+
+ if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */
+ && (sv && (strpos + SvCUR(sv) != strend)) )
+ goto fail;
+
+ s = (char*)HOP((U8*)strpos, prog->check_offset_min);
+ if (SvTAIL(prog->check_substr)) {
+ slen = SvCUR(prog->check_substr); /* >= 1 */
+
+ if ( strend - s > slen || strend - s < slen - 1 ) {
+ s = Nullch;
+ goto finish;
+ }
+ if ( strend - s == slen && strend[-1] != '\n') {
+ s = Nullch;
+ goto finish;
+ }
+ /* Now should match s[0..slen-2] */
+ slen--;
+ if (slen && (*SvPVX(prog->check_substr) != *s
+ || (slen > 1
+ && memNE(SvPVX(prog->check_substr), s, slen))))
+ s = Nullch;
+ }
+ else if (*SvPVX(prog->check_substr) != *s
+ || ((slen = SvCUR(prog->check_substr)) > 1
+ && memNE(SvPVX(prog->check_substr), s, slen)))
+ s = Nullch;
+ else
+ s = strpos;
+ goto finish;
+ }
+ s = strpos;
+ if (!ml_anch && (s + prog->check_offset_max < strend - prog->minlen))
+ end_shift += strend - s - prog->minlen - prog->check_offset_max;
+ }
+ else {
+ ml_anch = 0;
+ s = strpos;
+ }
+
+ restart:
+ if (flags & REXEC_SCREAM) {
+ SV *c = prog->check_substr;
+ char *strbeg = SvPVX(sv); /* XXXX Assume PV_force() on SCREAM! */
+ I32 p = -1; /* Internal iterator of scream. */
+ I32 *pp = data ? data->scream_pos : &p;
+
+ if (PL_screamfirst[BmRARE(c)] >= 0
+ || ( BmRARE(c) == '\n'
+ && (BmPREVIOUS(c) == SvCUR(c) - 1)
+ && SvTAIL(c) ))
+ s = screaminstr(sv, prog->check_substr,
+ start_shift + (strpos - strbeg), end_shift, pp, 0);
+ else
+ s = Nullch;
+ if (data)
+ *data->scream_olds = s;
+ }
+ else
+ s = fbm_instr((unsigned char*)s + start_shift,
+ (unsigned char*)strend - end_shift,
+ prog->check_substr, PL_multiline ? FBMrf_MULTILINE : 0);
+
+ /* Update the count-of-usability, remove useless subpatterns,
+ unshift s. */
+ finish:
+ if (!s) {
+ ++BmUSEFUL(prog->check_substr); /* hooray */
+ goto fail; /* not present */
+ }
+ else if (s - strpos > prog->check_offset_max &&
+ ((prog->reganch & ROPT_UTF8)
+ ? ((t = reghopmaybe_c(s, -(prog->check_offset_max)))
+ && t >= strpos)
+ : (t = s - prog->check_offset_max) != 0) ) {
+ if (ml_anch && t[-1] != '\n') {
+ find_anchor:
+ while (t < strend - end_shift - prog->minlen) {
+ if (*t == '\n') {
+ if (t < s - prog->check_offset_min) {
+ s = t + 1;
+ goto set_useful;
+ }
+ s = t + 1;
+ goto restart;
+ }
+ t++;
+ }
+ s = Nullch;
+ goto finish;
+ }
+ s = t;
+ set_useful:
+ ++BmUSEFUL(prog->check_substr); /* hooray/2 */
+ }
+ else {
+ if (ml_anch && sv
+ && (strpos + SvCUR(sv) != strend) && strpos[-1] != '\n') {
+ t = strpos;
+ goto find_anchor;
+ }
+ if (!(prog->reganch & ROPT_NAUGHTY)
+ && --BmUSEFUL(prog->check_substr) < 0
+ && prog->check_substr == prog->float_substr) { /* boo */
+ /* If flags & SOMETHING - do not do it many times on the same match */
+ SvREFCNT_dec(prog->check_substr);
+ prog->check_substr = Nullsv; /* disable */
+ prog->float_substr = Nullsv; /* clear */
+ s = strpos;
+ prog->reganch &= ~RE_USE_INTUIT;
+ }
+ else
+ s = strpos;
+ }
+
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "%sFound%s at offset %ld\n",
+ PL_colors[4],PL_colors[5], (long)(s - strpos)) );
+ return s;
+ fail:
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "%sNot found...%s\n",
+ PL_colors[4],PL_colors[5]));
+ return Nullch;
+}
+
/*
- regexec_flags - match a regexp against a string
*/
I32
-regexec_flags(register regexp *prog, char *stringarg, register char *strend,
- char *strbeg, I32 minend, SV *screamer, void *data, U32 flags)
+Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
+ char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
/* strend: pointer to null at end of string */
/* strbeg: real beginning of string */
/* minend: end of match must be >=minend after stringarg. */
/* Be paranoid... */
if (prog == NULL || startpos == NULL) {
- croak("NULL regexp parameter");
+ Perl_croak(aTHX_ "NULL regexp parameter");
return 0;
}
/* Check validity of program. */
if (UCHARAT(prog->program) != REG_MAGIC) {
- FAIL("corrupted regexp program");
+ Perl_croak(aTHX_ "corrupted regexp program");
}
PL_reg_flags = 0;
/* Mark beginning of line for ^ and lookbehind. */
PL_regbol = startpos;
PL_bostr = strbeg;
+ PL_reg_sv = sv;
/* Mark end of line for $ (and such) */
PL_regeol = strend;
/* If there is a "must appear" string, look for it. */
s = startpos;
- if (!(flags & REXEC_CHECKED)
- && prog->check_substr != Nullsv &&
- !(prog->reganch & ROPT_ANCH_GPOS) &&
- (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL))
- || (PL_multiline && prog->check_substr == prog->anchored_substr)) )
- {
- char *t;
- start_shift = prog->check_offset_min; /* okay to underestimate on CC */
- /* Should be nonnegative! */
- end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr);
- if (screamer) {
- if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0)
- s = screaminstr(screamer, prog->check_substr,
- start_shift + (stringarg - strbeg),
- end_shift, &scream_pos, 0);
- else
- s = Nullch;
- scream_olds = s;
- }
+
+ if (prog->reganch & ROPT_GPOS_SEEN) {
+ MAGIC *mg;
+
+ if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG
+ && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0)
+ PL_reg_ganch = strbeg + mg->mg_len;
else
- s = fbm_instr((unsigned char*)s + start_shift,
- (unsigned char*)strend - end_shift,
- prog->check_substr, 0);
- if (!s) {
- ++BmUSEFUL(prog->check_substr); /* hooray */
- goto phooey; /* not present */
- }
- else if (s - stringarg > prog->check_offset_max &&
- (UTF
- ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg)
- : (t = s - prog->check_offset_max) != 0
- )
- )
- {
- ++BmUSEFUL(prog->check_substr); /* hooray/2 */
- s = t;
- }
- else if (!(prog->reganch & ROPT_NAUGHTY)
- && --BmUSEFUL(prog->check_substr) < 0
- && prog->check_substr == prog->float_substr) { /* boo */
- SvREFCNT_dec(prog->check_substr);
- prog->check_substr = Nullsv; /* disable */
- prog->float_substr = Nullsv; /* clear */
- s = startpos;
+ PL_reg_ganch = startpos;
+ if (prog->reganch & ROPT_ANCH_GPOS) {
+ if (s > PL_reg_ganch)
+ goto phooey;
+ s = PL_reg_ganch;
}
- else
- s = startpos;
}
- DEBUG_r(
- PerlIO_printf(Perl_debug_log,
+ if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
+ re_scream_pos_data d;
+
+ d.scream_olds = &scream_olds;
+ d.scream_pos = &scream_pos;
+ s = re_intuit_start(prog, sv, s, strend, flags, &d);
+ if (!s)
+ goto phooey; /* not present */
+ }
+
+ DEBUG_r( if (!PL_colorset) reginitcolors() );
+ DEBUG_r(PerlIO_printf(Perl_debug_log,
"%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
PL_colors[4],PL_colors[5],PL_colors[0],
prog->precomp,
PL_colors[1],
(strlen(prog->precomp) > 60 ? "..." : ""),
- PL_colors[0],
+ PL_colors[0],
(strend - startpos > 60 ? 60 : strend - startpos),
startpos, PL_colors[1],
(strend - startpos > 60 ? "..." : ""))
/* Simplest case: anchored match need be tried only once. */
/* [unless only anchor is BOL and multiline is set] */
- if (prog->reganch & ROPT_ANCH) {
- if (regtry(prog, startpos))
+ if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
+ if (s == startpos && regtry(prog, startpos))
goto got_it;
- else if (!(prog->reganch & ROPT_ANCH_GPOS) &&
- (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
- || (prog->reganch & ROPT_ANCH_MBOL)))
+ else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
+ || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
{
+ char *end;
+
if (minlen)
dontbother = minlen - 1;
- strend = HOPc(strend, -dontbother);
+ end = HOPc(strend, -dontbother) - 1;
/* for multiline we only have to try after newlines */
- if (s > startpos)
- s--;
- while (s < strend) {
- if (*s++ == '\n') { /* don't need utf8skip here */
- if (s < strend && regtry(prog, s))
+ if (prog->check_substr) {
+ while (1) {
+ if (regtry(prog, s))
goto got_it;
- }
+ if (s >= end)
+ goto phooey;
+ s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
+ if (!s)
+ goto phooey;
+ }
+ } else {
+ if (s > startpos)
+ s--;
+ while (s < end) {
+ if (*s++ == '\n') { /* don't need PL_utf8skip here */
+ if (regtry(prog, s))
+ goto got_it;
+ }
+ }
}
}
goto phooey;
+ } else if (prog->reganch & ROPT_ANCH_GPOS) {
+ if (regtry(prog, PL_reg_ganch))
+ goto got_it;
+ goto phooey;
}
/* Messy cases: unanchored match. */
if (prog->anchored_substr && prog->reganch & ROPT_SKIP) {
/* we have /x+whatever/ */
- /* it must be a one character string */
+ /* it must be a one character string (XXXX Except UTF?) */
char ch = SvPVX(prog->anchored_substr)[0];
if (UTF) {
while (s < strend) {
I32 back_min =
prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
I32 delta = back_max - back_min;
- char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */
+ char *last = HOPc(strend, /* Cannot start after this */
+ -(I32)(CHR_SVLEN(must)
+ - (SvTAIL(must) != 0) + back_min));
char *last1; /* Last position checked before */
if (s > PL_bostr)
dontbother = end_shift;
strend = HOPc(strend, -dontbother);
while ( (s <= last) &&
- (screamer
- ? (s = screaminstr(screamer, must, HOPc(s, back_min) - strbeg,
+ ((flags & REXEC_SCREAM)
+ ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg,
end_shift, &scream_pos, 0))
: (s = fbm_instr((unsigned char*)HOP(s, back_min),
- (unsigned char*)strend, must, 0))) ) {
+ (unsigned char*)strend, must,
+ PL_multiline ? FBMrf_MULTILINE : 0))) ) {
if (HOPc(s, -back_max) > last1) {
last1 = HOPc(s, -back_min);
s = HOPc(s, -back_max);
s += UTF8SKIP(s);
}
break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case DIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NDIGIT:
while (s < strend) {
if (!isDIGIT(*s)) {
s += UTF8SKIP(s);
}
break;
- }
- }
- else {
- dontbother = 0;
- if (prog->float_substr != Nullsv) { /* Trim the end. */
- char *last;
- I32 oldpos = scream_pos;
-
- if (screamer) {
- last = screaminstr(screamer, prog->float_substr, s - strbeg,
- end_shift, &scream_pos, 1); /* last one */
- if (!last) {
- last = scream_olds; /* Only one occurence. */
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
}
+ else
+ tmp = 1;
+ s++;
}
- else {
- STRLEN len;
- char *little = SvPV(prog->float_substr, len);
- if (len)
- last = rninstr(s, strend, little, little + len);
+ break;
+ case NDIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
else
- last = strend; /* matching `$' */
+ tmp = 1;
+ s += UTF8SKIP(s);
}
- if (last == NULL) goto phooey; /* Should not happen! */
- dontbother = strend - last + prog->float_min_offset;
- }
- if (minlen && (dontbother < minlen))
- dontbother = minlen - 1;
- strend -= dontbother; /* this one's always in bytes! */
- /* We don't know much -- general case. */
- if (UTF) {
- for (;;) {
- if (regtry(prog, s))
- goto got_it;
- if (s >= strend)
- break;
+ break;
+ case ALNUMC:
+ while (s < strend) {
+ if (isALNUMC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case ALNUMCUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_alnumc, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
s += UTF8SKIP(s);
- };
- }
- else {
- do {
- if (regtry(prog, s))
- goto got_it;
- } while (s++ < strend);
- }
- }
-
- /* Failure. */
- goto phooey;
-
-got_it:
- prog->subbeg = strbeg;
- prog->subend = PL_regeol; /* strend may have been modified */
- RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
-
- /* make sure $`, $&, $', and $digit will work later */
- if (strbeg != prog->subbase) { /* second+ //g match. */
- if (!(flags & REXEC_COPY_STR)) {
- if (prog->subbase) {
- Safefree(prog->subbase);
- prog->subbase = Nullch;
}
- }
- else {
- I32 i = PL_regeol - startpos + (stringarg - strbeg);
- s = savepvn(strbeg, i);
- Safefree(prog->subbase);
- prog->subbase = s;
- prog->subbeg = prog->subbase;
- prog->subend = prog->subbase + i;
- s = prog->subbase + (stringarg - strbeg);
- for (i = 0; i <= prog->nparens; i++) {
- if (prog->endp[i]) {
- prog->startp[i] = s + (prog->startp[i] - startpos);
- prog->endp[i] = s + (prog->endp[i] - startpos);
+ break;
+ case ALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isALNUMC_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
}
+ else
+ tmp = 1;
+ s++;
}
- }
- }
- /* Preserve the current value of $^R */
- if (oreplsv != GvSV(PL_replgv)) {
- sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
- restored, the value remains
- the same. */
- }
- return 1;
-
-phooey:
- return 0;
-}
-
-/*
- - regtry - try match at specific point
- */
-STATIC I32 /* 0 failure, 1 success */
-regtry(regexp *prog, char *startpos)
-{
- dTHR;
- register I32 i;
- register char **sp;
- register char **ep;
- CHECKPOINT lastcp;
-
- if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
- PL_reg_eval_set = RS_init;
- DEBUG_r(DEBUG_s(
- PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
- PL_stack_sp - PL_stack_base);
- ));
- SAVEINT(cxstack[cxstack_ix].blk_oldsp);
- cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
- /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
- SAVETMPS;
- /* Apparently this is not needed, judging by wantarray. */
- /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
- cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
- }
- PL_reginput = startpos;
- PL_regstartp = prog->startp;
- PL_regendp = prog->endp;
- PL_reglastparen = &prog->lastparen;
- prog->lastparen = 0;
- PL_regsize = 0;
+ break;
+ case ALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isALNUMC_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NALNUMC:
+ while (s < strend) {
+ if (!isALNUMC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NALNUMCUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_alnumc, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isALNUMC_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isALNUMC_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case ASCII:
+ while (s < strend) {
+ if (isASCII(*(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NASCII:
+ while (s < strend) {
+ if (!isASCII(*(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRL:
+ while (s < strend) {
+ if (isCNTRL(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRLUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_cntrl,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case CNTRLL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isCNTRL_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isCNTRL_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NCNTRL:
+ while (s < strend) {
+ if (!isCNTRL(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NCNTRLUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_cntrl,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NCNTRLL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isCNTRL_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NCNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isCNTRL_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case GRAPH:
+ while (s < strend) {
+ if (isGRAPH(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case GRAPHUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_graph,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case GRAPHL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isGRAPH_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case GRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isGRAPH_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NGRAPH:
+ while (s < strend) {
+ if (!isGRAPH(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NGRAPHUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_graph,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NGRAPHL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isGRAPH_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NGRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isGRAPH_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case LOWER:
+ while (s < strend) {
+ if (isLOWER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case LOWERUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_lower,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case LOWERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isLOWER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case LOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isLOWER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NLOWER:
+ while (s < strend) {
+ if (!isLOWER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NLOWERUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_lower,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NLOWERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isLOWER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NLOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isLOWER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PRINT:
+ while (s < strend) {
+ if (isPRINT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PRINTUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_print,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PRINTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isPRINT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PRINTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isPRINT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPRINT:
+ while (s < strend) {
+ if (!isPRINT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPRINTUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_print,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPRINTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPRINT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPRINTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPRINT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PUNCT:
+ while (s < strend) {
+ if (isPUNCT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PUNCTUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_punct,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PUNCTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isPUNCT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PUNCTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isPUNCT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPUNCT:
+ while (s < strend) {
+ if (!isPUNCT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPUNCTUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_punct,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPUNCTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPUNCT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPUNCTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPUNCT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case UPPER:
+ while (s < strend) {
+ if (isUPPER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case UPPERUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_upper,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case UPPERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isUPPER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case UPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isUPPER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NUPPER:
+ while (s < strend) {
+ if (!isUPPER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NUPPERUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_upper,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NUPPERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isUPPER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NUPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isUPPER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case XDIGIT:
+ while (s < strend) {
+ if (isXDIGIT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NXDIGIT:
+ while (s < strend) {
+ if (!isXDIGIT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ }
+ }
+ else {
+ dontbother = 0;
+ if (prog->float_substr != Nullsv) { /* Trim the end. */
+ char *last;
+ I32 oldpos = scream_pos;
+
+ if (flags & REXEC_SCREAM) {
+ last = screaminstr(sv, prog->float_substr, s - strbeg,
+ end_shift, &scream_pos, 1); /* last one */
+ if (!last)
+ last = scream_olds; /* Only one occurence. */
+ }
+ else {
+ STRLEN len;
+ char *little = SvPV(prog->float_substr, len);
+
+ if (SvTAIL(prog->float_substr)) {
+ if (memEQ(strend - len + 1, little, len - 1))
+ last = strend - len + 1;
+ else if (!PL_multiline)
+ last = memEQ(strend - len, little, len)
+ ? strend - len : Nullch;
+ else
+ goto find_last;
+ } else {
+ find_last:
+ if (len)
+ last = rninstr(s, strend, little, little + len);
+ else
+ last = strend; /* matching `$' */
+ }
+ }
+ if (last == NULL) goto phooey; /* Should not happen! */
+ dontbother = strend - last + prog->float_min_offset;
+ }
+ if (minlen && (dontbother < minlen))
+ dontbother = minlen - 1;
+ strend -= dontbother; /* this one's always in bytes! */
+ /* We don't know much -- general case. */
+ if (UTF) {
+ for (;;) {
+ if (regtry(prog, s))
+ goto got_it;
+ if (s >= strend)
+ break;
+ s += UTF8SKIP(s);
+ };
+ }
+ else {
+ do {
+ if (regtry(prog, s))
+ goto got_it;
+ } while (s++ < strend);
+ }
+ }
+
+ /* Failure. */
+ goto phooey;
+
+got_it:
+ RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
+
+ if (PL_reg_eval_set) {
+ /* Preserve the current value of $^R */
+ if (oreplsv != GvSV(PL_replgv))
+ sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
+ restored, the value remains
+ the same. */
+ restore_pos(aTHXo_ 0);
+ }
+
+ /* make sure $`, $&, $', and $digit will work later */
+ if ( !(flags & REXEC_NOT_FIRST) ) {
+ if (RX_MATCH_COPIED(prog)) {
+ Safefree(prog->subbeg);
+ RX_MATCH_COPIED_off(prog);
+ }
+ if (flags & REXEC_COPY_STR) {
+ I32 i = PL_regeol - startpos + (stringarg - strbeg);
+
+ s = savepvn(strbeg, i);
+ prog->subbeg = s;
+ prog->sublen = i;
+ RX_MATCH_COPIED_on(prog);
+ }
+ else {
+ prog->subbeg = strbeg;
+ prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
+ }
+ }
+
+ return 1;
+
+phooey:
+ if (PL_reg_eval_set)
+ restore_pos(aTHXo_ 0);
+ return 0;
+}
+
+/*
+ - regtry - try match at specific point
+ */
+STATIC I32 /* 0 failure, 1 success */
+S_regtry(pTHX_ regexp *prog, char *startpos)
+{
+ dTHR;
+ register I32 i;
+ register I32 *sp;
+ register I32 *ep;
+ CHECKPOINT lastcp;
+
+ if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
+ MAGIC *mg;
+
+ PL_reg_eval_set = RS_init;
+ DEBUG_r(DEBUG_s(
+ PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
+ PL_stack_sp - PL_stack_base);
+ ));
+ SAVEINT(cxstack[cxstack_ix].blk_oldsp);
+ cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
+ /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
+ SAVETMPS;
+ /* Apparently this is not needed, judging by wantarray. */
+ /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
+ cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
+
+ if (PL_reg_sv) {
+ /* Make $_ available to executed code. */
+ if (PL_reg_sv != DEFSV) {
+ /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
+ SAVESPTR(DEFSV);
+ DEFSV = PL_reg_sv;
+ }
+
+ if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
+ && (mg = mg_find(PL_reg_sv, 'g')))) {
+ /* prepare for quick setting of pos */
+ sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
+ mg = mg_find(PL_reg_sv, 'g');
+ mg->mg_len = -1;
+ }
+ PL_reg_magic = mg;
+ PL_reg_oldpos = mg->mg_len;
+ SAVEDESTRUCTOR(restore_pos, 0);
+ }
+ if (!PL_reg_curpm)
+ New(22,PL_reg_curpm, 1, PMOP);
+ PL_reg_curpm->op_pmregexp = prog;
+ PL_reg_oldcurpm = PL_curpm;
+ PL_curpm = PL_reg_curpm;
+ if (RX_MATCH_COPIED(prog)) {
+ /* Here is a serious problem: we cannot rewrite subbeg,
+ since it may be needed if this match fails. Thus
+ $` inside (?{}) could fail... */
+ PL_reg_oldsaved = prog->subbeg;
+ PL_reg_oldsavedlen = prog->sublen;
+ RX_MATCH_COPIED_off(prog);
+ }
+ else
+ PL_reg_oldsaved = Nullch;
+ prog->subbeg = PL_bostr;
+ prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
+ }
+ prog->startp[0] = startpos - PL_bostr;
+ PL_reginput = startpos;
+ PL_regstartp = prog->startp;
+ PL_regendp = prog->endp;
+ PL_reglastparen = &prog->lastparen;
+ prog->lastparen = 0;
+ PL_regsize = 0;
+ DEBUG_r(PL_reg_starttry = startpos);
if (PL_reg_start_tmpl <= prog->nparens) {
PL_reg_start_tmpl = prog->nparens*3/2 + 3;
if(PL_reg_start_tmp)
New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*);
}
+ /* XXXX What this code is doing here?!!! There should be no need
+ to do this again and again, PL_reglastparen should take care of
+ this! */
sp = prog->startp;
ep = prog->endp;
if (prog->nparens) {
- for (i = prog->nparens; i >= 0; i--) {
- *sp++ = NULL;
- *ep++ = NULL;
+ for (i = prog->nparens; i >= 1; i--) {
+ *++sp = -1;
+ *++ep = -1;
}
}
REGCP_SET;
if (regmatch(prog->program + 1)) {
- prog->startp[0] = startpos;
- prog->endp[0] = PL_reginput;
+ prog->endp[0] = PL_reginput - PL_bostr;
return 1;
}
REGCP_UNWIND;
* advantage of machines that use a register save mask on subroutine entry.
*/
STATIC I32 /* 0 failure, 1 success */
-regmatch(regnode *prog)
+S_regmatch(pTHX_ regnode *prog)
{
dTHR;
register regnode *scan; /* Current node. */
int docolor = *PL_colors[0];
int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput);
+ /* The part of the string before starttry has one color
+ (pref0_len chars), between starttry and current
+ position another one (pref_len - pref0_len chars),
+ after the current position the third one.
+ We assume that pref0_len <= pref_len, otherwise we
+ decrease pref0_len. */
int pref_len = (locinput - PL_bostr > (5 + taill) - l
? (5 + taill) - l : locinput - PL_bostr);
- int pref0_len = pref_len - (locinput - PL_reginput);
+ int pref0_len = pref_len - (locinput - PL_reg_starttry);
if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
l = ( PL_regeol - locinput > (5 + taill) - pref_len
? (5 + taill) - pref_len : PL_regeol - locinput);
if (pref0_len < 0)
pref0_len = 0;
+ if (pref0_len > pref_len)
+ pref0_len = pref_len;
regprop(prop, scan);
PerlIO_printf(Perl_debug_log,
"%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
/* regtill = regbol; */
break;
}
- sayNO;
- case MBOL:
- if (locinput == PL_bostr
- ? PL_regprev == '\n'
- : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
- {
+ sayNO;
+ case MBOL:
+ if (locinput == PL_bostr
+ ? PL_regprev == '\n'
+ : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
+ {
+ break;
+ }
+ sayNO;
+ case SBOL:
+ if (locinput == PL_regbol && PL_regprev == '\n')
+ break;
+ sayNO;
+ case GPOS:
+ if (locinput == PL_reg_ganch)
+ break;
+ sayNO;
+ case EOL:
+ if (PL_multiline)
+ goto meol;
+ else
+ goto seol;
+ case MEOL:
+ meol:
+ if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ sayNO;
+ break;
+ case SEOL:
+ seol:
+ if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ sayNO;
+ if (PL_regeol - locinput > 1)
+ sayNO;
+ break;
+ case EOS:
+ if (PL_regeol != locinput)
+ sayNO;
+ break;
+ case SANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += PL_utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case SANY:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += PL_utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case REG_ANY:
+ if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case EXACT:
+ s = (char *) OPERAND(scan);
+ ln = UCHARAT(s++);
+ /* Inline the first character, for speed. */
+ if (UCHARAT(s) != nextchr)
+ sayNO;
+ if (PL_regeol - locinput < ln)
+ sayNO;
+ if (ln > 1 && memNE(s, locinput, ln))
+ sayNO;
+ locinput += ln;
+ nextchr = UCHARAT(locinput);
+ break;
+ case EXACTFL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case EXACTF:
+ s = (char *) OPERAND(scan);
+ ln = UCHARAT(s++);
+
+ if (UTF) {
+ char *l = locinput;
+ char *e = s + ln;
+ c1 = OP(scan) == EXACTF;
+ while (s < e) {
+ if (l >= PL_regeol)
+ sayNO;
+ if (utf8_to_uv((U8*)s, 0) != (c1 ?
+ toLOWER_utf8((U8*)l) :
+ toLOWER_LC_utf8((U8*)l)))
+ {
+ sayNO;
+ }
+ s += UTF8SKIP(s);
+ l += UTF8SKIP(l);
+ }
+ locinput = l;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+
+ /* Inline the first character, for speed. */
+ if (UCHARAT(s) != nextchr &&
+ UCHARAT(s) != ((OP(scan) == EXACTF)
+ ? PL_fold : PL_fold_locale)[nextchr])
+ sayNO;
+ if (PL_regeol - locinput < ln)
+ sayNO;
+ if (ln > 1 && (OP(scan) == EXACTF
+ ? ibcmp(s, locinput, ln)
+ : ibcmp_locale(s, locinput, ln)))
+ sayNO;
+ locinput += ln;
+ nextchr = UCHARAT(locinput);
+ break;
+ case ANYOFUTF8:
+ s = (char *) OPERAND(scan);
+ if (!REGINCLASSUTF8(scan, (U8*)locinput))
+ sayNO;
+ if (locinput >= PL_regeol)
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ case ANYOF:
+ s = (char *) OPERAND(scan);
+ if (nextchr < 0)
+ nextchr = UCHARAT(locinput);
+ if (!REGINCLASS(s, nextchr))
+ sayNO;
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALNUML:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUM:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == ALNUM
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUML:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUM:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (OP(scan) == NALNUM
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (OP(scan) == NALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case BOUNDL:
+ case NBOUNDL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUND:
+ case NBOUND:
+ /* was last char in word? */
+ ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
+ if (OP(scan) == BOUND || OP(scan) == NBOUND) {
+ ln = isALNUM(ln);
+ n = isALNUM(nextchr);
+ }
+ else {
+ ln = isALNUM_LC(ln);
+ n = isALNUM_LC(nextchr);
+ }
+ if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
+ sayNO;
+ break;
+ case BOUNDLUTF8:
+ case NBOUNDLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUNDUTF8:
+ case NBOUNDUTF8:
+ /* was last char in word? */
+ ln = (locinput != PL_regbol)
+ ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
+ if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
+ ln = isALNUM_uni(ln);
+ n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
+ }
+ else {
+ ln = isALNUM_LC_uni(ln);
+ n = isALNUM_LC_utf8((U8*)locinput);
+ }
+ if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
+ sayNO;
+ break;
+ case SPACEL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case SPACE:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!(OP(scan) == SPACE
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case SPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case SPACEUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == SPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == SPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NSPACEL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NSPACE:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == SPACE
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NSPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NSPACEUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NSPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
}
- sayNO;
- case SBOL:
- if (locinput == PL_regbol && PL_regprev == '\n')
+ if (OP(scan) == NSPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case DIGIT:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!(OP(scan) == DIGIT
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case DIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case DIGITUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NDIGITUTF8
+ ? swash_fetch(PL_utf8_digit,(U8*)locinput)
+ : isDIGIT_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
- sayNO;
- case GPOS:
- if (locinput == PL_regbol)
+ }
+ if (!isDIGIT(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NDIGIT:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == DIGIT
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NDIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NDIGITUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_digit,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
- sayNO;
- case EOL:
- if (PL_multiline)
- goto meol;
- else
- goto seol;
- case MEOL:
- meol:
- if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ }
+ if (isDIGIT(nextchr))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case SEOL:
- seol:
- if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ case ALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMC:
+ if (!nextchr)
sayNO;
- if (PL_regeol - locinput > 1)
+ if (!(OP(scan) == ALNUMC
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case EOS:
- if (PL_regeol != locinput)
+ case ALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMCUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALNUMCUTF8
+ ? swash_fetch(PL_utf8_alnumc, (U8*)locinput)
+ : isALNUMC_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALNUMCUTF8
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMC:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == ALNUMC
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMCUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_alnumc,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isALNUMC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALPHAL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALPHA:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == ALPHA
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALPHALUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALPHAUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALPHAUTF8
+ ? swash_fetch(PL_utf8_alpha, (U8*)locinput)
+ : isALPHA_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALPHAUTF8
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALPHAL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALPHA:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == ALPHA
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALPHALUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALPHAUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_alpha,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isALPHA(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ASCII:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!isASCII(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NASCII:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (isASCII(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case CNTRLL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case CNTRL:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == CNTRL
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case CNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case CNTRLUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == CNTRLUTF8
+ ? swash_fetch(PL_utf8_cntrl, (U8*)locinput)
+ : isCNTRL_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == CNTRLUTF8
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NCNTRLL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NCNTRL:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == CNTRL
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NCNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NCNTRLUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_cntrl,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isCNTRL(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case GRAPHL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case GRAPH:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == GRAPH
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case SANYUTF8:
+ case GRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case GRAPHUTF8:
+ if (!nextchr)
+ sayNO;
if (nextchr & 0x80) {
- locinput += utf8skip[nextchr];
- if (locinput > PL_regeol)
+ if (!(OP(scan) == GRAPHUTF8
+ ? swash_fetch(PL_utf8_graph, (U8*)locinput)
+ : isGRAPH_LC_utf8((U8*)locinput)))
+ {
sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!nextchr && locinput >= PL_regeol)
+ if (!(OP(scan) == GRAPHUTF8
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case SANY:
- if (!nextchr && locinput >= PL_regeol)
+ case NGRAPHL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NGRAPH:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == GRAPH
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ANYUTF8:
+ case NGRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NGRAPHUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
if (nextchr & 0x80) {
- locinput += utf8skip[nextchr];
- if (locinput > PL_regeol)
+ if (swash_fetch(PL_utf8_graph,(U8*)locinput))
sayNO;
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
- sayNO;
- nextchr = UCHARAT(++locinput);
- break;
- case REG_ANY:
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ if (isGRAPH(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case EXACT:
- s = (char *) OPERAND(scan);
- ln = UCHARAT(s++);
- /* Inline the first character, for speed. */
- if (UCHARAT(s) != nextchr)
- sayNO;
- if (PL_regeol - locinput < ln)
+ case LOWERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case LOWER:
+ if (!nextchr)
sayNO;
- if (ln > 1 && memNE(s, locinput, ln))
+ if (!(OP(scan) == LOWER
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr)))
sayNO;
- locinput += ln;
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case EXACTFL:
+ case LOWERLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case EXACTF:
- s = (char *) OPERAND(scan);
- ln = UCHARAT(s++);
-
- if (UTF) {
- char *l = locinput;
- char *e = s + ln;
- c1 = OP(scan) == EXACTF;
- while (s < e) {
- if (l >= PL_regeol)
- sayNO;
- if (utf8_to_uv((U8*)s, 0) != (c1 ?
- toLOWER_utf8((U8*)l) :
- toLOWER_LC_utf8((U8*)l)))
- {
- sayNO;
- }
- s += UTF8SKIP(s);
- l += UTF8SKIP(l);
+ case LOWERUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == LOWERUTF8
+ ? swash_fetch(PL_utf8_lower, (U8*)locinput)
+ : isLOWER_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
}
- locinput = l;
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
-
- /* Inline the first character, for speed. */
- if (UCHARAT(s) != nextchr &&
- UCHARAT(s) != ((OP(scan) == EXACTF)
- ? PL_fold : PL_fold_locale)[nextchr])
- sayNO;
- if (PL_regeol - locinput < ln)
- sayNO;
- if (ln > 1 && (OP(scan) == EXACTF
- ? ibcmp(s, locinput, ln)
- : ibcmp_locale(s, locinput, ln)))
+ if (!(OP(scan) == LOWERUTF8
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr)))
sayNO;
- locinput += ln;
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case ANYOFUTF8:
- s = (char *) OPERAND(scan);
- if (!REGINCLASSUTF8(scan, (U8*)locinput))
+ case NLOWERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NLOWER:
+ if (!nextchr)
sayNO;
- if (locinput >= PL_regeol)
+ if (OP(scan) == LOWER
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr))
sayNO;
- locinput += utf8skip[nextchr];
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case ANYOF:
- s = (char *) OPERAND(scan);
- if (nextchr < 0)
- nextchr = UCHARAT(locinput);
- if (!REGINCLASS(s, nextchr))
- sayNO;
+ case NLOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NLOWERUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_lower,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isLOWER(nextchr))
+ sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ALNUML:
+ case PRINTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case ALNUM:
+ case PRINT:
if (!nextchr)
sayNO;
- if (!(OP(scan) == ALNUM
- ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ if (!(OP(scan) == PRINT
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ALNUMLUTF8:
+ case PRINTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case ALNUMUTF8:
+ case PRINTUTF8:
if (!nextchr)
sayNO;
if (nextchr & 0x80) {
- if (!(OP(scan) == ALNUMUTF8
- ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
- : isALNUM_LC_utf8((U8*)locinput)))
+ if (!(OP(scan) == PRINTUTF8
+ ? swash_fetch(PL_utf8_print, (U8*)locinput)
+ : isPRINT_LC_utf8((U8*)locinput)))
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!(OP(scan) == ALNUMUTF8
- ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ if (!(OP(scan) == PRINTUTF8
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NALNUML:
+ case NPRINTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NALNUM:
- if (!nextchr && locinput >= PL_regeol)
+ case NPRINT:
+ if (!nextchr)
sayNO;
- if (OP(scan) == NALNUM
- ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ if (OP(scan) == PRINT
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NALNUMLUTF8:
+ case NPRINTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NALNUMUTF8:
+ case NPRINTUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
- if (OP(scan) == NALNUMUTF8
- ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
- : isALNUM_LC_utf8((U8*)locinput))
- {
+ if (swash_fetch(PL_utf8_print,(U8*)locinput))
sayNO;
- }
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (OP(scan) == NALNUMUTF8
- ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ if (isPRINT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case BOUNDL:
- case NBOUNDL:
+ case PUNCTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case BOUND:
- case NBOUND:
- /* was last char in word? */
- ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
- if (OP(scan) == BOUND || OP(scan) == NBOUND) {
- ln = isALNUM(ln);
- n = isALNUM(nextchr);
- }
- else {
- ln = isALNUM_LC(ln);
- n = isALNUM_LC(nextchr);
- }
- if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
+ case PUNCT:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == PUNCT
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case BOUNDLUTF8:
- case NBOUNDLUTF8:
+ case PUNCTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case BOUNDUTF8:
- case NBOUNDUTF8:
- /* was last char in word? */
- ln = (locinput != PL_regbol)
- ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
- if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
- ln = isALNUM_uni(ln);
- n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
- }
- else {
- ln = isALNUM_LC_uni(ln);
- n = isALNUM_LC_utf8((U8*)locinput);
+ case PUNCTUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == PUNCTUTF8
+ ? swash_fetch(PL_utf8_punct, (U8*)locinput)
+ : isPUNCT_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
}
- if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
+ if (!(OP(scan) == PUNCTUTF8
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case SPACEL:
+ case NPUNCTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case SPACE:
- if (!nextchr && locinput >= PL_regeol)
+ case NPUNCT:
+ if (!nextchr)
sayNO;
- if (!(OP(scan) == SPACE
- ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ if (OP(scan) == PUNCT
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case SPACELUTF8:
+ case NPUNCTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case SPACEUTF8:
+ case NPUNCTUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
- if (!(OP(scan) == SPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
- : isSPACE_LC_utf8((U8*)locinput)))
- {
+ if (swash_fetch(PL_utf8_punct,(U8*)locinput))
sayNO;
- }
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!(OP(scan) == SPACEUTF8
- ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ if (isPUNCT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NSPACEL:
+ case UPPERL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NSPACE:
+ case UPPER:
if (!nextchr)
sayNO;
- if (OP(scan) == SPACE
- ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ if (!(OP(scan) == UPPER
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NSPACELUTF8:
+ case UPPERLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NSPACEUTF8:
+ case UPPERUTF8:
if (!nextchr)
sayNO;
if (nextchr & 0x80) {
- if (OP(scan) == NSPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
- : isSPACE_LC_utf8((U8*)locinput))
+ if (!(OP(scan) == UPPERUTF8
+ ? swash_fetch(PL_utf8_upper, (U8*)locinput)
+ : isUPPER_LC_utf8((U8*)locinput)))
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (OP(scan) == NSPACEUTF8
- ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ if (!(OP(scan) == UPPERUTF8
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case DIGIT:
- if (!isDIGIT(nextchr))
+ case NUPPERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NUPPER:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == UPPER
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case DIGITUTF8:
+ case NUPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NUPPERUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
if (nextchr & 0x80) {
- if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
+ if (swash_fetch(PL_utf8_upper,(U8*)locinput))
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!isDIGIT(nextchr))
+ if (isUPPER(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NDIGIT:
+ case XDIGIT:
if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (isDIGIT(nextchr))
+ if (!isXDIGIT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NDIGITUTF8:
+ case NXDIGIT:
if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (nextchr & 0x80) {
- if (swash_fetch(PL_utf8_digit,(U8*)locinput))
- sayNO;
- locinput += utf8skip[nextchr];
- nextchr = UCHARAT(locinput);
- break;
- }
- if (isDIGIT(nextchr))
+ if (isXDIGIT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
case CLUMP:
if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
locinput += UTF8SKIP(locinput);
if (locinput > PL_regeol)
case REF:
case REFF:
n = ARG(scan); /* which paren pair */
- s = PL_regstartp[n];
- if (*PL_reglastparen < n || !s)
+ ln = PL_regstartp[n];
+ if (*PL_reglastparen < n || ln == -1)
sayNO; /* Do not match unless seen CLOSEn. */
- if (s == PL_regendp[n])
+ if (ln == PL_regendp[n])
break;
+ s = PL_bostr + ln;
if (UTF && OP(scan) != REF) { /* REF can do byte comparison */
char *l = locinput;
- char *e = PL_regendp[n];
+ char *e = PL_bostr + PL_regendp[n];
/*
* Note that we can't do the "other character" lookup trick as
* in the 8-bit case (no pun intended) because in Unicode we
(UCHARAT(s) != ((OP(scan) == REFF
? PL_fold : PL_fold_locale)[nextchr]))))
sayNO;
- ln = PL_regendp[n] - s;
+ ln = PL_regendp[n] - ln;
if (locinput + ln > PL_regeol)
sayNO;
if (ln > 1 && (OP(scan) == REF
n = ARG(scan);
PL_op = (OP_4tree*)PL_regdata->data[n];
DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) );
- PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]);
+ PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]);
+ PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
- CALLRUNOPS(); /* Scalar context. */
+ CALLRUNOPS(aTHX); /* Scalar context. */
SPAGAIN;
ret = POPs;
PUTBACK;
I32 onpar = PL_regnpar;
pm.op_pmflags = 0;
- re = CALLREGCOMP(t, t + len, &pm);
+ re = CALLREGCOMP(aTHX_ t, t + len, &pm);
if (!(SvFLAGS(ret)
& (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0);
break;
case CLOSE:
n = ARG(scan); /* which paren pair */
- PL_regstartp[n] = PL_reg_start_tmp[n];
- PL_regendp[n] = locinput;
+ PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
+ PL_regendp[n] = locinput - PL_bostr;
if (n > *PL_reglastparen)
*PL_reglastparen = n;
break;
case GROUPP:
n = ARG(scan); /* which paren pair */
- sw = (*PL_reglastparen >= n && PL_regendp[n] != NULL);
+ sw = (*PL_reglastparen >= n && PL_regendp[n] != -1);
break;
case IFTHEN:
if (sw)
if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
&& !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- warner(WARN_UNSAFE, "%s limit (%d) exceeded",
+ Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded",
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
&& !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- warner(WARN_UNSAFE, "%s limit (%d) exceeded",
+ Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded",
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
sayYES;
REGCP_UNWIND;
for (n = *PL_reglastparen; n > lastparen; n--)
- PL_regendp[n] = 0;
+ PL_regendp[n] = -1;
*PL_reglastparen = n;
scan = next;
/*SUPPRESS 560*/
{
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -l);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] =
+ HOPc(PL_reginput, -l) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
);
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -l);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
sayNO;
locinput = PL_reginput;
REGCP_SET;
+ if (c1 != -1000) {
+ char *e = locinput + n - ln; /* Should not check after this */
+ char *old = locinput;
+
+ if (e >= PL_regeol || (n == REG_INFTY))
+ e = PL_regeol - 1;
+ while (1) {
+ /* Find place 'next' could work */
+ if (c1 == c2) {
+ while (locinput <= e && *locinput != c1)
+ locinput++;
+ } else {
+ while (locinput <= e
+ && *locinput != c1
+ && *locinput != c2)
+ locinput++;
+ }
+ if (locinput > e)
+ sayNO;
+ /* PL_reginput == old now */
+ if (locinput != old) {
+ ln = 1; /* Did some */
+ if (regrepeat(scan, locinput - old) <
+ locinput - old)
+ sayNO;
+ }
+ /* PL_reginput == locinput now */
+ if (paren) {
+ if (ln) {
+ PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr;
+ PL_regendp[paren] = locinput - PL_bostr;
+ }
+ else
+ PL_regendp[paren] = -1;
+ }
+ if (regmatch(next))
+ sayYES;
+ PL_reginput = locinput; /* Could be reset... */
+ REGCP_UNWIND;
+ /* Couldn't or didn't -- move forward. */
+ old = locinput++;
+ }
+ }
+ else
while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
/* If it could work, try it. */
if (c1 == -1000 ||
{
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
{
if (paren && n) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
case UNLESSM:
n = 0;
if (scan->flags) {
- s = HOPMAYBEc(locinput, -scan->flags);
- if (!s)
- goto say_yes;
- PL_reginput = s;
+ if (UTF) { /* XXXX This is absolutely
+ broken, we read before
+ start of string. */
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s)
+ goto say_yes;
+ PL_reginput = s;
+ }
+ else {
+ if (locinput < PL_bostr + scan->flags)
+ goto say_yes;
+ PL_reginput = locinput - scan->flags;
+ goto do_ifmatch;
+ }
}
else
PL_reginput = locinput;
case IFMATCH:
n = 1;
if (scan->flags) {
- s = HOPMAYBEc(locinput, -scan->flags);
- if (!s)
- goto say_no;
- PL_reginput = s;
+ if (UTF) { /* XXXX This is absolutely
+ broken, we read before
+ start of string. */
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s || s < PL_bostr)
+ goto say_no;
+ PL_reginput = s;
+ }
+ else {
+ if (locinput < PL_bostr + scan->flags)
+ goto say_no;
+ PL_reginput = locinput - scan->flags;
+ goto do_ifmatch;
+ }
}
else
PL_reginput = locinput;
default:
PerlIO_printf(PerlIO_stderr(), "%lx %d\n",
(unsigned long)scan, OP(scan));
- FAIL("regexp memory corruption");
+ Perl_croak(aTHX_ "regexp memory corruption");
}
scan = next;
}
* We get here only if there's trouble -- normally "case END" is
* the terminating point.
*/
- FAIL("corrupted regexp pointers");
+ Perl_croak(aTHX_ "corrupted regexp pointers");
/*NOTREACHED*/
sayNO;
* rather than incrementing count on every character. [Er, except utf8.]]
*/
STATIC I32
-regrepeat(regnode *p, I32 max)
+S_regrepeat(pTHX_ regnode *p, I32 max)
{
dTHR;
register char *scan;
*/
STATIC I32
-regrepeat_hard(regnode *p, I32 max, I32 *lp)
+S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp)
{
dTHR;
register char *scan;
}
/*
- - regclass - determine if a character falls into a character class
+ - reginclass - determine if a character falls into a character class
*/
STATIC bool
-reginclass(register char *p, register I32 c)
+S_reginclass(pTHX_ register char *p, register I32 c)
{
dTHR;
- char flags = *p;
+ char flags = ANYOF_FLAGS(p);
bool match = FALSE;
c &= 0xFF;
- if (ANYOF_TEST(p, c))
+ if (ANYOF_BITMAP_TEST(p, c))
match = TRUE;
else if (flags & ANYOF_FOLD) {
I32 cf;
}
else
cf = PL_fold[c];
- if (ANYOF_TEST(p, cf))
+ if (ANYOF_BITMAP_TEST(p, cf))
match = TRUE;
}
- if (!match && (flags & ANYOF_ISA)) {
+ if (!match && (flags & ANYOF_CLASS)) {
PL_reg_flags |= RF_tainted;
-
- if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) ||
- ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) ||
- ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) ||
- ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c)))
+ if (
+ (ANYOF_CLASS_TEST(p, ANYOF_ALNUM) && isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_SPACE) && isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ALPHA) && isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ASCII) && isASCII(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NASCII) && !isASCII(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_LOWER) && isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_PRINT) && isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_UPPER) && isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_XDIGIT) && isXDIGIT(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c))
+ ) /* How's that for a conditional? */
{
match = TRUE;
}
}
STATIC bool
-reginclassutf8(regnode *f, U8 *p)
+S_reginclassutf8(pTHX_ regnode *f, U8 *p)
{
dTHR;
char flags = ARG1(f);
match = TRUE;
}
- if (!match && (flags & ANYOF_ISA)) {
- PL_reg_flags |= RF_tainted;
-
- if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) ||
- ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) ||
- ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) ||
- ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p)))
- {
- match = TRUE;
- }
- }
+ /* UTF8 combined with ANYOF_CLASS is ill-defined. */
return (flags & ANYOF_INVERT) ? !match : match;
}
STATIC U8 *
-reghop(U8 *s, I32 off)
+S_reghop(pTHX_ U8 *s, I32 off)
{
dTHR;
if (off >= 0) {
}
STATIC U8 *
-reghopmaybe(U8* s, I32 off)
+S_reghopmaybe(pTHX_ U8* s, I32 off)
{
dTHR;
if (off >= 0) {
}
return s;
}
+
+#ifdef PERL_OBJECT
+#define NO_XSLOCKS
+#include "XSUB.h"
+#endif
+
+static void
+restore_pos(pTHXo_ void *arg)
+{
+ dTHR;
+ if (PL_reg_eval_set) {
+ if (PL_reg_oldsaved) {
+ PL_reg_re->subbeg = PL_reg_oldsaved;
+ PL_reg_re->sublen = PL_reg_oldsavedlen;
+ RX_MATCH_COPIED_on(PL_reg_re);
+ }
+ PL_reg_magic->mg_len = PL_reg_oldpos;
+ PL_reg_eval_set = 0;
+ PL_curpm = PL_reg_oldcurpm;
+ }
+}
+
https://perl5.git.perl.org / perl5.git / blobdiff