* "One Ring to rule them all, One Ring to find them..."
*/
+/* This file contains functions for executing a regular expression. See
+ * also regcomp.c which funnily enough, contains functions for compiling
+ * a regular expression.
+ *
+ * This file is also copied at build time to ext/re/re_exec.c, where
+ * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT.
+ * This causes the main functions to be compiled under new names and with
+ * debugging support added, which makes "use re 'debug'" work.
+
+ */
+
/* NOTE: this is derived from Henry Spencer's regexp code, and should not
* confused with the original package (see point 3 below). Thanks, Henry!
*/
# define PERL_NO_GET_CONTEXT
#endif
-/*SUPPRESS 112*/
/*
* pregcomp and pregexec -- regsub and regerror are not used in perl
*
*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-2002, Larry Wall
+ **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+ **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others
****
**** You may distribute under the terms of either the GNU General Public
**** License or the Artistic License, as specified in the README file.
#define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv))
#define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b)
-#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
-#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
-#define HOP(pos,off) (PL_reg_match_utf8 ? reghop((U8*)pos, off) : (U8*)(pos + off))
-#define HOPMAYBE(pos,off) (PL_reg_match_utf8 ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
-#define HOPc(pos,off) ((char*)HOP(pos,off))
-#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
-
-#define HOPBACK(pos, off) ( \
- (PL_reg_match_utf8) \
- ? reghopmaybe((U8*)pos, -off) \
+#define HOPc(pos,off) ((char *)(PL_reg_match_utf8 \
+ ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \
+ : (U8*)(pos + off)))
+#define HOPBACKc(pos, off) ((char*) \
+ ((PL_reg_match_utf8) \
+ ? reghopmaybe3((U8*)pos, -off, ((U8*)(off < 0 ? PL_regeol : PL_bostr))) \
: (pos - off >= PL_bostr) \
? (U8*)(pos - off) \
- : (U8*)NULL \
+ : (U8*)NULL) \
)
-#define HOPBACKc(pos, off) (char*)HOPBACK(pos, off)
-#define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
#define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
#define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
-#define HOPMAYBE3(pos,off,lim) (PL_reg_match_utf8 ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
#define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim))
-#define HOPMAYBE3c(pos,off,lim) ((char*)HOPMAYBE3(pos,off,lim))
-#define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((U8*)b); LEAVE; } } STMT_END
+#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
+#define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a")
+#define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0")
+#define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ")
+#define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86")
/* for use after a quantifier and before an EXACT-like node -- japhy */
#define JUMPABLE(rn) ( \
STATIC CHECKPOINT
S_regcppush(pTHX_ I32 parenfloor)
{
- int retval = PL_savestack_ix;
+ dVAR;
+ const int retval = PL_savestack_ix;
#define REGCP_PAREN_ELEMS 4
- int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
+ const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS;
int p;
if (paren_elems_to_push < 0)
Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
#define REGCP_OTHER_ELEMS 6
- SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS);
+ SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
for (p = PL_regsize; p > parenfloor; p--) {
/* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
SSPUSHINT(PL_regendp[p]);
}
/* These are needed since we do not localize EVAL nodes: */
-# define REGCP_SET(cp) DEBUG_r(PerlIO_printf(Perl_debug_log, \
+# define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \
" Setting an EVAL scope, savestack=%"IVdf"\n", \
(IV)PL_savestack_ix)); cp = PL_savestack_ix
-# define REGCP_UNWIND(cp) DEBUG_r(cp != PL_savestack_ix ? \
+# define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \
PerlIO_printf(Perl_debug_log, \
" Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
(IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp)
STATIC char *
S_regcppop(pTHX)
{
+ dVAR;
I32 i;
U32 paren = 0;
char *input;
- I32 tmps;
+
+ GET_RE_DEBUG_FLAGS_DECL;
/* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */
i = SSPOPINT;
/* Now restore the parentheses context. */
for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
i > 0; i -= REGCP_PAREN_ELEMS) {
+ I32 tmps;
paren = (U32)SSPOPINT;
PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
PL_regstartp[paren] = SSPOPINT;
tmps = SSPOPINT;
if (paren <= *PL_reglastparen)
PL_regendp[paren] = tmps;
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
" restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n",
(UV)paren, (IV)PL_regstartp[paren],
(paren > *PL_reglastparen ? "(no)" : ""));
);
}
- DEBUG_r(
+ DEBUG_EXECUTE_r(
if ((I32)(*PL_reglastparen + 1) <= PL_regnpar) {
PerlIO_printf(Perl_debug_log,
" restoring \\%"IVdf"..\\%"IVdf" to undef\n",
return input;
}
-STATIC char *
-S_regcp_set_to(pTHX_ I32 ss)
-{
- I32 tmp = PL_savestack_ix;
-
- PL_savestack_ix = ss;
- regcppop();
- PL_savestack_ix = tmp;
- return Nullch;
-}
-
typedef struct re_cc_state
{
I32 ss;
#define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */
-#define TRYPAREN(paren, n, input) { \
+#define TRYPAREN(paren, n, input, where) { \
if (paren) { \
if (n) { \
PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \
else \
PL_regendp[paren] = -1; \
} \
- if (regmatch(next)) \
+ REGMATCH(next, where); \
+ if (result) \
sayYES; \
if (paren && n) \
PL_regendp[paren] = -1; \
STATIC void
S_cache_re(pTHX_ regexp *prog)
{
+ dVAR;
PL_regprecomp = prog->precomp; /* Needed for FAIL. */
#ifdef DEBUGGING
PL_regprogram = prog->program;
/* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */
-/* If SCREAM, then SvPVX(sv) should be compatible with strpos and strend.
+/* If SCREAM, then SvPVX_const(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. */
Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
char *strend, U32 flags, re_scream_pos_data *data)
{
+ dVAR;
register I32 start_shift = 0;
/* Should be nonnegative! */
register I32 end_shift = 0;
register SV *check;
char *strbeg;
char *t;
- int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */
+ const int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */
I32 ml_anch;
- register char *other_last = Nullch; /* other substr checked before this */
- char *check_at = Nullch; /* check substr found at this pos */
+ register char *other_last = NULL; /* other substr checked before this */
+ char *check_at = NULL; /* check substr found at this pos */
+ const I32 multiline = prog->reganch & PMf_MULTILINE;
#ifdef DEBUGGING
- char *i_strpos = strpos;
- SV *dsv = PERL_DEBUG_PAD_ZERO(0);
+ const char * const i_strpos = strpos;
+ SV * const dsv = PERL_DEBUG_PAD_ZERO(0);
#endif
+ GET_RE_DEBUG_FLAGS_DECL;
+
+ RX_MATCH_UTF8_set(prog,do_utf8);
+
if (prog->reganch & ROPT_UTF8) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"UTF-8 regex...\n"));
PL_reg_flags |= RF_utf8;
}
- DEBUG_r({
- char *s = PL_reg_match_utf8 ?
+ DEBUG_EXECUTE_r({
+ const char *s = PL_reg_match_utf8 ?
sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) :
strpos;
- int len = PL_reg_match_utf8 ?
+ const int len = PL_reg_match_utf8 ?
strlen(s) : strend - strpos;
if (!PL_colorset)
reginitcolors();
if (PL_reg_match_utf8)
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"UTF-8 target...\n"));
PerlIO_printf(Perl_debug_log,
- "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n",
- PL_colors[4],PL_colors[5],PL_colors[0],
+ "%sGuessing start of match, REx%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 ? "..." : ""),
/* CHR_DIST() would be more correct here but it makes things slow. */
if (prog->minlen > strend - strpos) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"String too short... [re_intuit_start]\n"));
goto fail;
}
check = prog->check_substr;
}
if (check == &PL_sv_undef) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"Non-utf string cannot match utf check string\n"));
goto fail;
}
if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */
ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE)
|| ( (prog->reganch & ROPT_ANCH_BOL)
- && !PL_multiline ) ); /* Check after \n? */
+ && !multiline ) ); /* Check after \n? */
if (!ml_anch) {
if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */
| ROPT_IMPLICIT)) /* not a real BOL */
- /* SvCUR is not set on references: SvRV and SvPVX overlap */
+ /* SvCUR is not set on references: SvRV and SvPVX_const overlap */
&& sv && !SvROK(sv)
&& (strpos != strbeg)) {
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
goto fail;
}
if (prog->check_offset_min == prog->check_offset_max &&
if ( strend - s > slen || strend - s < slen - 1
|| (strend - s == slen && strend[-1] != '\n')) {
- DEBUG_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n"));
goto fail_finish;
}
/* Now should match s[0..slen-2] */
slen--;
- if (slen && (*SvPVX(check) != *s
+ if (slen && (*SvPVX_const(check) != *s
|| (slen > 1
- && memNE(SvPVX(check), s, slen)))) {
+ && memNE(SvPVX_const(check), s, slen)))) {
report_neq:
- DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
goto fail_finish;
}
}
- else if (*SvPVX(check) != *s
+ else if (*SvPVX_const(check) != *s
|| ((slen = SvCUR(check)) > 1
- && memNE(SvPVX(check), s, slen)))
+ && memNE(SvPVX_const(check), s, slen)))
goto report_neq;
+ check_at = s;
goto success_at_start;
}
}
end_shift = prog->minlen - start_shift -
CHR_SVLEN(check) + (SvTAIL(check) != 0);
if (!ml_anch) {
- I32 end = prog->check_offset_max + CHR_SVLEN(check)
+ const I32 end = prog->check_offset_max + CHR_SVLEN(check)
- (SvTAIL(check) != 0);
- I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
+ const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
if (end_shift < eshift)
end_shift = eshift;
the "check" substring in the region corrected by start/end_shift. */
if (flags & REXEC_SCREAM) {
I32 p = -1; /* Internal iterator of scream. */
- I32 *pp = data ? data->scream_pos : &p;
+ I32 * const pp = data ? data->scream_pos : &p;
if (PL_screamfirst[BmRARE(check)] >= 0
|| ( BmRARE(check) == '\n'
start_shift + (s - strbeg), end_shift, pp, 0);
else
goto fail_finish;
+ /* we may be pointing at the wrong string */
+ if (s && RX_MATCH_COPIED(prog))
+ s = strbeg + (s - SvPVX_const(sv));
if (data)
*data->scream_olds = s;
}
else if (prog->reganch & ROPT_CANY_SEEN)
s = fbm_instr((U8*)(s + start_shift),
(U8*)(strend - end_shift),
- check, PL_multiline ? FBMrf_MULTILINE : 0);
+ check, multiline ? FBMrf_MULTILINE : 0);
else
s = fbm_instr(HOP3(s, start_shift, strend),
HOP3(strend, -end_shift, strbeg),
- check, PL_multiline ? FBMrf_MULTILINE : 0);
+ check, multiline ? FBMrf_MULTILINE : 0);
/* Update the count-of-usability, remove useless subpatterns,
unshift s. */
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s",
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr \"%s%.*s%s\"%s%s",
(s ? "Found" : "Did not find"),
(check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"),
PL_colors[0],
(int)(SvCUR(check) - (SvTAIL(check)!=0)),
- SvPVX(check),
+ SvPVX_const(check),
PL_colors[1], (SvTAIL(check) ? "$" : ""),
(s ? " at offset " : "...\n") ) );
check_at = s;
/* Finish the diagnostic message */
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
/* Got a candidate. Check MBOL anchoring, and the *other* substr.
Start with the other substr.
XXXX no SCREAM optimization yet - and a very coarse implementation
- XXXX /ttx+/ results in anchored=`ttx', floating=`x'. floating will
+ XXXX /ttx+/ results in anchored="ttx", floating="x". floating will
*always* match. Probably should be marked during compile...
Probably it is right to do no SCREAM here...
*/
if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) {
do_other_anchored:
{
- char *last = HOP3c(s, -start_shift, strbeg), *last1, *last2;
+ char * const last = HOP3c(s, -start_shift, strbeg);
+ char *last1, *last2;
char *s1 = s;
SV* must;
t = s - prog->check_offset_max;
if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
- && (!(prog->reganch & ROPT_UTF8)
+ && (!do_utf8
|| ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos))
&& t > strpos)))
/* EMPTY */;
must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr;
if (must == &PL_sv_undef) {
s = (char*)NULL;
- DEBUG_r(must = prog->anchored_utf8); /* for debug */
+ DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */
}
else
s = fbm_instr(
HOP3(HOP3(last1, prog->anchored_offset, strend)
+ SvCUR(must), -(SvTAIL(must)!=0), strbeg),
must,
- PL_multiline ? FBMrf_MULTILINE : 0
+ multiline ? FBMrf_MULTILINE : 0
);
- DEBUG_r(PerlIO_printf(Perl_debug_log,
- "%s anchored substr `%s%.*s%s'%s",
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
+ "%s anchored substr \"%s%.*s%s\"%s",
(s ? "Found" : "Contradicts"),
PL_colors[0],
(int)(SvCUR(must)
- (SvTAIL(must)!=0)),
- SvPVX(must),
+ SvPVX_const(must),
PL_colors[1], (SvTAIL(must) ? "$" : "")));
if (!s) {
if (last1 >= last2) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
", giving up...\n"));
goto fail_finish;
}
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
", trying floating at offset %ld...\n",
(long)(HOP3c(s1, 1, strend) - i_strpos)));
other_last = HOP3c(last1, prog->anchored_offset+1, strend);
goto restart;
}
else {
- DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
(long)(s - i_strpos)));
t = HOP3c(s, -prog->anchored_offset, strbeg);
other_last = HOP3c(s, 1, strend);
and end-of-str is not later than strend we are OK. */
if (must == &PL_sv_undef) {
s = (char*)NULL;
- DEBUG_r(must = prog->float_utf8); /* for debug message */
+ DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */
}
else
s = fbm_instr((unsigned char*)s,
(unsigned char*)last + SvCUR(must)
- (SvTAIL(must)!=0),
- must, PL_multiline ? FBMrf_MULTILINE : 0);
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s",
+ must, multiline ? FBMrf_MULTILINE : 0);
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr \"%s%.*s%s\"%s",
(s ? "Found" : "Contradicts"),
PL_colors[0],
(int)(SvCUR(must) - (SvTAIL(must)!=0)),
- SvPVX(must),
+ SvPVX_const(must),
PL_colors[1], (SvTAIL(must) ? "$" : "")));
if (!s) {
if (last1 == last) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
", giving up...\n"));
goto fail_finish;
}
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
", trying anchored starting at offset %ld...\n",
(long)(s1 + 1 - i_strpos)));
other_last = last;
goto restart;
}
else {
- DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
(long)(s - i_strpos)));
other_last = s; /* Fix this later. --Hugo */
s = s1;
t = s - prog->check_offset_max;
if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */
- && (!(prog->reganch & ROPT_UTF8)
+ && (!do_utf8
|| ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos))
&& t > strpos))) {
/* Fixed substring is found far enough so that the match
is float. Redo checking for "other"=="fixed".
*/
strpos = t + 1;
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
- PL_colors[0],PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n",
+ PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset)));
goto do_other_anchored;
}
/* We don't contradict the found floating substring. */
/* XXXX Why not check for STCLASS? */
s = t + 1;
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
- PL_colors[0],PL_colors[1], (long)(s - i_strpos)));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n",
+ PL_colors[0], PL_colors[1], (long)(s - i_strpos)));
goto set_useful;
}
/* Position contradicts check-string */
/* XXXX probably better to look for check-string
than for "\n", so one should lower the limit for t? */
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
- PL_colors[0],PL_colors[1], (long)(t + 1 - i_strpos)));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n",
+ PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos)));
other_last = strpos = s = t + 1;
goto restart;
}
t++;
}
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
- PL_colors[0],PL_colors[1]));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n",
+ PL_colors[0], PL_colors[1]));
goto fail_finish;
}
else {
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
- PL_colors[0],PL_colors[1]));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n",
+ PL_colors[0], PL_colors[1]));
}
s = t;
set_useful:
t = strpos;
goto find_anchor;
}
- DEBUG_r( if (ml_anch)
+ DEBUG_EXECUTE_r( if (ml_anch)
PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n",
- (long)(strpos - i_strpos), PL_colors[0],PL_colors[1]);
+ (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]);
);
success_at_start:
if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
)))
{
/* If flags & SOMETHING - do not do it many times on the same match */
- DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr);
if (do_utf8 ? prog->check_substr : prog->check_utf8)
SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8);
- prog->check_substr = prog->check_utf8 = Nullsv; /* disable */
- prog->float_substr = prog->float_utf8 = Nullsv; /* clear */
- check = Nullsv; /* abort */
+ prog->check_substr = prog->check_utf8 = NULL; /* disable */
+ prog->float_substr = prog->float_utf8 = NULL; /* clear */
+ check = NULL; /* abort */
s = strpos;
/* XXXX This is a remnant of the old implementation. It
looks wasteful, since now INTUIT can use many
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(). */
- U8* str = (U8*)STRING(prog->regstclass);
- int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
+ const U8* const str = (U8*)STRING(prog->regstclass);
+ const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
? CHR_DIST(str+STR_LEN(prog->regstclass), str)
: 1);
- char *endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
+ const char * const endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch)
? HOP3c(s, (prog->minlen ? cl_l : 0), strend)
: (prog->float_substr || prog->float_utf8
? HOP3c(HOP3c(check_at, -start_shift, strbeg),
cl_l, strend)
: strend);
- char *startpos = strbeg;
t = s;
- if (prog->reganch & ROPT_UTF8) {
- PL_regdata = prog->data;
- PL_bostr = startpos;
- }
- s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1);
+ cache_re(prog);
+ s = find_byclass(prog, prog->regstclass, s, endpos, 1);
if (!s) {
#ifdef DEBUGGING
- char *what = 0;
+ const char *what = NULL;
#endif
if (endpos == strend) {
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"Could not match STCLASS...\n") );
goto fail;
}
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"This position contradicts STCLASS...\n") );
if ((prog->reganch & ROPT_ANCH) && !ml_anch)
goto fail;
/* Contradict one of substrings */
if (prog->anchored_substr || prog->anchored_utf8) {
if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) {
- DEBUG_r( what = "anchored" );
+ DEBUG_EXECUTE_r( what = "anchored" );
hop_and_restart:
s = HOP3c(t, 1, strend);
if (s + start_shift + end_shift > strend) {
/* XXXX Should be taken into account earlier? */
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"Could not match STCLASS...\n") );
goto fail;
}
if (!check)
goto giveup;
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"Looking for %s substr starting at offset %ld...\n",
what, (long)(s + start_shift - i_strpos)) );
goto restart;
s = check_at;
if (!check)
goto giveup;
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"Looking for anchored substr starting at offset %ld...\n",
(long)(other_last - i_strpos)) );
goto do_other_anchored;
s = t = t + 1;
if (!check)
goto giveup;
- DEBUG_r( PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log,
"Looking for /%s^%s/m starting at offset %ld...\n",
- PL_colors[0],PL_colors[1], (long)(t - i_strpos)) );
+ PL_colors[0], PL_colors[1], (long)(t - i_strpos)) );
goto try_at_offset;
}
if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */
/* Check is floating subtring. */
retry_floating_check:
t = check_at - start_shift;
- DEBUG_r( what = "floating" );
+ DEBUG_EXECUTE_r( what = "floating" );
goto hop_and_restart;
}
if (t != s) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"By STCLASS: moving %ld --> %ld\n",
(long)(t - i_strpos), (long)(s - i_strpos))
);
}
else {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"Does not contradict STCLASS...\n");
);
}
}
giveup:
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n",
PL_colors[4], (check ? "Guessed" : "Giving up"),
PL_colors[5], (long)(s - i_strpos)) );
return s;
if (prog->check_substr || prog->check_utf8) /* could be removed already */
BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */
fail:
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
- PL_colors[4],PL_colors[5]));
- return Nullch;
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n",
+ PL_colors[4], PL_colors[5]));
+ return NULL;
}
/* We know what class REx starts with. Try to find this position... */
STATIC char *
-S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *startpos, I32 norun)
+S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32 norun)
{
- I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
+ dVAR;
+ const I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
char *m;
STRLEN ln;
+ STRLEN lnc;
+ register STRLEN uskip;
unsigned int c1;
unsigned int c2;
char *e;
register I32 tmp = 1; /* Scratch variable? */
- register bool do_utf8 = PL_reg_match_utf8;
+ register const bool do_utf8 = PL_reg_match_utf8;
/* We know what class it must start with. */
switch (OP(c)) {
case ANYOF:
- while (s < strend) {
- STRLEN skip = do_utf8 ? UTF8SKIP(s) : 1;
-
- if (do_utf8 ?
- reginclass(c, (U8*)s, 0, do_utf8) :
- REGINCLASS(c, (U8*)s) ||
- (ANYOF_FOLD_SHARP_S(c, s, strend) &&
- /* The assignment of 2 is intentional:
- * for the sharp s, the skip is 2. */
- (skip = SHARP_S_SKIP)
- )) {
- if (tmp && (norun || regtry(prog, s)))
- goto got_it;
- else
- tmp = doevery;
- }
- else
- tmp = 1;
- s += skip;
+ if (do_utf8) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
+ if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
+ !UTF8_IS_INVARIANT((U8)s[0]) ?
+ reginclass(c, (U8*)s, 0, do_utf8) :
+ REGINCLASS(c, (U8*)s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += uskip;
+ }
+ }
+ else {
+ while (s < strend) {
+ STRLEN skip = 1;
+
+ if (REGINCLASS(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 && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += skip;
+ }
}
break;
case CANY:
}
break;
case EXACTF:
- m = STRING(c);
- ln = STR_LEN(c);
+ m = STRING(c);
+ ln = STR_LEN(c); /* length to match in octets/bytes */
+ lnc = (I32) ln; /* length to match in characters */
if (UTF) {
STRLEN ulen1, ulen2;
- U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
- U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
+ U8 *sm = (U8 *) m;
+ U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
+ U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
+ const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
- c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN_UCLC,
- 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
- c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN_UCLC,
- 0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+ 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);
+ }
}
else {
c1 = *(U8*)m;
}
goto do_exactf;
case EXACTFL:
- m = STRING(c);
- ln = STR_LEN(c);
+ m = STRING(c);
+ ln = STR_LEN(c);
+ lnc = (I32) ln;
c1 = *(U8*)m;
c2 = PL_fold_locale[c1];
do_exactf:
- e = HOP3c(strend, -(I32)ln, s);
+ e = HOP3c(strend, -((I32)lnc), s);
if (norun && e < s)
e = s; /* Due to minlen logic of intuit() */
if (do_utf8) {
UV c, f;
- U8 tmpbuf [UTF8_MAXLEN+1];
- U8 foldbuf[UTF8_MAXLEN_FOLD+1];
+ U8 tmpbuf [UTF8_MAXBYTES+1];
STRLEN len, foldlen;
-
+ const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
if (c1 == c2) {
+ /* Upper and lower of 1st char are equal -
+ * probably not a "letter". */
while (s <= e) {
- c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &len,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
+ uniflags);
if ( c == c1
&& (ln == len ||
ibcmp_utf8(s, (char **)0, 0, do_utf8,
&& (norun || regtry(prog, 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
}
else {
while (s <= e) {
- c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &len,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len,
+ uniflags);
/* Handle some of the three Greek sigmas cases.
* Note that not all the possible combinations
&& (norun || regtry(prog, 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
if (s == PL_bostr)
tmp = '\n';
else {
- U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
-
+ U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
}
tmp = ((OP(c) == BOUND ?
isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
- LOAD_UTF8_CHARCLASS(alnum,"a");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_ALNUM();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (tmp == !(OP(c) == BOUND ?
swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
isALNUM_LC_utf8((U8*)s)))
if ((norun || regtry(prog, s)))
goto got_it;
}
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
if (s == PL_bostr)
tmp = '\n';
else {
- U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
-
+ U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr);
tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
}
tmp = ((OP(c) == NBOUND ?
isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
- LOAD_UTF8_CHARCLASS(alnum,"a");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_ALNUM();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (tmp == !(OP(c) == NBOUND ?
swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
isALNUM_LC_utf8((U8*)s)))
tmp = !tmp;
else if ((norun || regtry(prog, s)))
goto got_it;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case ALNUM:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(alnum,"a");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_ALNUM();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case ALNUML:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (isALNUM_LC_utf8((U8*)s)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case NALNUM:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(alnum,"a");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_ALNUM();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case NALNUML:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!isALNUM_LC_utf8((U8*)s)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case SPACE:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(space," ");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_SPACE();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case SPACEL:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case NSPACE:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(space," ");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_SPACE();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case NSPACEL:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case DIGIT:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(digit,"0");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_DIGIT();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case DIGITL:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (isDIGIT_LC_utf8((U8*)s)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
break;
case NDIGIT:
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(digit,"0");
- while (s < strend) {
+ LOAD_UTF8_CHARCLASS_DIGIT();
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
case NDIGITL:
PL_reg_flags |= RF_tainted;
if (do_utf8) {
- while (s < strend) {
+ while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!isDIGIT_LC_utf8((U8*)s)) {
if (tmp && (norun || regtry(prog, s)))
goto got_it;
}
else
tmp = 1;
- s += UTF8SKIP(s);
+ s += uskip;
}
}
else {
/* data: May be used for some additional optimizations. */
/* nosave: For optimizations. */
{
+ dVAR;
register char *s;
register regnode *c;
register char *startpos = stringarg;
I32 minlen; /* must match at least this many chars */
I32 dontbother = 0; /* how many characters not to try at end */
- /* I32 start_shift = 0; */ /* Offset of the start to find
- constant substr. */ /* CC */
I32 end_shift = 0; /* Same for the end. */ /* CC */
I32 scream_pos = -1; /* Internal iterator of scream. */
- char *scream_olds;
+ char *scream_olds = NULL;
SV* oreplsv = GvSV(PL_replgv);
- bool do_utf8 = DO_UTF8(sv);
+ const bool do_utf8 = DO_UTF8(sv);
+ const I32 multiline = prog->reganch & PMf_MULTILINE;
#ifdef DEBUGGING
- SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
- SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
+ SV * const dsv0 = PERL_DEBUG_PAD_ZERO(0);
+ SV * const dsv1 = PERL_DEBUG_PAD_ZERO(1);
#endif
- PL_regcc = 0;
+ GET_RE_DEBUG_FLAGS_DECL;
+
+ PERL_UNUSED_ARG(data);
+ RX_MATCH_UTF8_set(prog,do_utf8);
cache_re(prog);
#ifdef DEBUGGING
minlen = prog->minlen;
if (strend - startpos < minlen) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"String too short [regexec_flags]...\n"));
goto phooey;
}
PL_reg_ganch = strbeg;
}
- if (!(flags & REXEC_CHECKED) && (prog->check_substr != Nullsv || prog->check_utf8 != Nullsv)) {
+ if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
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) {
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n"));
goto phooey; /* not present */
}
}
- DEBUG_r({
- char *s0 = UTF ?
- pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60,
- UNI_DISPLAY_REGEX) :
- prog->precomp;
- int len0 = UTF ? SvCUR(dsv0) : prog->prelen;
- char *s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60,
+ DEBUG_EXECUTE_r({
+ const char * const s0 = UTF
+ ? pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60,
+ UNI_DISPLAY_REGEX)
+ : prog->precomp;
+ const int len0 = UTF ? SvCUR(dsv0) : prog->prelen;
+ const char * const s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60,
UNI_DISPLAY_REGEX) : startpos;
- int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos;
+ const int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos;
if (!PL_colorset)
reginitcolors();
PerlIO_printf(Perl_debug_log,
- "%sMatching REx%s `%s%*.*s%s%s' against `%s%.*s%s%s'\n",
- PL_colors[4],PL_colors[5],PL_colors[0],
+ "%sMatching REx%s \"%s%*.*s%s%s\" against \"%s%.*s%s%s\"\n",
+ PL_colors[4], PL_colors[5], PL_colors[0],
len0, len0, s0,
PL_colors[1],
len0 > 60 ? "..." : "",
if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
if (s == startpos && regtry(prog, startpos))
goto got_it;
- else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
+ else if (multiline || (prog->reganch & ROPT_IMPLICIT)
|| (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
{
char *end;
#endif
if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr))
do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog);
- ch = SvPVX(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
+ ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0];
if (do_utf8) {
while (s < strend) {
if (*s == ch) {
- DEBUG_r( did_match = 1 );
+ DEBUG_EXECUTE_r( did_match = 1 );
if (regtry(prog, s)) goto got_it;
s += UTF8SKIP(s);
while (s < strend && *s == ch)
else {
while (s < strend) {
if (*s == ch) {
- DEBUG_r( did_match = 1 );
+ DEBUG_EXECUTE_r( did_match = 1 );
if (regtry(prog, s)) goto got_it;
s++;
while (s < strend && *s == ch)
s++;
}
}
- DEBUG_r(if (!did_match)
+ DEBUG_EXECUTE_r(if (!did_match)
PerlIO_printf(Perl_debug_log,
"Did not find anchored character...\n")
);
}
- /*SUPPRESS 560*/
- else if (prog->anchored_substr != Nullsv
- || prog->anchored_utf8 != Nullsv
- || ((prog->float_substr != Nullsv || prog->float_utf8 != Nullsv)
+ else if (prog->anchored_substr != NULL
+ || prog->anchored_utf8 != NULL
+ || ((prog->float_substr != NULL || prog->float_utf8 != NULL)
&& prog->float_max_offset < strend - s)) {
SV *must;
I32 back_max;
else
last1 = s - 1; /* bogus */
- /* XXXX check_substr already used to find `s', can optimize if
+ /* XXXX check_substr already used to find "s", can optimize if
check_substr==must. */
scream_pos = -1;
dontbother = end_shift;
end_shift, &scream_pos, 0))
: (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend),
(unsigned char*)strend, must,
- PL_multiline ? FBMrf_MULTILINE : 0))) ) {
- DEBUG_r( did_match = 1 );
+ multiline ? FBMrf_MULTILINE : 0))) ) {
+ /* we may be pointing at the wrong string */
+ if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
+ s = strbeg + (s - SvPVX_const(sv));
+ DEBUG_EXECUTE_r( did_match = 1 );
if (HOPc(s, -back_max) > last1) {
last1 = HOPc(s, -back_min);
s = HOPc(s, -back_max);
}
else {
- char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
+ char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
last1 = HOPc(s, -back_min);
- s = t;
+ s = t;
}
if (do_utf8) {
while (s <= last1) {
}
}
}
- DEBUG_r(if (!did_match)
+ DEBUG_EXECUTE_r(if (!did_match)
PerlIO_printf(Perl_debug_log,
- "Did not find %s substr `%s%.*s%s'%s...\n",
+ "Did not find %s substr \"%s%.*s%s\"%s...\n",
((must == prog->anchored_substr || must == prog->anchored_utf8)
? "anchored" : "floating"),
PL_colors[0],
(int)(SvCUR(must) - (SvTAIL(must)!=0)),
- SvPVX(must),
+ SvPVX_const(must),
PL_colors[1], (SvTAIL(must) ? "$" : ""))
);
goto phooey;
}
else if ((c = prog->regstclass)) {
- if (minlen && PL_regkind[(U8)OP(prog->regstclass)] != EXACT)
+ if (minlen) {
+ I32 op = (U8)OP(prog->regstclass);
/* don't bother with what can't match */
- strend = HOPc(strend, -(minlen - 1));
- DEBUG_r({
+ if (PL_regkind[op] != EXACT && op != CANY)
+ strend = HOPc(strend, -(minlen - 1));
+ }
+ DEBUG_EXECUTE_r({
SV *prop = sv_newmortal();
- char *s0;
- char *s1;
+ const char *s0;
+ const char *s1;
int len0;
int len1;
regprop(prop, c);
s0 = UTF ?
- pv_uni_display(dsv0, (U8*)SvPVX(prop), SvCUR(prop), 60,
+ pv_uni_display(dsv0, (U8*)SvPVX_const(prop), SvCUR(prop), 60,
UNI_DISPLAY_REGEX) :
- SvPVX(prop);
+ SvPVX_const(prop);
len0 = UTF ? SvCUR(dsv0) : SvCUR(prop);
s1 = UTF ?
sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s;
len1 = UTF ? SvCUR(dsv1) : strend - s;
PerlIO_printf(Perl_debug_log,
- "Matching stclass `%*.*s' against `%*.*s'\n",
+ "Matching stclass \"%*.*s\" against \"%*.*s\"\n",
len0, len0, s0,
len1, len1, s1);
});
- if (find_byclass(prog, c, s, strend, startpos, 0))
+ if (find_byclass(prog, c, s, strend, 0))
goto got_it;
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
}
else {
dontbother = 0;
- if (prog->float_substr != Nullsv || prog->float_utf8 != Nullsv) {
+ if (prog->float_substr != NULL || prog->float_utf8 != NULL) {
/* Trim the end. */
char *last;
SV* float_real;
end_shift, &scream_pos, 1); /* last one */
if (!last)
last = scream_olds; /* Only one occurrence. */
+ /* we may be pointing at the wrong string */
+ else if (RX_MATCH_COPIED(prog))
+ s = strbeg + (s - SvPVX_const(sv));
}
else {
STRLEN len;
- char *little = SvPV(float_real, len);
+ const char * const little = SvPV_const(float_real, len);
if (SvTAIL(float_real)) {
if (memEQ(strend - len + 1, little, len - 1))
last = strend - len + 1;
- else if (!PL_multiline)
+ else if (!multiline)
last = memEQ(strend - len, little, len)
- ? strend - len : Nullch;
+ ? strend - len : NULL;
else
goto find_last;
} else {
if (len)
last = rninstr(s, strend, little, little + len);
else
- last = strend; /* matching `$' */
+ last = strend; /* matching "$" */
}
}
if (last == NULL) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"%sCan't trim the tail, match fails (should not happen)%s\n",
- PL_colors[4],PL_colors[5]));
+ PL_colors[4], PL_colors[5]));
goto phooey; /* Should not happen! */
}
dontbother = strend - last + prog->float_min_offset;
/* 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);
- }
+ RX_MATCH_COPY_FREE(prog);
if (flags & REXEC_COPY_STR) {
I32 i = PL_regeol - startpos + (stringarg - strbeg);
-
- s = savepvn(strbeg, i);
- prog->subbeg = s;
+#ifdef PERL_OLD_COPY_ON_WRITE
+ if ((SvIsCOW(sv)
+ || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) {
+ if (DEBUG_C_TEST) {
+ PerlIO_printf(Perl_debug_log,
+ "Copy on write: regexp capture, type %d\n",
+ (int) SvTYPE(sv));
+ }
+ prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv);
+ prog->subbeg = (char *)SvPVX_const(prog->saved_copy);
+ assert (SvPOKp(prog->saved_copy));
+ } else
+#endif
+ {
+ RX_MATCH_COPIED_on(prog);
+ s = savepvn(strbeg, i);
+ prog->subbeg = s;
+ }
prog->sublen = i;
- RX_MATCH_COPIED_on(prog);
}
else {
prog->subbeg = strbeg;
return 1;
phooey:
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
- PL_colors[4],PL_colors[5]));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
+ PL_colors[4], PL_colors[5]));
if (PL_reg_eval_set)
restore_pos(aTHX_ 0);
return 0;
STATIC I32 /* 0 failure, 1 success */
S_regtry(pTHX_ regexp *prog, char *startpos)
{
+ dVAR;
register I32 i;
register I32 *sp;
register I32 *ep;
CHECKPOINT lastcp;
+ GET_RE_DEBUG_FLAGS_DECL;
#ifdef DEBUGGING
PL_regindent = 0; /* XXXX Not good when matches are reenterable... */
MAGIC *mg;
PL_reg_eval_set = RS_init;
- DEBUG_r(DEBUG_s(
+ DEBUG_EXECUTE_r(DEBUG_s(
PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
(IV)(PL_stack_sp - PL_stack_base));
));
if (PL_reg_sv) {
/* Make $_ available to executed code. */
if (PL_reg_sv != DEFSV) {
- /* SAVE_DEFSV does *not* suffice here for USE_5005THREADS */
- SAVESPTR(DEFSV);
+ SAVE_DEFSV;
DEFSV = PL_reg_sv;
}
if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
&& (mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global)))) {
/* prepare for quick setting of pos */
- sv_magic(PL_reg_sv, (SV*)0,
- PERL_MAGIC_regex_global, Nullch, 0);
- mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global);
+#ifdef PERL_OLD_COPY_ON_WRITE
+ if (SvIsCOW(sv))
+ sv_force_normal_flags(sv, 0);
+#endif
+ mg = sv_magicext(PL_reg_sv, (SV*)0, PERL_MAGIC_regex_global,
+ &PL_vtbl_mglob, NULL, 0);
mg->mg_len = -1;
}
PL_reg_magic = mg;
SAVEDESTRUCTOR_X(restore_pos, 0);
}
if (!PL_reg_curpm) {
- Newz(22,PL_reg_curpm, 1, PMOP);
+ Newxz(PL_reg_curpm, 1, PMOP);
#ifdef USE_ITHREADS
{
SV* repointer = newSViv(0);
$` inside (?{}) could fail... */
PL_reg_oldsaved = prog->subbeg;
PL_reg_oldsavedlen = prog->sublen;
+#ifdef PERL_OLD_COPY_ON_WRITE
+ PL_nrs = prog->saved_copy;
+#endif
RX_MATCH_COPIED_off(prog);
}
else
- PL_reg_oldsaved = Nullch;
+ PL_reg_oldsaved = NULL;
prog->subbeg = PL_bostr;
prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
}
PL_reglastparen = &prog->lastparen;
PL_reglastcloseparen = &prog->lastcloseparen;
prog->lastparen = 0;
+ prog->lastcloseparen = 0;
PL_regsize = 0;
- DEBUG_r(PL_reg_starttry = startpos);
+ DEBUG_EXECUTE_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)
Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
else
- New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*);
+ Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*);
}
/* XXXX What this code is doing here?!!! There should be no need
#define sayNO_SILENT goto do_no
#define saySAME(x) if (x) goto yes; else goto no
-#define REPORT_CODE_OFF 24
+#define POSCACHE_SUCCESS 0 /* caching success rather than failure */
+#define POSCACHE_SEEN 1 /* we know what we're caching */
+#define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */
+#define CACHEsayYES STMT_START { \
+ if (cache_offset | cache_bit) { \
+ if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
+ PL_reg_poscache[0] |= (1<<POSCACHE_SUCCESS) || (1<<POSCACHE_SEEN); \
+ else if (!(PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
+ /* cache records failure, but this is success */ \
+ DEBUG_r( \
+ PerlIO_printf(Perl_debug_log, \
+ "%*s (remove success from failure cache)\n", \
+ REPORT_CODE_OFF+PL_regindent*2, "") \
+ ); \
+ PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
+ } \
+ } \
+ sayYES; \
+} STMT_END
+#define CACHEsayNO STMT_START { \
+ if (cache_offset | cache_bit) { \
+ if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) \
+ PL_reg_poscache[0] |= (1<<POSCACHE_SEEN); \
+ else if ((PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \
+ /* cache records success, but this is failure */ \
+ DEBUG_r( \
+ PerlIO_printf(Perl_debug_log, \
+ "%*s (remove failure from success cache)\n", \
+ REPORT_CODE_OFF+PL_regindent*2, "") \
+ ); \
+ PL_reg_poscache[cache_offset] &= ~(1<<cache_bit); \
+ } \
+ } \
+ sayNO; \
+} STMT_END
+
+/* this is used to determine how far from the left messages like
+ 'failed...' are printed. Currently 29 makes these messages line
+ up with the opcode they refer to. Earlier perls used 25 which
+ left these messages outdented making reviewing a debug output
+ quite difficult.
+*/
+#define REPORT_CODE_OFF 29
+
+
+/* Make sure there is a test for this +1 options in re_tests */
+#define TRIE_INITAL_ACCEPT_BUFFLEN 4;
+
+
+/* simulate a recursive call to regmatch */
+
+#define REGMATCH(ns, where) \
+ new_scan = (ns); \
+ resume_state = resume_##where; \
+ goto start_recurse; \
+ resume_point_##where:
+
+typedef enum {
+ resume_TRIE1,
+ resume_TRIE2,
+ resume_CURLYX,
+ resume_WHILEM1,
+ resume_WHILEM2,
+ resume_WHILEM3,
+ resume_WHILEM4,
+ resume_WHILEM5,
+ resume_WHILEM6,
+ resume_CURLYM1,
+ resume_CURLYM2,
+ resume_CURLYM3,
+ resume_CURLYM4,
+ resume_IFMATCH,
+ resume_PLUS1,
+ resume_PLUS2,
+ resume_PLUS3,
+ resume_PLUS4,
+ resume_END
+} resume_states;
+
+
+struct regmatch_state {
+ struct regmatch_state *prev_state;
+ resume_states resume_state;
+ regnode *scan;
+ regnode *next;
+ bool minmod;
+ bool sw;
+ int logical;
+ I32 unwind;
+ CURCUR *cc;
+ char *locinput;
+
+ I32 n;
+ I32 ln;
+ I32 c1, c2, paren;
+ CHECKPOINT cp, lastcp;
+ CURCUR *oldcc;
+ char *lastloc;
+ I32 cache_offset, cache_bit;
+ I32 curlym_l;
+ I32 matches;
+ I32 maxwanted;
+ char *e;
+ char *old;
+ int count;
+ re_cc_state *cur_call_cc;
+ regexp *end_re;
+ reg_trie_accepted *accept_buff;
+ U32 accepted;
+
+ re_cc_state *reg_call_cc; /* saved value of PL_reg_call_cc */
+};
/*
- regmatch - main matching routine
/* [lwall] I've hoisted the register declarations to the outer block in order to
* maybe save a little bit of pushing and popping on the stack. It also takes
* advantage of machines that use a register save mask on subroutine entry.
+ *
+ * This function used to be heavily recursive, but since this had the
+ * effect of blowing the CPU stack on complex regexes, it has been
+ * restructured to be iterative, and to save state onto the heap rather
+ * than the stack. Essentially whereever regmatch() used to be called, it
+ * pushes the current state, notes where to return, then jumps back into
+ * the main loop.
+ *
+ * Originally the structure of this function used to look something like
+
+ S_regmatch() {
+ int a = 1, b = 2;
+ ...
+ while (scan != NULL) {
+ ...
+ switch (OP(scan)) {
+ case FOO: {
+ int local = 3;
+ ...
+ if (regmatch(...)) // recurse
+ goto yes;
+ }
+ ...
+ }
+ }
+ yes:
+ return 1;
+ }
+
+ * Now it looks something like this:
+
+ struct regmatch_state {
+ int a, b, local;
+ int resume_state;
+ };
+
+ S_regmatch() {
+ int a = 1, b = 2;
+ int local;
+ ...
+ while (scan != NULL) {
+ ...
+ switch (OP(scan)) {
+ case FOO: {
+ local = 3;
+ ...
+ resume_state = resume_FOO;
+ new_scan = ...;
+ goto start_recurse;
+ resume_point_FOO:
+
+ if (result) // recurse
+ goto yes;
+ }
+ ...
+ }
+ start_recurse:
+ ...push a, b, local onto the heap
+ a = 1; b = 2;
+ scan = new_scan;
+ }
+ yes:
+ result = 1;
+ if (states pushed on heap) {
+ ... restore a, b, local from heap
+ switch (resume_state) {
+ case resume_FOO:
+ goto resume_point_FOO;
+ ...
+ }
+ }
+ return result
+ }
+
+ * WARNING: this means that any line in this function that contains a
+ * REGMATCH() or TRYPAREN() is actually simulating a recursive call to
+ * regmatch() using gotos instead. Thus the values of any local variables
+ * not saved in the regmatch_state structure will have been lost when
+ * execution resumes on the next line .
*/
+
+
STATIC I32 /* 0 failure, 1 success */
S_regmatch(pTHX_ regnode *prog)
{
+ dVAR;
+ register const bool do_utf8 = PL_reg_match_utf8;
+ const U32 uniflags = ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY;
+
+ /************************************************************
+ * the following variabes are saved and restored on each fake
+ * recursive call to regmatch:
+ *
+ * The first ones contain state that needs to be maintained
+ * across the main while loop: */
+
+ struct regmatch_state *prev_state = NULL; /* stack of pushed states */
+ resume_states resume_state; /* where to jump to on return */
register regnode *scan; /* Current node. */
regnode *next; /* Next node. */
- regnode *inner; /* Next node in internal branch. */
- register I32 nextchr; /* renamed nextchr - nextchar colides with
- function of same name */
- register I32 n; /* no or next */
- register I32 ln = 0; /* len or last */
- register char *s = Nullch; /* operand or save */
+ bool minmod = 0; /* the next {n.m} is a {n,m}? */
+ bool sw = 0; /* the condition value in (?(cond)a|b) */
+ int logical = 0;
+ I32 unwind = 0; /* savestack index of current unwind block */
+ CURCUR *cc = NULL; /* current innermost curly struct */
register char *locinput = PL_reginput;
- register I32 c1 = 0, c2 = 0, paren; /* case fold search, parenth */
- int minmod = 0, sw = 0, logical = 0;
- I32 unwind = 0;
-#if 0
- I32 firstcp = PL_savestack_ix;
-#endif
- register bool do_utf8 = PL_reg_match_utf8;
-#ifdef DEBUGGING
- SV *dsv0 = PERL_DEBUG_PAD_ZERO(0);
- SV *dsv1 = PERL_DEBUG_PAD_ZERO(1);
- SV *dsv2 = PERL_DEBUG_PAD_ZERO(2);
-#endif
+ /* while the rest of these are local to an individual branch, and
+ * have only been hoisted into this outer scope to allow for saving and
+ * restoration - thus they can be safely reused in other branches.
+ * Note that they are only initialized here to silence compiler
+ * warnings :-( */
+
+ register I32 n = 0; /* no or next */
+ register I32 ln = 0; /* len or last */
+ register I32 c1 = 0, c2 = 0, paren = 0; /* case fold search, parenth */
+ CHECKPOINT cp = 0; /* remember current savestack indexes */
+ CHECKPOINT lastcp = 0;
+ CURCUR *oldcc = NULL; /* tmp copy of cc */
+ char *lastloc = NULL; /* Detection of 0-len. */
+ I32 cache_offset = 0;
+ I32 cache_bit = 0;
+ I32 curlym_l = 0;
+ I32 matches = 0;
+ I32 maxwanted = 0;
+ char *e = NULL;
+ char *old = NULL;
+ int count = 0;
+ re_cc_state *cur_call_cc = NULL;
+ regexp *end_re = NULL;
+ reg_trie_accepted *accept_buff = NULL;
+ U32 accepted = 0; /* how many accepting states we have seen */
+
+ /************************************************************
+ * these variables are NOT saved: */
+
+ register I32 nextchr; /* is always set to UCHARAT(locinput) */
+ regnode *new_scan; /* node to begin exedcution at when recursing */
+ bool result; /* return value of S_regmatch */
+
+ regnode *inner; /* Next node in internal branch. */
+
#ifdef DEBUGGING
+ SV *re_debug_flags = NULL;
+ GET_RE_DEBUG_FLAGS;
PL_regindent++;
#endif
scan = prog;
while (scan != NULL) {
- DEBUG_r( {
- SV *prop = sv_newmortal();
- int docolor = *PL_colors[0];
- int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
+ DEBUG_EXECUTE_r( {
+ SV * const prop = sv_newmortal();
+ const int docolor = *PL_colors[0];
+ const 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
pref0_len = pref_len;
regprop(prop, scan);
{
- char *s0 =
- do_utf8 ?
- pv_uni_display(dsv0, (U8*)(locinput - pref_len),
+ const char * const s0 =
+ do_utf8 && OP(scan) != CANY ?
+ pv_uni_display(PERL_DEBUG_PAD(0), (U8*)(locinput - pref_len),
pref0_len, 60, UNI_DISPLAY_REGEX) :
locinput - pref_len;
- int len0 = do_utf8 ? strlen(s0) : pref0_len;
- char *s1 = do_utf8 ?
- pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len),
+ const int len0 = do_utf8 ? strlen(s0) : pref0_len;
+ const char * const s1 = do_utf8 && OP(scan) != CANY ?
+ pv_uni_display(PERL_DEBUG_PAD(1),
+ (U8*)(locinput - pref_len + pref0_len),
pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) :
locinput - pref_len + pref0_len;
- int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len;
- char *s2 = do_utf8 ?
- pv_uni_display(dsv2, (U8*)locinput,
+ const int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len;
+ const char * const s2 = do_utf8 && OP(scan) != CANY ?
+ pv_uni_display(PERL_DEBUG_PAD(2), (U8*)locinput,
PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) :
locinput;
- int len2 = do_utf8 ? strlen(s2) : l;
+ const int len2 = do_utf8 ? strlen(s2) : l;
PerlIO_printf(Perl_debug_log,
"%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n",
(IV)(locinput - PL_bostr),
15 - l - pref_len + 1,
"",
(IV)(scan - PL_regprogram), PL_regindent*2, "",
- SvPVX(prop));
+ SvPVX_const(prop));
}
});
switch (OP(scan)) {
case BOL:
- if (locinput == PL_bostr || (PL_multiline &&
- (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
+ if (locinput == PL_bostr)
{
/* regtill = regbol; */
break;
break;
sayNO;
case EOL:
- if (PL_multiline)
- goto meol;
- else
goto seol;
case MEOL:
- meol:
if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
sayNO;
break;
else
nextchr = UCHARAT(++locinput);
break;
- case EXACT:
- s = STRING(scan);
+
+
+
+ /*
+ traverse the TRIE keeping track of all accepting states
+ we transition through until we get to a failing node.
+
+
+ */
+ case TRIE:
+ case TRIEF:
+ case TRIEFL:
+ {
+ U8 *uc = ( U8* )locinput;
+ U32 state = 1;
+ U16 charid = 0;
+ U32 base = 0;
+ UV uvc = 0;
+ STRLEN len = 0;
+ STRLEN foldlen = 0;
+ U8 *uscan = (U8*)NULL;
+ STRLEN bufflen=0;
+ SV *sv_accept_buff = NULL;
+ const enum { trie_plain, trie_utf8, trie_uft8_fold }
+ trie_type = do_utf8 ?
+ (OP(scan) == TRIE ? trie_utf8 : trie_uft8_fold)
+ : trie_plain;
+
+ /* what trie are we using right now */
+ reg_trie_data *trie
+ = (reg_trie_data*)PL_regdata->data[ ARG( scan ) ];
+ accepted = 0; /* how many accepting states we have seen */
+ result = 0;
+
+ while ( state && uc <= (U8*)PL_regeol ) {
+
+ if (trie->states[ state ].wordnum) {
+ if (!accepted ) {
+ ENTER;
+ SAVETMPS;
+ bufflen = TRIE_INITAL_ACCEPT_BUFFLEN;
+ sv_accept_buff=newSV(bufflen *
+ sizeof(reg_trie_accepted) - 1);
+ SvCUR_set(sv_accept_buff,
+ sizeof(reg_trie_accepted));
+ SvPOK_on(sv_accept_buff);
+ sv_2mortal(sv_accept_buff);
+ accept_buff =
+ (reg_trie_accepted*)SvPV_nolen(sv_accept_buff );
+ }
+ else {
+ if (accepted >= bufflen) {
+ bufflen *= 2;
+ accept_buff =(reg_trie_accepted*)
+ SvGROW(sv_accept_buff,
+ bufflen * sizeof(reg_trie_accepted));
+ }
+ SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff)
+ + sizeof(reg_trie_accepted));
+ }
+ accept_buff[accepted].wordnum = trie->states[state].wordnum;
+ accept_buff[accepted].endpos = uc;
+ ++accepted;
+ }
+
+ base = trie->states[ state ].trans.base;
+
+ DEBUG_TRIE_EXECUTE_r(
+ PerlIO_printf( Perl_debug_log,
+ "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ",
+ REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
+ (UV)state, (UV)base, (UV)accepted );
+ );
+
+ if ( base ) {
+ switch (trie_type) {
+ case trie_uft8_fold:
+ if ( foldlen>0 ) {
+ uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags );
+ foldlen -= len;
+ uscan += len;
+ len=0;
+ } else {
+ U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ];
+ uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags );
+ uvc = to_uni_fold( uvc, foldbuf, &foldlen );
+ foldlen -= UNISKIP( uvc );
+ uscan = foldbuf + UNISKIP( uvc );
+ }
+ break;
+ case trie_utf8:
+ uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN,
+ &len, uniflags );
+ break;
+ case trie_plain:
+ uvc = (UV)*uc;
+ len = 1;
+ }
+
+ if (uvc < 256) {
+ charid = trie->charmap[ uvc ];
+ }
+ else {
+ charid = 0;
+ if (trie->widecharmap) {
+ SV** svpp = (SV**)NULL;
+ svpp = hv_fetch(trie->widecharmap,
+ (char*)&uvc, sizeof(UV), 0);
+ if (svpp)
+ charid = (U16)SvIV(*svpp);
+ }
+ }
+
+ if (charid &&
+ (base + charid > trie->uniquecharcount )
+ && (base + charid - 1 - trie->uniquecharcount
+ < trie->lasttrans)
+ && trie->trans[base + charid - 1 -
+ trie->uniquecharcount].check == state)
+ {
+ state = trie->trans[base + charid - 1 -
+ trie->uniquecharcount ].next;
+ }
+ else {
+ state = 0;
+ }
+ uc += len;
+
+ }
+ else {
+ state = 0;
+ }
+ DEBUG_TRIE_EXECUTE_r(
+ PerlIO_printf( Perl_debug_log,
+ "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n",
+ charid, uvc, (UV)state, PL_colors[5] );
+ );
+ }
+ if (!accepted )
+ sayNO;
+
+ /*
+ There was at least one accepting state that we
+ transitioned through. Presumably the number of accepting
+ states is going to be low, typically one or two. So we
+ simply scan through to find the one with lowest wordnum.
+ Once we find it, we swap the last state into its place
+ and decrement the size. We then try to match the rest of
+ the pattern at the point where the word ends, if we
+ succeed then we end the loop, otherwise the loop
+ eventually terminates once all of the accepting states
+ have been tried.
+ */
+
+ if ( accepted == 1 ) {
+ DEBUG_EXECUTE_r({
+ SV **tmp = av_fetch( trie->words, accept_buff[ 0 ].wordnum-1, 0 );
+ PerlIO_printf( Perl_debug_log,
+ "%*s %sonly one match : #%d <%s>%s\n",
+ REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
+ accept_buff[ 0 ].wordnum,
+ tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr",
+ PL_colors[5] );
+ });
+ PL_reginput = (char *)accept_buff[ 0 ].endpos;
+ /* in this case we free tmps/leave before we call regmatch
+ as we wont be using accept_buff again. */
+ FREETMPS;
+ LEAVE;
+ REGMATCH(scan + NEXT_OFF(scan), TRIE1);
+ /*** all unsaved local vars undefined at this point */
+ } else {
+ DEBUG_EXECUTE_r(
+ PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n",
+ REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)accepted,
+ PL_colors[5] );
+ );
+ while ( !result && accepted-- ) {
+ U32 best = 0;
+ U32 cur;
+ for( cur = 1 ; cur <= accepted ; cur++ ) {
+ DEBUG_TRIE_EXECUTE_r(
+ PerlIO_printf( Perl_debug_log,
+ "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n",
+ REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4],
+ (IV)best, accept_buff[ best ].wordnum, (IV)cur,
+ accept_buff[ cur ].wordnum, PL_colors[5] );
+ );
+
+ if ( accept_buff[ cur ].wordnum < accept_buff[ best ].wordnum )
+ best = cur;
+ }
+ DEBUG_EXECUTE_r({
+ SV ** const tmp = av_fetch( trie->words, accept_buff[ best ].wordnum - 1, 0 );
+ PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n",
+ REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],
+ accept_buff[best].wordnum,
+ tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr",scan,
+ PL_colors[5] );
+ });
+ if ( best<accepted ) {
+ reg_trie_accepted tmp = accept_buff[ best ];
+ accept_buff[ best ] = accept_buff[ accepted ];
+ accept_buff[ accepted ] = tmp;
+ best = accepted;
+ }
+ PL_reginput = (char *)accept_buff[ best ].endpos;
+
+ /*
+ as far as I can tell we only need the SAVETMPS/FREETMPS
+ for re's with EVAL in them but I'm leaving them in for
+ all until I can be sure.
+ */
+ SAVETMPS;
+ REGMATCH(scan + NEXT_OFF(scan), TRIE2);
+ /*** all unsaved local vars undefined at this point */
+ FREETMPS;
+ }
+ FREETMPS;
+ LEAVE;
+ }
+
+ if (result) {
+ sayYES;
+ } else {
+ sayNO;
+ }
+ }
+ /* unreached codepoint */
+ case EXACT: {
+ char *s = STRING(scan);
ln = STR_LEN(scan);
if (do_utf8 != UTF) {
/* The target and the pattern have differing utf8ness. */
char *l = locinput;
- char *e = s + ln;
- STRLEN ulen;
+ const char *e = s + ln;
if (do_utf8) {
/* The target is utf8, the pattern is not utf8. */
while (s < e) {
+ STRLEN ulen;
if (l >= PL_regeol)
sayNO;
if (NATIVE_TO_UNI(*(U8*)s) !=
- utf8n_to_uvuni((U8*)l, UTF8_MAXLEN, &ulen,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY))
+ utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen,
+ uniflags))
sayNO;
l += ulen;
s ++;
else {
/* The target is not utf8, the pattern is utf8. */
while (s < e) {
+ STRLEN ulen;
if (l >= PL_regeol)
sayNO;
if (NATIVE_TO_UNI(*((U8*)l)) !=
- utf8n_to_uvuni((U8*)s, UTF8_MAXLEN, &ulen,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY))
+ utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen,
+ uniflags))
sayNO;
s += ulen;
l ++;
locinput += ln;
nextchr = UCHARAT(locinput);
break;
+ }
case EXACTFL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case EXACTF:
- s = STRING(scan);
+ case EXACTF: {
+ char *s = STRING(scan);
ln = STR_LEN(scan);
if (do_utf8 || UTF) {
locinput += ln;
nextchr = UCHARAT(locinput);
break;
+ }
case ANYOF:
if (do_utf8) {
STRLEN inclasslen = PL_regeol - locinput;
if (!nextchr)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(alnum,"a");
+ LOAD_UTF8_CHARCLASS_ALNUM();
if (!(OP(scan) == ALNUM
? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
: isALNUM_LC_utf8((U8*)locinput)))
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(alnum,"a");
+ LOAD_UTF8_CHARCLASS_ALNUM();
if (OP(scan) == NALNUM
? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
: isALNUM_LC_utf8((U8*)locinput))
if (locinput == PL_bostr)
ln = '\n';
else {
- U8 *r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
+ const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, 0);
}
if (OP(scan) == BOUND || OP(scan) == NBOUND) {
ln = isALNUM_uni(ln);
- LOAD_UTF8_CHARCLASS(alnum,"a");
+ LOAD_UTF8_CHARCLASS_ALNUM();
n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
}
else {
sayNO;
if (do_utf8) {
if (UTF8_IS_CONTINUED(nextchr)) {
- LOAD_UTF8_CHARCLASS(space," ");
+ LOAD_UTF8_CHARCLASS_SPACE();
if (!(OP(scan) == SPACE
? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
: isSPACE_LC_utf8((U8*)locinput)))
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(space," ");
+ LOAD_UTF8_CHARCLASS_SPACE();
if (OP(scan) == NSPACE
? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
: isSPACE_LC_utf8((U8*)locinput))
if (!nextchr)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(digit,"0");
+ LOAD_UTF8_CHARCLASS_DIGIT();
if (!(OP(scan) == DIGIT
? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
: isDIGIT_LC_utf8((U8*)locinput)))
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(digit,"0");
+ LOAD_UTF8_CHARCLASS_DIGIT();
if (OP(scan) == NDIGIT
? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
: isDIGIT_LC_utf8((U8*)locinput))
if (locinput >= PL_regeol)
sayNO;
if (do_utf8) {
- LOAD_UTF8_CHARCLASS(mark,"~");
+ LOAD_UTF8_CHARCLASS_MARK();
if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
sayNO;
locinput += PL_utf8skip[nextchr];
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case REF:
- case REFF:
+ case REFF: {
+ char *s;
n = ARG(scan); /* which paren pair */
ln = PL_regstartp[n];
PL_reg_leftiter = PL_reg_maxiter; /* Void cache */
s = PL_bostr + ln;
if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */
char *l = locinput;
- char *e = PL_bostr + PL_regendp[n];
+ const 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
* have to map both upper and title case to lower case.
*/
if (OP(scan) == REFF) {
- STRLEN ulen1, ulen2;
- U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
- U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
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);
locinput += ln;
nextchr = UCHARAT(locinput);
break;
+ }
case NOTHING:
case TAIL:
case EVAL:
{
dSP;
- OP_4tree *oop = PL_op;
- COP *ocurcop = PL_curcop;
- SV **ocurpad = PL_curpad;
+ OP_4tree * const oop = PL_op;
+ COP * const ocurcop = PL_curcop;
+ PAD *old_comppad;
SV *ret;
+ struct regexp * const oreg = PL_reg_re;
n = ARG(scan);
PL_op = (OP_4tree*)PL_regdata->data[n];
- DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
- PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]);
+ DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
+ PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_regdata->data[n + 2]);
PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
{
- SV **before = SP;
+ SV ** const before = SP;
CALLRUNOPS(aTHX); /* Scalar context. */
SPAGAIN;
if (SP == before)
- ret = Nullsv; /* protect against empty (?{}) blocks. */
+ ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */
else {
ret = POPs;
PUTBACK;
}
PL_op = oop;
- PL_curpad = ocurpad;
+ PAD_RESTORE_LOCAL(old_comppad);
PL_curcop = ocurcop;
if (logical) {
if (logical == 2) { /* Postponed subexpression. */
regexp *re;
- MAGIC *mg = Null(MAGIC*);
+ MAGIC *mg = NULL;
re_cc_state state;
- CHECKPOINT cp, lastcp;
-
- if(SvROK(ret) || SvRMAGICAL(ret)) {
- SV *sv = SvROK(ret) ? SvRV(ret) : ret;
-
- if(SvMAGICAL(sv))
- mg = mg_find(sv, PERL_MAGIC_qr);
+ int toggleutf;
+ register SV *sv;
+
+ if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret)))
+ mg = mg_find(sv, PERL_MAGIC_qr);
+ else if (SvSMAGICAL(ret)) {
+ if (SvGMAGICAL(ret))
+ sv_unmagic(ret, PERL_MAGIC_qr);
+ else
+ mg = mg_find(ret, PERL_MAGIC_qr);
}
+
if (mg) {
re = (regexp *)mg->mg_obj;
(void)ReREFCNT_inc(re);
}
else {
STRLEN len;
- char *t = SvPV(ret, len);
+ const char * const t = SvPV_const(ret, len);
PMOP pm;
- char *oprecomp = PL_regprecomp;
- I32 osize = PL_regsize;
- I32 onpar = PL_regnpar;
+ char * const oprecomp = PL_regprecomp;
+ const I32 osize = PL_regsize;
+ const I32 onpar = PL_regnpar;
Zero(&pm, 1, PMOP);
- re = CALLREGCOMP(aTHX_ t, t + len, &pm);
+ if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8;
+ re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm);
if (!(SvFLAGS(ret)
- & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
+ & (SVs_TEMP | SVs_PADTMP | SVf_READONLY
+ | SVs_GMG)))
sv_magic(ret,(SV*)ReREFCNT_inc(re),
PERL_MAGIC_qr,0,0);
PL_regprecomp = oprecomp;
PL_regsize = osize;
PL_regnpar = onpar;
}
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
- "Entering embedded `%s%.60s%s%s'\n",
+ "Entering embedded \"%s%.60s%s%s\"\n",
PL_colors[0],
re->precomp,
PL_colors[1],
);
state.node = next;
state.prev = PL_reg_call_cc;
- state.cc = PL_regcc;
+ state.cc = cc;
state.re = PL_reg_re;
- PL_regcc = 0;
+ cc = 0;
cp = regcppush(0); /* Save *all* the positions. */
REGCP_SET(lastcp);
*PL_reglastcloseparen = 0;
PL_reg_call_cc = &state;
PL_reginput = locinput;
+ toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^
+ ((re->reganch & ROPT_UTF8) != 0);
+ if (toggleutf) PL_reg_flags ^= RF_utf8;
/* XXXX This is too dramatic a measure... */
PL_reg_maxiter = 0;
+ /* XXX the only recursion left in regmatch() */
if (regmatch(re->program + 1)) {
/* Even though we succeeded, we need to restore
global variables, since we may be wrapped inside
/* XXXX Do this only if SUSPENDed? */
PL_reg_call_cc = state.prev;
- PL_regcc = state.cc;
+ cc = state.cc;
PL_reg_re = state.re;
cache_re(PL_reg_re);
+ if (toggleutf) PL_reg_flags ^= RF_utf8;
/* XXXX This is too dramatic a measure... */
PL_reg_maxiter = 0;
REGCP_UNWIND(lastcp);
regcppop();
PL_reg_call_cc = state.prev;
- PL_regcc = state.cc;
+ cc = state.cc;
PL_reg_re = state.re;
cache_re(PL_reg_re);
+ if (toggleutf) PL_reg_flags ^= RF_utf8;
/* XXXX This is too dramatic a measure... */
PL_reg_maxiter = 0;
sw = SvTRUE(ret);
logical = 0;
}
- else
+ else {
sv_setsv(save_scalar(PL_replgv), ret);
+ cache_re(oreg);
+ }
break;
}
case OPEN:
logical = scan->flags;
break;
/*******************************************************************
- PL_regcc contains infoblock about the innermost (...)* loop, and
+ cc contains infoblock about the innermost (...)* loop, and
a pointer to the next outer infoblock.
Here is how Y(A)*Z is processed (if it is compiled into CURLYX/WHILEM):
- 1) After matching X, regnode for CURLYX is processed;
+ 1) After matching Y, regnode for CURLYX is processed;
- 2) This regnode creates infoblock on the stack, and calls
- regmatch() recursively with the starting point at WHILEM node;
+ 2) This regnode mallocs an infoblock, and calls regmatch() recursively
+ with the starting point at WHILEM node;
3) Each hit of WHILEM node tries to match A and Z (in the order
depending on the current iteration, min/max of {min,max} and
4) After A matches, the same WHILEM node is hit again.
- 5) Each time WHILEM is hit, PL_regcc is the infoblock created by CURLYX
+ 5) Each time WHILEM is hit, cc is the infoblock created by CURLYX
of the same pair. Thus when WHILEM tries to match Z, it temporarily
- resets PL_regcc, since this Y(A)*Z can be a part of some other loop:
+ resets cc, since this Y(A)*Z can be a part of some other loop:
as in (Y(A)*Z)*. If Z matches, the automaton will hit the WHILEM node
of the external loop.
and whatever it mentions via ->next, and additional attached trees
corresponding to temporarily unset infoblocks as in "5" above.
- In the following picture infoblocks for outer loop of
+ In the following picture, infoblocks for outer loop of
(Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block
is denoted by x. The matched string is YAAZYAZT. Temporarily postponed
infoblocks are drawn below the "reset" infoblock.
O
I,I
*******************************************************************/
+
case CURLYX: {
- CURCUR cc;
- CHECKPOINT cp = PL_savestack_ix;
/* No need to save/restore up to this paren */
I32 parenfloor = scan->flags;
+ {
+ CURCUR *newcc;
+ Newx(newcc, 1, CURCUR);
+ oldcc = cc;
+ newcc->oldcc = cc;
+ cc = newcc;
+ }
+ cp = PL_savestack_ix;
if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
next += ARG(next);
- cc.oldcc = PL_regcc;
- PL_regcc = &cc;
/* XXXX Probably it is better to teach regpush to support
parenfloor > PL_regsize... */
if (parenfloor > (I32)*PL_reglastparen)
parenfloor = *PL_reglastparen; /* Pessimization... */
- cc.parenfloor = parenfloor;
- cc.cur = -1;
- cc.min = ARG1(scan);
- cc.max = ARG2(scan);
- cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
- cc.next = next;
- cc.minmod = minmod;
- cc.lastloc = 0;
+ cc->parenfloor = parenfloor;
+ cc->cur = -1;
+ cc->min = ARG1(scan);
+ cc->max = ARG2(scan);
+ cc->scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS;
+ cc->next = next;
+ cc->minmod = minmod;
+ cc->lastloc = 0;
PL_reginput = locinput;
- n = regmatch(PREVOPER(next)); /* start on the WHILEM */
+ REGMATCH(PREVOPER(next), CURLYX); /* start on the WHILEM */
+ /*** all unsaved local vars undefined at this point */
regcpblow(cp);
- PL_regcc = cc.oldcc;
- saySAME(n);
+ Safefree(cc);
+ cc = oldcc;
+ saySAME(result);
}
- /* NOT REACHED */
+ /* NOTREACHED */
case WHILEM: {
/*
* This is really hard to understand, because after we match
* that we can try again after backing off.
*/
- CHECKPOINT cp, lastcp;
- CURCUR* cc = PL_regcc;
- char *lastloc = cc->lastloc; /* Detection of 0-len. */
+ lastloc = cc->lastloc; /* Detection of 0-len. */
+ cache_offset = 0;
+ cache_bit = 0;
n = cc->cur + 1; /* how many we know we matched */
PL_reginput = locinput;
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s %ld out of %ld..%ld cc=%"UVxf"\n",
REPORT_CODE_OFF+PL_regindent*2, "",
/* If degenerate scan matches "", assume scan done. */
if (locinput == cc->lastloc && n >= cc->min) {
- PL_regcc = cc->oldcc;
- if (PL_regcc)
- ln = PL_regcc->cur;
- DEBUG_r(
+ oldcc = cc;
+ cc = cc->oldcc;
+ if (cc)
+ ln = cc->cur;
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s empty match detected, try continuation...\n",
REPORT_CODE_OFF+PL_regindent*2, "")
);
- if (regmatch(cc->next))
+ REGMATCH(oldcc->next, WHILEM1);
+ /*** all unsaved local vars undefined at this point */
+ cc = oldcc;
+ if (result)
sayYES;
- if (PL_regcc)
- PL_regcc->cur = ln;
- PL_regcc = cc;
+ if (cc->oldcc)
+ cc->oldcc->cur = ln;
sayNO;
}
if (n < cc->min) {
cc->cur = n;
cc->lastloc = locinput;
- if (regmatch(cc->scan))
+ REGMATCH(cc->scan, WHILEM2);
+ /*** all unsaved local vars undefined at this point */
+ if (result)
sayYES;
cc->cur = n - 1;
cc->lastloc = lastloc;
PL_reg_leftiter = PL_reg_maxiter;
}
if (PL_reg_leftiter-- == 0) {
- I32 size = (PL_reg_maxiter + 7)/8;
+ const I32 size = (PL_reg_maxiter + 7 + POSCACHE_START)/8;
if (PL_reg_poscache) {
if ((I32)PL_reg_poscache_size < size) {
Renew(PL_reg_poscache, size, char);
}
else {
PL_reg_poscache_size = size;
- Newz(29, PL_reg_poscache, size, char);
+ Newxz(PL_reg_poscache, size, char);
}
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%sDetected a super-linear match, switching on caching%s...\n",
PL_colors[4], PL_colors[5])
);
}
if (PL_reg_leftiter < 0) {
- I32 o = locinput - PL_bostr, b;
-
- o = (scan->flags & 0xf) - 1 + o * (scan->flags>>4);
- b = o % 8;
- o /= 8;
- if (PL_reg_poscache[o] & (1<<b)) {
- DEBUG_r(
+ cache_offset = locinput - PL_bostr;
+
+ cache_offset = (scan->flags & 0xf) - 1 + POSCACHE_START
+ + cache_offset * (scan->flags>>4);
+ cache_bit = cache_offset % 8;
+ cache_offset /= 8;
+ if (PL_reg_poscache[cache_offset] & (1<<cache_bit)) {
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s already tried at this position...\n",
REPORT_CODE_OFF+PL_regindent*2, "")
);
- sayNO_SILENT;
+ if (PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))
+ /* cache records success */
+ sayYES;
+ else
+ /* cache records failure */
+ sayNO_SILENT;
}
- PL_reg_poscache[o] |= (1<<b);
+ PL_reg_poscache[cache_offset] |= (1<<cache_bit);
}
}
/* Prefer next over scan for minimal matching. */
if (cc->minmod) {
- PL_regcc = cc->oldcc;
- if (PL_regcc)
- ln = PL_regcc->cur;
- cp = regcppush(cc->parenfloor);
+ oldcc = cc;
+ cc = cc->oldcc;
+ if (cc)
+ ln = cc->cur;
+ cp = regcppush(oldcc->parenfloor);
REGCP_SET(lastcp);
- if (regmatch(cc->next)) {
+ REGMATCH(oldcc->next, WHILEM3);
+ /*** all unsaved local vars undefined at this point */
+ cc = oldcc;
+ if (result) {
regcpblow(cp);
- sayYES; /* All done. */
+ CACHEsayYES; /* All done. */
}
REGCP_UNWIND(lastcp);
regcppop();
- if (PL_regcc)
- PL_regcc->cur = ln;
- PL_regcc = cc;
+ if (cc->oldcc)
+ cc->oldcc->cur = ln;
if (n >= cc->max) { /* Maximum greed exceeded? */
if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
- sayNO;
+ CACHEsayNO;
}
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s trying longer...\n",
REPORT_CODE_OFF+PL_regindent*2, "")
cc->lastloc = locinput;
cp = regcppush(cc->parenfloor);
REGCP_SET(lastcp);
- if (regmatch(cc->scan)) {
+ REGMATCH(cc->scan, WHILEM4);
+ /*** all unsaved local vars undefined at this point */
+ if (result) {
regcpblow(cp);
- sayYES;
+ CACHEsayYES;
}
REGCP_UNWIND(lastcp);
regcppop();
cc->cur = n - 1;
cc->lastloc = lastloc;
- sayNO;
+ CACHEsayNO;
}
/* Prefer scan over next for maximal matching. */
cc->cur = n;
cc->lastloc = locinput;
REGCP_SET(lastcp);
- if (regmatch(cc->scan)) {
+ REGMATCH(cc->scan, WHILEM5);
+ /*** all unsaved local vars undefined at this point */
+ if (result) {
regcpblow(cp);
- sayYES;
+ CACHEsayYES;
}
REGCP_UNWIND(lastcp);
regcppop(); /* Restore some previous $<digit>s? */
PL_reginput = locinput;
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s failed, try continuation...\n",
REPORT_CODE_OFF+PL_regindent*2, "")
}
/* Failed deeper matches of scan, so see if this one works. */
- PL_regcc = cc->oldcc;
- if (PL_regcc)
- ln = PL_regcc->cur;
- if (regmatch(cc->next))
- sayYES;
- if (PL_regcc)
- PL_regcc->cur = ln;
- PL_regcc = cc;
+ oldcc = cc;
+ cc = cc->oldcc;
+ if (cc)
+ ln = cc->cur;
+ REGMATCH(oldcc->next, WHILEM6);
+ /*** all unsaved local vars undefined at this point */
+ cc = oldcc;
+ if (result)
+ CACHEsayYES;
+ if (cc->oldcc)
+ cc->oldcc->cur = ln;
cc->cur = n - 1;
cc->lastloc = lastloc;
- sayNO;
+ CACHEsayNO;
}
- /* NOT REACHED */
+ /* NOTREACHED */
case BRANCHJ:
next = scan + ARG(scan);
if (next == scan)
if (OP(next) != c1) /* No choice. */
next = inner; /* Avoid recursion. */
else {
- I32 lastparen = *PL_reglastparen;
- I32 unwind1;
- re_unwind_branch_t *uw;
-
+ const I32 lastparen = *PL_reglastparen;
/* Put unwinding data on stack */
- unwind1 = SSNEWt(1,re_unwind_branch_t);
- uw = SSPTRt(unwind1,re_unwind_branch_t);
+ const I32 unwind1 = SSNEWt(1,re_unwind_branch_t);
+ re_unwind_branch_t * const uw = SSPTRt(unwind1,re_unwind_branch_t);
+
uw->prev = unwind;
unwind = unwind1;
uw->type = ((c1 == BRANCH)
break;
case CURLYM:
{
- I32 l = 0;
- CHECKPOINT lastcp;
+ curlym_l = matches = 0;
/* We suppose that the next guy does not need
- backtracking: in particular, it is of constant length,
+ backtracking: in particular, it is of constant non-zero length,
and has no parenths to influence future backrefs. */
ln = ARG1(scan); /* min to match */
n = ARG2(scan); /* max to match */
if (paren)
scan += NEXT_OFF(scan); /* Skip former OPEN. */
PL_reginput = locinput;
+ maxwanted = minmod ? ln : n;
+ if (maxwanted) {
+ while (PL_reginput < PL_regeol && matches < maxwanted) {
+ REGMATCH(scan, CURLYM1);
+ /*** all unsaved local vars undefined at this point */
+ if (!result)
+ break;
+ /* on first match, determine length, curlym_l */
+ if (!matches++) {
+ if (PL_reg_match_utf8) {
+ char *s = locinput;
+ while (s < PL_reginput) {
+ curlym_l++;
+ s += UTF8SKIP(s);
+ }
+ }
+ else {
+ curlym_l = PL_reginput - locinput;
+ }
+ if (curlym_l == 0) {
+ matches = maxwanted;
+ break;
+ }
+ }
+ locinput = PL_reginput;
+ }
+ }
+
+ PL_reginput = locinput;
+
if (minmod) {
minmod = 0;
- if (ln && regrepeat_hard(scan, ln, &l) < ln)
+ if (ln && matches < ln)
sayNO;
- /* if we matched something zero-length we don't need to
- backtrack - capturing parens are already defined, so
- the caveat in the maximal case doesn't apply
-
- XXXX if ln == 0, we can redo this check first time
- through the following loop
- */
- if (ln && l == 0)
- n = ln; /* don't backtrack */
- locinput = PL_reginput;
if (HAS_TEXT(next) || JUMPABLE(next)) {
regnode *text_node = next;
if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
else {
if (PL_regkind[(U8)OP(text_node)] == REF) {
- I32 n, ln;
- n = ARG(text_node); /* which paren pair */
- ln = PL_regstartp[n];
- /* assume yes if we haven't seen CLOSEn */
- if (
- (I32)*PL_reglastparen < n ||
- ln == -1 ||
- ln == PL_regendp[n]
- ) {
- c1 = c2 = -1000;
- goto assume_ok_MM;
- }
- c1 = *(PL_bostr + ln);
+ c1 = c2 = -1000;
+ goto assume_ok_MM;
}
else { c1 = (U8)*STRING(text_node); }
if (OP(text_node) == EXACTF || OP(text_node) == REFF)
c1 = c2 = -1000;
assume_ok_MM:
REGCP_SET(lastcp);
- /* This may be improved if l == 0. */
- while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */
+ while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
/* If it could work, try it. */
if (c1 == -1000 ||
UCHARAT(PL_reginput) == c1 ||
if (paren) {
if (ln) {
PL_regstartp[paren] =
- HOPc(PL_reginput, -l) - PL_bostr;
+ HOPc(PL_reginput, -curlym_l) - PL_bostr;
PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
PL_regendp[paren] = -1;
}
- if (regmatch(next))
+ REGMATCH(next, CURLYM2);
+ /*** all unsaved local vars undefined at this point */
+ if (result)
sayYES;
REGCP_UNWIND(lastcp);
}
/* Couldn't or didn't -- move forward. */
PL_reginput = locinput;
- if (regrepeat_hard(scan, 1, &l)) {
+ REGMATCH(scan, CURLYM3);
+ /*** all unsaved local vars undefined at this point */
+ if (result) {
ln++;
locinput = PL_reginput;
}
}
}
else {
- n = regrepeat_hard(scan, n, &l);
- /* if we matched something zero-length we don't need to
- backtrack, unless the minimum count is zero and we
- are capturing the result - in that case the capture
- being defined or not may affect later execution
- */
- if (n != 0 && l == 0 && !(paren && ln == 0))
- ln = n; /* don't backtrack */
- locinput = PL_reginput;
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
- "%*s matched %"IVdf" times, len=%"IVdf"...\n",
- (int)(REPORT_CODE_OFF+PL_regindent*2), "",
- (IV) n, (IV)l)
+ "%*s matched %"IVdf" times, len=%"IVdf"...\n",
+ (int)(REPORT_CODE_OFF+PL_regindent*2), "",
+ (IV) matches, (IV)curlym_l)
);
- if (n >= ln) {
+ if (matches >= ln) {
if (HAS_TEXT(next) || JUMPABLE(next)) {
regnode *text_node = next;
if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
else {
if (PL_regkind[(U8)OP(text_node)] == REF) {
- I32 n, ln;
- n = ARG(text_node); /* which paren pair */
- ln = PL_regstartp[n];
- /* assume yes if we haven't seen CLOSEn */
- if (
- (I32)*PL_reglastparen < n ||
- ln == -1 ||
- ln == PL_regendp[n]
- ) {
- c1 = c2 = -1000;
- goto assume_ok_REG;
- }
- c1 = *(PL_bostr + ln);
+ c1 = c2 = -1000;
+ goto assume_ok_REG;
}
else { c1 = (U8)*STRING(text_node); }
}
assume_ok_REG:
REGCP_SET(lastcp);
- while (n >= ln) {
+ while (matches >= ln) {
/* If it could work, try it. */
if (c1 == -1000 ||
UCHARAT(PL_reginput) == c1 ||
UCHARAT(PL_reginput) == c2)
{
- DEBUG_r(
- PerlIO_printf(Perl_debug_log,
- "%*s trying tail with n=%"IVdf"...\n",
- (int)(REPORT_CODE_OFF+PL_regindent*2), "", (IV)n)
+ DEBUG_EXECUTE_r(
+ PerlIO_printf(Perl_debug_log,
+ "%*s trying tail with matches=%"IVdf"...\n",
+ (int)(REPORT_CODE_OFF+PL_regindent*2),
+ "", (IV)matches)
);
if (paren) {
- if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr;
+ if (matches) {
+ PL_regstartp[paren]
+ = HOPc(PL_reginput, -curlym_l) - PL_bostr;
PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
PL_regendp[paren] = -1;
}
- if (regmatch(next))
+ REGMATCH(next, CURLYM4);
+ /*** all unsaved local vars undefined at this point */
+ if (result)
sayYES;
REGCP_UNWIND(lastcp);
}
/* Couldn't or didn't -- back up. */
- n--;
- locinput = HOPc(locinput, -l);
+ matches--;
+ locinput = HOPc(locinput, -curlym_l);
PL_reginput = locinput;
}
}
sayNO;
+ /* NOTREACHED */
break;
}
case CURLYN:
if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
else {
if (PL_regkind[(U8)OP(text_node)] == REF) {
- I32 n, ln;
- n = ARG(text_node); /* which paren pair */
- ln = PL_regstartp[n];
- /* assume yes if we haven't seen CLOSEn */
- if (
- (I32)*PL_reglastparen < n ||
- ln == -1 ||
- ln == PL_regendp[n]
- ) {
- c1 = c2 = -1000;
- goto assume_ok_easy;
- }
- s = (U8*)PL_bostr + ln;
+ c1 = c2 = -1000;
+ goto assume_ok_easy;
}
else { s = (U8*)STRING(text_node); }
else { /* UTF */
if (OP(text_node) == EXACTF || OP(text_node) == REFF) {
STRLEN ulen1, ulen2;
- U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
- U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
+ U8 tmpbuf1[UTF8_MAXBYTES_CASE+1];
+ U8 tmpbuf2[UTF8_MAXBYTES_CASE+1];
to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
- c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
- c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0,
+ uniflags);
+ c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0,
+ uniflags);
}
else {
- c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ c2 = c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0,
+ uniflags);
}
}
}
assume_ok_easy:
PL_reginput = locinput;
if (minmod) {
- CHECKPOINT lastcp;
minmod = 0;
if (ln && regrepeat(scan, ln) < ln)
sayNO;
locinput = PL_reginput;
REGCP_SET(lastcp);
if (c1 != -1000) {
- char *e; /* Should not check after this */
- char *old = locinput;
- int count = 0;
+ old = locinput;
+ count = 0;
if (n == REG_INFTY) {
e = PL_regeol - 1;
count = locinput - old;
}
else {
- STRLEN len;
if (c1 == c2) {
+ STRLEN len;
/* count initialised to
* utf8_distance(old, locinput) */
while (locinput <= e &&
utf8n_to_uvchr((U8*)locinput,
- UTF8_MAXLEN, &len,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY) != (UV)c1) {
+ UTF8_MAXBYTES, &len,
+ uniflags) != (UV)c1) {
locinput += len;
count++;
}
} else {
+ STRLEN len;
/* count initialised to
* utf8_distance(old, locinput) */
while (locinput <= e) {
UV c = utf8n_to_uvchr((U8*)locinput,
- UTF8_MAXLEN, &len,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ UTF8_MAXBYTES, &len,
+ uniflags);
if (c == (UV)c1 || c == (UV)c2)
break;
locinput += len;
sayNO;
}
/* PL_reginput == locinput now */
- TRYPAREN(paren, ln, locinput);
+ TRYPAREN(paren, ln, locinput, PLUS1);
+ /*** all unsaved local vars undefined at this point */
PL_reginput = locinput; /* Could be reset... */
REGCP_UNWIND(lastcp);
/* Couldn't or didn't -- move forward. */
if (c1 != -1000) {
if (do_utf8)
c = utf8n_to_uvchr((U8*)PL_reginput,
- UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ UTF8_MAXBYTES, 0,
+ uniflags);
else
c = UCHARAT(PL_reginput);
/* If it could work, try it. */
if (c == (UV)c1 || c == (UV)c2)
{
- TRYPAREN(paren, n, PL_reginput);
+ TRYPAREN(paren, ln, PL_reginput, PLUS2);
+ /*** all unsaved local vars undefined at this point */
REGCP_UNWIND(lastcp);
}
}
/* If it could work, try it. */
else if (c1 == -1000)
{
- TRYPAREN(paren, n, PL_reginput);
+ TRYPAREN(paren, ln, PL_reginput, PLUS3);
+ /*** all unsaved local vars undefined at this point */
REGCP_UNWIND(lastcp);
}
/* Couldn't or didn't -- move forward. */
}
}
else {
- CHECKPOINT lastcp;
n = regrepeat(scan, n);
locinput = PL_reginput;
if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
- ((!PL_multiline && OP(next) != MEOL) ||
+ (OP(next) != MEOL ||
OP(next) == SEOL || OP(next) == EOS))
{
ln = n; /* why back off? */
ln--;
}
REGCP_SET(lastcp);
- if (paren) {
- UV c = 0;
- while (n >= ln) {
- if (c1 != -1000) {
- if (do_utf8)
- c = utf8n_to_uvchr((U8*)PL_reginput,
- UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
- else
- c = UCHARAT(PL_reginput);
- }
- /* If it could work, try it. */
- if (c1 == -1000 || c == (UV)c1 || c == (UV)c2)
- {
- TRYPAREN(paren, n, PL_reginput);
- REGCP_UNWIND(lastcp);
- }
- /* Couldn't or didn't -- back up. */
- n--;
- PL_reginput = locinput = HOPc(locinput, -1);
- }
- }
- else {
+ {
UV c = 0;
while (n >= ln) {
if (c1 != -1000) {
if (do_utf8)
c = utf8n_to_uvchr((U8*)PL_reginput,
- UTF8_MAXLEN, 0,
- ckWARN(WARN_UTF8) ?
- 0 : UTF8_ALLOW_ANY);
+ UTF8_MAXBYTES, 0,
+ uniflags);
else
c = UCHARAT(PL_reginput);
}
/* If it could work, try it. */
if (c1 == -1000 || c == (UV)c1 || c == (UV)c2)
{
- TRYPAREN(paren, n, PL_reginput);
+ TRYPAREN(paren, n, PL_reginput, PLUS4);
+ /*** all unsaved local vars undefined at this point */
REGCP_UNWIND(lastcp);
}
/* Couldn't or didn't -- back up. */
break;
case END:
if (PL_reg_call_cc) {
- re_cc_state *cur_call_cc = PL_reg_call_cc;
- CURCUR *cctmp = PL_regcc;
- regexp *re = PL_reg_re;
- CHECKPOINT cp, lastcp;
-
- cp = regcppush(0); /* Save *all* the positions. */
+ cur_call_cc = PL_reg_call_cc;
+ end_re = PL_reg_re;
+
+ /* Save *all* the positions. */
+ cp = regcppush(0);
REGCP_SET(lastcp);
- regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of
- the caller. */
- PL_reginput = locinput; /* Make position available to
- the callcc. */
+
+ /* Restore parens of the caller. */
+ {
+ I32 tmp = PL_savestack_ix;
+ PL_savestack_ix = PL_reg_call_cc->ss;
+ regcppop();
+ PL_savestack_ix = tmp;
+ }
+
+ /* Make position available to the callcc. */
+ PL_reginput = locinput;
+
cache_re(PL_reg_call_cc->re);
- PL_regcc = PL_reg_call_cc->cc;
+ oldcc = cc;
+ cc = PL_reg_call_cc->cc;
PL_reg_call_cc = PL_reg_call_cc->prev;
- if (regmatch(cur_call_cc->node)) {
+ REGMATCH(cur_call_cc->node, END);
+ /*** all unsaved local vars undefined at this point */
+ if (result) {
PL_reg_call_cc = cur_call_cc;
regcpblow(cp);
sayYES;
REGCP_UNWIND(lastcp);
regcppop();
PL_reg_call_cc = cur_call_cc;
- PL_regcc = cctmp;
- PL_reg_re = re;
- cache_re(re);
+ cc = oldcc;
+ PL_reg_re = end_re;
+ cache_re(end_re);
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s continuation failed...\n",
REPORT_CODE_OFF+PL_regindent*2, "")
sayNO_SILENT;
}
if (locinput < PL_regtill) {
- DEBUG_r(PerlIO_printf(Perl_debug_log,
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
"%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n",
PL_colors[4],
(long)(locinput - PL_reg_starttry),
case UNLESSM:
n = 0;
if (scan->flags) {
- s = HOPBACKc(locinput, scan->flags);
+ char * const s = HOPBACKc(locinput, scan->flags);
if (!s)
goto say_yes;
PL_reginput = s;
case IFMATCH:
n = 1;
if (scan->flags) {
- s = HOPBACKc(locinput, scan->flags);
+ char * const s = HOPBACKc(locinput, scan->flags);
if (!s)
goto say_no;
PL_reginput = s;
PL_reginput = locinput;
do_ifmatch:
- inner = NEXTOPER(NEXTOPER(scan));
- if (regmatch(inner) != n) {
+ REGMATCH(NEXTOPER(NEXTOPER(scan)), IFMATCH);
+ /*** all unsaved local vars undefined at this point */
+ if (result != n) {
say_no:
if (logical) {
logical = 0;
PTR2UV(scan), OP(scan));
Perl_croak(aTHX_ "regexp memory corruption");
}
+
reenter:
scan = next;
+ continue;
+ /* NOTREACHED */
+
+ /* simulate recursively calling regmatch(), but without actually
+ * recursing - ie save the current state on the heap rather than on
+ * the stack, then re-enter the loop. This avoids complex regexes
+ * blowing the processor stack */
+
+ start_recurse:
+ {
+ /* save existing local variables */
+ struct regmatch_state *p;
+
+ Newx(p, 1, struct regmatch_state);
+ p->prev_state = prev_state;
+ prev_state = p;
+ p->resume_state = resume_state;
+ p->scan = scan;
+ p->next = next;
+ p->minmod = minmod;
+ p->sw = sw;
+ p->logical = logical;
+ p->unwind = unwind;
+ p->cc = cc;
+ p->locinput = locinput;
+ p->n = n;
+ p->ln = ln;
+ p->c1 = c1;
+ p->c2 = c2;
+ p->paren = paren;
+ p->cp = cp;
+ p->lastcp = lastcp;
+ p->oldcc = oldcc;
+ p->lastloc = lastloc;
+ p->cache_offset = cache_offset;
+ p->cache_bit = cache_bit;
+ p->curlym_l = curlym_l;
+ p->matches = matches;
+ p->maxwanted = maxwanted;
+ p->e = e;
+ p->old = old;
+ p->count = count;
+ p->cur_call_cc = cur_call_cc;
+ p->end_re = end_re;
+ p->accept_buff = accept_buff;
+ p->accepted = accepted;
+ p->reg_call_cc = PL_reg_call_cc;
+
+ scan = new_scan;
+ locinput = PL_reginput;
+ nextchr = UCHARAT(locinput);
+ minmod = 0;
+ sw = 0;
+ logical = 0;
+ unwind = 0;
+#ifdef DEBUGGING
+ PL_regindent++;
+#endif
+ }
}
/*
sayNO;
yes_loud:
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s %scould match...%s\n",
- REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4],PL_colors[5])
+ REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5])
);
goto yes;
yes_final:
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
- PL_colors[4],PL_colors[5]));
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n",
+ PL_colors[4], PL_colors[5]));
yes:
#ifdef DEBUGGING
PL_regindent--;
#endif
-#if 0 /* Breaks $^R */
- if (unwind)
- regcpblow(firstcp);
-#endif
- return 1;
+ result = 1;
+ goto exit_level;
no:
- DEBUG_r(
+ DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"%*s %sfailed...%s\n",
- REPORT_CODE_OFF+PL_regindent*2, "",PL_colors[4],PL_colors[5])
+ REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], PL_colors[5])
);
goto do_no;
no_final:
do_no:
if (unwind) {
- re_unwind_t *uw = SSPTRt(unwind,re_unwind_t);
+ re_unwind_t * const uw = SSPTRt(unwind,re_unwind_t);
switch (uw->type) {
case RE_UNWIND_BRANCH:
case RE_UNWIND_BRANCHJ:
{
- re_unwind_branch_t *uwb = &(uw->branch);
- I32 lastparen = uwb->lastparen;
+ re_unwind_branch_t * const uwb = &(uw->branch);
+ const I32 lastparen = uwb->lastparen;
REGCP_UNWIND(uwb->lastcp);
for (n = *PL_reglastparen; n > lastparen; n--)
goto do_no;
}
/* Have more choice yet. Reuse the same uwb. */
- /*SUPPRESS 560*/
if ((n = (uwb->type == RE_UNWIND_BRANCH
? NEXT_OFF(next) : ARG(next))))
next += n;
goto reenter;
}
- /* NOT REACHED */
+ /* NOTREACHED */
default:
Perl_croak(aTHX_ "regexp unwind memory corruption");
}
- /* NOT REACHED */
+ /* NOTREACHED */
}
#ifdef DEBUGGING
PL_regindent--;
#endif
- return 0;
+ result = 0;
+exit_level:
+ if (prev_state) {
+ /* restore previous state and re-enter */
+
+ struct regmatch_state *p = prev_state;
+ resume_state = p->resume_state;
+ scan = p->scan;
+ next = p->next;
+ minmod = p->minmod;
+ sw = p->sw;
+ logical = p->logical;
+ unwind = p->unwind;
+ cc = p->cc;
+ locinput = p->locinput;
+ nextchr = UCHARAT(locinput);
+ n = p->n;
+ ln = p->ln;
+ c1 = p->c1;
+ c2 = p->c2;
+ paren = p->paren;
+ cp = p->cp;
+ lastcp = p->lastcp;
+ oldcc = p->oldcc;
+ lastloc = p->lastloc;
+ cache_offset = p->cache_offset;
+ cache_bit = p->cache_bit;
+ curlym_l = p->curlym_l;
+ matches = p->matches;
+ maxwanted = p->maxwanted;
+ e = p->e;
+ old = p->old;
+ count = p->count;
+ cur_call_cc = p->cur_call_cc;
+ end_re = p->end_re;
+ accept_buff = p->accept_buff;
+ accepted = p->accepted;
+ PL_reg_call_cc = p->reg_call_cc;
+ prev_state = p->prev_state;
+ Safefree(p);
+
+ switch (resume_state) {
+ case resume_TRIE1:
+ goto resume_point_TRIE1;
+ case resume_TRIE2:
+ goto resume_point_TRIE2;
+ case resume_CURLYX:
+ goto resume_point_CURLYX;
+ case resume_WHILEM1:
+ goto resume_point_WHILEM1;
+ case resume_WHILEM2:
+ goto resume_point_WHILEM2;
+ case resume_WHILEM3:
+ goto resume_point_WHILEM3;
+ case resume_WHILEM4:
+ goto resume_point_WHILEM4;
+ case resume_WHILEM5:
+ goto resume_point_WHILEM5;
+ case resume_WHILEM6:
+ goto resume_point_WHILEM6;
+ case resume_CURLYM1:
+ goto resume_point_CURLYM1;
+ case resume_CURLYM2:
+ goto resume_point_CURLYM2;
+ case resume_CURLYM3:
+ goto resume_point_CURLYM3;
+ case resume_CURLYM4:
+ goto resume_point_CURLYM4;
+ case resume_IFMATCH:
+ goto resume_point_IFMATCH;
+ case resume_PLUS1:
+ goto resume_point_PLUS1;
+ case resume_PLUS2:
+ goto resume_point_PLUS2;
+ case resume_PLUS3:
+ goto resume_point_PLUS3;
+ case resume_PLUS4:
+ goto resume_point_PLUS4;
+ case resume_END:
+ goto resume_point_END;
+ default:
+ Perl_croak(aTHX_ "regexp resume memory corruption");
+ }
+ /* NOTREACHED */
+ }
+ return result;
+
}
/*
* rather than incrementing count on every character. [Er, except utf8.]]
*/
STATIC I32
-S_regrepeat(pTHX_ regnode *p, I32 max)
+S_regrepeat(pTHX_ const regnode *p, I32 max)
{
+ dVAR;
register char *scan;
register I32 c;
register char *loceol = PL_regeol;
register bool do_utf8 = PL_reg_match_utf8;
scan = PL_reginput;
- if (max != REG_INFTY && max < loceol - scan)
+ if (max == REG_INFTY)
+ max = I32_MAX;
+ else if (max < loceol - scan)
loceol = scan + max;
switch (OP(p)) {
case REG_ANY:
case ALNUM:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(alnum,"a");
+ LOAD_UTF8_CHARCLASS_ALNUM();
while (hardcount < max && scan < loceol &&
swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
scan += UTF8SKIP(scan);
case NALNUM:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(alnum,"a");
+ LOAD_UTF8_CHARCLASS_ALNUM();
while (hardcount < max && scan < loceol &&
!swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
scan += UTF8SKIP(scan);
case SPACE:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(space," ");
+ LOAD_UTF8_CHARCLASS_SPACE();
while (hardcount < max && scan < loceol &&
(*scan == ' ' ||
swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
case NSPACE:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(space," ");
+ LOAD_UTF8_CHARCLASS_SPACE();
while (hardcount < max && scan < loceol &&
!(*scan == ' ' ||
swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
case DIGIT:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(digit,"0");
+ LOAD_UTF8_CHARCLASS_DIGIT();
while (hardcount < max && scan < loceol &&
swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
scan += UTF8SKIP(scan);
case NDIGIT:
if (do_utf8) {
loceol = PL_regeol;
- LOAD_UTF8_CHARCLASS(digit,"0");
+ LOAD_UTF8_CHARCLASS_DIGIT();
while (hardcount < max && scan < loceol &&
!swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
scan += UTF8SKIP(scan);
c = scan - PL_reginput;
PL_reginput = scan;
- DEBUG_r(
- {
- SV *prop = sv_newmortal();
-
+ DEBUG_r({
+ SV *re_debug_flags = NULL;
+ SV * const prop = sv_newmortal();
+ GET_RE_DEBUG_FLAGS;
+ DEBUG_EXECUTE_r({
regprop(prop, p);
PerlIO_printf(Perl_debug_log,
"%*s %s can match %"IVdf" times out of %"IVdf"...\n",
- REPORT_CODE_OFF+1, "", SvPVX(prop),(IV)c,(IV)max);
+ REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max);
+ });
});
return(c);
}
-/*
- - regrepeat_hard - repeatedly match something, report total lenth and length
- *
- * The repeater is supposed to have constant length.
- */
-
-STATIC I32
-S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp)
-{
- register char *scan = Nullch;
- register char *start;
- register char *loceol = PL_regeol;
- I32 l = 0;
- I32 count = 0, res = 1;
-
- if (!max)
- return 0;
-
- start = PL_reginput;
- if (PL_reg_match_utf8) {
- while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
- if (!count++) {
- l = 0;
- while (start < PL_reginput) {
- l++;
- start += UTF8SKIP(start);
- }
- *lp = l;
- if (l == 0)
- return max;
- }
- if (count == max)
- return count;
- }
- }
- else {
- while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
- if (!count++) {
- *lp = l = PL_reginput - start;
- if (max != REG_INFTY && l*max < loceol - scan)
- loceol = scan + l*max;
- if (l == 0)
- return max;
- }
- }
- }
- if (!res)
- PL_reginput = scan;
-
- return count;
-}
/*
- regclass_swash - prepare the utf8 swash
*/
SV *
-Perl_regclass_swash(pTHX_ register regnode* node, bool doinit, SV** listsvp, SV **altsvp)
+Perl_regclass_swash(pTHX_ register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
{
+ dVAR;
SV *sw = NULL;
SV *si = NULL;
SV *alt = NULL;
if (PL_regdata && PL_regdata->count) {
- U32 n = ARG(node);
+ const U32 n = ARG(node);
if (PL_regdata->what[n] == 's') {
- SV *rv = (SV*)PL_regdata->data[n];
- AV *av = (AV*)SvRV((SV*)rv);
- SV **ary = AvARRAY(av);
+ SV * const rv = (SV*)PL_regdata->data[n];
+ AV * const av = (AV*)SvRV((SV*)rv);
+ SV **const ary = AvARRAY(av);
SV **a, **b;
- /* See the end of regcomp.c:S_reglass() for
+ /* See the end of regcomp.c:S_regclass() for
* documentation of these array elements. */
si = *ary;
- a = SvTYPE(ary[1]) == SVt_RV ? &ary[1] : 0;
+ a = SvROK(ary[1]) ? &ary[1] : 0;
b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0;
if (a)
*/
STATIC bool
-S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register bool do_utf8)
+S_reginclass(pTHX_ register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8)
{
- char flags = ANYOF_FLAGS(n);
+ dVAR;
+ const char flags = ANYOF_FLAGS(n);
bool match = FALSE;
UV c = *p;
STRLEN len = 0;
STRLEN plen;
- if (do_utf8 && !UTF8_IS_INVARIANT(c))
- c = utf8n_to_uvchr(p, UTF8_MAXLEN, &len,
- ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+ if (do_utf8 && !UTF8_IS_INVARIANT(c)) {
+ c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len,
+ ckWARN(WARN_UTF8) ? UTF8_CHECK_ONLY :
+ UTF8_ALLOW_ANYUV|UTF8_CHECK_ONLY);
+ if (len == (STRLEN)-1)
+ Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)");
+ }
plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c));
if (do_utf8 || (flags & ANYOF_UNICODE)) {
match = TRUE;
if (!match) {
AV *av;
- SV *sw = regclass_swash(n, TRUE, 0, (SV**)&av);
+ SV * const sw = regclass_swash(n, TRUE, 0, (SV**)&av);
if (sw) {
if (swash_fetch(sw, p, do_utf8))
else if (flags & ANYOF_FOLD) {
if (!match && lenp && av) {
I32 i;
-
for (i = 0; i <= av_len(av); i++) {
- SV* sv = *av_fetch(av, i, FALSE);
+ SV* const sv = *av_fetch(av, i, FALSE);
STRLEN len;
- char *s = SvPV(sv, len);
+ const char * const s = SvPV_const(sv, len);
if (len <= plen && memEQ(s, (char*)p, len)) {
*lenp = len;
}
}
if (!match) {
- U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
+ U8 tmpbuf[UTF8_MAXBYTES_CASE+1];
STRLEN tmplen;
to_utf8_fold(p, tmpbuf, &tmplen);
}
STATIC U8 *
-S_reghop(pTHX_ U8 *s, I32 off)
-{
- return S_reghop3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr));
-}
-
-STATIC U8 *
-S_reghop3(pTHX_ U8 *s, I32 off, U8* lim)
+S_reghop3(U8 *s, I32 off, U8* lim)
{
+ dVAR;
if (off >= 0) {
while (off-- && s < lim) {
/* XXX could check well-formedness here */
}
STATIC U8 *
-S_reghopmaybe(pTHX_ U8 *s, I32 off)
-{
- return S_reghopmaybe3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr));
-}
-
-STATIC U8 *
-S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim)
+S_reghopmaybe3(U8* s, I32 off, U8* lim)
{
+ dVAR;
if (off >= 0) {
while (off-- && s < lim) {
/* XXX could check well-formedness here */
static void
restore_pos(pTHX_ void *arg)
{
+ dVAR;
+ PERL_UNUSED_ARG(arg);
if (PL_reg_eval_set) {
if (PL_reg_oldsaved) {
PL_reg_re->subbeg = PL_reg_oldsaved;
PL_reg_re->sublen = PL_reg_oldsavedlen;
+#ifdef PERL_OLD_COPY_ON_WRITE
+ PL_reg_re->saved_copy = PL_nrs;
+#endif
RX_MATCH_COPIED_on(PL_reg_re);
}
PL_reg_magic->mg_len = PL_reg_oldpos;
STATIC void
S_to_utf8_substr(pTHX_ register regexp *prog)
{
- SV* sv;
if (prog->float_substr && !prog->float_utf8) {
- prog->float_utf8 = sv = NEWSV(117, 0);
- SvSetSV(sv, prog->float_substr);
+ SV* sv;
+ prog->float_utf8 = sv = newSVsv(prog->float_substr);
sv_utf8_upgrade(sv);
if (SvTAIL(prog->float_substr))
SvTAIL_on(sv);
prog->check_utf8 = sv;
}
if (prog->anchored_substr && !prog->anchored_utf8) {
- prog->anchored_utf8 = sv = NEWSV(118, 0);
- SvSetSV(sv, prog->anchored_substr);
+ SV* sv;
+ prog->anchored_utf8 = sv = newSVsv(prog->anchored_substr);
sv_utf8_upgrade(sv);
if (SvTAIL(prog->anchored_substr))
SvTAIL_on(sv);
STATIC void
S_to_byte_substr(pTHX_ register regexp *prog)
{
- SV* sv;
+ dVAR;
if (prog->float_utf8 && !prog->float_substr) {
- prog->float_substr = sv = NEWSV(117, 0);
- SvSetSV(sv, prog->float_utf8);
+ SV* sv;
+ prog->float_substr = sv = newSVsv(prog->float_utf8);
if (sv_utf8_downgrade(sv, TRUE)) {
if (SvTAIL(prog->float_utf8))
SvTAIL_on(sv);
prog->check_substr = sv;
}
if (prog->anchored_utf8 && !prog->anchored_substr) {
- prog->anchored_substr = sv = NEWSV(118, 0);
- SvSetSV(sv, prog->anchored_utf8);
+ SV* sv;
+ prog->anchored_substr = sv = newSVsv(prog->anchored_utf8);
if (sv_utf8_downgrade(sv, TRUE)) {
if (SvTAIL(prog->anchored_utf8))
SvTAIL_on(sv);
prog->check_substr = sv;
}
}
+
+/*
+ * Local variables:
+ * c-indentation-style: bsd
+ * c-basic-offset: 4
+ * indent-tabs-mode: t
+ * End:
+ *
+ * ex: set ts=8 sts=4 sw=4 noet:
+ */
projects / p5sagit/p5-mst-13.2.git / blobdiff