*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-1999, Larry Wall
+ **** Copyright (c) 1991-2000, Larry Wall
****
**** You may distribute under the terms of either the GNU General Public
**** License or the Artistic License, as specified in the README file.
/* These are needed since we do not localize EVAL nodes: */
# define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \
- " Setting an EVAL scope, savestack=%i\n", \
- PL_savestack_ix)); lastcp = PL_savestack_ix
+ " Setting an EVAL scope, savestack=%"IVdf"\n", \
+ (IV)PL_savestack_ix)); lastcp = PL_savestack_ix
# define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \
PerlIO_printf(Perl_debug_log, \
- " Clearing an EVAL scope, savestack=%i..%i\n", \
- lastcp, PL_savestack_ix) : 0); regcpblow(lastcp)
+ " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \
+ (IV)lastcp, (IV)PL_savestack_ix) : 0); regcpblow(lastcp)
STATIC char *
S_regcppop(pTHX)
DEBUG_r(
if (*PL_reglastparen + 1 <= PL_regnpar) {
PerlIO_printf(Perl_debug_log,
- " restoring \\%d..\\%d to undef\n",
- *PL_reglastparen + 1, PL_regnpar);
+ " restoring \\%"IVdf"..\\%"IVdf" to undef\n",
+ (IV)(*PL_reglastparen + 1), (IV)PL_regnpar);
}
);
for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) {
#define regcpblow(cp) LEAVE_SCOPE(cp)
+#define TRYPAREN(paren, n, input) { \
+ if (paren) { \
+ if (n) { \
+ PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \
+ PL_regendp[paren] = input - PL_bostr; \
+ } \
+ else \
+ PL_regendp[paren] = -1; \
+ } \
+ if (regmatch(next)) \
+ sayYES; \
+ if (paren && n) \
+ PL_regendp[paren] = -1; \
+}
+
+
/*
* pregexec and friends
*/
/* XXXX We assume that strpos is strbeg unless sv. */
+/* XXXX Some places assume that there is a fixed substring.
+ An update may be needed if optimizer marks as "INTUITable"
+ RExen without fixed substrings. Similarly, it is assumed that
+ lengths of all the strings are no more than minlen, thus they
+ cannot come from lookahead.
+ (Or minlen should take into account lookahead.) */
+
/* A failure to find a constant substring means that there is no need to make
an expensive call to REx engine, thus we celebrate a failure. Similarly,
finding a substring too deep into the string means that less calls to
b) Fixed substring;
c) Whether we are anchored (beginning-of-line or \G);
d) First node (of those at offset 0) which may distingush positions;
- We use 'a', 'b', multiline-part of 'c', and try to find a position in the
+ We use a)b)d) and multiline-part of c), and try to find a position in the
string which does not contradict any of them.
*/
+/* Most of decisions we do here should have been done at compile time.
+ The nodes of the REx which we used for the search should have been
+ deleted from the finite automaton. */
+
char *
Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
char *strend, U32 flags, re_scream_pos_data *data)
register I32 end_shift;
register char *s;
register SV *check;
+ char *strbeg;
char *t;
I32 ml_anch;
char *tmp;
- register char *other_last = Nullch;
+ register char *other_last = Nullch; /* other substr checked before this */
+ char *check_at; /* check substr found at this pos */
#ifdef DEBUGGING
char *i_strpos = strpos;
#endif
DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n"));
goto fail;
}
+ strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos;
+ check = prog->check_substr;
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? */
- if ((prog->check_offset_min == prog->check_offset_max) && !ml_anch) {
+ if (!ml_anch) {
+ if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */
+ /* SvCUR is not set on references: SvRV and SvPVX overlap */
+ && sv && !SvROK(sv)
+ && (strpos != strbeg)) {
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
+ goto fail;
+ }
+ if (prog->check_offset_min == prog->check_offset_max) {
/* Substring at constant offset from beg-of-str... */
I32 slen;
- if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */
- && (sv && (strpos + SvCUR(sv) != strend)) ) {
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
- goto fail;
- }
PL_regeol = strend; /* Used in HOP() */
s = HOPc(strpos, prog->check_offset_min);
- if (SvTAIL(prog->check_substr)) {
- slen = SvCUR(prog->check_substr); /* >= 1 */
+ if (SvTAIL(check)) {
+ slen = SvCUR(check); /* >= 1 */
if ( strend - s > slen || strend - s < slen - 1
|| (strend - s == slen && strend[-1] != '\n')) {
}
/* Now should match s[0..slen-2] */
slen--;
- if (slen && (*SvPVX(prog->check_substr) != *s
+ if (slen && (*SvPVX(check) != *s
|| (slen > 1
- && memNE(SvPVX(prog->check_substr), s, slen)))) {
+ && memNE(SvPVX(check), s, slen)))) {
report_neq:
DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n"));
goto fail_finish;
}
}
- else if (*SvPVX(prog->check_substr) != *s
- || ((slen = SvCUR(prog->check_substr)) > 1
- && memNE(SvPVX(prog->check_substr), s, slen)))
+ else if (*SvPVX(check) != *s
+ || ((slen = SvCUR(check)) > 1
+ && memNE(SvPVX(check), s, slen)))
goto report_neq;
goto success_at_start;
+ }
}
/* Match is anchored, but substr is not anchored wrt beg-of-str. */
s = strpos;
start_shift = prog->check_offset_min; /* okay to underestimate on CC */
- /* Should be nonnegative! */
end_shift = prog->minlen - start_shift -
- CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0);
+ CHR_SVLEN(check) + (SvTAIL(check) != 0);
if (!ml_anch) {
- I32 end = prog->check_offset_max + CHR_SVLEN(prog->check_substr)
- - (SvTAIL(prog->check_substr) != 0);
+ I32 end = prog->check_offset_max + CHR_SVLEN(check)
+ - (SvTAIL(check) != 0);
I32 eshift = strend - s - end;
if (end_shift < eshift)
start_shift = prog->check_offset_min; /* okay to underestimate on CC */
/* Should be nonnegative! */
end_shift = prog->minlen - start_shift -
- CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0);
+ CHR_SVLEN(check) + (SvTAIL(check) != 0);
}
#ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */
Perl_croak(aTHX_ "panic: end_shift");
#endif
- check = prog->check_substr;
restart:
/* Find a possible match in the region s..strend by looking for
the "check" substring in the region corrected by start/end_shift. */
if (flags & REXEC_SCREAM) {
- char *strbeg = SvPVX(sv); /* XXXX Assume PV_force() on SCREAM! */
I32 p = -1; /* Internal iterator of scream. */
I32 *pp = data ? data->scream_pos : &p;
if (!s)
goto fail_finish;
+ check_at = s;
+
/* Finish the diagnostic message */
DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) );
/* Take into account the "other" substring. */
/* XXXX May be hopelessly wrong for UTF... */
if (!other_last)
- other_last = strpos - 1;
+ other_last = strpos;
if (check == prog->float_substr) {
do_other_anchored:
{
else
t = strpos;
t += prog->anchored_offset;
- if (t <= other_last)
- t = other_last + 1;
+ if (t < other_last) /* These positions already checked */
+ t = other_last;
PL_bostr = tmp;
last2 = last1 = strend - prog->minlen;
if (last < last1)
", trying floating at offset %ld...\n",
(long)(s1 + 1 - i_strpos)));
PL_regeol = strend; /* Used in HOP() */
- other_last = last1 + prog->anchored_offset;
+ other_last = last1 + prog->anchored_offset + 1;
s = HOPc(last, 1);
goto restart;
}
DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
(long)(s - i_strpos)));
t = s - prog->anchored_offset;
- other_last = s - 1;
+ other_last = s + 1;
s = s1;
if (t == strpos)
goto try_at_start;
if (last - t > prog->float_max_offset)
last = t + prog->float_max_offset;
s = t + prog->float_min_offset;
- if (s <= other_last)
- s = other_last + 1;
+ if (s < other_last)
+ s = other_last;
/* XXXX It is not documented what units *_offsets are in. Assume bytes. */
/* fbm_instr() takes into account exact value of end-of-str
if the check is SvTAIL(ed). Since false positives are OK,
DEBUG_r(PerlIO_printf(Perl_debug_log,
", trying anchored starting at offset %ld...\n",
(long)(s1 + 1 - i_strpos)));
- other_last = last;
+ other_last = last + 1;
PL_regeol = strend; /* Used in HOP() */
s = HOPc(t, 1);
goto restart;
else {
DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n",
(long)(s - i_strpos)));
- other_last = s - 1;
+ other_last = s + 1;
s = s1;
if (t == strpos)
goto try_at_start;
find_anchor:
while (t < strend - prog->minlen) {
if (*t == '\n') {
- if (t < s - prog->check_offset_min) {
+ if (t < check_at - prog->check_offset_min) {
if (prog->anchored_substr) {
- /* We definitely contradict the found anchored
+ /* Since we moved from the found position,
+ we definitely contradict the found anchored
substr. Due to the above check we do not
contradict "check" substr.
Thus we can arrive here only if check substr
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)));
goto set_useful;
}
- DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting at offset %ld...\n",
+ /* 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)));
- strpos = s = t + 1;
+ other_last = strpos = s = t + 1;
goto restart;
}
t++;
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]));
+ }
s = t;
set_useful:
++BmUSEFUL(prog->check_substr); /* hooray/5 */
}
else {
PL_bostr = tmp;
- /* The found string does not prohibit matching at beg-of-str
+ /* The found string does not prohibit matching at strpos,
- no optimization of calling REx engine can be performed,
- unless it was an MBOL and we are not after MBOL. */
+ unless it was an MBOL and we are not after MBOL,
+ or a future STCLASS check will fail this. */
try_at_start:
/* Even in this situation we may use MBOL flag if strpos is offset
wrt the start of the string. */
- if (ml_anch && sv
- && (strpos + SvCUR(sv) != strend) && strpos[-1] != '\n') {
+ if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */
+ && (strpos != strbeg) && strpos[-1] != '\n'
+ /* May be due to an implicit anchor of m{.*foo} */
+ && !(prog->reganch & ROPT_IMPLICIT))
+ {
t = strpos;
goto find_anchor;
}
DEBUG_r( if (ml_anch)
- PerlIO_printf(Perl_debug_log, "Does not contradict /%s^%s/m...\n",
- PL_colors[0],PL_colors[1]);
+ 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]);
);
success_at_start:
if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */
+ && prog->check_substr /* Could be deleted already */
&& --BmUSEFUL(prog->check_substr) < 0
- && prog->check_substr == prog->float_substr) { /* boo */
+ && prog->check_substr == prog->float_substr)
+ {
/* If flags & SOMETHING - do not do it many times on the same match */
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n"));
SvREFCNT_dec(prog->check_substr);
prog->check_substr = Nullsv; /* disable */
prog->float_substr = Nullsv; /* clear */
+ check = Nullsv; /* abort */
s = strpos;
+ /* XXXX This is a remnant of the old implementation. It
+ looks wasteful, since now INTUIT can use many
+ other heuristics. */
prog->reganch &= ~RE_USE_INTUIT;
}
else
s = strpos;
}
- DEBUG_r(PerlIO_printf(Perl_debug_log, "%sGuessed:%s match at offset %ld\n",
- PL_colors[4], PL_colors[5], (long)(s - i_strpos)) );
+ /* Last resort... */
+ /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */
+ if (prog->regstclass) {
+ /* minlen == 0 is possible if regstclass is \b or \B,
+ and the fixed substr is ''$.
+ Since minlen is already taken into account, s+1 is before strend;
+ accidentally, minlen >= 1 guaranties no false positives at s + 1
+ even for \b or \B. But (minlen? 1 : 0) below assumes that
+ regstclass does not come from lookahead... */
+ /* If regstclass takes bytelength more than 1: If charlength==1, OK.
+ This leaves EXACTF only, which is dealt with in find_byclass(). */
+ int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
+ ? STR_LEN(prog->regstclass)
+ : 1);
+ char *endpos = (prog->anchored_substr || ml_anch)
+ ? s + (prog->minlen? cl_l : 0)
+ : (prog->float_substr ? check_at - start_shift + cl_l
+ : strend) ;
+ char *startpos = strbeg;
+
+ t = s;
+ if (prog->reganch & ROPT_UTF8) {
+ PL_regdata = prog->data; /* Used by REGINCLASS UTF logic */
+ PL_bostr = startpos;
+ }
+ s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1);
+ if (!s) {
+#ifdef DEBUGGING
+ char *what;
+#endif
+ if (endpos == strend) {
+ DEBUG_r( PerlIO_printf(Perl_debug_log,
+ "Could not match STCLASS...\n") );
+ goto fail;
+ }
+ DEBUG_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) {
+ if (prog->anchored_substr == check) {
+ DEBUG_r( what = "anchored" );
+ hop_and_restart:
+ PL_regeol = strend; /* Used in HOP() */
+ s = HOPc(t, 1);
+ if (s + start_shift + end_shift > strend) {
+ /* XXXX Should be taken into account earlier? */
+ DEBUG_r( PerlIO_printf(Perl_debug_log,
+ "Could not match STCLASS...\n") );
+ goto fail;
+ }
+ if (!check)
+ goto giveup;
+ DEBUG_r( PerlIO_printf(Perl_debug_log,
+ "Looking for %s substr starting at offset %ld...\n",
+ what, (long)(s + start_shift - i_strpos)) );
+ goto restart;
+ }
+ /* Have both, check_string is floating */
+ if (t + start_shift >= check_at) /* Contradicts floating=check */
+ goto retry_floating_check;
+ /* Recheck anchored substring, but not floating... */
+ s = check_at;
+ if (!check)
+ goto giveup;
+ DEBUG_r( PerlIO_printf(Perl_debug_log,
+ "Looking for anchored substr starting at offset %ld...\n",
+ (long)(other_last - i_strpos)) );
+ goto do_other_anchored;
+ }
+ /* Another way we could have checked stclass at the
+ current position only: */
+ if (ml_anch) {
+ s = t = t + 1;
+ if (!check)
+ goto giveup;
+ DEBUG_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)) );
+ goto try_at_offset;
+ }
+ if (!prog->float_substr) /* Could have been deleted */
+ goto fail;
+ /* Check is floating subtring. */
+ retry_floating_check:
+ t = check_at - start_shift;
+ DEBUG_r( what = "floating" );
+ goto hop_and_restart;
+ }
+ DEBUG_r( if (t != s)
+ PerlIO_printf(Perl_debug_log,
+ "By STCLASS: moving %ld --> %ld\n",
+ (long)(t - i_strpos), (long)(s - i_strpos));
+ else
+ 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",
+ PL_colors[4], (check ? "Guessed" : "Giving up"),
+ PL_colors[5], (long)(s - i_strpos)) );
return s;
fail_finish: /* Substring not found */
- BmUSEFUL(prog->check_substr) += 5; /* hooray */
+ if (prog->check_substr) /* could be removed already */
+ BmUSEFUL(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;
}
-/*
- - regexec_flags - match a regexp against a string
- */
-I32
-Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
- char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
-/* strend: pointer to null at end of string */
-/* strbeg: real beginning of string */
-/* minend: end of match must be >=minend after stringarg. */
-/* data: May be used for some additional optimizations. */
-/* nosave: For optimizations. */
+/* 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)
{
- dTHR;
- register char *s;
- register regnode *c;
- register char *startpos = stringarg;
- register I32 tmp;
- 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;
- SV* oreplsv = GvSV(PL_replgv);
-
- PL_regcc = 0;
-
- cache_re(prog);
-#ifdef DEBUGGING
- PL_regnarrate = PL_debug & 512;
-#endif
-
- /* Be paranoid... */
- if (prog == NULL || startpos == NULL) {
- Perl_croak(aTHX_ "NULL regexp parameter");
- return 0;
- }
-
- minlen = prog->minlen;
- if (strend - startpos < minlen) goto phooey;
-
- if (startpos == strbeg) /* is ^ valid at stringarg? */
- PL_regprev = '\n';
- else {
- PL_regprev = (U32)stringarg[-1];
- if (!PL_multiline && PL_regprev == '\n')
- PL_regprev = '\0'; /* force ^ to NOT match */
- }
-
- /* Check validity of program. */
- if (UCHARAT(prog->program) != REG_MAGIC) {
- Perl_croak(aTHX_ "corrupted regexp program");
- }
-
- PL_reg_flags = 0;
- PL_reg_eval_set = 0;
- PL_reg_maxiter = 0;
-
- if (prog->reganch & ROPT_UTF8)
- PL_reg_flags |= RF_utf8;
-
- /* Mark beginning of line for ^ and lookbehind. */
- PL_regbol = startpos;
- PL_bostr = strbeg;
- PL_reg_sv = sv;
-
- /* Mark end of line for $ (and such) */
- PL_regeol = strend;
-
- /* see how far we have to get to not match where we matched before */
- PL_regtill = startpos+minend;
-
- /* We start without call_cc context. */
- PL_reg_call_cc = 0;
-
- /* If there is a "must appear" string, look for it. */
- s = startpos;
-
- if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
- MAGIC *mg;
-
- if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
- PL_reg_ganch = startpos;
- else if (sv && SvTYPE(sv) >= SVt_PVMG
- && SvMAGIC(sv)
- && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) {
- PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
- if (prog->reganch & ROPT_ANCH_GPOS) {
- if (s > PL_reg_ganch)
- goto phooey;
- s = PL_reg_ganch;
- }
- }
- else /* pos() not defined */
- PL_reg_ganch = strbeg;
- }
-
- if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
- re_scream_pos_data d;
-
- d.scream_olds = &scream_olds;
- d.scream_pos = &scream_pos;
- s = re_intuit_start(prog, sv, s, strend, flags, &d);
- if (!s)
- goto phooey; /* not present */
- }
-
- DEBUG_r( if (!PL_colorset) reginitcolors() );
- DEBUG_r(PerlIO_printf(Perl_debug_log,
- "%sMatching 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 ? "..." : ""),
- PL_colors[0],
- (int)(strend - startpos > 60 ? 60 : strend - startpos),
- startpos, PL_colors[1],
- (strend - startpos > 60 ? "..." : ""))
- );
-
- /* Simplest case: anchored match need be tried only once. */
- /* [unless only anchor is BOL and multiline is set] */
- if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
- if (s == startpos && regtry(prog, startpos))
- goto got_it;
- else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
- || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
- {
- char *end;
-
- if (minlen)
- dontbother = minlen - 1;
- end = HOPc(strend, -dontbother) - 1;
- /* for multiline we only have to try after newlines */
- if (prog->check_substr) {
- if (s == startpos)
- goto after_try;
- while (1) {
- if (regtry(prog, s))
- goto got_it;
- after_try:
- if (s >= end)
- goto phooey;
- s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
- if (!s)
- goto phooey;
- }
- } else {
- if (s > startpos)
- s--;
- while (s < end) {
- if (*s++ == '\n') { /* don't need PL_utf8skip here */
- if (regtry(prog, s))
- goto got_it;
- }
- }
- }
- }
- goto phooey;
- } else if (prog->reganch & ROPT_ANCH_GPOS) {
- if (regtry(prog, PL_reg_ganch))
- goto got_it;
- goto phooey;
- }
-
- /* Messy cases: unanchored match. */
- if (prog->anchored_substr && prog->reganch & ROPT_SKIP) {
- /* we have /x+whatever/ */
- /* it must be a one character string (XXXX Except UTF?) */
- char ch = SvPVX(prog->anchored_substr)[0];
- if (UTF) {
- while (s < strend) {
- if (*s == ch) {
- if (regtry(prog, s)) goto got_it;
- s += UTF8SKIP(s);
- while (s < strend && *s == ch)
- s += UTF8SKIP(s);
- }
- s += UTF8SKIP(s);
- }
- }
- else {
- while (s < strend) {
- if (*s == ch) {
- if (regtry(prog, s)) goto got_it;
- s++;
- while (s < strend && *s == ch)
- s++;
- }
- s++;
- }
- }
- }
- /*SUPPRESS 560*/
- else if (prog->anchored_substr != Nullsv
- || (prog->float_substr != Nullsv
- && prog->float_max_offset < strend - s)) {
- SV *must = prog->anchored_substr
- ? prog->anchored_substr : prog->float_substr;
- I32 back_max =
- prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset;
- I32 back_min =
- prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
- I32 delta = back_max - back_min;
- char *last = HOPc(strend, /* Cannot start after this */
- -(I32)(CHR_SVLEN(must)
- - (SvTAIL(must) != 0) + back_min));
- char *last1; /* Last position checked before */
-
- if (s > PL_bostr)
- last1 = HOPc(s, -1);
- else
- last1 = s - 1; /* bogus */
-
- /* XXXX check_substr already used to find `s', can optimize if
- check_substr==must. */
- scream_pos = -1;
- dontbother = end_shift;
- strend = HOPc(strend, -dontbother);
- while ( (s <= last) &&
- ((flags & REXEC_SCREAM)
- ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg,
- end_shift, &scream_pos, 0))
- : (s = fbm_instr((unsigned char*)HOP(s, back_min),
- (unsigned char*)strend, must,
- PL_multiline ? FBMrf_MULTILINE : 0))) ) {
- 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;
-
- last1 = HOPc(s, -back_min);
- s = t;
- }
- if (UTF) {
- while (s <= last1) {
- if (regtry(prog, s))
- goto got_it;
- s += UTF8SKIP(s);
- }
- }
- else {
- while (s <= last1) {
- if (regtry(prog, s))
- goto got_it;
- s++;
- }
- }
- }
- goto phooey;
- }
- else if (c = prog->regstclass) {
I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
- char *cc;
char *m;
- int ln;
- int c1;
- int c2;
+ STRLEN ln;
+ unsigned int c1;
+ unsigned int c2;
char *e;
+ register I32 tmp = 1; /* Scratch variable? */
- if (minlen)
- dontbother = minlen - 1;
- strend = HOPc(strend, -dontbother); /* don't bother with what can't match */
- tmp = 1;
/* We know what class it must start with. */
switch (OP(c)) {
case ANYOFUTF8:
- cc = MASK(c);
while (s < strend) {
if (REGINCLASSUTF8(c, (U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
}
break;
case ANYOF:
- cc = MASK(c);
while (s < strend) {
- if (REGINCLASS(cc, *s)) {
- if (tmp && regtry(prog, s))
+ if (REGINCLASS(c, *(U8*)s)) {
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case EXACTF:
m = STRING(c);
ln = STR_LEN(c);
- c1 = *m;
+ c1 = *(U8*)m;
c2 = PL_fold[c1];
goto do_exactf;
case EXACTFL:
m = STRING(c);
ln = STR_LEN(c);
- c1 = *m;
+ c1 = *(U8*)m;
c2 = PL_fold_locale[c1];
do_exactf:
e = strend - ln;
+ if (norun && e < s)
+ e = s; /* Due to minlen logic of intuit() */
/* Here it is NOT UTF! */
if (c1 == c2) {
while (s <= e) {
- if ( *s == c1
- && (ln == 1 || (OP(c) == EXACTF
- ? ibcmp(s, m, ln)
- : ibcmp_locale(s, m, ln)))
- && regtry(prog, s) )
+ if ( *(U8*)s == c1
+ && (ln == 1 || !(OP(c) == EXACTF
+ ? ibcmp(s, m, ln)
+ : ibcmp_locale(s, m, ln)))
+ && (norun || regtry(prog, s)) )
goto got_it;
s++;
}
} else {
while (s <= e) {
- if ( (*s == c1 || *s == c2)
- && (ln == 1 || (OP(c) == EXACTF
- ? ibcmp(s, m, ln)
- : ibcmp_locale(s, m, ln)))
- && regtry(prog, s) )
+ if ( (*(U8*)s == c1 || *(U8*)s == c2)
+ && (ln == 1 || !(OP(c) == EXACTF
+ ? ibcmp(s, m, ln)
+ : ibcmp_locale(s, m, ln)))
+ && (norun || regtry(prog, s)) )
goto got_it;
s++;
}
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case BOUND:
- if (minlen) {
- dontbother++;
- strend -= 1;
- }
- tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
+ tmp = (s != startpos) ? UCHARAT(s - 1) : '\n';
tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
while (s < strend) {
if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
tmp = !tmp;
- if (regtry(prog, s))
+ if ((norun || regtry(prog, s)))
goto got_it;
}
s++;
}
- if ((minlen || tmp) && regtry(prog,s))
+ if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
goto got_it;
break;
case BOUNDLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case BOUNDUTF8:
- if (minlen) {
- dontbother++;
- strend = reghop_c(strend, -1);
- }
- tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
- tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
+ tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1),
+ strend - s,
+ 0, 0) : '\n';
+ tmp = ((OP(c) == BOUNDUTF8 ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
while (s < strend) {
- if (tmp == !(OP(c) == BOUND ?
+ if (tmp == !(OP(c) == BOUNDUTF8 ?
swash_fetch(PL_utf8_alnum, (U8*)s) :
isALNUM_LC_utf8((U8*)s)))
{
tmp = !tmp;
- if (regtry(prog, s))
+ if ((norun || regtry(prog, s)))
goto got_it;
}
s += UTF8SKIP(s);
}
- if ((minlen || tmp) && regtry(prog,s))
+ if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
goto got_it;
break;
case NBOUNDL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NBOUND:
- if (minlen) {
- dontbother++;
- strend -= 1;
- }
- tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
+ tmp = (s != startpos) ? UCHARAT(s - 1) : '\n';
tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
while (s < strend) {
if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
tmp = !tmp;
- else if (regtry(prog, s))
+ else if ((norun || regtry(prog, s)))
goto got_it;
s++;
}
- if ((minlen || !tmp) && regtry(prog,s))
+ if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
goto got_it;
break;
case NBOUNDLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NBOUNDUTF8:
- if (minlen) {
- dontbother++;
- strend = reghop_c(strend, -1);
- }
- tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
- tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
+ tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1),
+ strend - s,
+ 0, 0) : '\n';
+ tmp = ((OP(c) == NBOUNDUTF8 ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
while (s < strend) {
- if (tmp == !(OP(c) == NBOUND ?
+ if (tmp == !(OP(c) == NBOUNDUTF8 ?
swash_fetch(PL_utf8_alnum, (U8*)s) :
isALNUM_LC_utf8((U8*)s)))
tmp = !tmp;
- else if (regtry(prog, s))
+ else if ((norun || regtry(prog, s)))
goto got_it;
s += UTF8SKIP(s);
}
- if ((minlen || !tmp) && regtry(prog,s))
+ if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
goto got_it;
break;
case ALNUM:
while (s < strend) {
if (isALNUM(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case ALNUMUTF8:
while (s < strend) {
if (swash_fetch(PL_utf8_alnum, (U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (isALNUM_LC(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (isALNUM_LC_utf8((U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case NALNUM:
while (s < strend) {
if (!isALNUM(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case NALNUMUTF8:
while (s < strend) {
if (!swash_fetch(PL_utf8_alnum, (U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (!isALNUM_LC(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (!isALNUM_LC_utf8((U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case SPACE:
while (s < strend) {
if (isSPACE(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case SPACEUTF8:
while (s < strend) {
if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (isSPACE_LC(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case NSPACE:
while (s < strend) {
if (!isSPACE(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case NSPACEUTF8:
while (s < strend) {
if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (!isSPACE_LC(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
PL_reg_flags |= RF_tainted;
while (s < strend) {
if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case DIGIT:
while (s < strend) {
if (isDIGIT(*s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
case DIGITUTF8:
while (s < strend) {
if (swash_fetch(PL_utf8_digit,(U8*)s)) {
- if (tmp && regtry(prog, s))
+ if (tmp && (norun || regtry(prog, s)))
goto got_it;
else
tmp = doevery;
tmp = 1;
s += UTF8SKIP(s);
}
- break;
- case DIGITL:
- PL_reg_flags |= RF_tainted;
- while (s < strend) {
- if (isDIGIT_LC(*s)) {
- if (tmp && regtry(prog, s))
+ break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC(*s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case DIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NDIGIT:
+ while (s < strend) {
+ if (!isDIGIT(*s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NDIGITUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC(*s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NDIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && (norun || regtry(prog, s)))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ default:
+ Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c));
+ break;
+ }
+ return 0;
+ got_it:
+ return s;
+}
+
+/*
+ - regexec_flags - match a regexp against a string
+ */
+I32
+Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend,
+ char *strbeg, I32 minend, SV *sv, void *data, U32 flags)
+/* strend: pointer to null at end of string */
+/* strbeg: real beginning of string */
+/* minend: end of match must be >=minend after stringarg. */
+/* data: May be used for some additional optimizations. */
+/* nosave: For optimizations. */
+{
+ dTHR;
+ 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;
+ SV* oreplsv = GvSV(PL_replgv);
+
+ PL_regcc = 0;
+
+ cache_re(prog);
+#ifdef DEBUGGING
+ PL_regnarrate = PL_debug & 512;
+#endif
+
+ /* Be paranoid... */
+ if (prog == NULL || startpos == NULL) {
+ Perl_croak(aTHX_ "NULL regexp parameter");
+ return 0;
+ }
+
+ minlen = prog->minlen;
+ if (strend - startpos < minlen) goto phooey;
+
+ if (startpos == strbeg) /* is ^ valid at stringarg? */
+ PL_regprev = '\n';
+ else {
+ PL_regprev = (U32)stringarg[-1];
+ if (!PL_multiline && PL_regprev == '\n')
+ PL_regprev = '\0'; /* force ^ to NOT match */
+ }
+
+ /* Check validity of program. */
+ if (UCHARAT(prog->program) != REG_MAGIC) {
+ Perl_croak(aTHX_ "corrupted regexp program");
+ }
+
+ PL_reg_flags = 0;
+ PL_reg_eval_set = 0;
+ PL_reg_maxiter = 0;
+
+ if (prog->reganch & ROPT_UTF8)
+ PL_reg_flags |= RF_utf8;
+
+ /* Mark beginning of line for ^ and lookbehind. */
+ PL_regbol = startpos;
+ PL_bostr = strbeg;
+ PL_reg_sv = sv;
+
+ /* Mark end of line for $ (and such) */
+ PL_regeol = strend;
+
+ /* see how far we have to get to not match where we matched before */
+ PL_regtill = startpos+minend;
+
+ /* We start without call_cc context. */
+ PL_reg_call_cc = 0;
+
+ /* If there is a "must appear" string, look for it. */
+ s = startpos;
+
+ if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
+ MAGIC *mg;
+
+ if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
+ PL_reg_ganch = startpos;
+ else if (sv && SvTYPE(sv) >= SVt_PVMG
+ && SvMAGIC(sv)
+ && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) {
+ PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
+ if (prog->reganch & ROPT_ANCH_GPOS) {
+ if (s > PL_reg_ganch)
+ goto phooey;
+ s = PL_reg_ganch;
+ }
+ }
+ else /* pos() not defined */
+ PL_reg_ganch = strbeg;
+ }
+
+ if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) {
+ re_scream_pos_data d;
+
+ d.scream_olds = &scream_olds;
+ d.scream_pos = &scream_pos;
+ s = re_intuit_start(prog, sv, s, strend, flags, &d);
+ if (!s)
+ goto phooey; /* not present */
+ }
+
+ DEBUG_r( if (!PL_colorset) reginitcolors() );
+ DEBUG_r(PerlIO_printf(Perl_debug_log,
+ "%sMatching 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 ? "..." : ""),
+ PL_colors[0],
+ (int)(strend - startpos > 60 ? 60 : strend - startpos),
+ startpos, PL_colors[1],
+ (strend - startpos > 60 ? "..." : ""))
+ );
+
+ /* Simplest case: anchored match need be tried only once. */
+ /* [unless only anchor is BOL and multiline is set] */
+ if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
+ if (s == startpos && regtry(prog, startpos))
+ goto got_it;
+ else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
+ || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
+ {
+ char *end;
+
+ if (minlen)
+ dontbother = minlen - 1;
+ end = HOPc(strend, -dontbother) - 1;
+ /* for multiline we only have to try after newlines */
+ if (prog->check_substr) {
+ if (s == startpos)
+ goto after_try;
+ while (1) {
+ if (regtry(prog, s))
goto got_it;
+ after_try:
+ if (s >= end)
+ goto phooey;
+ if (prog->reganch & RE_USE_INTUIT) {
+ s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
+ if (!s)
+ goto phooey;
+ }
else
- tmp = doevery;
- }
- else
- tmp = 1;
- s++;
+ s++;
+ }
+ } else {
+ if (s > startpos)
+ s--;
+ while (s < end) {
+ if (*s++ == '\n') { /* don't need PL_utf8skip here */
+ if (regtry(prog, s))
+ goto got_it;
+ }
+ }
}
- break;
- case DIGITLUTF8:
- PL_reg_flags |= RF_tainted;
+ }
+ goto phooey;
+ } else if (prog->reganch & ROPT_ANCH_GPOS) {
+ if (regtry(prog, PL_reg_ganch))
+ goto got_it;
+ goto phooey;
+ }
+
+ /* Messy cases: unanchored match. */
+ if (prog->anchored_substr && prog->reganch & ROPT_SKIP) {
+ /* we have /x+whatever/ */
+ /* it must be a one character string (XXXX Except UTF?) */
+ char ch = SvPVX(prog->anchored_substr)[0];
+#ifdef DEBUGGING
+ int did_match = 0;
+#endif
+
+ if (UTF) {
while (s < strend) {
- if (isDIGIT_LC_utf8((U8*)s)) {
- if (tmp && regtry(prog, s))
- goto got_it;
- else
- tmp = doevery;
+ if (*s == ch) {
+ DEBUG_r( did_match = 1 );
+ if (regtry(prog, s)) goto got_it;
+ s += UTF8SKIP(s);
+ while (s < strend && *s == ch)
+ s += UTF8SKIP(s);
}
- else
- tmp = 1;
s += UTF8SKIP(s);
}
- break;
- case NDIGIT:
+ }
+ else {
while (s < strend) {
- if (!isDIGIT(*s)) {
- if (tmp && regtry(prog, s))
- goto got_it;
- else
- tmp = doevery;
+ if (*s == ch) {
+ DEBUG_r( did_match = 1 );
+ if (regtry(prog, s)) goto got_it;
+ s++;
+ while (s < strend && *s == ch)
+ s++;
}
- else
- tmp = 1;
s++;
}
- break;
- case NDIGITUTF8:
- while (s < strend) {
- if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
- if (tmp && regtry(prog, s))
- goto got_it;
- else
- tmp = doevery;
- }
- else
- tmp = 1;
- s += UTF8SKIP(s);
+ }
+ DEBUG_r(did_match ||
+ PerlIO_printf(Perl_debug_log,
+ "Did not find anchored character...\n"));
+ }
+ /*SUPPRESS 560*/
+ else if (prog->anchored_substr != Nullsv
+ || (prog->float_substr != Nullsv
+ && prog->float_max_offset < strend - s)) {
+ SV *must = prog->anchored_substr
+ ? prog->anchored_substr : prog->float_substr;
+ I32 back_max =
+ prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset;
+ I32 back_min =
+ prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
+ char *last = HOPc(strend, /* Cannot start after this */
+ -(I32)(CHR_SVLEN(must)
+ - (SvTAIL(must) != 0) + back_min));
+ char *last1; /* Last position checked before */
+#ifdef DEBUGGING
+ int did_match = 0;
+#endif
+
+ if (s > PL_bostr)
+ last1 = HOPc(s, -1);
+ else
+ last1 = s - 1; /* bogus */
+
+ /* XXXX check_substr already used to find `s', can optimize if
+ check_substr==must. */
+ scream_pos = -1;
+ dontbother = end_shift;
+ strend = HOPc(strend, -dontbother);
+ while ( (s <= last) &&
+ ((flags & REXEC_SCREAM)
+ ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg,
+ end_shift, &scream_pos, 0))
+ : (s = fbm_instr((unsigned char*)HOP(s, back_min),
+ (unsigned char*)strend, must,
+ PL_multiline ? FBMrf_MULTILINE : 0))) ) {
+ DEBUG_r( did_match = 1 );
+ if (HOPc(s, -back_max) > last1) {
+ last1 = HOPc(s, -back_min);
+ s = HOPc(s, -back_max);
}
- break;
- case NDIGITL:
- PL_reg_flags |= RF_tainted;
- while (s < strend) {
- if (!isDIGIT_LC(*s)) {
- if (tmp && regtry(prog, s))
+ else {
+ char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
+
+ last1 = HOPc(s, -back_min);
+ s = t;
+ }
+ if (UTF) {
+ while (s <= last1) {
+ if (regtry(prog, s))
goto got_it;
- else
- tmp = doevery;
+ s += UTF8SKIP(s);
}
- else
- tmp = 1;
- s++;
}
- break;
- case NDIGITLUTF8:
- PL_reg_flags |= RF_tainted;
- while (s < strend) {
- if (!isDIGIT_LC_utf8((U8*)s)) {
- if (tmp && regtry(prog, s))
+ else {
+ while (s <= last1) {
+ if (regtry(prog, s))
goto got_it;
- else
- tmp = doevery;
+ s++;
}
- else
- tmp = 1;
- s += UTF8SKIP(s);
}
- break;
- default:
- croak("panic: unknown regstclass %d", (int)OP(c));
- break;
}
+ DEBUG_r(did_match ||
+ PerlIO_printf(Perl_debug_log, "Did not find %s substr `%s%.*s%s'%s...\n",
+ ((must == prog->anchored_substr)
+ ? "anchored" : "floating"),
+ PL_colors[0],
+ (int)(SvCUR(must) - (SvTAIL(must)!=0)),
+ SvPVX(must),
+ PL_colors[1], (SvTAIL(must) ? "$" : "")));
+ goto phooey;
+ }
+ else if ((c = prog->regstclass)) {
+ if (minlen && PL_regkind[(U8)OP(prog->regstclass)] != EXACT)
+ /* don't bother with what can't match */
+ strend = HOPc(strend, -(minlen - 1));
+ if (find_byclass(prog, c, s, strend, startpos, 0))
+ goto got_it;
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
}
else {
dontbother = 0;
if (prog->float_substr != Nullsv) { /* Trim the end. */
char *last;
- I32 oldpos = scream_pos;
if (flags & REXEC_SCREAM) {
last = screaminstr(sv, prog->float_substr, s - strbeg,
last = strend; /* matching `$' */
}
}
- if (last == NULL) goto phooey; /* Should not happen! */
+ if (last == NULL) {
+ DEBUG_r(PerlIO_printf(Perl_debug_log,
+ "%sCan't trim the tail, match fails (should not happen)%s\n",
+ PL_colors[4],PL_colors[5]));
+ goto phooey; /* Should not happen! */
+ }
dontbother = strend - last + prog->float_min_offset;
}
if (minlen && (dontbother < minlen))
return 1;
phooey:
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n",
+ PL_colors[4],PL_colors[5]));
if (PL_reg_eval_set)
restore_pos(aTHXo_ 0);
return 0;
PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n",
(IV)(PL_stack_sp - PL_stack_base));
));
- SAVEINT(cxstack[cxstack_ix].blk_oldsp);
+ SAVEI32(cxstack[cxstack_ix].blk_oldsp);
cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
/* Otherwise OP_NEXTSTATE will free whatever on stack now. */
SAVETMPS;
/* Apparently this is not needed, judging by wantarray. */
- /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
+ /* SAVEI8(cxstack[cxstack_ix].blk_gimme);
cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
if (PL_reg_sv) {
SAVEDESTRUCTOR_X(restore_pos, 0);
}
if (!PL_reg_curpm)
- New(22,PL_reg_curpm, 1, PMOP);
+ Newz(22,PL_reg_curpm, 1, PMOP);
PL_reg_curpm->op_pmregexp = prog;
PL_reg_oldcurpm = PL_curpm;
PL_curpm = PL_reg_curpm;
}
sayNO;
case SBOL:
- if (locinput == PL_regbol && PL_regprev == '\n')
+ if (locinput == PL_bostr)
break;
sayNO;
case GPOS:
nextchr = UCHARAT(locinput);
break;
}
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
sayNO;
nextchr = UCHARAT(++locinput);
break;
case REG_ANY:
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n')
sayNO;
nextchr = UCHARAT(++locinput);
break;
while (s < e) {
if (l >= PL_regeol)
sayNO;
- if (utf8_to_uv((U8*)s, 0) != (c1 ?
+ if (utf8_to_uv((U8*)s, e - s, 0, 0) != (c1 ?
toLOWER_utf8((U8*)l) :
toLOWER_LC_utf8((U8*)l)))
{
nextchr = UCHARAT(locinput);
break;
case ANYOFUTF8:
- s = MASK(scan);
if (!REGINCLASSUTF8(scan, (U8*)locinput))
sayNO;
if (locinput >= PL_regeol)
nextchr = UCHARAT(locinput);
break;
case ANYOF:
- s = MASK(scan);
if (nextchr < 0)
nextchr = UCHARAT(locinput);
- if (!REGINCLASS(s, nextchr))
+ if (!REGINCLASS(scan, nextchr))
sayNO;
if (!nextchr && locinput >= PL_regeol)
sayNO;
case NBOUNDUTF8:
/* was last char in word? */
ln = (locinput != PL_regbol)
- ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
+ ? utf8_to_uv(reghop((U8*)locinput, -1),
+ PL_regeol - locinput, 0, 0) : PL_regprev;
if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
ln = isALNUM_uni(ln);
n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case SPACE:
- if (!nextchr && locinput >= PL_regeol)
+ if (!nextchr)
sayNO;
if (!(OP(scan) == SPACE
? isSPACE(nextchr) : isSPACE_LC(nextchr)))
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case SPACEUTF8:
- if (!nextchr && locinput >= PL_regeol)
+ if (!nextchr)
sayNO;
if (nextchr & 0x80) {
if (!(OP(scan) == SPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ ? swash_fetch(PL_utf8_space, (U8*)locinput)
: isSPACE_LC_utf8((U8*)locinput)))
{
sayNO;
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NSPACE:
- if (!nextchr)
+ if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (OP(scan) == SPACE
+ if (OP(scan) == NSPACE
? isSPACE(nextchr) : isSPACE_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NSPACEUTF8:
- if (!nextchr)
+ if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
if (OP(scan) == NSPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ ? swash_fetch(PL_utf8_space, (U8*)locinput)
: isSPACE_LC_utf8((U8*)locinput))
{
sayNO;
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case DIGIT:
- if (!nextchr && locinput >= PL_regeol)
+ if (!nextchr)
sayNO;
if (!(OP(scan) == DIGIT
? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
if (!nextchr)
sayNO;
if (nextchr & 0x80) {
- if (OP(scan) == NDIGITUTF8
- ? swash_fetch(PL_utf8_digit,(U8*)locinput)
- : isDIGIT_LC_utf8((U8*)locinput))
+ if (!(OP(scan) == DIGITUTF8
+ ? swash_fetch(PL_utf8_digit, (U8*)locinput)
+ : isDIGIT_LC_utf8((U8*)locinput)))
{
sayNO;
}
nextchr = UCHARAT(locinput);
break;
}
- if (!isDIGIT(nextchr))
+ if (!(OP(scan) == DIGITUTF8
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NDIGIT:
- if (!nextchr)
+ if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (OP(scan) == DIGIT
+ if (OP(scan) == NDIGIT
? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
- if (swash_fetch(PL_utf8_digit,(U8*)locinput))
+ if (OP(scan) == NDIGITUTF8
+ ? swash_fetch(PL_utf8_digit, (U8*)locinput)
+ : isDIGIT_LC_utf8((U8*)locinput))
+ {
sayNO;
+ }
locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (isDIGIT(nextchr))
+ if (OP(scan) == NDIGITUTF8
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
n = ARG(scan);
PL_op = (OP_4tree*)PL_regdata->data[n];
- DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", (UV)PL_op) );
+ DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]);
PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
I32 onpar = PL_regnpar;
pm.op_pmflags = 0;
+ pm.op_pmdynflags = (UTF||DO_UTF8(ret) ? PMdf_UTF8 : 0);
re = CALLREGCOMP(aTHX_ t, t + len, &pm);
if (!(SvFLAGS(ret)
& (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
case CURLYX: {
CURCUR cc;
CHECKPOINT cp = PL_savestack_ix;
+ /* No need to save/restore up to this paren */
+ I32 parenfloor = scan->flags;
if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
next += ARG(next);
cc.oldcc = PL_regcc;
PL_regcc = &cc;
- cc.parenfloor = *PL_reglastparen;
+ /* XXXX Probably it is better to teach regpush to support
+ parenfloor > PL_regsize... */
+ if (parenfloor > *PL_reglastparen)
+ parenfloor = *PL_reglastparen; /* Pessimization... */
+ cc.parenfloor = parenfloor;
cc.cur = -1;
cc.min = ARG1(scan);
cc.max = ARG2(scan);
PL_regcc = cc;
if (n >= cc->max) { /* Maximum greed exceeded? */
- if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
+ if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
&& !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded",
+ Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded",
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
REPORT_CODE_OFF+PL_regindent*2, "")
);
}
- if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
+ if (ckWARN(WARN_REGEXP) && n >= REG_INFTY
&& !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded",
+ Perl_warner(aTHX_ WARN_REGEXP, "%s limit (%d) exceeded",
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
*PL_reglastparen = n;
scan = next;
/*SUPPRESS 560*/
- if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next)))
+ if ((n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))))
next += n;
else
next = NULL;
locinput = PL_reginput;
DEBUG_r(
PerlIO_printf(Perl_debug_log,
- "%*s matched %d times, len=%"IVdf"...\n",
+ "%*s matched %"IVdf" times, len=%"IVdf"...\n",
(int)(REPORT_CODE_OFF+PL_regindent*2), "",
- n, (IV)l)
+ (IV) n, (IV)l)
);
if (n >= ln) {
if (PL_regkind[(U8)OP(next)] == EXACT) {
sayNO;
}
/* PL_reginput == locinput now */
- if (paren) {
- if (ln) {
- PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr;
- PL_regendp[paren] = locinput - PL_bostr;
- }
- else
- PL_regendp[paren] = -1;
- }
- if (regmatch(next))
- sayYES;
+ TRYPAREN(paren, ln, locinput);
PL_reginput = locinput; /* Could be reset... */
REGCP_UNWIND;
/* Couldn't or didn't -- move forward. */
UCHARAT(PL_reginput) == c1 ||
UCHARAT(PL_reginput) == c2)
{
- if (paren) {
- if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
- PL_regendp[paren] = PL_reginput - PL_bostr;
- }
- else
- PL_regendp[paren] = -1;
- }
- if (regmatch(next))
- sayYES;
+ TRYPAREN(paren, n, PL_reginput);
REGCP_UNWIND;
}
/* Couldn't or didn't -- move forward. */
n = regrepeat(scan, n);
locinput = PL_reginput;
if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
- (!PL_multiline || OP(next) == SEOL))
+ (!PL_multiline || OP(next) == SEOL || OP(next) == EOS)) {
ln = n; /* why back off? */
+ /* ...because $ and \Z can match before *and* after
+ newline at the end. Consider "\n\n" =~ /\n+\Z\n/.
+ We should back off by one in this case. */
+ if (UCHARAT(PL_reginput - 1) == '\n' && OP(next) != EOS)
+ ln--;
+ }
REGCP_SET;
if (paren) {
while (n >= ln) {
UCHARAT(PL_reginput) == c1 ||
UCHARAT(PL_reginput) == c2)
{
- if (paren && n) {
- if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
- PL_regendp[paren] = PL_reginput - PL_bostr;
- }
- else
- PL_regendp[paren] = -1;
- }
- if (regmatch(next))
- sayYES;
+ TRYPAREN(paren, n, PL_reginput);
REGCP_UNWIND;
}
/* Couldn't or didn't -- back up. */
UCHARAT(PL_reginput) == c1 ||
UCHARAT(PL_reginput) == c2)
{
- if (regmatch(next))
- sayYES;
+ TRYPAREN(paren, n, PL_reginput);
REGCP_UNWIND;
}
/* Couldn't or didn't -- back up. */
break;
default:
PerlIO_printf(Perl_error_log, "%"UVxf" %d\n",
- (UV)scan, OP(scan));
+ PTR2UV(scan), OP(scan));
Perl_croak(aTHX_ "regexp memory corruption");
}
scan = next;
{
dTHR;
register char *scan;
- register char *opnd;
register I32 c;
register char *loceol = PL_regeol;
register I32 hardcount = 0;
}
break;
case ANYOF:
- opnd = MASK(p);
- while (scan < loceol && REGINCLASS(opnd, *scan))
+ while (scan < loceol && REGINCLASS(p, *scan))
scan++;
break;
case ALNUM:
*/
STATIC bool
-S_reginclass(pTHX_ register char *p, register I32 c)
+S_reginclass(pTHX_ register regnode *p, register I32 c)
{
dTHR;
char flags = ANYOF_FLAGS(p);
(ANYOF_CLASS_TEST(p, ANYOF_UPPER) && isUPPER_LC(c)) ||
(ANYOF_CLASS_TEST(p, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
(ANYOF_CLASS_TEST(p, ANYOF_XDIGIT) && isXDIGIT(c)) ||
- (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c))
+ (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_PSXSPC) && isPSXSPC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NPSXSPC) && !isPSXSPC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_BLANK) && isBLANK(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NBLANK) && !isBLANK(c))
) /* How's that for a conditional? */
{
match = TRUE;
if (swash_fetch(sv, p))
match = TRUE;
else if (flags & ANYOF_FOLD) {
- I32 cf;
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
if (flags & ANYOF_LOCALE) {
PL_reg_flags |= RF_tainted;
uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p));
}
#ifdef PERL_OBJECT
-#define NO_XSLOCKS
#include "XSUB.h"
#endif
PL_curpm = PL_reg_oldcurpm;
}
}
-