#define STATIC static
#endif
-#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
+#define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c)))
/*
* Forwards.
(IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp)
STATIC char *
-S_regcppop(pTHX_ regexp *rex)
+S_regcppop(pTHX_ const regexp *rex)
{
dVAR;
I32 i;
- U32 paren = 0;
char *input;
GET_RE_DEBUG_FLAGS_DECL;
for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
i > 0; i -= REGCP_PAREN_ELEMS) {
I32 tmps;
- paren = (U32)SSPOPINT;
+ U32 paren = (U32)SSPOPINT;
PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
PL_regstartp[paren] = SSPOPINT;
tmps = SSPOPINT;
* building DynaLoader will fail:
* "Error: '*' not in typemap in DynaLoader.xs, line 164"
* --jhi */
- for (paren = *PL_reglastparen + 1; (I32)paren <= rex->nparens; paren++) {
- if ((I32)paren > PL_regsize)
- PL_regstartp[paren] = -1;
- PL_regendp[paren] = -1;
+ for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) {
+ if (i > PL_regsize)
+ PL_regstartp[i] = -1;
+ PL_regendp[i] = -1;
}
#endif
return input;
: strend);
t = s;
- PL_reg_re = prog;
- s = find_byclass(prog, prog->regstclass, s, endpos, 1);
+ s = find_byclass(prog, prog->regstclass, s, endpos, NULL);
if (!s) {
#ifdef DEBUGGING
const char *what = NULL;
}
/* We know what class REx starts with. Try to find this position... */
+/* if reginfo is NULL, its a dryrun */
+
STATIC char *
-S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, const char *strend, I32 norun)
+S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, const char
+*strend, const regmatch_info *reginfo)
{
dVAR;
const I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
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)))
+ reginclass(prog, c, (U8*)s, 0, do_utf8) :
+ REGINCLASS(prog, c, (U8*)s)) {
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
while (s < strend) {
STRLEN skip = 1;
- if (REGINCLASS(c, (U8*)s) ||
+ if (REGINCLASS(prog, c, (U8*)s) ||
(ANYOF_FOLD_SHARP_S(c, s, strend) &&
/* The assignment of 2 is intentional:
* for the folded sharp s, the skip is 2. */
(skip = SHARP_S_SKIP))) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
break;
case CANY:
while (s < strend) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
do_exactf:
e = HOP3c(strend, -((I32)lnc), s);
- if (norun && e < s)
+ if (!reginfo && e < s)
e = s; /* Due to minlen logic of intuit() */
/* The idea in the EXACTF* cases is to first find the
&& (ln == len ||
ibcmp_utf8(s, (char **)0, 0, do_utf8,
m, (char **)0, ln, (bool)UTF))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
else {
U8 foldbuf[UTF8_MAXBYTES_CASE+1];
(char **)0, foldlen, do_utf8,
m,
(char **)0, ln, (bool)UTF))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
}
s += len;
&& (ln == len ||
ibcmp_utf8(s, (char **)0, 0, do_utf8,
m, (char **)0, ln, (bool)UTF))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
else {
U8 foldbuf[UTF8_MAXBYTES_CASE+1];
(char **)0, foldlen, do_utf8,
m,
(char **)0, ln, (bool)UTF))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
}
s += len;
&& (ln == 1 || !(OP(c) == EXACTF
? ibcmp(s, m, ln)
: ibcmp_locale(s, m, ln)))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
s++;
}
&& (ln == 1 || !(OP(c) == EXACTF
? ibcmp(s, m, ln)
: ibcmp_locale(s, m, ln)))
- && (norun || regtry(prog, s)) )
+ && (!reginfo || regtry(reginfo, s)) )
goto got_it;
s++;
}
isALNUM_LC_utf8((U8*)s)))
{
tmp = !tmp;
- if ((norun || regtry(prog, s)))
+ if ((!reginfo || regtry(reginfo, s)))
goto got_it;
}
s += uskip;
if (tmp ==
!(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) {
tmp = !tmp;
- if ((norun || regtry(prog, s)))
+ if ((!reginfo || regtry(reginfo, s)))
goto got_it;
}
s++;
}
}
- if ((!prog->minlen && tmp) && (norun || regtry(prog, s)))
+ if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, s)))
goto got_it;
break;
case NBOUNDL:
swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
isALNUM_LC_utf8((U8*)s)))
tmp = !tmp;
- else if ((norun || regtry(prog, s)))
+ else if ((!reginfo || regtry(reginfo, s)))
goto got_it;
s += uskip;
}
if (tmp ==
!(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s)))
tmp = !tmp;
- else if ((norun || regtry(prog, s)))
+ else if ((!reginfo || regtry(reginfo, s)))
goto got_it;
s++;
}
}
- if ((!prog->minlen && !tmp) && (norun || regtry(prog, s)))
+ if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, s)))
goto got_it;
break;
case ALNUM:
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isALNUM(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (isALNUM_LC_utf8((U8*)s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isALNUM_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isALNUM(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!isALNUM_LC_utf8((U8*)s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isALNUM_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isSPACE(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isSPACE_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isSPACE(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isSPACE_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isDIGIT(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (isDIGIT_LC_utf8((U8*)s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (isDIGIT_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
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)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isDIGIT(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
if (do_utf8) {
while (s + (uskip = UTF8SKIP(s)) <= strend) {
if (!isDIGIT_LC_utf8((U8*)s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
else {
while (s < strend) {
if (!isDIGIT_LC(*s)) {
- if (tmp && (norun || regtry(prog, s)))
+ if (tmp && (!reginfo || regtry(reginfo, s)))
goto got_it;
else
tmp = doevery;
char *scream_olds = NULL;
SV* oreplsv = GvSV(PL_replgv);
const bool do_utf8 = DO_UTF8(sv);
- const I32 multiline = prog->reganch & PMf_MULTILINE;
+ I32 multiline;
#ifdef DEBUGGING
- SV * const dsv0 = PERL_DEBUG_PAD_ZERO(0);
- SV * const dsv1 = PERL_DEBUG_PAD_ZERO(1);
+ SV* dsv0;
+ SV* dsv1;
#endif
+ regmatch_info reginfo; /* create some info to pass to regtry etc */
GET_RE_DEBUG_FLAGS_DECL;
PERL_UNUSED_ARG(data);
- RX_MATCH_UTF8_set(prog,do_utf8);
-
- PL_reg_re = prog;
-#ifdef DEBUGGING
- PL_regnarrate = DEBUG_r_TEST;
-#endif
/* Be paranoid... */
if (prog == NULL || startpos == NULL) {
return 0;
}
+ multiline = prog->reganch & PMf_MULTILINE;
+ reginfo.prog = prog;
+
+#ifdef DEBUGGING
+ dsv0 = PERL_DEBUG_PAD_ZERO(0);
+ dsv1 = PERL_DEBUG_PAD_ZERO(1);
+#endif
+
+ RX_MATCH_UTF8_set(prog, do_utf8);
+
minlen = prog->minlen;
if (strend - startpos < minlen) {
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log,
PL_reg_flags |= RF_utf8;
/* Mark beginning of line for ^ and lookbehind. */
- PL_regbol = startpos;
+ reginfo.bol = startpos; /* XXX not used ??? */
PL_bostr = strbeg;
- PL_reg_sv = sv;
+ reginfo.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;
+ reginfo.till = startpos+minend;
/* If there is a "must appear" string, look for it. */
s = startpos;
- if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to have PL_reg_ganch */
+ if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */
MAGIC *mg;
if (flags & REXEC_IGNOREPOS) /* Means: check only at start */
- PL_reg_ganch = startpos;
+ reginfo.ganch = startpos;
else if (sv && SvTYPE(sv) >= SVt_PVMG
&& SvMAGIC(sv)
&& (mg = mg_find(sv, PERL_MAGIC_regex_global))
&& mg->mg_len >= 0) {
- PL_reg_ganch = strbeg + mg->mg_len; /* Defined pos() */
+ reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */
if (prog->reganch & ROPT_ANCH_GPOS) {
- if (s > PL_reg_ganch)
+ if (s > reginfo.ganch)
goto phooey;
- s = PL_reg_ganch;
+ s = reginfo.ganch;
}
}
else /* pos() not defined */
- PL_reg_ganch = strbeg;
+ reginfo.ganch = strbeg;
}
if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) {
/* 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))
+ if (s == startpos && regtry(®info, startpos))
goto got_it;
else if (multiline || (prog->reganch & ROPT_IMPLICIT)
|| (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
if (s == startpos)
goto after_try;
while (1) {
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
after_try:
if (s >= end)
s--;
while (s < end) {
if (*s++ == '\n') { /* don't need PL_utf8skip here */
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
}
}
}
goto phooey;
} else if (prog->reganch & ROPT_ANCH_GPOS) {
- if (regtry(prog, PL_reg_ganch))
+ if (regtry(®info, reginfo.ganch))
goto got_it;
goto phooey;
}
while (s < strend) {
if (*s == ch) {
DEBUG_EXECUTE_r( did_match = 1 );
- if (regtry(prog, s)) goto got_it;
+ if (regtry(®info, s)) goto got_it;
s += UTF8SKIP(s);
while (s < strend && *s == ch)
s += UTF8SKIP(s);
while (s < strend) {
if (*s == ch) {
DEBUG_EXECUTE_r( did_match = 1 );
- if (regtry(prog, s)) goto got_it;
+ if (regtry(®info, s)) goto got_it;
s++;
while (s < strend && *s == ch)
s++;
}
if (do_utf8) {
while (s <= last1) {
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
s += UTF8SKIP(s);
}
}
else {
while (s <= last1) {
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
s++;
}
int len0;
int len1;
- regprop(prop, c);
+ regprop(prog, prop, c);
s0 = UTF ?
pv_uni_display(dsv0, (U8*)SvPVX_const(prop), SvCUR(prop), 60,
UNI_DISPLAY_REGEX) :
len0, len0, s0,
len1, len1, s1);
});
- if (find_byclass(prog, c, s, strend, 0))
+ if (find_byclass(prog, c, s, strend, ®info))
goto got_it;
DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n"));
}
/* We don't know much -- general case. */
if (do_utf8) {
for (;;) {
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
if (s >= strend)
break;
}
else {
do {
- if (regtry(prog, s))
+ if (regtry(®info, s))
goto got_it;
} while (s++ < strend);
}
- regtry - try match at specific point
*/
STATIC I32 /* 0 failure, 1 success */
-S_regtry(pTHX_ regexp *prog, char *startpos)
+S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos)
{
dVAR;
- register I32 i;
register I32 *sp;
register I32 *ep;
CHECKPOINT lastcp;
+ regexp *prog = reginfo->prog;
GET_RE_DEBUG_FLAGS_DECL;
#ifdef DEBUGGING
/* SAVEI8(cxstack[cxstack_ix].blk_gimme);
cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
- if (PL_reg_sv) {
+ if (reginfo->sv) {
/* Make $_ available to executed code. */
- if (PL_reg_sv != DEFSV) {
+ if (reginfo->sv != DEFSV) {
SAVE_DEFSV;
- DEFSV = PL_reg_sv;
+ DEFSV = reginfo->sv;
}
- if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
- && (mg = mg_find(PL_reg_sv, PERL_MAGIC_regex_global)))) {
+ if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv)
+ && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) {
/* prepare for quick setting of pos */
#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,
+ mg = sv_magicext(reginfo->sv, (SV*)0, PERL_MAGIC_regex_global,
&PL_vtbl_mglob, NULL, 0);
mg->mg_len = -1;
}
sp = prog->startp;
ep = prog->endp;
if (prog->nparens) {
+ register I32 i;
for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) {
*++sp = -1;
*++ep = -1;
}
#endif
REGCP_SET(lastcp);
- if (regmatch(prog, prog->program + 1)) {
+ if (regmatch(reginfo, prog->program + 1)) {
prog->endp[0] = PL_reginput - PL_bostr;
return 1;
}
regnode *next;
char *locinput;
I32 nextchr;
+ int minmod;
#ifdef DEBUGGING
int regindent;
#endif
goto start_recurse; \
resume_point_##where:
+
+/* push a new regex state. Set newst to point to it */
+
+#define PUSH_STATE(newst, resume) \
+ depth++; \
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "PUSH STATE(%d)\n", depth)); \
+ st->scan = scan; \
+ st->next = next; \
+ st->n = n; \
+ st->locinput = locinput; \
+ st->resume_state = resume; \
+ newst = st+1; \
+ if (newst > &(PL_regmatch_slab->states[PERL_REGMATCH_SLAB_SLOTS-1])) \
+ newst = S_push_slab(aTHX); \
+ PL_regmatch_state = newst; \
+ newst->cc = 0; \
+ newst->minmod = 0; \
+ newst->sw = 0; \
+ newst->logical = 0; \
+ newst->unwind = 0; \
+ locinput = PL_reginput; \
+ nextchr = UCHARAT(locinput);
+
+#define POP_STATE \
+ DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "POP STATE(%d)\n", depth)); \
+ depth--; \
+ st--; \
+ if (st < &PL_regmatch_slab->states[0]) { \
+ PL_regmatch_slab = PL_regmatch_slab->prev; \
+ st = &PL_regmatch_slab->states[PERL_REGMATCH_SLAB_SLOTS-1]; \
+ } \
+ PL_regmatch_state = st; \
+ scan = st->scan; \
+ next = st->next; \
+ n = st->n; \
+ locinput = st->locinput; \
+ nextchr = UCHARAT(locinput);
+
/*
- regmatch - main matching routine
*
STATIC I32 /* 0 failure, 1 success */
-S_regmatch(pTHX_ regexp *rex, regnode *prog)
+S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
{
dVAR;
register const bool do_utf8 = PL_reg_match_utf8;
const U32 uniflags = UTF8_ALLOW_DEFAULT;
+ regexp *rex = reginfo->prog;
+
regmatch_slab *orig_slab;
regmatch_state *orig_state;
bool result; /* return value of S_regmatch */
regnode *inner; /* Next node in internal branch. */
int depth = 0; /* depth of recursion */
+ regmatch_state *newst; /* when pushing a state, this is the new one */
+ regmatch_state *cur_eval = NULL; /* most recent (??{}) state */
#ifdef DEBUGGING
SV *re_debug_flags = NULL;
pref0_len = 0;
if (pref0_len > pref_len)
pref0_len = pref_len;
- regprop(prop, scan);
+ regprop(rex, prop, scan);
{
const char * const s0 =
do_utf8 && OP(scan) != CANY ?
case BOL:
if (locinput == PL_bostr)
{
- /* regtill = regbol; */
+ /* reginfo->till = reginfo->bol; */
break;
}
sayNO;
break;
sayNO;
case GPOS:
- if (locinput == PL_reg_ganch)
+ if (locinput == reginfo->ganch)
break;
sayNO;
case EOL:
/* what trie are we using right now */
reg_trie_data *trie
- = (reg_trie_data*)PL_reg_re->data->data[ ARG( scan ) ];
+ = (reg_trie_data*)rex->data->data[ ARG( scan ) ];
st->u.trie.accepted = 0; /* how many accepting states we have seen */
result = 0;
if ( st->u.trie.accepted == 1 ) {
DEBUG_EXECUTE_r({
- SV **tmp = av_fetch( trie->words, st->u.trie.accept_buff[ 0 ].wordnum-1, 0 );
+ SV ** const tmp = av_fetch( trie->words, st->u.trie.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],
}
DEBUG_EXECUTE_r({
reg_trie_data * const trie = (reg_trie_data*)
- PL_reg_re->data->data[ARG(scan)];
+ rex->data->data[ARG(scan)];
SV ** const tmp = av_fetch( trie->words, st->u.trie.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],
if (do_utf8) {
STRLEN inclasslen = PL_regeol - locinput;
- if (!reginclass(scan, (U8*)locinput, &inclasslen, do_utf8))
+ if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8))
sayNO_ANYOF;
if (locinput >= PL_regeol)
sayNO;
else {
if (nextchr < 0)
nextchr = UCHARAT(locinput);
- if (!REGINCLASS(scan, (U8*)locinput))
+ if (!REGINCLASS(rex, scan, (U8*)locinput))
sayNO_ANYOF;
if (!nextchr && locinput >= PL_regeol)
sayNO;
OP_4tree * const oop = PL_op;
COP * const ocurcop = PL_curcop;
PAD *old_comppad;
- struct regexp * const oreg = PL_reg_re;
n = ARG(scan);
- PL_op = (OP_4tree*)PL_reg_re->data->data[n];
+ PL_op = (OP_4tree*)rex->data->data[n];
DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) );
- PAD_SAVE_LOCAL(old_comppad, (PAD*)PL_reg_re->data->data[n + 2]);
+ PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]);
PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
CALLRUNOPS(aTHX); /* Scalar context. */
PL_op = oop;
PAD_RESTORE_LOCAL(old_comppad);
PL_curcop = ocurcop;
- PL_reg_re = oreg;
if (!st->logical) {
/* /(?{...})/ */
sv_setsv(save_scalar(PL_replgv), ret);
}
if (st->logical == 2) { /* Postponed subexpression: /(??{...})/ */
regexp *re;
- re_cc_state state;
- int toggleutf;
-
{
/* extract RE object from returned value; compiling if
* necessary */
PL_regsize = osize;
}
}
+
+ /* run the pattern returned from (??{...}) */
+
DEBUG_EXECUTE_r(
PerlIO_printf(Perl_debug_log,
"Entering embedded \"%s%.60s%s%s\"\n",
PL_colors[1],
(strlen(re->precomp) > 60 ? "..." : ""))
);
- state.node = next;
- state.prev = PL_reg_call_cc;
- state.cc = st->cc;
- state.re = PL_reg_re;
- st->cc = 0;
-
st->u.eval.cp = regcppush(0); /* Save *all* the positions. */
REGCP_SET(st->u.eval.lastcp);
- PL_reg_re = re;
- state.ss = PL_savestack_ix;
*PL_reglastparen = 0;
*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, re->program + 1)) {
- /* Even though we succeeded, we need to restore
- global variables, since we may be wrapped inside
- SUSPEND, thus the match may be not finished yet. */
-
- /* XXXX Do this only if SUSPENDed? */
- PL_reg_call_cc = state.prev;
- st->cc = state.cc;
- PL_reg_re = state.re;
-
- if (toggleutf) PL_reg_flags ^= RF_utf8;
-
- /* XXXX This is too dramatic a measure... */
- PL_reg_maxiter = 0;
-
- /* These are needed even if not SUSPEND. */
- ReREFCNT_dec(re);
- regcpblow(st->u.eval.cp);
- sayYES;
- }
- ReREFCNT_dec(re);
- REGCP_UNWIND(st->u.eval.lastcp);
- regcppop(rex);
- PL_reg_call_cc = state.prev;
- st->cc = state.cc;
- PL_reg_re = state.re;
- if (toggleutf) PL_reg_flags ^= RF_utf8;
/* XXXX This is too dramatic a measure... */
PL_reg_maxiter = 0;
st->logical = 0;
- sayNO;
+ st->u.eval.toggleutf = ((PL_reg_flags & RF_utf8) != 0) ^
+ ((re->reganch & ROPT_UTF8) != 0);
+ if (st->u.eval.toggleutf) PL_reg_flags ^= RF_utf8;
+ st->u.eval.prev_rex = rex;
+ rex = re;
+
+ st->u.eval.prev_eval = cur_eval;
+ st->u.eval.prev_slab = PL_regmatch_slab;
+ st->u.eval.depth = depth;
+ cur_eval = st;
+ PUSH_STATE(newst, resume_EVAL);
+ st = newst;
+
+ /* now continue from first node in postoned RE */
+ next = re->program + 1;
+ break;
/* NOTREACHED */
}
/* /(?(?{...})X|Y)/ */
/* No need to save/restore up to this paren */
I32 parenfloor = scan->flags;
- if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */
+ if (next && (OP(PREVOPER(next)) == NOTHING)) /* LONGJMP */
next += ARG(next);
/* XXXX Probably it is better to teach regpush to support
parenfloor > PL_regsize... */
{
I32 type;
type = OP(scan);
- if (OP(next) != type) /* No choice. */
+ if (!next || OP(next) != type) /* No choice. */
next = inner; /* Avoid recursion. */
else {
const I32 lastparen = *PL_reglastparen;
uw->next = next;
uw->locinput = locinput;
uw->nextchr = nextchr;
+ uw->minmod = st->minmod;
#ifdef DEBUGGING
uw->regindent = ++PL_regindent;
#endif
PL_reginput = locinput;
if (st->minmod) {
st->minmod = 0;
- if (st->ln && regrepeat(scan, st->ln) < st->ln)
+ if (st->ln && regrepeat(rex, scan, st->ln) < st->ln)
sayNO;
locinput = PL_reginput;
REGCP_SET(st->u.plus.lastcp);
/* PL_reginput == old now */
if (locinput != st->u.plus.old) {
st->ln = 1; /* Did some */
- if (regrepeat(scan, st->u.plus.count) < st->u.plus.count)
+ if (regrepeat(rex, scan, st->u.plus.count) < st->u.plus.count)
sayNO;
}
/* PL_reginput == locinput now */
}
/* Couldn't or didn't -- move forward. */
PL_reginput = locinput;
- if (regrepeat(scan, 1)) {
+ if (regrepeat(rex, scan, 1)) {
st->ln++;
locinput = PL_reginput;
}
}
}
else {
- n = regrepeat(scan, n);
+ n = regrepeat(rex, scan, n);
locinput = PL_reginput;
if (st->ln < n && PL_regkind[(U8)OP(next)] == EOL &&
(OP(next) != MEOL ||
sayNO;
break;
case END:
- if (PL_reg_call_cc) {
- st->u.end.cur_call_cc = PL_reg_call_cc;
- st->u.end.end_re = PL_reg_re;
+ if (cur_eval) {
+ /* we have successfully completed the execution of a
+ * postponed re. Pop all states back to the last EVAL
+ * then continue with the node following the (??{...})
+ */
+
+ /* this simulates a POP_STATE, except that it pops several
+ * levels, and doesn't restore locinput */
- /* Save *all* the positions. */
- st->u.end.cp = regcppush(0);
- REGCP_SET(st->u.end.lastcp);
+ st = cur_eval;
+ PL_regmatch_slab = st->u.eval.prev_slab;
+ cur_eval = st->u.eval.prev_eval;
+ depth = st->u.eval.depth;
- /* Restore parens of the caller. */
+ PL_regmatch_state = st;
+ scan = st->scan;
+ next = st->next;
+ n = st->n;
+
+ if (st->u.eval.toggleutf)
+ PL_reg_flags ^= RF_utf8;
+ ReREFCNT_dec(rex);
+ rex = st->u.eval.prev_rex;
+ /* XXXX This is too dramatic a measure... */
+ PL_reg_maxiter = 0;
+
+ /* Restore parens of the caller without popping the
+ * savestack */
{
I32 tmp = PL_savestack_ix;
- PL_savestack_ix = PL_reg_call_cc->ss;
+ PL_savestack_ix = st->u.eval.lastcp;
regcppop(rex);
PL_savestack_ix = tmp;
}
- /* Make position available to the callcc. */
- PL_reginput = locinput;
- PL_reg_re = PL_reg_call_cc->re;
- st->u.end.savecc = st->cc;
- st->cc = PL_reg_call_cc->cc;
- PL_reg_call_cc = PL_reg_call_cc->prev;
- REGMATCH(st->u.end.cur_call_cc->node, END);
- /*** all unsaved local vars undefined at this point */
- if (result) {
- PL_reg_call_cc = st->u.end.cur_call_cc;
- regcpblow(st->u.end.cp);
- sayYES;
- }
- REGCP_UNWIND(st->u.end.lastcp);
- regcppop(rex);
- PL_reg_call_cc = st->u.end.cur_call_cc;
- st->cc = st->u.end.savecc;
- PL_reg_re = st->u.end.end_re;
+ PL_reginput = locinput;
+ /* resume at node following the (??{...}) */
+ break;
- DEBUG_EXECUTE_r(
- PerlIO_printf(Perl_debug_log,
- "%*s continuation failed...\n",
- REPORT_CODE_OFF+PL_regindent*2, "")
- );
- sayNO_SILENT;
}
- if (locinput < PL_regtill) {
+
+ if (locinput < reginfo->till) {
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),
- (long)(PL_regtill - PL_reg_starttry),
+ (long)(reginfo->till - PL_reg_starttry),
PL_colors[5]));
sayNO_FINAL; /* Cannot match: too short. */
}
oldst->next = next;
oldst->n = n;
oldst->locinput = locinput;
- oldst->reg_call_cc = PL_reg_call_cc;
st->cc = oldst->cc;
locinput = PL_reginput;
}
}
+
+
/*
* We get here only if there's trouble -- normally "case END" is
* the terminating point.
#endif
result = 1;
- goto exit_level;
+ /* XXX this is duplicate(ish) code to that in the do_no section.
+ * eventually a yes should just pop the whole stack */
+ if (depth) {
+ /* restore previous state and re-enter */
+ POP_STATE;
+
+ switch (st->resume_state) {
+ case resume_TRIE1:
+ goto resume_point_TRIE1;
+ case resume_TRIE2:
+ goto resume_point_TRIE2;
+ case resume_EVAL:
+ break;
+ 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;
+ default:
+ Perl_croak(aTHX_ "regexp resume memory corruption");
+ }
+ }
+ goto final_exit;
no:
DEBUG_EXECUTE_r(
PL_regendp[n] = -1;
*PL_reglastparen = n;
scan = next = uwb->next;
+ st->minmod = uwb->minmod;
if ( !scan ||
OP(scan) != (uwb->type == RE_UNWIND_BRANCH
? BRANCH : BRANCHJ) ) { /* Failure */
}
/* NOTREACHED */
}
+
#ifdef DEBUGGING
PL_regindent--;
#endif
result = 0;
-exit_level:
-
- if (depth--) {
- /* restore previous state and re-enter */
- st--;
- if (st < &PL_regmatch_slab->states[0]) {
- PL_regmatch_slab = PL_regmatch_slab->prev;
- st = &PL_regmatch_slab->states[PERL_REGMATCH_SLAB_SLOTS-1];
- }
- PL_regmatch_state = st;
-
- PL_reg_call_cc = st->reg_call_cc;
- scan = st->scan;
- next = st->next;
- n = st->n;
- locinput = st->locinput;
- nextchr = UCHARAT(locinput);
+ if (depth) {
+ /* there's a previous state to backtrack to */
+ POP_STATE;
switch (st->resume_state) {
case resume_TRIE1:
goto resume_point_TRIE1;
case resume_TRIE2:
goto resume_point_TRIE2;
+ case resume_EVAL:
+ /* we have failed an (??{...}). Restore state to the outer re
+ * then re-throw the failure */
+ if (st->u.eval.toggleutf)
+ PL_reg_flags ^= RF_utf8;
+ ReREFCNT_dec(rex);
+ rex = st->u.eval.prev_rex;
+ cur_eval = st->u.eval.prev_eval;
+
+ /* XXXX This is too dramatic a measure... */
+ PL_reg_maxiter = 0;
+
+ PL_reginput = locinput;
+ REGCP_UNWIND(st->u.eval.lastcp);
+ regcppop(rex);
+ goto do_no;
+
case resume_CURLYX:
goto resume_point_CURLYX;
case resume_WHILEM1:
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 */
}
+
+final_exit:
+
/* restore original high-water mark */
PL_regmatch_slab = orig_slab;
PL_regmatch_state = orig_state;
* rather than incrementing count on every character. [Er, except utf8.]]
*/
STATIC I32
-S_regrepeat(pTHX_ const regnode *p, I32 max)
+S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max)
{
dVAR;
register char *scan;
if (do_utf8) {
loceol = PL_regeol;
while (hardcount < max && scan < loceol &&
- reginclass(p, (U8*)scan, 0, do_utf8)) {
+ reginclass(prog, p, (U8*)scan, 0, do_utf8)) {
scan += UTF8SKIP(scan);
hardcount++;
}
} else {
- while (scan < loceol && REGINCLASS(p, (U8*)scan))
+ while (scan < loceol && REGINCLASS(prog, p, (U8*)scan))
scan++;
}
break;
SV * const prop = sv_newmortal();
GET_RE_DEBUG_FLAGS;
DEBUG_EXECUTE_r({
- regprop(prop, p);
+ regprop(prog, prop, p);
PerlIO_printf(Perl_debug_log,
"%*s %s can match %"IVdf" times out of %"IVdf"...\n",
REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max);
*/
SV *
-Perl_regclass_swash(pTHX_ register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
+Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp)
{
dVAR;
SV *sw = NULL;
SV *si = NULL;
SV *alt = NULL;
- const struct reg_data *data = PL_reg_re->data;
+ const struct reg_data *data = prog ? prog->data : NULL;
if (data && data->count) {
const U32 n = ARG(node);
*/
STATIC bool
-S_reginclass(pTHX_ register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8)
+S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8)
{
dVAR;
const char flags = ANYOF_FLAGS(n);
match = TRUE;
if (!match) {
AV *av;
- SV * const sw = regclass_swash(n, TRUE, 0, (SV**)&av);
+ SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av);
if (sw) {
if (swash_fetch(sw, p, do_utf8))
restore_pos(pTHX_ void *arg)
{
dVAR;
- regexp *rex = (regexp *)arg;
+ regexp * const rex = (regexp *)arg;
if (PL_reg_eval_set) {
if (PL_reg_oldsaved) {
rex->subbeg = PL_reg_oldsaved;
S_to_utf8_substr(pTHX_ register regexp *prog)
{
if (prog->float_substr && !prog->float_utf8) {
- SV* sv;
- prog->float_utf8 = sv = newSVsv(prog->float_substr);
+ SV* const sv = newSVsv(prog->float_substr);
+ prog->float_utf8 = sv;
sv_utf8_upgrade(sv);
if (SvTAIL(prog->float_substr))
SvTAIL_on(sv);
prog->check_utf8 = sv;
}
if (prog->anchored_substr && !prog->anchored_utf8) {
- SV* sv;
- prog->anchored_utf8 = sv = newSVsv(prog->anchored_substr);
+ SV* const sv = newSVsv(prog->anchored_substr);
+ prog->anchored_utf8 = sv;
sv_utf8_upgrade(sv);
if (SvTAIL(prog->anchored_substr))
SvTAIL_on(sv);
{
dVAR;
if (prog->float_utf8 && !prog->float_substr) {
- SV* sv;
- prog->float_substr = sv = newSVsv(prog->float_utf8);
+ SV* sv = newSVsv(prog->float_utf8);
+ prog->float_substr = sv;
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) {
- SV* sv;
- prog->anchored_substr = sv = newSVsv(prog->anchored_utf8);
+ SV* sv = newSVsv(prog->anchored_utf8);
+ prog->anchored_substr = sv;
if (sv_utf8_downgrade(sv, TRUE)) {
if (SvTAIL(prog->anchored_utf8))
SvTAIL_on(sv);
https://perl5.git.perl.org / perl5.git / blobdiff