# define Perl_regprop my_regprop
/* *These* symbols are masked to allow static link. */
# define Perl_pregexec my_pregexec
+# define Perl_reginitcolors my_reginitcolors
#endif
/*SUPPRESS 112*/
*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-1998, Larry Wall
+ **** Copyright (c) 1991-1999, Larry Wall
****
**** You may distribute under the terms of either the GNU General Public
**** License or the Artistic License, as specified in the README file.
*/
#include "EXTERN.h"
#include "perl.h"
-typedef MAGIC *my_magic;
#include "regcomp.h"
static bool reginclassutf8 _((regnode *f, U8* p));
static CHECKPOINT regcppush _((I32 parenfloor));
static char * regcppop _((void));
+static char * regcp_set_to _((I32 ss));
+static void cache_re _((regexp *prog));
+static void restore_pos _((void *arg));
#endif
#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c))
SSCHECK(i + 5);
for (p = PL_regsize; p > parenfloor; p--) {
- SSPUSHPTR(PL_regendp[p]);
- SSPUSHPTR(PL_regstartp[p]);
+ SSPUSHINT(PL_regendp[p]);
+ SSPUSHINT(PL_regstartp[p]);
SSPUSHPTR(PL_reg_start_tmp[p]);
SSPUSHINT(p);
}
I32 i = SSPOPINT;
U32 paren = 0;
char *input;
- char *tmps;
+ I32 tmps;
assert(i == SAVEt_REGCONTEXT);
i = SSPOPINT;
input = (char *) SSPOPPTR;
for (i -= 3; i > 0; i -= 4) {
paren = (U32)SSPOPINT;
PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
- PL_regstartp[paren] = (char *) SSPOPPTR;
- tmps = (char*)SSPOPPTR;
+ PL_regstartp[paren] = SSPOPINT;
+ tmps = SSPOPINT;
if (paren <= *PL_reglastparen)
PL_regendp[paren] = tmps;
DEBUG_r(
PerlIO_printf(Perl_debug_log,
" restoring \\%d to %d(%d)..%d%s\n",
- paren, PL_regstartp[paren] - PL_regbol,
- PL_reg_start_tmp[paren] - PL_regbol,
- PL_regendp[paren] - PL_regbol,
+ paren, PL_regstartp[paren],
+ PL_reg_start_tmp[paren] - PL_bostr,
+ PL_regendp[paren],
(paren > *PL_reglastparen ? "(no)" : ""));
);
}
);
for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) {
if (paren > PL_regsize)
- PL_regstartp[paren] = Nullch;
- PL_regendp[paren] = Nullch;
+ PL_regstartp[paren] = -1;
+ PL_regendp[paren] = -1;
}
return input;
}
STATIC char *
regcp_set_to(I32 ss)
{
+ dTHR;
I32 tmp = PL_savestack_ix;
PL_savestack_ix = ss;
regcppop();
PL_savestack_ix = tmp;
+ return Nullch;
}
typedef struct re_cc_state
STATIC void
cache_re(regexp *prog)
{
+ dTHR;
PL_regprecomp = prog->precomp; /* Needed for FAIL. */
#ifdef DEBUGGING
PL_regprogram = prog->program;
PL_regdata = prog->data;
PL_reg_re = prog;
}
-
+
+STATIC void
+restore_pos(void *arg)
+{
+ dTHR;
+ if (PL_reg_eval_set) {
+ if (PL_reg_oldsaved) {
+ PL_reg_re->subbeg = PL_reg_oldsaved;
+ PL_reg_re->sublen = PL_reg_oldsavedlen;
+ RX_MATCH_COPIED_on(PL_reg_re);
+ }
+ PL_reg_magic->mg_len = PL_reg_oldpos;
+ PL_reg_eval_set = 0;
+ PL_curpm = PL_reg_oldcurpm;
+ }
+}
+
+
/*
- regexec_flags - match a regexp against a string
*/
I32
regexec_flags(register regexp *prog, char *stringarg, register char *strend,
- char *strbeg, I32 minend, SV *screamer, void *data, U32 flags)
+ 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. */
}
/* Check validity of program. */
- if (UCHARAT(prog->program) != MAGIC) {
- FAIL("corrupted regexp program");
+ if (UCHARAT(prog->program) != REG_MAGIC) {
+ croak("corrupted regexp program");
}
PL_reg_flags = 0;
/* Mark beginning of line for ^ and lookbehind. */
PL_regbol = startpos;
PL_bostr = strbeg;
+ PL_reg_sv = sv;
/* Mark end of line for $ (and such) */
PL_regeol = strend;
char *t;
start_shift = prog->check_offset_min; /* okay to underestimate on CC */
/* Should be nonnegative! */
- end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr);
- if (screamer) {
- if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0)
- s = screaminstr(screamer, prog->check_substr,
+ end_shift = minlen - start_shift -
+ CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0);
+ if (flags & REXEC_SCREAM) {
+ SV *c = prog->check_substr;
+
+ if (PL_screamfirst[BmRARE(c)] >= 0
+ || ( BmRARE(c) == '\n'
+ && (BmPREVIOUS(c) == SvCUR(c) - 1)
+ && SvTAIL(c) ))
+ s = screaminstr(sv, prog->check_substr,
start_shift + (stringarg - strbeg),
end_shift, &scream_pos, 0);
else
else
s = fbm_instr((unsigned char*)s + start_shift,
(unsigned char*)strend - end_shift,
- prog->check_substr, 0);
+ prog->check_substr, PL_multiline ? FBMrf_MULTILINE : 0);
if (!s) {
++BmUSEFUL(prog->check_substr); /* hooray */
goto phooey; /* not present */
s = startpos;
}
- DEBUG_r(
- PerlIO_printf(Perl_debug_log,
+ DEBUG_r(if (!PL_colorset) reginitcolors());
+ DEBUG_r(PerlIO_printf(Perl_debug_log,
"%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
PL_colors[4],PL_colors[5],PL_colors[0],
prog->precomp,
(strend - startpos > 60 ? "..." : ""))
);
+ if (prog->reganch & ROPT_GPOS_SEEN) {
+ MAGIC *mg;
+
+ if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG
+ && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0)
+ PL_reg_ganch = strbeg + mg->mg_len;
+ else
+ PL_reg_ganch = startpos;
+ }
+
/* Simplest case: anchored match need be tried only once. */
/* [unless only anchor is BOL and multiline is set] */
- if (prog->reganch & ROPT_ANCH) {
+ if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) {
if (regtry(prog, startpos))
goto got_it;
- else if (!(prog->reganch & ROPT_ANCH_GPOS) &&
- (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
- || (prog->reganch & ROPT_ANCH_MBOL)))
+ else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT)
+ || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */
{
if (minlen)
dontbother = minlen - 1;
if (s > startpos)
s--;
while (s < strend) {
- if (*s++ == '\n') { /* don't need utf8skip here */
+ if (*s++ == '\n') { /* don't need PL_utf8skip here */
if (s < strend && 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. */
I32 back_min =
prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
I32 delta = back_max - back_min;
- char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */
+ char *last = HOPc(strend, /* Cannot start after this */
+ -(CHR_SVLEN(must) - (SvTAIL(must) != 0) + back_min));
char *last1; /* Last position checked before */
if (s > PL_bostr)
dontbother = end_shift;
strend = HOPc(strend, -dontbother);
while ( (s <= last) &&
- (screamer
- ? (s = screaminstr(screamer, must, HOPc(s, back_min) - strbeg,
+ ((flags & REXEC_SCREAM)
+ ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg,
end_shift, &scream_pos, 0))
: (s = fbm_instr((unsigned char*)HOP(s, back_min),
- (unsigned char*)strend, must, 0))) ) {
+ (unsigned char*)strend, must,
+ PL_multiline ? FBMrf_MULTILINE : 0))) ) {
if (HOPc(s, -back_max) > last1) {
last1 = HOPc(s, -back_min);
s = HOPc(s, -back_max);
char *last;
I32 oldpos = scream_pos;
- if (screamer) {
- last = screaminstr(screamer, prog->float_substr, s - strbeg,
+ if (flags & REXEC_SCREAM) {
+ last = screaminstr(sv, prog->float_substr, s - strbeg,
end_shift, &scream_pos, 1); /* last one */
- if (!last) {
+ if (!last)
last = scream_olds; /* Only one occurence. */
- }
}
else {
STRLEN len;
char *little = SvPV(prog->float_substr, len);
- if (len)
- last = rninstr(s, strend, little, little + len);
- else
- last = strend; /* matching `$' */
+
+ if (SvTAIL(prog->float_substr)) {
+ if (memEQ(strend - len + 1, little, len - 1))
+ last = strend - len + 1;
+ else if (!PL_multiline)
+ last = memEQ(strend - len, little, len)
+ ? strend - len : Nullch;
+ else
+ goto find_last;
+ } else {
+ find_last:
+ if (len)
+ last = rninstr(s, strend, little, little + len);
+ else
+ last = strend; /* matching `$' */
+ }
}
if (last == NULL) goto phooey; /* Should not happen! */
dontbother = strend - last + prog->float_min_offset;
goto phooey;
got_it:
- prog->subbeg = strbeg;
- prog->subend = PL_regeol; /* strend may have been modified */
RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
+ if (PL_reg_eval_set) {
+ /* Preserve the current value of $^R */
+ if (oreplsv != GvSV(PL_replgv))
+ sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
+ restored, the value remains
+ the same. */
+ restore_pos(0);
+ }
+
/* make sure $`, $&, $', and $digit will work later */
- if (strbeg != prog->subbase) { /* second+ //g match. */
- if (!(flags & REXEC_COPY_STR)) {
- if (prog->subbase) {
- Safefree(prog->subbase);
- prog->subbase = Nullch;
- }
+ if ( !(flags & REXEC_NOT_FIRST) ) {
+ if (RX_MATCH_COPIED(prog)) {
+ Safefree(prog->subbeg);
+ RX_MATCH_COPIED_off(prog);
}
- else {
+ if (flags & REXEC_COPY_STR) {
I32 i = PL_regeol - startpos + (stringarg - strbeg);
+
s = savepvn(strbeg, i);
- Safefree(prog->subbase);
- prog->subbase = s;
- prog->subbeg = prog->subbase;
- prog->subend = prog->subbase + i;
- s = prog->subbase + (stringarg - strbeg);
- for (i = 0; i <= prog->nparens; i++) {
- if (prog->endp[i]) {
- prog->startp[i] = s + (prog->startp[i] - startpos);
- prog->endp[i] = s + (prog->endp[i] - startpos);
- }
- }
+ prog->subbeg = s;
+ prog->sublen = i;
+ RX_MATCH_COPIED_on(prog);
+ }
+ else {
+ prog->subbeg = strbeg;
+ prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
}
}
- /* Preserve the current value of $^R */
- if (oreplsv != GvSV(PL_replgv)) {
- sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
- restored, the value remains
- the same. */
- }
+
return 1;
phooey:
+ if (PL_reg_eval_set)
+ restore_pos(0);
return 0;
}
{
dTHR;
register I32 i;
- register char **sp;
- register char **ep;
+ register I32 *sp;
+ register I32 *ep;
CHECKPOINT lastcp;
if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
+ MAGIC *mg;
+
PL_reg_eval_set = RS_init;
DEBUG_r(DEBUG_s(
PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
/* Apparently this is not needed, judging by wantarray. */
/* SAVEINT(cxstack[cxstack_ix].blk_gimme);
cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
+
+ if (PL_reg_sv) {
+ /* Make $_ available to executed code. */
+ if (PL_reg_sv != DEFSV) {
+ /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
+ SAVESPTR(DEFSV);
+ DEFSV = PL_reg_sv;
+ }
+
+ if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
+ && (mg = mg_find(PL_reg_sv, 'g')))) {
+ /* prepare for quick setting of pos */
+ sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
+ mg = mg_find(PL_reg_sv, 'g');
+ mg->mg_len = -1;
+ }
+ PL_reg_magic = mg;
+ PL_reg_oldpos = mg->mg_len;
+ SAVEDESTRUCTOR(restore_pos, 0);
+ }
+ if (!PL_reg_curpm)
+ New(22,PL_reg_curpm, 1, PMOP);
+ PL_reg_curpm->op_pmregexp = prog;
+ PL_reg_oldcurpm = PL_curpm;
+ PL_curpm = PL_reg_curpm;
+ if (RX_MATCH_COPIED(prog)) {
+ /* Here is a serious problem: we cannot rewrite subbeg,
+ since it may be needed if this match fails. Thus
+ $` inside (?{}) could fail... */
+ PL_reg_oldsaved = prog->subbeg;
+ PL_reg_oldsavedlen = prog->sublen;
+ RX_MATCH_COPIED_off(prog);
+ }
+ else
+ PL_reg_oldsaved = Nullch;
+ prog->subbeg = PL_bostr;
+ prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
}
+ prog->startp[0] = startpos - PL_bostr;
PL_reginput = startpos;
PL_regstartp = prog->startp;
PL_regendp = prog->endp;
PL_reglastparen = &prog->lastparen;
prog->lastparen = 0;
PL_regsize = 0;
+ DEBUG_r(PL_reg_starttry = startpos);
if (PL_reg_start_tmpl <= prog->nparens) {
PL_reg_start_tmpl = prog->nparens*3/2 + 3;
if(PL_reg_start_tmp)
New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*);
}
+ /* XXXX What this code is doing here?!!! There should be no need
+ to do this again and again, PL_reglastparen should take care of
+ this! */
sp = prog->startp;
ep = prog->endp;
if (prog->nparens) {
- for (i = prog->nparens; i >= 0; i--) {
- *sp++ = NULL;
- *ep++ = NULL;
+ for (i = prog->nparens; i >= 1; i--) {
+ *++sp = -1;
+ *++ep = -1;
}
}
REGCP_SET;
if (regmatch(prog->program + 1)) {
- prog->startp[0] = startpos;
- prog->endp[0] = PL_reginput;
+ prog->endp[0] = PL_reginput - PL_bostr;
return 1;
}
REGCP_UNWIND;
int docolor = *PL_colors[0];
int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */
int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput);
+ /* The part of the string before starttry has one color
+ (pref0_len chars), between starttry and current
+ position another one (pref_len - pref0_len chars),
+ after the current position the third one.
+ We assume that pref0_len <= pref_len, otherwise we
+ decrease pref0_len. */
int pref_len = (locinput - PL_bostr > (5 + taill) - l
? (5 + taill) - l : locinput - PL_bostr);
- int pref0_len = pref_len - (locinput - PL_reginput);
+ int pref0_len = pref_len - (locinput - PL_reg_starttry);
if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
l = ( PL_regeol - locinput > (5 + taill) - pref_len
? (5 + taill) - pref_len : PL_regeol - locinput);
if (pref0_len < 0)
pref0_len = 0;
+ if (pref0_len > pref_len)
+ pref0_len = pref_len;
regprop(prop, scan);
PerlIO_printf(Perl_debug_log,
"%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
break;
sayNO;
case GPOS:
- if (locinput == PL_regbol)
+ if (locinput == PL_reg_ganch)
break;
sayNO;
case EOL:
break;
case SANYUTF8:
if (nextchr & 0x80) {
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
if (locinput > PL_regeol)
sayNO;
nextchr = UCHARAT(locinput);
break;
case ANYUTF8:
if (nextchr & 0x80) {
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
if (locinput > PL_regeol)
sayNO;
nextchr = UCHARAT(locinput);
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ANY:
+ case REG_ANY:
if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
sayNO;
nextchr = UCHARAT(++locinput);
/* Inline the first character, for speed. */
if (UCHARAT(s) != nextchr &&
UCHARAT(s) != ((OP(scan) == EXACTF)
- ? fold : fold_locale)[nextchr])
+ ? PL_fold : PL_fold_locale)[nextchr])
sayNO;
if (PL_regeol - locinput < ln)
sayNO;
sayNO;
if (locinput >= PL_regeol)
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
case ANYOF:
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
{
sayNO;
}
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
if (nextchr & 0x80) {
if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
if (nextchr & 0x80) {
if (swash_fetch(PL_utf8_digit,(U8*)locinput))
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
case CLUMP:
if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
sayNO;
- locinput += utf8skip[nextchr];
+ locinput += PL_utf8skip[nextchr];
while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
locinput += UTF8SKIP(locinput);
if (locinput > PL_regeol)
case REF:
case REFF:
n = ARG(scan); /* which paren pair */
- s = PL_regstartp[n];
- if (*PL_reglastparen < n || !s)
+ ln = PL_regstartp[n];
+ if (*PL_reglastparen < n || ln == -1)
sayNO; /* Do not match unless seen CLOSEn. */
- if (s == PL_regendp[n])
+ if (ln == PL_regendp[n])
break;
+ s = PL_bostr + ln;
if (UTF && OP(scan) != REF) { /* REF can do byte comparison */
char *l = locinput;
- char *e = PL_regendp[n];
+ char *e = PL_bostr + PL_regendp[n];
/*
* Note that we can't do the "other character" lookup trick as
* in the 8-bit case (no pun intended) because in Unicode we
if (UCHARAT(s) != nextchr &&
(OP(scan) == REF ||
(UCHARAT(s) != ((OP(scan) == REFF
- ? fold : fold_locale)[nextchr]))))
+ ? PL_fold : PL_fold_locale)[nextchr]))))
sayNO;
- ln = PL_regendp[n] - s;
+ ln = PL_regendp[n] - ln;
if (locinput + ln > PL_regeol)
sayNO;
if (ln > 1 && (OP(scan) == REF
n = ARG(scan);
PL_op = (OP_4tree*)PL_regdata->data[n];
DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) );
- PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 1]);
+ PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]);
+ PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr;
CALLRUNOPS(); /* Scalar context. */
SPAGAIN;
if (logical) {
if (logical == 2) { /* Postponed subexpression. */
regexp *re;
- my_magic mg = Null(my_magic);
+ MAGIC *mg = Null(MAGIC*);
re_cc_state state;
CURCUR cctmp;
CHECKPOINT cp, lastcp;
}
if (mg) {
re = (regexp *)mg->mg_obj;
- ReREFCNT_inc(re);
+ (void)ReREFCNT_inc(re);
}
else {
STRLEN len;
break;
case CLOSE:
n = ARG(scan); /* which paren pair */
- PL_regstartp[n] = PL_reg_start_tmp[n];
- PL_regendp[n] = locinput;
+ PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr;
+ PL_regendp[n] = locinput - PL_bostr;
if (n > *PL_reglastparen)
*PL_reglastparen = n;
break;
case GROUPP:
n = ARG(scan); /* which paren pair */
- sw = (*PL_reglastparen >= n && PL_regendp[n] != NULL);
+ sw = (*PL_reglastparen >= n && PL_regendp[n] != -1);
break;
case IFTHEN:
if (sw)
sayYES;
REGCP_UNWIND;
for (n = *PL_reglastparen; n > lastparen; n--)
- PL_regendp[n] = 0;
+ PL_regendp[n] = -1;
*PL_reglastparen = n;
scan = next;
/*SUPPRESS 560*/
&& !(paren && ln == 0))
ln = n;
locinput = PL_reginput;
- if (regkind[(U8)OP(next)] == EXACT) {
+ if (PL_regkind[(U8)OP(next)] == EXACT) {
c1 = UCHARAT(OPERAND(next) + 1);
if (OP(next) == EXACTF)
- c2 = fold[c1];
+ c2 = PL_fold[c1];
else if (OP(next) == EXACTFL)
- c2 = fold_locale[c1];
+ c2 = PL_fold_locale[c1];
else
c2 = c1;
}
{
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -l);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] =
+ HOPc(PL_reginput, -l) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
REPORT_CODE_OFF+PL_regindent*2, "", n, l)
);
if (n >= ln) {
- if (regkind[(U8)OP(next)] == EXACT) {
+ if (PL_regkind[(U8)OP(next)] == EXACT) {
c1 = UCHARAT(OPERAND(next) + 1);
if (OP(next) == EXACTF)
- c2 = fold[c1];
+ c2 = PL_fold[c1];
else if (OP(next) == EXACTFL)
- c2 = fold_locale[c1];
+ c2 = PL_fold_locale[c1];
else
c2 = c1;
}
);
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -l);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
* Lookahead to avoid useless match attempts
* when we know what character comes next.
*/
- if (regkind[(U8)OP(next)] == EXACT) {
+ if (PL_regkind[(U8)OP(next)] == EXACT) {
c1 = UCHARAT(OPERAND(next) + 1);
if (OP(next) == EXACTF)
- c2 = fold[c1];
+ c2 = PL_fold[c1];
else if (OP(next) == EXACTFL)
- c2 = fold_locale[c1];
+ c2 = PL_fold_locale[c1];
else
c2 = c1;
}
sayNO;
locinput = PL_reginput;
REGCP_SET;
+ if (c1 != -1000) {
+ char *e = locinput + n - ln; /* Should not check after this */
+ char *old = locinput;
+
+ if (e >= PL_regeol || (n == REG_INFTY))
+ e = PL_regeol - 1;
+ while (1) {
+ /* Find place 'next' could work */
+ if (c1 == c2) {
+ while (locinput <= e && *locinput != c1)
+ locinput++;
+ } else {
+ while (locinput <= e
+ && *locinput != c1
+ && *locinput != c2)
+ locinput++;
+ }
+ if (locinput > e)
+ sayNO;
+ /* PL_reginput == old now */
+ if (locinput != old) {
+ ln = 1; /* Did some */
+ if (regrepeat(scan, locinput - old) <
+ locinput - old)
+ sayNO;
+ }
+ /* PL_reginput == locinput now */
+ if (paren) {
+ if (ln) {
+ PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr;
+ PL_regendp[paren] = locinput - PL_bostr;
+ }
+ else
+ PL_regendp[paren] = -1;
+ }
+ if (regmatch(next))
+ sayYES;
+ PL_reginput = locinput; /* Could be reset... */
+ REGCP_UNWIND;
+ /* Couldn't or didn't -- move forward. */
+ old = locinput++;
+ }
+ }
+ else
while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
/* If it could work, try it. */
if (c1 == -1000 ||
{
if (paren) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
CHECKPOINT lastcp;
n = regrepeat(scan, n);
locinput = PL_reginput;
- if (ln < n && regkind[(U8)OP(next)] == EOL &&
+ if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
(!PL_multiline || OP(next) == SEOL))
ln = n; /* why back off? */
REGCP_SET;
{
if (paren && n) {
if (n) {
- PL_regstartp[paren] = HOPc(PL_reginput, -1);
- PL_regendp[paren] = PL_reginput;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr;
+ PL_regendp[paren] = PL_reginput - PL_bostr;
}
else
- PL_regendp[paren] = NULL;
+ PL_regendp[paren] = -1;
}
if (regmatch(next))
sayYES;
case UNLESSM:
n = 0;
if (scan->flags) {
- s = HOPMAYBEc(locinput, -scan->flags);
- if (!s)
- goto say_yes;
- PL_reginput = s;
+ if (UTF) { /* XXXX This is absolutely
+ broken, we read before
+ start of string. */
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s)
+ goto say_yes;
+ PL_reginput = s;
+ }
+ else {
+ if (locinput < PL_bostr + scan->flags)
+ goto say_yes;
+ PL_reginput = locinput - scan->flags;
+ goto do_ifmatch;
+ }
}
else
PL_reginput = locinput;
case IFMATCH:
n = 1;
if (scan->flags) {
- s = HOPMAYBEc(locinput, -scan->flags);
- if (!s)
- goto say_no;
- PL_reginput = s;
+ if (UTF) { /* XXXX This is absolutely
+ broken, we read before
+ start of string. */
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s || s < PL_bostr)
+ goto say_no;
+ PL_reginput = s;
+ }
+ else {
+ if (locinput < PL_bostr + scan->flags)
+ goto say_no;
+ PL_reginput = locinput - scan->flags;
+ goto do_ifmatch;
+ }
}
else
PL_reginput = locinput;
default:
PerlIO_printf(PerlIO_stderr(), "%lx %d\n",
(unsigned long)scan, OP(scan));
- FAIL("regexp memory corruption");
+ croak("regexp memory corruption");
}
scan = next;
}
* We get here only if there's trouble -- normally "case END" is
* the terminating point.
*/
- FAIL("corrupted regexp pointers");
+ croak("corrupted regexp pointers");
/*NOTREACHED*/
sayNO;
loceol = scan + max;
opnd = (char *) OPERAND(p);
switch (OP(p)) {
- case ANY:
+ case REG_ANY:
while (scan < loceol && *scan != '\n')
scan++;
break;
case EXACTF: /* length of string is 1 */
c = UCHARAT(++opnd);
while (scan < loceol &&
- (UCHARAT(scan) == c || UCHARAT(scan) == fold[c]))
+ (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c]))
scan++;
break;
case EXACTFL: /* length of string is 1 */
PL_reg_flags |= RF_tainted;
c = UCHARAT(++opnd);
while (scan < loceol &&
- (UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c]))
+ (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c]))
scan++;
break;
case ANYOFUTF8:
}
/*
- - regclass - determine if a character falls into a character class
+ - reginclass - determine if a character falls into a character class
*/
STATIC bool
I32 cf;
if (flags & ANYOF_LOCALE) {
PL_reg_flags |= RF_tainted;
- cf = fold_locale[c];
+ cf = PL_fold_locale[c];
}
else
- cf = fold[c];
+ cf = PL_fold[c];
if (ANYOF_TEST(p, cf))
match = TRUE;
}
projects / p5sagit/p5-mst-13.2.git / blobdiff