/* *These* symbols are masked to allow static link. */
# define Perl_pregexec my_pregexec
# define Perl_reginitcolors my_reginitcolors
+
+# define PERL_NO_GET_CONTEXT
#endif
/*SUPPRESS 112*/
#define PERL_IN_REGEXEC_C
#include "perl.h"
+#ifdef PERL_IN_XSUB_RE
+# if defined(PERL_CAPI) || defined(PERL_OBJECT)
+# include "XSUB.h"
+# endif
+#endif
+
#include "regcomp.h"
#define RF_tainted 1 /* tainted information used? */
* Forwards.
*/
-#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c))
+#define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c) : ANYOF_BITMAP_TEST(p,c))
#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p))
#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
#define HOPc(pos,off) ((char*)HOP(pos,off))
#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
+static void restore_pos(pTHXo_ void *arg);
+
+
STATIC CHECKPOINT
S_regcppush(pTHX_ I32 parenfloor)
{
PL_reg_re = prog;
}
-STATIC void
-S_restore_pos(pTHX_ 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;
- }
-}
-
/*
* Need to implement the following flags for reg_anch:
*
&& (sv && (strpos + SvCUR(sv) != strend)) )
goto fail;
+ PL_regeol = strend; /* Used in HOP() */
s = (char*)HOP((U8*)strpos, prog->check_offset_min);
if (SvTAIL(prog->check_substr)) {
slen = SvCUR(prog->check_substr); /* >= 1 */
}
restart:
+ if (end_shift < 0)
+ end_shift = 0; /* can happen when strend == strpos */
if (flags & REXEC_SCREAM) {
SV *c = prog->check_substr;
char *strbeg = SvPVX(sv); /* XXXX Assume PV_force() on SCREAM! */
s += UTF8SKIP(s);
}
break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case DIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NDIGIT:
while (s < strend) {
if (!isDIGIT(*s)) {
s += UTF8SKIP(s);
}
break;
- }
- }
- else {
- dontbother = 0;
- if (prog->float_substr != Nullsv) { /* Trim the end. */
- char *last;
- I32 oldpos = scream_pos;
-
- if (flags & REXEC_SCREAM) {
- last = screaminstr(sv, prog->float_substr, s - strbeg,
- end_shift, &scream_pos, 1); /* last one */
- if (!last)
- last = scream_olds; /* Only one occurence. */
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
}
- else {
- STRLEN len;
- char *little = SvPV(prog->float_substr, len);
-
- if (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;
+ break;
+ case NDIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isDIGIT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
else
- goto find_last;
- } else {
- find_last:
- if (len)
- last = rninstr(s, strend, little, little + len);
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case ALNUMC:
+ while (s < strend) {
+ if (isALNUMC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
else
- last = strend; /* matching `$' */
+ tmp = doevery;
}
+ else
+ tmp = 1;
+ s++;
}
- if (last == NULL) goto phooey; /* Should not happen! */
- dontbother = strend - last + prog->float_min_offset;
- }
- if (minlen && (dontbother < minlen))
- dontbother = minlen - 1;
- strend -= dontbother; /* this one's always in bytes! */
- /* We don't know much -- general case. */
- if (UTF) {
- for (;;) {
- if (regtry(prog, s))
- goto got_it;
- if (s >= strend)
- break;
+ break;
+ case ALNUMCUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_alnumc, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
s += UTF8SKIP(s);
- };
- }
- else {
- do {
- if (regtry(prog, s))
- goto got_it;
- } while (s++ < strend);
- }
- }
-
- /* Failure. */
- goto phooey;
-
-got_it:
- 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 ( !(flags & REXEC_NOT_FIRST) ) {
- if (RX_MATCH_COPIED(prog)) {
- Safefree(prog->subbeg);
- RX_MATCH_COPIED_off(prog);
- }
- if (flags & REXEC_COPY_STR) {
- I32 i = PL_regeol - startpos + (stringarg - strbeg);
-
- s = savepvn(strbeg, i);
- prog->subbeg = s;
- prog->sublen = i;
- RX_MATCH_COPIED_on(prog);
- }
- else {
- prog->subbeg = strbeg;
- prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
- }
- }
-
- return 1;
-
-phooey:
- if (PL_reg_eval_set)
- restore_pos(0);
- return 0;
-}
-
-/*
- - regtry - try match at specific point
- */
-STATIC I32 /* 0 failure, 1 success */
-S_regtry(pTHX_ regexp *prog, char *startpos)
-{
- dTHR;
- register I32 i;
- register I32 *sp;
- register I32 *ep;
- CHECKPOINT lastcp;
-
- if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
- MAGIC *mg;
-
- PL_reg_eval_set = RS_init;
- DEBUG_r(DEBUG_s(
- PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
- PL_stack_sp - PL_stack_base);
- ));
- SAVEINT(cxstack[cxstack_ix].blk_oldsp);
- cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
- /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
- SAVETMPS;
- /* Apparently this is not needed, judging by wantarray. */
- /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
- cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
-
- if (PL_reg_sv) {
- /* Make $_ available to executed code. */
- if (PL_reg_sv != DEFSV) {
- /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
- SAVESPTR(DEFSV);
- DEFSV = PL_reg_sv;
}
-
- if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
- && (mg = mg_find(PL_reg_sv, 'g')))) {
- /* prepare for quick setting of pos */
- sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
- mg = mg_find(PL_reg_sv, 'g');
- mg->mg_len = -1;
+ break;
+ case ALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isALNUMC_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
}
- PL_reg_magic = mg;
- PL_reg_oldpos = mg->mg_len;
- SAVEDESTRUCTOR(S_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;
+ break;
+ case ALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isALNUMC_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NALNUMC:
+ while (s < strend) {
+ if (!isALNUMC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NALNUMCUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_alnumc, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isALNUMC_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isALNUMC_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case ASCII:
+ while (s < strend) {
+ if (isASCII(*(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NASCII:
+ while (s < strend) {
+ if (!isASCII(*(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRL:
+ while (s < strend) {
+ if (isCNTRL(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRLUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_cntrl,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case CNTRLL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isCNTRL_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case CNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isCNTRL_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NCNTRL:
+ while (s < strend) {
+ if (!isCNTRL(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NCNTRLUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_cntrl,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NCNTRLL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isCNTRL_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NCNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isCNTRL_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case GRAPH:
+ while (s < strend) {
+ if (isGRAPH(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case GRAPHUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_graph,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case GRAPHL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isGRAPH_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case GRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isGRAPH_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NGRAPH:
+ while (s < strend) {
+ if (!isGRAPH(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NGRAPHUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_graph,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NGRAPHL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isGRAPH_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NGRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isGRAPH_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case LOWER:
+ while (s < strend) {
+ if (isLOWER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case LOWERUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_lower,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case LOWERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isLOWER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case LOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isLOWER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NLOWER:
+ while (s < strend) {
+ if (!isLOWER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NLOWERUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_lower,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NLOWERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isLOWER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NLOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isLOWER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PRINT:
+ while (s < strend) {
+ if (isPRINT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PRINTUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_print,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PRINTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isPRINT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PRINTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isPRINT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPRINT:
+ while (s < strend) {
+ if (!isPRINT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPRINTUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_print,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPRINTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPRINT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPRINTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPRINT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PUNCT:
+ while (s < strend) {
+ if (isPUNCT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PUNCTUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_punct,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case PUNCTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isPUNCT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case PUNCTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isPUNCT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPUNCT:
+ while (s < strend) {
+ if (!isPUNCT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPUNCTUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_punct,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NPUNCTL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPUNCT_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NPUNCTLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isPUNCT_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case UPPER:
+ while (s < strend) {
+ if (isUPPER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case UPPERUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_upper,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case UPPERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isUPPER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case UPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isUPPER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NUPPER:
+ while (s < strend) {
+ if (!isUPPER(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NUPPERUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_upper,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case NUPPERL:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isUPPER_LC(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NUPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isUPPER_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
+ case XDIGIT:
+ while (s < strend) {
+ if (isXDIGIT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ case NXDIGIT:
+ while (s < strend) {
+ if (!isXDIGIT(*s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s++;
+ }
+ break;
+ }
+ }
+ else {
+ dontbother = 0;
+ if (prog->float_substr != Nullsv) { /* Trim the end. */
+ char *last;
+ I32 oldpos = scream_pos;
+
+ if (flags & REXEC_SCREAM) {
+ last = screaminstr(sv, prog->float_substr, s - strbeg,
+ end_shift, &scream_pos, 1); /* last one */
+ if (!last)
+ last = scream_olds; /* Only one occurence. */
+ }
+ else {
+ STRLEN len;
+ char *little = SvPV(prog->float_substr, len);
+
+ if (SvTAIL(prog->float_substr)) {
+ if (memEQ(strend - len + 1, little, len - 1))
+ last = strend - len + 1;
+ else if (!PL_multiline)
+ last = memEQ(strend - len, little, len)
+ ? strend - len : Nullch;
+ else
+ goto find_last;
+ } else {
+ find_last:
+ if (len)
+ last = rninstr(s, strend, little, little + len);
+ else
+ last = strend; /* matching `$' */
+ }
+ }
+ if (last == NULL) goto phooey; /* Should not happen! */
+ dontbother = strend - last + prog->float_min_offset;
+ }
+ if (minlen && (dontbother < minlen))
+ dontbother = minlen - 1;
+ strend -= dontbother; /* this one's always in bytes! */
+ /* We don't know much -- general case. */
+ if (UTF) {
+ for (;;) {
+ if (regtry(prog, s))
+ goto got_it;
+ if (s >= strend)
+ break;
+ s += UTF8SKIP(s);
+ };
+ }
+ else {
+ do {
+ if (regtry(prog, s))
+ goto got_it;
+ } while (s++ < strend);
+ }
+ }
+
+ /* Failure. */
+ goto phooey;
+
+got_it:
+ RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted);
+
+ if (PL_reg_eval_set) {
+ /* Preserve the current value of $^R */
+ if (oreplsv != GvSV(PL_replgv))
+ sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is
+ restored, the value remains
+ the same. */
+ restore_pos(aTHXo_ 0);
+ }
+
+ /* make sure $`, $&, $', and $digit will work later */
+ if ( !(flags & REXEC_NOT_FIRST) ) {
+ if (RX_MATCH_COPIED(prog)) {
+ Safefree(prog->subbeg);
+ RX_MATCH_COPIED_off(prog);
+ }
+ if (flags & REXEC_COPY_STR) {
+ I32 i = PL_regeol - startpos + (stringarg - strbeg);
+
+ s = savepvn(strbeg, i);
+ prog->subbeg = s;
+ prog->sublen = i;
+ RX_MATCH_COPIED_on(prog);
+ }
+ else {
+ prog->subbeg = strbeg;
+ prog->sublen = PL_regeol - strbeg; /* strend may have been modified */
+ }
+ }
+
+ return 1;
+
+phooey:
+ if (PL_reg_eval_set)
+ restore_pos(aTHXo_ 0);
+ return 0;
+}
+
+/*
+ - regtry - try match at specific point
+ */
+STATIC I32 /* 0 failure, 1 success */
+S_regtry(pTHX_ regexp *prog, char *startpos)
+{
+ dTHR;
+ register I32 i;
+ register I32 *sp;
+ register I32 *ep;
+ CHECKPOINT lastcp;
+
+ if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) {
+ MAGIC *mg;
+
+ PL_reg_eval_set = RS_init;
+ DEBUG_r(DEBUG_s(
+ PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n",
+ PL_stack_sp - PL_stack_base);
+ ));
+ SAVEINT(cxstack[cxstack_ix].blk_oldsp);
+ cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base;
+ /* Otherwise OP_NEXTSTATE will free whatever on stack now. */
+ SAVETMPS;
+ /* Apparently this is not needed, judging by wantarray. */
+ /* SAVEINT(cxstack[cxstack_ix].blk_gimme);
+ cxstack[cxstack_ix].blk_gimme = G_SCALAR; */
+
+ if (PL_reg_sv) {
+ /* Make $_ available to executed code. */
+ if (PL_reg_sv != DEFSV) {
+ /* SAVE_DEFSV does *not* suffice here for USE_THREADS */
+ SAVESPTR(DEFSV);
+ DEFSV = PL_reg_sv;
+ }
+
+ if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv)
+ && (mg = mg_find(PL_reg_sv, 'g')))) {
+ /* prepare for quick setting of pos */
+ sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0);
+ mg = mg_find(PL_reg_sv, 'g');
+ mg->mg_len = -1;
+ }
+ PL_reg_magic = mg;
+ PL_reg_oldpos = mg->mg_len;
+ SAVEDESTRUCTOR(restore_pos, 0);
+ }
+ if (!PL_reg_curpm)
+ New(22,PL_reg_curpm, 1, PMOP);
+ PL_reg_curpm->op_pmregexp = prog;
+ PL_reg_oldcurpm = PL_curpm;
+ PL_curpm = PL_reg_curpm;
+ if (RX_MATCH_COPIED(prog)) {
+ /* Here is a serious problem: we cannot rewrite subbeg,
+ since it may be needed if this match fails. Thus
+ $` inside (?{}) could fail... */
+ PL_reg_oldsaved = prog->subbeg;
+ PL_reg_oldsavedlen = prog->sublen;
+ RX_MATCH_COPIED_off(prog);
+ }
+ else
+ PL_reg_oldsaved = Nullch;
+ prog->subbeg = PL_bostr;
+ prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */
+ }
+ prog->startp[0] = startpos - PL_bostr;
+ PL_reginput = startpos;
+ PL_regstartp = prog->startp;
+ PL_regendp = prog->endp;
+ PL_reglastparen = &prog->lastparen;
+ prog->lastparen = 0;
PL_regsize = 0;
DEBUG_r(PL_reg_starttry = startpos);
if (PL_reg_start_tmpl <= prog->nparens) {
SvPVX(prop));
} );
- next = scan + NEXT_OFF(scan);
- if (next == scan)
- next = NULL;
+ next = scan + NEXT_OFF(scan);
+ if (next == scan)
+ next = NULL;
+
+ switch (OP(scan)) {
+ case BOL:
+ if (locinput == PL_bostr
+ ? PL_regprev == '\n'
+ : (PL_multiline &&
+ (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
+ {
+ /* regtill = regbol; */
+ break;
+ }
+ sayNO;
+ case MBOL:
+ if (locinput == PL_bostr
+ ? PL_regprev == '\n'
+ : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
+ {
+ break;
+ }
+ sayNO;
+ case SBOL:
+ if (locinput == PL_regbol && PL_regprev == '\n')
+ break;
+ sayNO;
+ case GPOS:
+ if (locinput == PL_reg_ganch)
+ break;
+ sayNO;
+ case EOL:
+ if (PL_multiline)
+ goto meol;
+ else
+ goto seol;
+ case MEOL:
+ meol:
+ if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ sayNO;
+ break;
+ case SEOL:
+ seol:
+ if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ sayNO;
+ if (PL_regeol - locinput > 1)
+ sayNO;
+ break;
+ case EOS:
+ if (PL_regeol != locinput)
+ sayNO;
+ break;
+ case SANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += PL_utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case SANY:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += PL_utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case REG_ANY:
+ if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case EXACT:
+ s = (char *) OPERAND(scan);
+ ln = UCHARAT(s++);
+ /* Inline the first character, for speed. */
+ if (UCHARAT(s) != nextchr)
+ sayNO;
+ if (PL_regeol - locinput < ln)
+ sayNO;
+ if (ln > 1 && memNE(s, locinput, ln))
+ sayNO;
+ locinput += ln;
+ nextchr = UCHARAT(locinput);
+ break;
+ case EXACTFL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case EXACTF:
+ s = (char *) OPERAND(scan);
+ ln = UCHARAT(s++);
+
+ if (UTF) {
+ char *l = locinput;
+ char *e = s + ln;
+ c1 = OP(scan) == EXACTF;
+ while (s < e) {
+ if (l >= PL_regeol)
+ sayNO;
+ if (utf8_to_uv((U8*)s, 0) != (c1 ?
+ toLOWER_utf8((U8*)l) :
+ toLOWER_LC_utf8((U8*)l)))
+ {
+ sayNO;
+ }
+ s += UTF8SKIP(s);
+ l += UTF8SKIP(l);
+ }
+ locinput = l;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
- switch (OP(scan)) {
- case BOL:
- if (locinput == PL_bostr
- ? PL_regprev == '\n'
- : (PL_multiline &&
- (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
- {
- /* regtill = regbol; */
+ /* Inline the first character, for speed. */
+ if (UCHARAT(s) != nextchr &&
+ UCHARAT(s) != ((OP(scan) == EXACTF)
+ ? PL_fold : PL_fold_locale)[nextchr])
+ sayNO;
+ if (PL_regeol - locinput < ln)
+ sayNO;
+ if (ln > 1 && (OP(scan) == EXACTF
+ ? ibcmp(s, locinput, ln)
+ : ibcmp_locale(s, locinput, ln)))
+ sayNO;
+ locinput += ln;
+ nextchr = UCHARAT(locinput);
+ break;
+ case ANYOFUTF8:
+ s = (char *) OPERAND(scan);
+ if (!REGINCLASSUTF8(scan, (U8*)locinput))
+ sayNO;
+ if (locinput >= PL_regeol)
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ case ANYOF:
+ s = (char *) OPERAND(scan);
+ if (nextchr < 0)
+ nextchr = UCHARAT(locinput);
+ if (!REGINCLASS(s, nextchr))
+ sayNO;
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALNUML:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUM:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == ALNUM
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUML:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUM:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (OP(scan) == NALNUM
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (OP(scan) == NALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case BOUNDL:
+ case NBOUNDL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUND:
+ case NBOUND:
+ /* was last char in word? */
+ ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
+ if (OP(scan) == BOUND || OP(scan) == NBOUND) {
+ ln = isALNUM(ln);
+ n = isALNUM(nextchr);
+ }
+ else {
+ ln = isALNUM_LC(ln);
+ n = isALNUM_LC(nextchr);
+ }
+ if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
+ sayNO;
+ break;
+ case BOUNDLUTF8:
+ case NBOUNDLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUNDUTF8:
+ case NBOUNDUTF8:
+ /* was last char in word? */
+ ln = (locinput != PL_regbol)
+ ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
+ if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
+ ln = isALNUM_uni(ln);
+ n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
+ }
+ else {
+ ln = isALNUM_LC_uni(ln);
+ n = isALNUM_LC_utf8((U8*)locinput);
+ }
+ if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
+ sayNO;
+ break;
+ case SPACEL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case SPACE:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!(OP(scan) == SPACE
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case SPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case SPACEUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == SPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
}
- sayNO;
- case MBOL:
- if (locinput == PL_bostr
- ? PL_regprev == '\n'
- : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
- {
+ if (!(OP(scan) == SPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NSPACEL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NSPACE:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == SPACE
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NSPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NSPACEUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NSPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
}
- sayNO;
- case SBOL:
- if (locinput == PL_regbol && PL_regprev == '\n')
+ if (OP(scan) == NSPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case DIGITL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case DIGIT:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!(OP(scan) == DIGIT
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case DIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case DIGITUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NDIGITUTF8
+ ? swash_fetch(PL_utf8_digit,(U8*)locinput)
+ : isDIGIT_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
- sayNO;
- case GPOS:
- if (locinput == PL_reg_ganch)
+ }
+ if (!isDIGIT(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NDIGITL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NDIGIT:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == DIGIT
+ ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NDIGITLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NDIGITUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_digit,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
break;
- sayNO;
- case EOL:
- if (PL_multiline)
- goto meol;
- else
- goto seol;
- case MEOL:
- meol:
- if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ }
+ if (isDIGIT(nextchr))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case SEOL:
- seol:
- if ((nextchr || locinput < PL_regeol) && nextchr != '\n')
+ case ALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMC:
+ if (!nextchr)
sayNO;
- if (PL_regeol - locinput > 1)
+ if (!(OP(scan) == ALNUMC
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMCUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALNUMCUTF8
+ ? swash_fetch(PL_utf8_alnumc, (U8*)locinput)
+ : isALNUMC_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALNUMCUTF8
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMCL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMC:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == ALNUMC
+ ? isALNUMC(nextchr) : isALNUMC_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALNUMCLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMCUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_alnumc,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isALNUMC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALPHAL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALPHA:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == ALPHA
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ALPHALUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALPHAUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALPHAUTF8
+ ? swash_fetch(PL_utf8_alpha, (U8*)locinput)
+ : isALPHA_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALPHAUTF8
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALPHAL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALPHA:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == ALPHA
+ ? isALPHA(nextchr) : isALPHA_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NALPHALUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALPHAUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_alpha,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isALPHA(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case ASCII:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (!isASCII(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NASCII:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (isASCII(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case CNTRLL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case CNTRL:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == CNTRL
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case CNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case CNTRLUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == CNTRLUTF8
+ ? swash_fetch(PL_utf8_cntrl, (U8*)locinput)
+ : isCNTRL_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == CNTRLUTF8
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NCNTRLL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NCNTRL:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == CNTRL
+ ? isCNTRL(nextchr) : isCNTRL_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case NCNTRLLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NCNTRLUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_cntrl,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isCNTRL(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case GRAPHL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case GRAPH:
+ if (!nextchr)
+ sayNO;
+ if (!(OP(scan) == GRAPH
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case EOS:
- if (PL_regeol != locinput)
+ case GRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case GRAPHUTF8:
+ if (!nextchr)
sayNO;
- break;
- case SANYUTF8:
if (nextchr & 0x80) {
- locinput += PL_utf8skip[nextchr];
- if (locinput > PL_regeol)
+ if (!(OP(scan) == GRAPHUTF8
+ ? swash_fetch(PL_utf8_graph, (U8*)locinput)
+ : isGRAPH_LC_utf8((U8*)locinput)))
+ {
sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!nextchr && locinput >= PL_regeol)
+ if (!(OP(scan) == GRAPHUTF8
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case SANY:
- if (!nextchr && locinput >= PL_regeol)
+ case NGRAPHL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NGRAPH:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == GRAPH
+ ? isGRAPH(nextchr) : isGRAPH_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ANYUTF8:
+ case NGRAPHLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NGRAPHUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
if (nextchr & 0x80) {
- locinput += PL_utf8skip[nextchr];
- if (locinput > PL_regeol)
+ if (swash_fetch(PL_utf8_graph,(U8*)locinput))
sayNO;
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
- sayNO;
- nextchr = UCHARAT(++locinput);
- break;
- case REG_ANY:
- if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ if (isGRAPH(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case EXACT:
- s = (char *) OPERAND(scan);
- ln = UCHARAT(s++);
- /* Inline the first character, for speed. */
- if (UCHARAT(s) != nextchr)
- sayNO;
- if (PL_regeol - locinput < ln)
+ case LOWERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case LOWER:
+ if (!nextchr)
sayNO;
- if (ln > 1 && memNE(s, locinput, ln))
+ if (!(OP(scan) == LOWER
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr)))
sayNO;
- locinput += ln;
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case EXACTFL:
+ case LOWERLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case EXACTF:
- s = (char *) OPERAND(scan);
- ln = UCHARAT(s++);
-
- if (UTF) {
- char *l = locinput;
- char *e = s + ln;
- c1 = OP(scan) == EXACTF;
- while (s < e) {
- if (l >= PL_regeol)
- sayNO;
- if (utf8_to_uv((U8*)s, 0) != (c1 ?
- toLOWER_utf8((U8*)l) :
- toLOWER_LC_utf8((U8*)l)))
- {
- sayNO;
- }
- s += UTF8SKIP(s);
- l += UTF8SKIP(l);
+ case LOWERUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == LOWERUTF8
+ ? swash_fetch(PL_utf8_lower, (U8*)locinput)
+ : isLOWER_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
}
- locinput = l;
+ locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
-
- /* Inline the first character, for speed. */
- if (UCHARAT(s) != nextchr &&
- UCHARAT(s) != ((OP(scan) == EXACTF)
- ? PL_fold : PL_fold_locale)[nextchr])
+ if (!(OP(scan) == LOWERUTF8
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr)))
sayNO;
- if (PL_regeol - locinput < ln)
- sayNO;
- if (ln > 1 && (OP(scan) == EXACTF
- ? ibcmp(s, locinput, ln)
- : ibcmp_locale(s, locinput, ln)))
- sayNO;
- locinput += ln;
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case ANYOFUTF8:
- s = (char *) OPERAND(scan);
- if (!REGINCLASSUTF8(scan, (U8*)locinput))
+ case NLOWERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NLOWER:
+ if (!nextchr)
sayNO;
- if (locinput >= PL_regeol)
+ if (OP(scan) == LOWER
+ ? isLOWER(nextchr) : isLOWER_LC(nextchr))
sayNO;
- locinput += PL_utf8skip[nextchr];
- nextchr = UCHARAT(locinput);
+ nextchr = UCHARAT(++locinput);
break;
- case ANYOF:
- s = (char *) OPERAND(scan);
- if (nextchr < 0)
- nextchr = UCHARAT(locinput);
- if (!REGINCLASS(s, nextchr))
- sayNO;
+ case NLOWERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NLOWERUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_lower,(U8*)locinput))
+ sayNO;
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isLOWER(nextchr))
+ sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ALNUML:
+ case PRINTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case ALNUM:
+ case PRINT:
if (!nextchr)
sayNO;
- if (!(OP(scan) == ALNUM
- ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ if (!(OP(scan) == PRINT
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case ALNUMLUTF8:
+ case PRINTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case ALNUMUTF8:
+ case PRINTUTF8:
if (!nextchr)
sayNO;
if (nextchr & 0x80) {
- if (!(OP(scan) == ALNUMUTF8
- ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
- : isALNUM_LC_utf8((U8*)locinput)))
+ if (!(OP(scan) == PRINTUTF8
+ ? swash_fetch(PL_utf8_print, (U8*)locinput)
+ : isPRINT_LC_utf8((U8*)locinput)))
{
sayNO;
}
nextchr = UCHARAT(locinput);
break;
}
- if (!(OP(scan) == ALNUMUTF8
- ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ if (!(OP(scan) == PRINTUTF8
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NALNUML:
+ case NPRINTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NALNUM:
- if (!nextchr && locinput >= PL_regeol)
+ case NPRINT:
+ if (!nextchr)
sayNO;
- if (OP(scan) == NALNUM
- ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ if (OP(scan) == PRINT
+ ? isPRINT(nextchr) : isPRINT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NALNUMLUTF8:
+ case NPRINTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NALNUMUTF8:
+ case NPRINTUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
- if (OP(scan) == NALNUMUTF8
- ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
- : isALNUM_LC_utf8((U8*)locinput))
- {
+ if (swash_fetch(PL_utf8_print,(U8*)locinput))
sayNO;
- }
locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (OP(scan) == NALNUMUTF8
- ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ if (isPRINT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case BOUNDL:
- case NBOUNDL:
+ case PUNCTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case BOUND:
- case NBOUND:
- /* was last char in word? */
- ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev;
- if (OP(scan) == BOUND || OP(scan) == NBOUND) {
- ln = isALNUM(ln);
- n = isALNUM(nextchr);
- }
- else {
- ln = isALNUM_LC(ln);
- n = isALNUM_LC(nextchr);
- }
- if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
+ case PUNCT:
+ if (!nextchr)
sayNO;
+ if (!(OP(scan) == PUNCT
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case BOUNDLUTF8:
- case NBOUNDLUTF8:
+ case PUNCTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case BOUNDUTF8:
- case NBOUNDUTF8:
- /* was last char in word? */
- ln = (locinput != PL_regbol)
- ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
- if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
- ln = isALNUM_uni(ln);
- n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
- }
- else {
- ln = isALNUM_LC_uni(ln);
- n = isALNUM_LC_utf8((U8*)locinput);
+ case PUNCTUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == PUNCTUTF8
+ ? swash_fetch(PL_utf8_punct, (U8*)locinput)
+ : isPUNCT_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += PL_utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
}
- if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
+ if (!(OP(scan) == PUNCTUTF8
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr)))
sayNO;
+ nextchr = UCHARAT(++locinput);
break;
- case SPACEL:
+ case NPUNCTL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case SPACE:
- if (!nextchr && locinput >= PL_regeol)
+ case NPUNCT:
+ if (!nextchr)
sayNO;
- if (!(OP(scan) == SPACE
- ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ if (OP(scan) == PUNCT
+ ? isPUNCT(nextchr) : isPUNCT_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case SPACELUTF8:
+ case NPUNCTLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case SPACEUTF8:
+ case NPUNCTUTF8:
if (!nextchr && locinput >= PL_regeol)
sayNO;
if (nextchr & 0x80) {
- if (!(OP(scan) == SPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
- : isSPACE_LC_utf8((U8*)locinput)))
- {
+ if (swash_fetch(PL_utf8_punct,(U8*)locinput))
sayNO;
- }
locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!(OP(scan) == SPACEUTF8
- ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ if (isPUNCT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NSPACEL:
+ case UPPERL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NSPACE:
+ case UPPER:
if (!nextchr)
sayNO;
- if (OP(scan) == SPACE
- ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ if (!(OP(scan) == UPPER
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NSPACELUTF8:
+ case UPPERLUTF8:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
- case NSPACEUTF8:
+ case UPPERUTF8:
if (!nextchr)
sayNO;
if (nextchr & 0x80) {
- if (OP(scan) == NSPACEUTF8
- ? swash_fetch(PL_utf8_space,(U8*)locinput)
- : isSPACE_LC_utf8((U8*)locinput))
+ if (!(OP(scan) == UPPERUTF8
+ ? swash_fetch(PL_utf8_upper, (U8*)locinput)
+ : isUPPER_LC_utf8((U8*)locinput)))
{
sayNO;
}
nextchr = UCHARAT(locinput);
break;
}
- if (OP(scan) == NSPACEUTF8
- ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ if (!(OP(scan) == UPPERUTF8
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr)))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case DIGIT:
- if (!isDIGIT(nextchr))
+ case NUPPERL:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NUPPER:
+ if (!nextchr)
+ sayNO;
+ if (OP(scan) == UPPER
+ ? isUPPER(nextchr) : isUPPER_LC(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case DIGITUTF8:
+ case NUPPERLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NUPPERUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
if (nextchr & 0x80) {
- if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
+ if (swash_fetch(PL_utf8_upper,(U8*)locinput))
sayNO;
locinput += PL_utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
}
- if (!isDIGIT(nextchr))
+ if (isUPPER(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NDIGIT:
+ case XDIGIT:
if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (isDIGIT(nextchr))
+ if (!isXDIGIT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
- case NDIGITUTF8:
+ case NXDIGIT:
if (!nextchr && locinput >= PL_regeol)
sayNO;
- if (nextchr & 0x80) {
- if (swash_fetch(PL_utf8_digit,(U8*)locinput))
- sayNO;
- locinput += PL_utf8skip[nextchr];
- nextchr = UCHARAT(locinput);
- break;
- }
- if (isDIGIT(nextchr))
+ if (isXDIGIT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
S_reginclass(pTHX_ register char *p, register I32 c)
{
dTHR;
- char flags = *p;
+ char flags = ANYOF_FLAGS(p);
bool match = FALSE;
c &= 0xFF;
- if (ANYOF_TEST(p, c))
+ if (ANYOF_BITMAP_TEST(p, c))
match = TRUE;
else if (flags & ANYOF_FOLD) {
I32 cf;
}
else
cf = PL_fold[c];
- if (ANYOF_TEST(p, cf))
+ if (ANYOF_BITMAP_TEST(p, cf))
match = TRUE;
}
- if (!match && (flags & ANYOF_ISA)) {
+ if (!match && (flags & ANYOF_CLASS)) {
PL_reg_flags |= RF_tainted;
-
- if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) ||
- ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) ||
- ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) ||
- ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c)))
+ if (
+ (ANYOF_CLASS_TEST(p, ANYOF_ALNUM) && isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALNUM) && !isALNUM_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_SPACE) && isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NSPACE) && !isSPACE_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_DIGIT) && isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NDIGIT) && !isDIGIT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ALNUMC) && isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALNUMC) && !isALNUMC_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ALPHA) && isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NALPHA) && !isALPHA_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_ASCII) && isASCII(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NASCII) && !isASCII(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_CNTRL) && isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NCNTRL) && !isCNTRL_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_GRAPH) && isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NGRAPH) && !isGRAPH_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_LOWER) && isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NLOWER) && !isLOWER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_PRINT) && isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NPRINT) && !isPRINT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_PUNCT) && isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NPUNCT) && !isPUNCT_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_UPPER) && isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NUPPER) && !isUPPER_LC(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_XDIGIT) && isXDIGIT(c)) ||
+ (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c))
+ ) /* How's that for a conditional? */
{
match = TRUE;
}
match = TRUE;
}
- if (!match && (flags & ANYOF_ISA)) {
- PL_reg_flags |= RF_tainted;
-
- if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) ||
- ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) ||
- ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) ||
- ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p)))
- {
- match = TRUE;
- }
- }
+ /* UTF8 combined with ANYOF_CLASS is ill-defined. */
return (flags & ANYOF_INVERT) ? !match : match;
}
}
return s;
}
+
+#ifdef PERL_OBJECT
+#define NO_XSLOCKS
+#include "XSUB.h"
+#endif
+
+static void
+restore_pos(pTHXo_ void *arg)
+{
+ dTHR;
+ if (PL_reg_eval_set) {
+ if (PL_reg_oldsaved) {
+ PL_reg_re->subbeg = PL_reg_oldsaved;
+ PL_reg_re->sublen = PL_reg_oldsavedlen;
+ RX_MATCH_COPIED_on(PL_reg_re);
+ }
+ PL_reg_magic->mg_len = PL_reg_oldpos;
+ PL_reg_eval_set = 0;
+ PL_curpm = PL_reg_oldcurpm;
+ }
+}
+
projects / p5sagit/p5-mst-13.2.git / blobdiff