/* *These* symbols are masked to allow static link. */
# define Perl_pregfree my_regfree
# define Perl_regnext my_regnext
-# define save_re_context my_save_re_context
+# define Perl_save_re_context my_save_re_context
+# 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.
* regular-expression syntax might require a total rethink.
*/
#include "EXTERN.h"
+#define PERL_IN_REGCOMP_C
#include "perl.h"
#ifndef PERL_IN_XSUB_RE
* Forward declarations for pregcomp()'s friends.
*/
-#ifndef PERL_OBJECT
-static regnode *reg _((I32, I32 *));
-static regnode *reganode _((U8, U32));
-static regnode *regatom _((I32 *));
-static regnode *regbranch _((I32 *, I32));
-static void regc _((U8, char *));
-static void reguni _((UV, char *, I32*));
-static regnode *regclass _((void));
-static regnode *regclassutf8 _((void));
-STATIC I32 regcurly _((char *));
-static regnode *reg_node _((U8));
-static regnode *regpiece _((I32 *));
-static void reginsert _((U8, regnode *));
-static void regoptail _((regnode *, regnode *));
-static void regtail _((regnode *, regnode *));
-static char* regwhite _((char *, char *));
-static char* nextchar _((void));
-static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
-#endif
-
-/* Length of a variant. */
-
-#ifndef PERL_OBJECT
-typedef struct {
- I32 len_min;
- I32 len_delta;
- I32 pos_min; /* CC */
- I32 pos_delta; /* CC */
- SV *last_found;
- I32 last_end; /* min value, <0 unless valid. */
- I32 last_start_min; /* CC */
- I32 last_start_max; /* CC */
- SV **longest; /* Either &l_fixed, or &l_float. */
- SV *longest_fixed;
- I32 offset_fixed; /* CC */
- SV *longest_float;
- I32 offset_float_min; /* CC */
- I32 offset_float_max; /* CC */
- I32 flags;
-} scan_data_t;
-#endif
-
-static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
+static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+ 0, 0, 0 };
#define SF_BEFORE_EOL (SF_BEFORE_SEOL|SF_BEFORE_MEOL)
#define SF_BEFORE_SEOL 0x1
#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
STATIC void
-scan_commit(scan_data_t *data)
+S_clear_re(pTHX_ void *r)
+{
+ ReREFCNT_dec((regexp *)r);
+}
+
+STATIC void
+S_scan_commit(pTHX_ scan_data_t *data)
{
dTHR;
STRLEN l = CHR_SVLEN(data->last_found);
to the position after last scanned or to NULL. */
STATIC I32
-study_chunk(regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
+S_study_chunk(pTHX_ regnode **scanp, I32 *deltap, regnode *last, scan_data_t *data, U32 flags)
/* scanp: Start here (read-write). */
/* deltap: Write maxlen-minlen here. */
/* last: Stop before this one. */
while (scan && OP(scan) != END && scan < last) {
/* Peephole optimizer: */
- if (regkind[(U8)OP(scan)] == EXACT) {
+ if (PL_regkind[(U8)OP(scan)] == EXACT) {
regnode *n = regnext(scan);
U32 stringok = 1;
#ifdef DEBUGGING
next = scan + (*OPERAND(scan) + 2 - 1)/sizeof(regnode) + 2;
/* Skip NOTHING, merge EXACT*. */
while (n &&
- ( regkind[(U8)OP(n)] == NOTHING ||
+ ( PL_regkind[(U8)OP(n)] == NOTHING ||
(stringok && (OP(n) == OP(scan))))
&& NEXT_OFF(n)
&& NEXT_OFF(scan) + NEXT_OFF(n) < I16_MAX) {
if (OP(n) == TAIL || n > next)
stringok = 0;
- if (regkind[(U8)OP(n)] == NOTHING) {
+ if (PL_regkind[(U8)OP(n)] == NOTHING) {
NEXT_OFF(scan) += NEXT_OFF(n);
next = n + NODE_STEP_REGNODE;
#ifdef DEBUGGING
* don't initialize the OP() slot of a node when that node
* is occupied by just the trailing null of the string in
* an EXACT node */
- if (regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
+ if (PL_regkind[(U8)OP(n)] != NOTHING || OP(n) == NOTHING) {
OP(n) = OPTIMIZED;
NEXT_OFF(n) = 0;
}
/* Skip NOTHING and LONGJMP. */
while ((n = regnext(n))
- && ((regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
+ && ((PL_regkind[(U8)OP(n)] == NOTHING && (noff = NEXT_OFF(n)))
|| ((OP(n) == LONGJMP) && (noff = ARG(n))))
&& off + noff < max)
off += noff;
data->flags &= ~SF_BEFORE_EOL;
}
}
- else if (regkind[(U8)OP(scan)] == EXACT) {
+ else if (PL_regkind[(U8)OP(scan)] == EXACT) {
I32 l = *OPERAND(scan);
if (flags & SCF_DO_SUBSTR)
scan_commit(data);
if (data && (flags & SCF_DO_SUBSTR))
data->pos_min += l;
}
- else if (strchr(varies,OP(scan))) {
+ else if (strchr(PL_varies,OP(scan))) {
I32 mincount, maxcount, minnext, deltanext, pos_before, fl;
regnode *oscan = scan;
- switch (regkind[(U8)OP(scan)]) {
+ switch (PL_regkind[(U8)OP(scan)]) {
case WHILEM:
scan = NEXTOPER(scan);
goto finish;
scan = next;
if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0)
&& !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
- && maxcount <= 10000) /* Complement check for big count */
- warner(WARN_UNSAFE, "Strange *+?{} on zero-length expression");
+ && maxcount <= REG_INFTY/3) /* Complement check for big count */
+ Perl_warner(aTHX_ WARN_UNSAFE, "Strange *+?{} on zero-length expression");
min += minnext * mincount;
is_inf_internal |= (maxcount == REG_INFTY
&& (minnext + deltanext) > 0
/* Skip open. */
nxt = regnext(nxt);
- if (!strchr(simple,OP(nxt))
- && !(regkind[(U8)OP(nxt)] == EXACT
+ if (!strchr(PL_simple,OP(nxt))
+ && !(PL_regkind[(U8)OP(nxt)] == EXACT
&& *OPERAND(nxt) == 1))
goto nogo;
nxt2 = nxt;
l -= old;
/* Get the added string: */
- last_str = newSVpv(s + old, l);
+ last_str = newSVpvn(s + old, l);
if (deltanext == 0 && pos_before == b) {
/* What was added is a constant string */
if (mincount > 1) {
data->flags |= SF_HAS_EVAL;
optimize_curly_tail:
if (OP(oscan) != CURLYX) {
- while (regkind[(U8)OP(next = regnext(oscan))] == NOTHING
+ while (PL_regkind[(U8)OP(next = regnext(oscan))] == NOTHING
&& NEXT_OFF(next))
NEXT_OFF(oscan) += NEXT_OFF(next);
}
break;
}
}
- else if (strchr(simple,OP(scan)) || regkind[(U8)OP(scan)] == ANYUTF8) {
+ else if (strchr(PL_simple,OP(scan)) || PL_regkind[(U8)OP(scan)] == ANYUTF8) {
if (flags & SCF_DO_SUBSTR) {
scan_commit(data);
data->pos_min++;
}
min++;
}
- else if (regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
+ else if (PL_regkind[(U8)OP(scan)] == EOL && flags & SCF_DO_SUBSTR) {
data->flags |= (OP(scan) == MEOL
? SF_BEFORE_MEOL
: SF_BEFORE_SEOL);
}
- else if (regkind[(U8)OP(scan)] == BRANCHJ
+ else if (PL_regkind[(U8)OP(scan)] == BRANCHJ
&& (scan->flags || data)
&& (OP(scan) == IFMATCH || OP(scan) == UNLESSM)) {
I32 deltanext, minnext;
}
STATIC I32
-add_data(I32 n, char *s)
+S_add_data(pTHX_ I32 n, char *s)
{
dTHR;
if (PL_regcomp_rx->data) {
return PL_regcomp_rx->data->count - n;
}
+void
+Perl_reginitcolors(pTHX)
+{
+ dTHR;
+ int i = 0;
+ char *s = PerlEnv_getenv("PERL_RE_COLORS");
+
+ if (s) {
+ PL_colors[0] = s = savepv(s);
+ while (++i < 6) {
+ s = strchr(s, '\t');
+ if (s) {
+ *s = '\0';
+ PL_colors[i] = ++s;
+ }
+ else
+ PL_colors[i] = s = "";
+ }
+ } else {
+ while (i < 6)
+ PL_colors[i++] = "";
+ }
+ PL_colorset = 1;
+}
+
/*
- pregcomp - compile a regular expression into internal code
*
* of the structure of the compiled regexp. [I'll say.]
*/
regexp *
-pregcomp(char *exp, char *xend, PMOP *pm)
+Perl_pregcomp(pTHX_ char *exp, char *xend, PMOP *pm)
{
dTHR;
register regexp *r;
if (exp == NULL)
FAIL("NULL regexp argument");
- if (PL_curcop == &compiling ? (PL_hints & HINT_UTF8) : IN_UTF8)
+ if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8)
PL_reg_flags |= RF_utf8;
else
PL_reg_flags = 0;
PL_regprecomp = savepvn(exp, xend - exp);
- DEBUG_r(PerlIO_printf(Perl_debug_log, "compiling RE `%*s'\n",
- xend - exp, PL_regprecomp));
+ DEBUG_r(if (!PL_colorset) reginitcolors());
+ DEBUG_r(PerlIO_printf(Perl_debug_log, "%sCompiling%s RE `%s%*s%s'\n",
+ PL_colors[4],PL_colors[5],PL_colors[0],
+ xend - exp, PL_regprecomp, PL_colors[1]));
PL_regflags = pm->op_pmflags;
PL_regsawback = 0;
PL_regnpar = 1;
PL_regsize = 0L;
PL_regcode = &PL_regdummy;
- regc((U8)MAGIC, (char*)PL_regcode);
+ regc((U8)REG_MAGIC, (char*)PL_regcode);
if (reg(0, &flags) == NULL) {
Safefree(PL_regprecomp);
PL_regprecomp = Nullch;
}
DEBUG_r(PerlIO_printf(Perl_debug_log, "size %d ", PL_regsize));
- DEBUG_r(
- if (!PL_colorset) {
- int i = 0;
- char *s = PerlEnv_getenv("TERMCAP_COLORS");
-
- PL_colorset = 1;
- if (s) {
- PL_colors[0] = s = savepv(s);
- while (++i < 4) {
- s = strchr(s, '\t');
- if (!s)
- FAIL("Not enough TABs in TERMCAP_COLORS");
- *s = '\0';
- PL_colors[i] = ++s;
- }
- }
- else {
- while (i < 4)
- PL_colors[i++] = "";
- }
- /* Reset colors: */
- PerlIO_printf(Perl_debug_log, "%s%s%s%s",
- PL_colors[0],PL_colors[1],PL_colors[2],PL_colors[3]);
- }
- );
-
/* Small enough for pointer-storage convention?
If extralen==0, this means that we will not need long jumps. */
if (PL_regsize >= 0x10000L && PL_extralen)
r->refcnt = 1;
r->prelen = xend - exp;
r->precomp = PL_regprecomp;
- r->subbeg = r->subbase = NULL;
- r->nparens = PL_regnpar - 1; /* set early to validate backrefs */
+ r->subbeg = NULL;
+ r->reganch = pm->op_pmflags & PMf_COMPILETIME;
+ r->nparens = PL_regnpar - 1; /* set early to validate backrefs */
+
+ r->substrs = 0; /* Useful during FAIL. */
+ r->startp = 0; /* Useful during FAIL. */
+ r->endp = 0; /* Useful during FAIL. */
+
PL_regcomp_rx = r;
/* Second pass: emit code. */
PL_regcode = r->program;
/* Store the count of eval-groups for security checks: */
PL_regcode->next_off = ((PL_seen_evals > U16_MAX) ? U16_MAX : PL_seen_evals);
- regc((U8)MAGIC, (char*) PL_regcode++);
+ regc((U8)REG_MAGIC, (char*) PL_regcode++);
r->data = 0;
if (reg(0, &flags) == NULL)
return(NULL);
/* Dig out information for optimizations. */
- r->reganch = pm->op_pmflags & PMf_COMPILETIME;
+ r->reganch = pm->op_pmflags & PMf_COMPILETIME; /* Again? */
pm->op_pmflags = PL_regflags;
if (UTF)
r->reganch |= ROPT_UTF8;
(OP(first) == BRANCH && OP(regnext(first)) != BRANCH) ||
(OP(first) == PLUS) ||
(OP(first) == MINMOD) ||
- (regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
+ (PL_regkind[(U8)OP(first)] == CURLY && ARG1(first) > 0) ) {
if (OP(first) == PLUS)
sawplus = 1;
else
/* Starting-point info. */
again:
if (OP(first) == EXACT); /* Empty, get anchored substr later. */
- else if (strchr(simple+4,OP(first)))
+ else if (strchr(PL_simple+4,OP(first)))
r->regstclass = first;
- else if (regkind[(U8)OP(first)] == BOUND ||
- regkind[(U8)OP(first)] == NBOUND)
+ else if (PL_regkind[(U8)OP(first)] == BOUND ||
+ PL_regkind[(U8)OP(first)] == NBOUND)
r->regstclass = first;
- else if (regkind[(U8)OP(first)] == BOL) {
+ else if (PL_regkind[(U8)OP(first)] == BOL) {
r->reganch |= (OP(first) == MBOL ? ROPT_ANCH_MBOL: ROPT_ANCH_BOL);
first = NEXTOPER(first);
goto again;
goto again;
}
else if ((OP(first) == STAR &&
- regkind[(U8)OP(NEXTOPER(first))] == ANY) &&
+ PL_regkind[(U8)OP(NEXTOPER(first))] == REG_ANY) &&
!(r->reganch & ROPT_ANCH) )
{
/* turn .* into ^.* with an implied $*=1 */
*/
minlen = 0;
- data.longest_fixed = newSVpv("",0);
- data.longest_float = newSVpv("",0);
- data.last_found = newSVpv("",0);
+ data.longest_fixed = newSVpvn("",0);
+ data.longest_float = newSVpvn("",0);
+ data.last_found = newSVpvn("",0);
data.longest = &(data.longest_fixed);
first = scan;
|| (data.flags & SF_FL_BEFORE_EOL
&& (!(data.flags & SF_FL_BEFORE_MEOL)
|| (PL_regflags & PMf_MULTILINE)))) {
+ int t;
+
if (SvCUR(data.longest_fixed) /* ok to leave SvCUR */
&& data.offset_fixed == data.offset_float_min
&& SvCUR(data.longest_fixed) == SvCUR(data.longest_float))
r->float_substr = data.longest_float;
r->float_min_offset = data.offset_float_min;
r->float_max_offset = data.offset_float_max;
- fbm_compile(r->float_substr, 0);
- BmUSEFUL(r->float_substr) = 100;
- if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
- && (!(data.flags & SF_FL_BEFORE_MEOL)
- || (PL_regflags & PMf_MULTILINE)))
- SvTAIL_on(r->float_substr);
+ t = (data.flags & SF_FL_BEFORE_EOL /* Can't have SEOL and MULTI */
+ && (!(data.flags & SF_FL_BEFORE_MEOL)
+ || (PL_regflags & PMf_MULTILINE)));
+ fbm_compile(r->float_substr, t ? FBMcf_TAIL : 0);
}
else {
remove_float:
|| (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
&& (!(data.flags & SF_FIX_BEFORE_MEOL)
|| (PL_regflags & PMf_MULTILINE)))) {
+ int t;
+
r->anchored_substr = data.longest_fixed;
r->anchored_offset = data.offset_fixed;
- fbm_compile(r->anchored_substr, 0);
- BmUSEFUL(r->anchored_substr) = 100;
- if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
- && (!(data.flags & SF_FIX_BEFORE_MEOL)
- || (PL_regflags & PMf_MULTILINE)))
- SvTAIL_on(r->anchored_substr);
+ t = (data.flags & SF_FIX_BEFORE_EOL /* Can't have SEOL and MULTI */
+ && (!(data.flags & SF_FIX_BEFORE_MEOL)
+ || (PL_regflags & PMf_MULTILINE)));
+ fbm_compile(r->anchored_substr, t ? FBMcf_TAIL : 0);
}
else {
r->anchored_substr = Nullsv;
r->reganch |= ROPT_LOOKBEHIND_SEEN;
if (PL_regseen & REG_SEEN_EVAL)
r->reganch |= ROPT_EVAL_SEEN;
- Newz(1002, r->startp, PL_regnpar, char*);
- Newz(1002, r->endp, PL_regnpar, char*);
+ Newz(1002, r->startp, PL_regnpar, I32);
+ Newz(1002, r->endp, PL_regnpar, I32);
DEBUG_r(regdump(r));
return(r);
}
* follows makes it hard to avoid.
*/
STATIC regnode *
-reg(I32 paren, I32 *flagp)
+S_reg(pTHX_ I32 paren, I32 *flagp)
/* paren: Parenthesized? 0=top, 1=(, inside: changed to letter. */
{
dTHR;
AV *av;
if (PL_regcomp_parse - 1 - s)
- sv = newSVpv(s, PL_regcomp_parse - 1 - s);
+ sv = newSVpvn(s, PL_regcomp_parse - 1 - s);
else
- sv = newSVpv("", 0);
+ sv = newSVpvn("", 0);
rop = sv_compile_2op(sv, &sop, "re", &av);
- n = add_data(3, "nso");
+ n = add_data(3, "nop");
PL_regcomp_rx->data->data[n] = (void*)rop;
- PL_regcomp_rx->data->data[n+1] = (void*)av;
- PL_regcomp_rx->data->data[n+2] = (void*)sop;
+ PL_regcomp_rx->data->data[n+1] = (void*)sop;
+ PL_regcomp_rx->data->data[n+2] = (void*)av;
SvREFCNT_dec(sv);
}
- else { /* First pass */
- if (PL_reginterp_cnt < ++PL_seen_evals && PL_curcop != &compiling)
+ else { /* First pass */
+ if (PL_reginterp_cnt < ++PL_seen_evals
+ && PL_curcop != &PL_compiling)
/* No compiled RE interpolated, has runtime
components ===> unsafe. */
FAIL("Eval-group not allowed at runtime, use re 'eval'");
else
regtail(br, reganode(LONGJMP, 0));
c = *nextchar();
+ if (flags&HASWIDTH)
+ *flagp |= HASWIDTH;
if (c == '|') {
lastbr = reganode(IFTHEN, 0); /* Fake one for optimizer. */
regbranch(&flags, 1);
regtail(ret, lastbr);
+ if (flags&HASWIDTH)
+ *flagp |= HASWIDTH;
c = *nextchar();
}
else
}
/* Check for proper termination. */
- if (paren && (PL_regcomp_parse >= PL_regxend || *nextchar() != ')')) {
- FAIL("unmatched () in regexp");
+ if (paren) {
+ PL_regflags = oregflags;
+ if (PL_regcomp_parse >= PL_regxend || *nextchar() != ')') {
+ FAIL("unmatched () in regexp");
+ }
}
else if (!paren && PL_regcomp_parse < PL_regxend) {
if (*PL_regcomp_parse == ')') {
FAIL("junk on end of regexp"); /* "Can't happen". */
/* NOTREACHED */
}
- if (paren != 0) {
- PL_regflags = oregflags;
- }
return(ret);
}
* Implements the concatenation operator.
*/
STATIC regnode *
-regbranch(I32 *flagp, I32 first)
+S_regbranch(pTHX_ I32 *flagp, I32 first)
{
dTHR;
register regnode *ret;
* endmarker role is not redundant.
*/
STATIC regnode *
-regpiece(I32 *flagp)
+S_regpiece(pTHX_ I32 *flagp)
{
dTHR;
register regnode *ret;
goto do_curly;
}
nest_check:
- if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > 10000) {
- warner(WARN_UNSAFE, "%.*s matches null string many times",
+ if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
+ Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times",
PL_regcomp_parse - origparse, origparse);
}
* [Yes, it is worth fixing, some scripts can run twice the speed.]
*/
STATIC regnode *
-regatom(I32 *flagp)
+S_regatom(pTHX_ I32 *flagp)
{
dTHR;
register regnode *ret = 0;
if (PL_regflags & PMf_SINGLELINE)
ret = reg_node(SANY);
else
- ret = reg_node(ANY);
+ ret = reg_node(REG_ANY);
*flagp |= HASWIDTH|SIMPLE;
}
PL_regnaughty++;
break;
case 'b':
PL_seen_zerolen++;
+ PL_regseen |= REG_SEEN_LOOKBEHIND;
ret = reg_node(
UTF
? (LOC ? BOUNDLUTF8 : BOUNDUTF8)
break;
case 'B':
PL_seen_zerolen++;
+ PL_regseen |= REG_SEEN_LOOKBEHIND;
ret = reg_node(
UTF
? (LOC ? NBOUNDLUTF8 : NBOUNDUTF8)
FAIL("trailing \\ in regexp");
/* FALL THROUGH */
default:
+ /* Do not generate `unrecognized' warnings here, we fall
+ back into the quick-grab loop below */
goto defchar;
}
break;
FAIL("trailing \\ in regexp");
/* FALL THROUGH */
default:
+ if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p))
+ Perl_warner(aTHX_ WARN_UNSAFE,
+ "/%.127s/: Unrecognized escape \\%c passed through",
+ PL_regprecomp,
+ *p);
goto normal_default;
}
break;
}
STATIC char *
-regwhite(char *p, char *e)
+S_regwhite(pTHX_ char *p, char *e)
{
while (p < e) {
if (isSPACE(*p))
return p;
}
+/* parse POSIX character classes like [[:foo:]] */
+STATIC char*
+S_regpposixcc(pTHX_ I32 value)
+{
+ dTHR;
+ char *posixcc = 0;
+
+ if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
+ /* I smell either [: or [= or [. -- POSIX has been here, right? */
+ (*PL_regcomp_parse == ':' ||
+ *PL_regcomp_parse == '=' ||
+ *PL_regcomp_parse == '.')) {
+ char c = *PL_regcomp_parse;
+ char* s = PL_regcomp_parse++;
+
+ while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != c)
+ PL_regcomp_parse++;
+ if (PL_regcomp_parse == PL_regxend)
+ /* Grandfather lone [:, [=, [. */
+ PL_regcomp_parse = s;
+ else {
+ PL_regcomp_parse++; /* skip over the c */
+ if (*PL_regcomp_parse == ']') {
+ /* Not Implemented Yet.
+ * (POSIX Extended Character Classes, that is)
+ * The text between e.g. [: and :] would start
+ * at s + 1 and stop at regcomp_parse - 2. */
+ if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
+ Perl_warner(aTHX_ WARN_UNSAFE,
+ "Character class syntax [%c %c] is reserved for future extensions", c, c);
+ PL_regcomp_parse++; /* skip over the ending ] */
+ posixcc = s + 1;
+ }
+ else {
+ /* maternal grandfather */
+ PL_regcomp_parse = s;
+ }
+ }
+ }
+
+ return posixcc;
+}
+
STATIC regnode *
-regclass(void)
+S_regclass(pTHX)
{
dTHR;
register char *opnd, *s;
while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != ']') {
skipcond:
value = UCHARAT(PL_regcomp_parse++);
- if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
- /* I smell either [: or [= or [. -- POSIX has been here, right? */
- (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) {
- char posixccc = *PL_regcomp_parse;
- char* posixccs = PL_regcomp_parse++;
-
- while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc)
- PL_regcomp_parse++;
- if (PL_regcomp_parse == PL_regxend)
- /* Grandfather lone [:, [=, [. */
- PL_regcomp_parse = posixccs;
- else {
- PL_regcomp_parse++; /* skip over the posixccc */
- if (*PL_regcomp_parse == ']') {
- /* Not Implemented Yet.
- * (POSIX Extended Character Classes, that is)
- * The text between e.g. [: and :] would start
- * at posixccs + 1 and stop at regcomp_parse - 2. */
- if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
- warner(WARN_UNSAFE,
- "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
- PL_regcomp_parse++; /* skip over the ending ] */
- }
- }
- }
- if (value == '\\') {
+ if (value == '[')
+ (void)regpposixcc(value); /* ignore the return value for now */
+ else if (value == '\\') {
value = UCHARAT(PL_regcomp_parse++);
switch (value) {
case 'w':
if (!SIZE_ONLY) {
#ifndef ASCIIish
if ((isLOWER(lastvalue) && isLOWER(value)) ||
- (isUPPER(lastvalue) && isUPPER(value))) {
+ (isUPPER(lastvalue) && isUPPER(value)))
+ {
+ I32 i;
if (isLOWER(lastvalue)) {
for (i = lastvalue; i <= value; i++)
if (isLOWER(i))
if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
for (value = 0; value < 256; ++value) {
if (ANYOF_TEST(opnd, value)) {
- I32 cf = fold[value];
+ I32 cf = PL_fold[value];
ANYOF_SET(opnd, cf);
}
}
}
STATIC regnode *
-regclassutf8(void)
+S_regclassutf8(pTHX)
{
register char *opnd, *e;
register U32 value;
flags |= ANYOF_FOLD;
if (LOC)
flags |= ANYOF_LOCALE;
- listsv = newSVpv("# comment\n",0);
+ listsv = newSVpvn("# comment\n",10);
}
if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
PL_regcomp_parse += numlen;
- if (value == '[' && PL_regcomp_parse + 1 < PL_regxend &&
- /* I smell either [: or [= or [. -- POSIX has been here, right? */
- (*PL_regcomp_parse == ':' || *PL_regcomp_parse == '=' || *PL_regcomp_parse == '.')) {
- char posixccc = *PL_regcomp_parse;
- char* posixccs = PL_regcomp_parse++;
-
- while (PL_regcomp_parse < PL_regxend && *PL_regcomp_parse != posixccc)
- PL_regcomp_parse++;
- if (PL_regcomp_parse == PL_regxend)
- /* Grandfather lone [:, [=, [. */
- PL_regcomp_parse = posixccs;
- else {
- PL_regcomp_parse++; /* skip over the posixccc */
- if (*PL_regcomp_parse == ']') {
- /* Not Implemented Yet.
- * (POSIX Extended Character Classes, that is)
- * The text between e.g. [: and :] would start
- * at posixccs + 1 and stop at regcomp_parse - 2. */
- if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
- warner(WARN_UNSAFE,
- "Character class syntax [%c %c] is reserved for future extensions", posixccc, posixccc);
- PL_regcomp_parse++; /* skip over the ending ] */
- }
- }
- }
-
- if (value == '\\') {
+ if (value == '[')
+ (void)regpposixcc(value); /* ignore the return value for now */
+ else if (value == '\\') {
value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
PL_regcomp_parse += numlen;
switch (value) {
if (LOC)
flags |= ANYOF_ALNUML;
- sv_catpvf(listsv, "+utf8::IsAlnum\n");
+ Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsAlnum\n");
}
lastvalue = 123456;
continue;
if (LOC)
flags |= ANYOF_NALNUML;
- sv_catpvf(listsv,
+ Perl_sv_catpvf(aTHX_ listsv,
"-utf8::IsAlpha\n-utf8::IsDigit\n0000\t%04x\n%04x\tffff\n",
'_' - 1,
'_' + 1);
if (!SIZE_ONLY) {
if (LOC)
flags |= ANYOF_SPACEL;
- sv_catpvf(listsv, "+utf8::IsSpace\n");
+ Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsSpace\n");
if (!PL_utf8_space)
is_utf8_space((U8*)" ");
}
if (!SIZE_ONLY) {
if (LOC)
flags |= ANYOF_NSPACEL;
- sv_catpvf(listsv,
+ Perl_sv_catpvf(aTHX_ listsv,
"!utf8::IsSpace\n");
if (!PL_utf8_space)
is_utf8_space((U8*)" ");
continue;
case 'd':
if (!SIZE_ONLY) {
- sv_catpvf(listsv, "+utf8::IsDigit\n");
+ Perl_sv_catpvf(aTHX_ listsv, "+utf8::IsDigit\n");
}
lastvalue = 123456;
continue;
case 'D':
if (!SIZE_ONLY) {
- sv_catpvf(listsv,
+ Perl_sv_catpvf(aTHX_ listsv,
"!utf8::IsDigit\n");
}
lastvalue = 123456;
}
if (!SIZE_ONLY) {
if (value == 'p')
- sv_catpvf(listsv, "+utf8::%.*s\n", n, PL_regcomp_parse);
+ Perl_sv_catpvf(aTHX_ listsv, "+utf8::%.*s\n", n, PL_regcomp_parse);
else
- sv_catpvf(listsv,
+ Perl_sv_catpvf(aTHX_ listsv,
"!utf8::%.*s\n", n, PL_regcomp_parse);
}
PL_regcomp_parse = e + 1;
e = strchr(PL_regcomp_parse++, '}');
if (!e)
FAIL("Missing right brace on \\x{}");
- value = scan_hex(PL_regcomp_parse + 1, e - PL_regcomp_parse, &numlen);
+ value = scan_hex(PL_regcomp_parse, e - PL_regcomp_parse, &numlen);
PL_regcomp_parse = e + 1;
}
else {
if (lastvalue > value)
FAIL("invalid [] range in regexp");
if (!SIZE_ONLY)
- sv_catpvf(listsv, "%04x\t%04x\n", lastvalue, value);
+ Perl_sv_catpvf(aTHX_ listsv, "%04x\t%04x\n", lastvalue, value);
lastvalue = value;
range = 0;
}
continue; /* do it next time */
}
if (!SIZE_ONLY)
- sv_catpvf(listsv, "%04x\n", value);
+ Perl_sv_catpvf(aTHX_ listsv, "%04x\n", value);
}
}
}
STATIC char*
-nextchar(void)
+S_nextchar(pTHX)
{
dTHR;
char* retval = PL_regcomp_parse++;
- reg_node - emit a node
*/
STATIC regnode * /* Location. */
-reg_node(U8 op)
+S_reg_node(pTHX_ U8 op)
{
dTHR;
register regnode *ret;
- reganode - emit a node with an argument
*/
STATIC regnode * /* Location. */
-reganode(U8 op, U32 arg)
+S_reganode(pTHX_ U8 op, U32 arg)
{
dTHR;
register regnode *ret;
- regc - emit (if appropriate) a Unicode character
*/
STATIC void
-reguni(UV uv, char* s, I32* lenp)
+S_reguni(pTHX_ UV uv, char* s, I32* lenp)
{
dTHR;
if (SIZE_ONLY) {
- regc - emit (if appropriate) a byte of code
*/
STATIC void
-regc(U8 b, char* s)
+S_regc(pTHX_ U8 b, char* s)
{
dTHR;
if (!SIZE_ONLY)
* Means relocating the operand.
*/
STATIC void
-reginsert(U8 op, regnode *opnd)
+S_reginsert(pTHX_ U8 op, regnode *opnd)
{
dTHR;
register regnode *src;
register regnode *place;
register int offset = regarglen[(U8)op];
-/* (regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
+/* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
if (SIZE_ONLY) {
PL_regsize += NODE_STEP_REGNODE + offset;
- regtail - set the next-pointer at the end of a node chain of p to val.
*/
STATIC void
-regtail(regnode *p, regnode *val)
+S_regtail(pTHX_ regnode *p, regnode *val)
{
dTHR;
register regnode *scan;
- regoptail - regtail on operand of first argument; nop if operandless
*/
STATIC void
-regoptail(regnode *p, regnode *val)
+S_regoptail(pTHX_ regnode *p, regnode *val)
{
dTHR;
/* "Operandless" and "op != BRANCH" are synonymous in practice. */
if (p == NULL || SIZE_ONLY)
return;
- if (regkind[(U8)OP(p)] == BRANCH) {
+ if (PL_regkind[(U8)OP(p)] == BRANCH) {
regtail(NEXTOPER(p), val);
}
- else if ( regkind[(U8)OP(p)] == BRANCHJ) {
+ else if ( PL_regkind[(U8)OP(p)] == BRANCHJ) {
regtail(NEXTOPER(NEXTOPER(p)), val);
}
else
- regcurly - a little FSA that accepts {\d+,?\d*}
*/
STATIC I32
-regcurly(register char *s)
+S_regcurly(pTHX_ register char *s)
{
if (*s++ != '{')
return FALSE;
STATIC regnode *
-dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
+S_dumpuntil(pTHX_ regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
{
#ifdef DEBUGGING
register char op = EXACT; /* Arbitrary non-END op. */
PerlIO_printf(Perl_debug_log, "(%d)", next - start);
(void)PerlIO_putc(Perl_debug_log, '\n');
after_print:
- if (regkind[(U8)op] == BRANCHJ) {
+ if (PL_regkind[(U8)op] == BRANCHJ) {
register regnode *nnode = (OP(next) == LONGJMP
? regnext(next)
: next);
nnode = last;
node = dumpuntil(start, NEXTOPER(NEXTOPER(node)), nnode, sv, l + 1);
}
- else if (regkind[(U8)op] == BRANCH) {
+ else if (PL_regkind[(U8)op] == BRANCH) {
node = dumpuntil(start, NEXTOPER(node), next, sv, l + 1);
}
else if ( op == CURLY) { /* `next' might be very big: optimizer */
node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
NEXTOPER(node) + EXTRA_STEP_2ARGS + 1, sv, l + 1);
}
- else if (regkind[(U8)op] == CURLY && op != CURLYX) {
+ else if (PL_regkind[(U8)op] == CURLY && op != CURLYX) {
node = dumpuntil(start, NEXTOPER(node) + EXTRA_STEP_2ARGS,
next, sv, l + 1);
}
node = NEXTOPER(node);
node += ANY_SKIP;
}
- else if (regkind[(U8)op] == EXACT) {
+ else if (PL_regkind[(U8)op] == EXACT) {
/* Literal string, where present. */
node += ((*OPERAND(node)) + 2 + sizeof(regnode) - 1) / sizeof(regnode);
node = NEXTOPER(node);
- regdump - dump a regexp onto Perl_debug_log in vaguely comprehensible form
*/
void
-regdump(regexp *r)
+Perl_regdump(pTHX_ regexp *r)
{
#ifdef DEBUGGING
dTHR;
- regprop - printable representation of opcode
*/
void
-regprop(SV *sv, regnode *o)
+Perl_regprop(pTHX_ SV *sv, regnode *o)
{
#ifdef DEBUGGING
dTHR;
- register char *p = 0;
+ register int k;
sv_setpvn(sv, "", 0);
- switch (OP(o)) {
- case BOL:
- p = "BOL";
- break;
- case MBOL:
- p = "MBOL";
- break;
- case SBOL:
- p = "SBOL";
- break;
- case EOL:
- p = "EOL";
- break;
- case EOS:
- p = "EOS";
- break;
- case MEOL:
- p = "MEOL";
- break;
- case SEOL:
- p = "SEOL";
- break;
- case ANY:
- p = "ANY";
- break;
- case SANY:
- p = "SANY";
- break;
- case ANYUTF8:
- p = "ANYUTF8";
- break;
- case SANYUTF8:
- p = "SANYUTF8";
- break;
- case ANYOFUTF8:
- p = "ANYOFUTF8";
- break;
- case ANYOF:
- p = "ANYOF";
- break;
- case BRANCH:
- p = "BRANCH";
- break;
- case EXACT:
- sv_catpvf(sv, "EXACT <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
- break;
- case EXACTF:
- sv_catpvf(sv, "EXACTF <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
- break;
- case EXACTFL:
- sv_catpvf(sv, "EXACTFL <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
- break;
- case NOTHING:
- p = "NOTHING";
- break;
- case TAIL:
- p = "TAIL";
- break;
- case BACK:
- p = "BACK";
- break;
- case END:
- p = "END";
- break;
- case BOUND:
- p = "BOUND";
- break;
- case BOUNDL:
- p = "BOUNDL";
- break;
- case NBOUND:
- p = "NBOUND";
- break;
- case NBOUNDL:
- p = "NBOUNDL";
- break;
- case CURLY:
- sv_catpvf(sv, "CURLY {%d,%d}", ARG1(o), ARG2(o));
- break;
- case CURLYM:
- sv_catpvf(sv, "CURLYM[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
- break;
- case CURLYN:
- sv_catpvf(sv, "CURLYN[%d] {%d,%d}", o->flags, ARG1(o), ARG2(o));
- break;
- case CURLYX:
- sv_catpvf(sv, "CURLYX {%d,%d}", ARG1(o), ARG2(o));
- break;
- case REF:
- sv_catpvf(sv, "REF%d", ARG(o));
- break;
- case REFF:
- sv_catpvf(sv, "REFF%d", ARG(o));
- break;
- case REFFL:
- sv_catpvf(sv, "REFFL%d", ARG(o));
- break;
- case OPEN:
- sv_catpvf(sv, "OPEN%d", ARG(o));
- break;
- case CLOSE:
- sv_catpvf(sv, "CLOSE%d", ARG(o));
- p = NULL;
- break;
- case STAR:
- p = "STAR";
- break;
- case PLUS:
- p = "PLUS";
- break;
- case MINMOD:
- p = "MINMOD";
- break;
- case GPOS:
- p = "GPOS";
- break;
- case UNLESSM:
- sv_catpvf(sv, "UNLESSM[-%d]", o->flags);
- break;
- case IFMATCH:
- sv_catpvf(sv, "IFMATCH[-%d]", o->flags);
- break;
- case SUCCEED:
- p = "SUCCEED";
- break;
- case WHILEM:
- p = "WHILEM";
- break;
- case DIGIT:
- p = "DIGIT";
- break;
- case NDIGIT:
- p = "NDIGIT";
- break;
- case ALNUM:
- p = "ALNUM";
- break;
- case NALNUM:
- p = "NALNUM";
- break;
- case SPACE:
- p = "SPACE";
- break;
- case NSPACE:
- p = "NSPACE";
- break;
- case ALNUML:
- p = "ALNUML";
- break;
- case NALNUML:
- p = "NALNUML";
- break;
- case SPACEL:
- p = "SPACEL";
- break;
- case NSPACEL:
- p = "NSPACEL";
- break;
- case EVAL:
- p = "EVAL";
- break;
- case LONGJMP:
- p = "LONGJMP";
- break;
- case BRANCHJ:
- p = "BRANCHJ";
- break;
- case IFTHEN:
- p = "IFTHEN";
- break;
- case GROUPP:
- sv_catpvf(sv, "GROUPP%d", ARG(o));
- break;
- case LOGICAL:
- sv_catpvf(sv, "LOGICAL[%d]", o->flags);
- break;
- case SUSPEND:
- p = "SUSPEND";
- break;
- case RENUM:
- p = "RENUM";
- break;
- case OPTIMIZED:
- p = "OPTIMIZED";
- break;
- default:
+ if (OP(o) >= reg_num) /* regnode.type is unsigned */
FAIL("corrupted regexp opcode");
+ sv_catpv(sv, (char*)reg_name[OP(o)]); /* Take off const! */
+
+ k = PL_regkind[(U8)OP(o)];
+
+ if (k == EXACT)
+ Perl_sv_catpvf(aTHX_ sv, " <%s%s%s>", PL_colors[0], OPERAND(o) + 1, PL_colors[1]);
+ else if (k == CURLY) {
+ if (OP(o) == CURLYM || OP(o) == CURLYN)
+ Perl_sv_catpvf(aTHX_ sv, "[%d]", o->flags); /* Parenth number */
+ Perl_sv_catpvf(aTHX_ sv, " {%d,%d}", ARG1(o), ARG2(o));
}
- if (p)
- sv_catpv(sv, p);
+ else if (k == REF || k == OPEN || k == CLOSE || k == GROUPP )
+ Perl_sv_catpvf(aTHX_ sv, "%d", ARG(o)); /* Parenth number */
+ else if (k == LOGICAL)
+ Perl_sv_catpvf(aTHX_ sv, "[%d]", ARG(o)); /* 2: embedded, otherwise 1 */
+ else if (k == BRANCHJ && (OP(o) == UNLESSM || OP(o) == IFMATCH))
+ Perl_sv_catpvf(aTHX_ sv, "[-%d]", o->flags);
#endif /* DEBUGGING */
}
void
-pregfree(struct regexp *r)
+Perl_pregfree(pTHX_ struct regexp *r)
{
dTHR;
if (!r || (--r->refcnt > 0))
return;
if (r->precomp)
Safefree(r->precomp);
- if (r->subbase)
- Safefree(r->subbase);
+ if (RX_MATCH_COPIED(r))
+ Safefree(r->subbeg);
if (r->substrs) {
if (r->anchored_substr)
SvREFCNT_dec(r->anchored_substr);
}
if (r->data) {
int n = r->data->count;
+ AV* new_comppad = NULL;
+ AV* old_comppad;
+ SV** old_curpad;
+
while (--n >= 0) {
switch (r->data->what[n]) {
case 's':
SvREFCNT_dec((SV*)r->data->data[n]);
break;
+ case 'p':
+ new_comppad = (AV*)r->data->data[n];
+ break;
case 'o':
+ if (new_comppad == NULL)
+ Perl_croak(aTHX_ "panic: pregfree comppad");
+ old_comppad = PL_comppad;
+ old_curpad = PL_curpad;
+ PL_comppad = new_comppad;
+ PL_curpad = AvARRAY(new_comppad);
op_free((OP_4tree*)r->data->data[n]);
+ PL_comppad = old_comppad;
+ PL_curpad = old_curpad;
+ SvREFCNT_dec((SV*)new_comppad);
+ new_comppad = NULL;
break;
case 'n':
break;
* that bypass this code for speed.]
*/
regnode *
-regnext(register regnode *p)
+Perl_regnext(pTHX_ register regnode *p)
{
dTHR;
register I32 offset;
}
STATIC void
-re_croak2(const char* pat1,const char* pat2,...)
+S_re_croak2(pTHX_ const char* pat1,const char* pat2,...)
{
va_list args;
STRLEN l1 = strlen(pat1);
STRLEN l2 = strlen(pat2);
char buf[512];
+ SV *msv;
char *message;
if (l1 > 510)
Copy(pat2, buf + l1, l2 , char);
buf[l1 + l2] = '\n';
buf[l1 + l2 + 1] = '\0';
+#ifdef I_STDARG
+ /* ANSI variant takes additional second argument */
va_start(args, pat2);
- message = mess(buf, &args);
+#else
+ va_start(args);
+#endif
+ msv = mess(buf, &args);
va_end(args);
- l1 = strlen(message);
+ message = SvPV(msv,l1);
if (l1 > 512)
l1 = 512;
Copy(message, buf, l1 , char);
buf[l1] = '\0'; /* Overwrite \n */
- croak("%s", buf);
+ Perl_croak(aTHX_ "%s", buf);
}
/* XXX Here's a total kludge. But we need to re-enter for swash routines. */
void
-save_re_context(void)
+Perl_save_re_context(pTHX)
{
dTHR;
SAVEPPTR(PL_bostr);
SAVESPTR(PL_regcode); /* Code-emit pointer; ®dummy = don't */
SAVEPPTR(PL_regxend); /* End of input for compile */
SAVEPPTR(PL_regcomp_parse); /* Input-scan pointer. */
+ SAVESPTR(PL_reg_call_cc); /* from regexec.c */
+ SAVESPTR(PL_reg_re); /* from regexec.c */
+ SAVEPPTR(PL_reg_ganch); /* from regexec.c */
+ SAVESPTR(PL_reg_sv); /* from regexec.c */
+ SAVESPTR(PL_reg_magic); /* from regexec.c */
+ SAVEI32(PL_reg_oldpos); /* from regexec.c */
+ SAVESPTR(PL_reg_oldcurpm); /* from regexec.c */
+ SAVESPTR(PL_reg_curpm); /* from regexec.c */
+#ifdef DEBUGGING
+ SAVEPPTR(PL_reg_starttry); /* from regexec.c */
+#endif
}
projects / p5sagit/p5-mst-13.2.git / blobdiff