char *parse; /* Input-scan pointer. */
I32 whilem_seen; /* number of WHILEM in this expr */
regnode *emit_start; /* Start of emitted-code area */
+ regnode *emit_bound; /* First regnode outside of the allocated space */
regnode *emit; /* Code-emit pointer; ®dummy = don't = compiling */
I32 naughty; /* How bad is this pattern? */
I32 sawback; /* Did we see \1, ...? */
#endif
#define RExC_emit (pRExC_state->emit)
#define RExC_emit_start (pRExC_state->emit_start)
+#define RExC_emit_bound (pRExC_state->emit_bound)
#define RExC_naughty (pRExC_state->naughty)
#define RExC_sawback (pRExC_state->sawback)
#define RExC_seen (pRExC_state->seen)
* Flags to be passed up and down.
*/
#define WORST 0 /* Worst case. */
-#define HASWIDTH 0x1 /* Known to match non-null strings. */
-#define SIMPLE 0x2 /* Simple enough to be STAR/PLUS operand. */
-#define SPSTART 0x4 /* Starts with * or +. */
-#define TRYAGAIN 0x8 /* Weeded out a declaration. */
+#define HASWIDTH 0x01 /* Known to match non-null strings. */
+#define SIMPLE 0x02 /* Simple enough to be STAR/PLUS operand. */
+#define SPSTART 0x04 /* Starts with * or +. */
+#define TRYAGAIN 0x08 /* Weeded out a declaration. */
+#define POSTPONED 0x10 /* (?1),(?&name), (??{...}) or similar */
#define REG_NODE_NUM(x) ((x) ? (int)((x)-RExC_emit_start) : -1)
if (RExC_whilem_seen > 15)
RExC_whilem_seen = 15;
-#ifdef DEBUGGING
- /* Make room for a sentinel value at the end of the program */
- RExC_size++;
-#endif
-
/* Allocate space and zero-initialize. Note, the two step process
of zeroing when in debug mode, thus anything assigned has to
happen after that */
if (has_runon)
*p++='\n';
*p=')';
-
-
- if (0)
- PerlIO_printf(Perl_debug_log,
- "RExC_precomp: %.*s\nr->precomp: %.*s\nr->wrapped:%.*s\n",
- r->prelen,
- RExC_precomp,
- r->prelen,
- r->precomp,
- r->wraplen,
- r->wrapped
- );
-
-
}
r->intflags = 0;
RExC_npar = 1;
RExC_emit_start = ri->program;
RExC_emit = ri->program;
-#ifdef DEBUGGING
- /* put a sentinal on the end of the program so we can check for
- overwrites */
- ri->program[RExC_size].type = 255;
-#endif
+ RExC_emit_bound = ri->program + RExC_size + 1;
+
/* Store the count of eval-groups for security checks: */
RExC_rx->seen_evals = RExC_seen_evals;
REGC((U8)REG_MAGIC, (char*) RExC_emit++);
return(r);
}
-#undef CORE_ONLY_BLOCK
#undef RE_ENGINE_PTR
-#ifndef PERL_IN_XSUB_RE
+
SV*
-Perl_reg_named_buff_get(pTHX_ SV* namesv, const REGEXP * const from_re, U32 flags)
+Perl_reg_named_buff_get(pTHX_ const REGEXP * const rx, SV* namesv, U32 flags)
{
AV *retarray = NULL;
SV *ret;
if (flags & 1)
retarray=newAV();
-
- if (from_re || PL_curpm) {
- const REGEXP * const rx = from_re ? from_re : PM_GETRE(PL_curpm);
- if (rx && rx->paren_names) {
- HE *he_str = hv_fetch_ent( rx->paren_names, namesv, 0, 0 );
- if (he_str) {
- IV i;
- SV* sv_dat=HeVAL(he_str);
- I32 *nums=(I32*)SvPVX(sv_dat);
- for ( i=0; i<SvIVX(sv_dat); i++ ) {
- if ((I32)(rx->nparens) >= nums[i]
- && rx->startp[nums[i]] != -1
- && rx->endp[nums[i]] != -1)
- {
- ret = reg_numbered_buff_get(nums[i],rx,NULL,0);
- if (!retarray)
- return ret;
- } else {
- ret = newSVsv(&PL_sv_undef);
- }
- if (retarray) {
- SvREFCNT_inc(ret);
- av_push(retarray, ret);
- }
+
+ if (rx && rx->paren_names) {
+ HE *he_str = hv_fetch_ent( rx->paren_names, namesv, 0, 0 );
+ if (he_str) {
+ IV i;
+ SV* sv_dat=HeVAL(he_str);
+ I32 *nums=(I32*)SvPVX(sv_dat);
+ for ( i=0; i<SvIVX(sv_dat); i++ ) {
+ if ((I32)(rx->nparens) >= nums[i]
+ && rx->startp[nums[i]] != -1
+ && rx->endp[nums[i]] != -1)
+ {
+ ret = CALLREG_NUMBUF(rx,nums[i],NULL);
+ if (!retarray)
+ return ret;
+ } else {
+ ret = newSVsv(&PL_sv_undef);
+ }
+ if (retarray) {
+ SvREFCNT_inc(ret);
+ av_push(retarray, ret);
}
- if (retarray)
- return (SV*)retarray;
}
+ if (retarray)
+ return (SV*)retarray;
}
}
return NULL;
}
SV*
-Perl_reg_numbered_buff_get(pTHX_ I32 paren, const REGEXP * const rx, SV* usesv, U32 flags)
+Perl_reg_numbered_buff_get(pTHX_ const REGEXP * const rx, I32 paren, SV* usesv)
{
char *s = NULL;
I32 i = 0;
I32 s1, t1;
SV *sv = usesv ? usesv : newSVpvs("");
- PERL_UNUSED_ARG(flags);
if (!rx->subbeg) {
sv_setsv(sv,&PL_sv_undef);
}
return sv;
}
-#endif
+
/* Scans the name of a named buffer from the pattern.
* If flags is REG_RSN_RETURN_NULL returns null.
PerlIO_printf(Perl_debug_log,"%16s",""); \
\
if (SIZE_ONLY) \
- num=RExC_size; \
+ num = RExC_size + 1; \
else \
num=REG_NODE_NUM(RExC_emit); \
if (RExC_lastnum!=num) \
const I32 oregflags = RExC_flags;
bool have_branch = 0;
bool is_open = 0;
+ I32 freeze_paren = 0;
+ I32 after_freeze = 0;
/* for (?g), (?gc), and (?o) warnings; warning
about (?c) will warn about (?g) -- japhy */
Perl_croak(aTHX_
"panic: paren_name hash element allocation failed");
} else if ( SvPOK(sv_dat) ) {
- IV count=SvIV(sv_dat);
- I32 *pv=(I32*)SvGROW(sv_dat,SvCUR(sv_dat)+sizeof(I32)+1);
- SvCUR_set(sv_dat,SvCUR(sv_dat)+sizeof(I32));
- pv[count]=RExC_npar;
- SvIVX(sv_dat)++;
+ /* (?|...) can mean we have dupes so scan to check
+ its already been stored. Maybe a flag indicating
+ we are inside such a construct would be useful,
+ but the arrays are likely to be quite small, so
+ for now we punt -- dmq */
+ IV count = SvIV(sv_dat);
+ I32 *pv = (I32*)SvPVX(sv_dat);
+ IV i;
+ for ( i = 0 ; i < count ; i++ ) {
+ if ( pv[i] == RExC_npar ) {
+ count = 0;
+ break;
+ }
+ }
+ if ( count ) {
+ pv = (I32*)SvGROW(sv_dat, SvCUR(sv_dat) + sizeof(I32)+1);
+ SvCUR_set(sv_dat, SvCUR(sv_dat) + sizeof(I32));
+ pv[count] = RExC_npar;
+ SvIVX(sv_dat)++;
+ }
} else {
(void)SvUPGRADE(sv_dat,SVt_PVNV);
sv_setpvn(sv_dat, (char *)&(RExC_npar), sizeof(I32));
nextchar(pRExC_state);
return ret;
}
+ break;
+ case '|': /* (?|...) */
+ /* branch reset, behave like a (?:...) except that
+ buffers in alternations share the same numbers */
+ paren = ':';
+ after_freeze = freeze_paren = RExC_npar;
+ break;
case ':': /* (?:...) */
case '>': /* (?>...) */
break;
if (*RExC_parse != ')')
FAIL("Sequence (?R) not terminated");
ret = reg_node(pRExC_state, GOSTART);
+ *flagp |= POSTPONED;
nextchar(pRExC_state);
return ret;
/*notreached*/
Set_Node_Length(ret, 1 + regarglen[OP(ret)]); /* MJD */
Set_Node_Offset(ret, parse_start); /* MJD */
+ *flagp |= POSTPONED;
nextchar(pRExC_state);
return ret;
} /* named and numeric backreferences */
/* NOT REACHED */
- case 'p': /* (?p...) */
- if (SIZE_ONLY && ckWARN2(WARN_DEPRECATED, WARN_REGEXP))
- vWARNdep(RExC_parse, "(?p{}) is deprecated - use (??{})");
- /* FALL THROUGH*/
case '?': /* (??...) */
is_logical = 1;
if (*RExC_parse != '{') {
vFAIL3("Sequence (%.*s...) not recognized", RExC_parse-seqstart, seqstart);
/*NOTREACHED*/
}
+ *flagp |= POSTPONED;
paren = *RExC_parse++;
/* FALL THROUGH */
case '{': /* (?{...}) */
}
else
REGTAIL(pRExC_state, ret, ender);
+ RExC_size++; /* XXX WHY do we need this?!!
+ For large programs it seems to be required
+ but I can't figure out why. -- dmq*/
return ret;
}
else {
if (!SIZE_ONLY ){
if (!RExC_nestroot)
RExC_nestroot = parno;
- if (RExC_seen & REG_SEEN_RECURSE) {
+ if (RExC_seen & REG_SEEN_RECURSE
+ && !RExC_open_parens[parno-1])
+ {
DEBUG_OPTIMISE_MORE_r(PerlIO_printf(Perl_debug_log,
"Setting open paren #%"IVdf" to %d\n",
(IV)parno, REG_NODE_NUM(ret)));
}
else if (paren != '?') /* Not Conditional */
ret = br;
- *flagp |= flags & (SPSTART | HASWIDTH);
+ *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
lastbr = br;
while (*RExC_parse == '|') {
if (!SIZE_ONLY && RExC_extralen) {
if (SIZE_ONLY)
RExC_extralen += 2; /* Account for LONGJMP. */
nextchar(pRExC_state);
+ if (freeze_paren) {
+ if (RExC_npar > after_freeze)
+ after_freeze = RExC_npar;
+ RExC_npar = freeze_paren;
+ }
br = regbranch(pRExC_state, &flags, 0, depth+1);
if (br == NULL)
return(NULL);
REGTAIL(pRExC_state, lastbr, br); /* BRANCH -> BRANCH. */
lastbr = br;
- if (flags&HASWIDTH)
- *flagp |= HASWIDTH;
- *flagp |= flags&SPSTART;
+ *flagp |= flags & (SPSTART | HASWIDTH | POSTPONED);
}
if (have_branch || paren != ':') {
FAIL("Junk on end of regexp"); /* "Can't happen". */
/* NOTREACHED */
}
-
+ if (after_freeze)
+ RExC_npar = after_freeze;
return(ret);
}
}
else if (ret == NULL)
ret = latest;
- *flagp |= flags&HASWIDTH;
+ *flagp |= flags&(HASWIDTH|POSTPONED);
if (chain == NULL) /* First piece. */
*flagp |= flags&SPSTART;
else {
goto do_curly;
}
nest_check:
- if (!SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3 && ckWARN(WARN_REGEXP)) {
+ if (!SIZE_ONLY && !(flags&(HASWIDTH|POSTPONED)) && max > REG_INFTY/3 && ckWARN(WARN_REGEXP)) {
vWARN3(RExC_parse,
"%.*s matches null string many times",
(int)(RExC_parse >= origparse ? RExC_parse - origparse : 0),
}
return(NULL);
}
- *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
+ *flagp |= flags&(HASWIDTH|SPSTART|SIMPLE|POSTPONED);
break;
case '|':
case ')':
return ret;
/****** !SIZE_ONLY AFTER HERE *********/
- if( stored == 1 && value < 256
+ if( stored == 1 && (value < 128 || (value < 256 && !UTF))
&& !( ANYOF_FLAGS(ret) & ( ANYOF_FLAGS_ALL ^ ANYOF_FOLD ) )
) {
/* optimize single char class to an EXACT node
RExC_size += 1;
return(ret);
}
-#ifdef DEBUGGING
- if (OP(RExC_emit) == 255)
- Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %s: %d ",
- reg_name[op], OP(RExC_emit));
-#endif
+ if (RExC_emit >= RExC_emit_bound)
+ Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
+
NODE_ALIGN_FILL(ret);
ptr = ret;
FILL_ADVANCE_NODE(ptr, op);
if (RExC_offsets) { /* MJD */
MJD_OFFSET_DEBUG(("%s:%d: (op %s) %s %"UVuf" (len %"UVuf") (max %"UVuf").\n",
"reg_node", __LINE__,
- reg_name[op],
+ PL_reg_name[op],
(UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0]
? "Overwriting end of array!\n" : "OK",
(UV)(RExC_emit - RExC_emit_start),
*/
return(ret);
}
-#ifdef DEBUGGING
- if (OP(RExC_emit) == 255)
- Perl_croak(aTHX_ "panic: reganode overwriting end of allocated program space");
-#endif
+ if (RExC_emit >= RExC_emit_bound)
+ Perl_croak(aTHX_ "panic: reg_node overrun trying to emit %d", op);
+
NODE_ALIGN_FILL(ret);
ptr = ret;
FILL_ADVANCE_NODE_ARG(ptr, op, arg);
MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
"reganode",
__LINE__,
- reg_name[op],
+ PL_reg_name[op],
(UV)(RExC_emit - RExC_emit_start) > RExC_offsets[0] ?
"Overwriting end of array!\n" : "OK",
(UV)(RExC_emit - RExC_emit_start),
const int offset = regarglen[(U8)op];
const int size = NODE_STEP_REGNODE + offset;
GET_RE_DEBUG_FLAGS_DECL;
+ PERL_UNUSED_ARG(depth);
/* (PL_regkind[(U8)op] == CURLY ? EXTRA_STEP_2ARGS : 0); */
- DEBUG_PARSE_FMT("inst"," - %s",reg_name[op]);
+ DEBUG_PARSE_FMT("inst"," - %s",PL_reg_name[op]);
if (SIZE_ONLY) {
RExC_size += size;
return;
dst = RExC_emit;
if (RExC_open_parens) {
int paren;
- DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);
+ /*DEBUG_PARSE_FMT("inst"," - %"IVdf, (IV)RExC_npar);*/
for ( paren=0 ; paren < RExC_npar ; paren++ ) {
if ( RExC_open_parens[paren] >= opnd ) {
- DEBUG_PARSE_FMT("open"," - %d",size);
+ /*DEBUG_PARSE_FMT("open"," - %d",size);*/
RExC_open_parens[paren] += size;
} else {
- DEBUG_PARSE_FMT("open"," - %s","ok");
+ /*DEBUG_PARSE_FMT("open"," - %s","ok");*/
}
if ( RExC_close_parens[paren] >= opnd ) {
- DEBUG_PARSE_FMT("close"," - %d",size);
+ /*DEBUG_PARSE_FMT("close"," - %d",size);*/
RExC_close_parens[paren] += size;
} else {
- DEBUG_PARSE_FMT("close"," - %s","ok");
+ /*DEBUG_PARSE_FMT("close"," - %s","ok");*/
}
}
}
MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s copy %"UVuf" -> %"UVuf" (max %"UVuf").\n",
"reg_insert",
__LINE__,
- reg_name[op],
+ PL_reg_name[op],
(UV)(dst - RExC_emit_start) > RExC_offsets[0]
? "Overwriting end of array!\n" : "OK",
(UV)(src - RExC_emit_start),
MJD_OFFSET_DEBUG(("%s(%d): (op %s) %s %"UVuf" <- %"UVuf" (max %"UVuf").\n",
"reginsert",
__LINE__,
- reg_name[op],
+ PL_reg_name[op],
(UV)(place - RExC_emit_start) > RExC_offsets[0]
? "Overwriting end of array!\n" : "OK",
(UV)(place - RExC_emit_start),
PerlIO_printf(Perl_debug_log, "~ %s (%d) %s %s\n",
SvPV_nolen_const(mysv), REG_NODE_NUM(scan),
(temp == NULL ? "->" : ""),
- (temp == NULL ? reg_name[OP(val)] : "")
+ (temp == NULL ? PL_reg_name[OP(val)] : "")
);
});
if (temp == NULL)
PerlIO_printf(Perl_debug_log, "~ %s (%d) -> %s\n",
SvPV_nolen_const(mysv),
REG_NODE_NUM(scan),
- reg_name[exact]);
+ PL_reg_name[exact]);
});
if (temp == NULL)
break;
/* It would be nice to FAIL() here, but this may be called from
regexec.c, and it would be hard to supply pRExC_state. */
Perl_croak(aTHX_ "Corrupted regexp opcode %d > %d", (int)OP(o), (int)REGNODE_MAX);
- sv_catpv(sv, reg_name[OP(o)]); /* Take off const! */
+ sv_catpv(sv, PL_reg_name[OP(o)]); /* Take off const! */
k = PL_regkind[OP(o)];
const reg_trie_data * const trie
= (reg_trie_data*)progi->data->data[!IS_TRIE_AC(op) ? n : ac->trie];
- Perl_sv_catpvf(aTHX_ sv, "-%s",reg_name[o->flags]);
+ Perl_sv_catpvf(aTHX_ sv, "-%s",PL_reg_name[o->flags]);
DEBUG_TRIE_COMPILE_r(
Perl_sv_catpvf(aTHX_ sv,
"<S:%"UVuf"/%"IVdf" W:%"UVuf" L:%"UVuf"/%"UVuf" C:%"UVuf"/%"UVuf">",
https://perl5.git.perl.org / perl5.git / blobdiff