*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-1999, Larry Wall
+ **** Copyright (c) 1991-2000, 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.
STATIC void
S_cl_init(pTHX_ struct regnode_charclass_class *cl)
{
+ Zero(cl, 1, struct regnode_charclass_class);
cl->type = ANYOF;
cl_anything(cl);
}
STATIC void
S_cl_init_zero(pTHX_ struct regnode_charclass_class *cl)
{
+ Zero(cl, 1, struct regnode_charclass_class);
cl->type = ANYOF;
cl_anything(cl);
- ANYOF_CLASS_ZERO(cl);
- ANYOF_BITMAP_ZERO(cl);
if (LOC)
cl->flags |= ANYOF_LOCALE;
}
S_cl_and(pTHX_ struct regnode_charclass_class *cl,
struct regnode_charclass_class *and_with)
{
- int value;
-
if (!(and_with->flags & ANYOF_CLASS)
&& !(cl->flags & ANYOF_CLASS)
&& (and_with->flags & ANYOF_LOCALE) == (cl->flags & ANYOF_LOCALE)
STATIC void
S_cl_or(pTHX_ struct regnode_charclass_class *cl, struct regnode_charclass_class *or_with)
{
- int value;
-
if (or_with->flags & ANYOF_INVERT) {
/* We do not use
* (B1 | CL1) | (!B2 & !CL2) = (B1 | !B2 & !CL2) | (CL1 | (!B2 & !CL2))
&& !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan))
&& (!(data->start_class->flags & ANYOF_FOLD)
|| !ANYOF_BITMAP_TEST(data->start_class,
- PL_fold[*STRING(scan)])))
+ PL_fold[*(U8*)STRING(scan)])))
compat = 0;
ANYOF_CLASS_ZERO(data->start_class);
ANYOF_BITMAP_ZERO(data->start_class);
if (!(data->start_class->flags & (ANYOF_CLASS | ANYOF_LOCALE))
&& !ANYOF_BITMAP_TEST(data->start_class, *STRING(scan))
&& !ANYOF_BITMAP_TEST(data->start_class,
- PL_fold[*STRING(scan)]))
+ PL_fold[*(U8*)STRING(scan)]))
compat = 0;
ANYOF_CLASS_ZERO(data->start_class);
ANYOF_BITMAP_ZERO(data->start_class);
}
if (!scan) /* It was not CURLYX, but CURLY. */
scan = next;
- if (ckWARN(WARN_UNSAFE) && (minnext + deltanext == 0)
+ if (ckWARN(WARN_REGEXP) && (minnext + deltanext == 0)
&& !(data->flags & (SF_HAS_PAR|SF_IN_PAR))
&& maxcount <= REG_INFTY/3) /* Complement check for big count */
- Perl_warner(aTHX_ WARN_UNSAFE,
+ Perl_warner(aTHX_ WARN_REGEXP,
"Strange *+?{} on zero-length expression");
min += minnext * mincount;
- is_inf_internal |= (maxcount == REG_INFTY
- && (minnext + deltanext) > 0
- || deltanext == I32_MAX);
+ is_inf_internal |= ((maxcount == REG_INFTY
+ && (minnext + deltanext) > 0)
+ || deltanext == I32_MAX);
is_inf |= is_inf_internal;
delta += (minnext + deltanext) * maxcount - minnext * mincount;
if (data)
data->flags |= SF_HAS_EVAL;
}
- else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded */
+ else if (OP(scan) == LOGICAL && scan->flags == 2) { /* Embedded follows */
if (flags & SCF_DO_SUBSTR) {
scan_commit(data);
data->longest = &(data->longest_float);
is_inf = is_inf_internal = 1;
if (flags & SCF_DO_STCLASS_OR) /* Allow everything */
cl_anything(data->start_class);
+ flags &= ~SCF_DO_STCLASS;
}
/* Else: zero-length, ignore. */
scan = regnext(scan);
dTHR;
register regexp *r;
regnode *scan;
- SV **longest;
- SV *longest_fixed;
- SV *longest_float;
regnode *first;
I32 flags;
I32 minlen = 0;
if (exp == NULL)
FAIL("NULL regexp argument");
- if (PL_curcop == &PL_compiling ? (PL_hints & HINT_UTF8) : IN_UTF8)
+ if (pm->op_pmdynflags & PMdf_UTF8) {
PL_reg_flags |= RF_utf8;
+ }
else
PL_reg_flags = 0;
PL_regsize = 0L;
PL_regcode = &PL_regdummy;
PL_reg_whilem_seen = 0;
+#if 0 /* REGC() is (currently) a NOP at the first pass.
+ * Clever compilers notice this and complain. --jhi */
REGC((U8)REG_MAGIC, (char*)PL_regcode);
+#endif
if (reg(0, &flags) == NULL) {
Safefree(PL_regprecomp);
PL_regprecomp = Nullch;
char, regexp);
if (r == NULL)
FAIL("regexp out of space");
+#ifdef DEBUGGING
+ /* avoid reading uninitialized memory in DEBUGGING code in study_chunk() */
+ Zero(r, sizeof(regexp) + (unsigned)PL_regsize * sizeof(regnode), char);
+#endif
r->refcnt = 1;
r->prelen = xend - exp;
r->precomp = PL_regprecomp;
*flagp = TRYAGAIN;
return NULL;
case 'p':
+ if (SIZE_ONLY)
+ Perl_warner(aTHX_ WARN_REGEXP,
+ "(?p{}) is deprecated - use (??{})");
+ /* FALL THROUGH*/
+ case '?':
logical = 1;
paren = *PL_regcomp_parse++;
/* FALL THROUGH */
goto do_curly;
}
nest_check:
- if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
- Perl_warner(aTHX_ WARN_UNSAFE, "%.*s matches null string many times",
+ if (ckWARN(WARN_REGEXP) && !SIZE_ONLY && !(flags&HASWIDTH) && max > REG_INFTY/3) {
+ Perl_warner(aTHX_ WARN_REGEXP, "%.*s matches null string many times",
PL_regcomp_parse - origparse, origparse);
}
nextchar();
ret = reg(1, &flags);
if (ret == NULL) {
- if (flags & TRYAGAIN)
+ if (flags & TRYAGAIN) {
+ if (PL_regcomp_parse == PL_regxend) {
+ /* Make parent create an empty node if needed. */
+ *flagp |= TRYAGAIN;
+ return(NULL);
+ }
goto tryagain;
+ }
return(NULL);
}
*flagp |= flags&(HASWIDTH|SPSTART|SIMPLE);
if (!e)
FAIL("Missing right brace on \\x{}");
else if (UTF) {
- ender = (UV)scan_hex(p + 1, e - p, &numlen);
- if (numlen + len >= 127) { /* numlen is generous */
+ numlen = 1; /* allow underscores */
+ ender = (UV)scan_hex(p + 1, e - p - 1, &numlen);
+ /* numlen is generous */
+ if (numlen + len >= 127) {
p--;
goto loopdone;
}
FAIL("Can't use \\x{} without 'use utf8' declaration");
}
else {
+ numlen = 0; /* disallow underscores */
ender = (UV)scan_hex(p, 2, &numlen);
p += numlen;
}
case '5': case '6': case '7': case '8':case '9':
if (*p == '0' ||
(isDIGIT(p[1]) && atoi(p) >= PL_regnpar) ) {
+ numlen = 0; /* disallow underscores */
ender = (UV)scan_oct(p, 3, &numlen);
p += numlen;
}
FAIL("trailing \\ in regexp");
/* FALL THROUGH */
default:
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(*p))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(*p))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: Unrecognized escape \\%c passed through",
PL_regprecomp,
*p);
}
break;
}
- if ((namedclass == OOB_NAMEDCLASS ||
- !(posixcc + skip + 2 < PL_regxend &&
- (posixcc[skip] == ':' &&
- posixcc[skip + 1] == ']'))))
- Perl_croak(aTHX_ "Character class [:%.*s:] unknown",
- t - s - 1, s + 1);
- } else if (ckWARN(WARN_UNSAFE) && !SIZE_ONLY)
+ if (namedclass == OOB_NAMEDCLASS ||
+ posixcc[skip] != ':' ||
+ posixcc[skip+1] != ']')
+ Perl_croak(aTHX_
+ "Character class [:%.*s:] unknown",
+ t - s - 1, s + 1);
+ } else if (ckWARN(WARN_REGEXP) && !SIZE_ONLY)
/* [[=foo=]] and [[.foo.]] are still future. */
- Perl_warner(aTHX_ WARN_UNSAFE,
+ Perl_warner(aTHX_ WARN_REGEXP,
"Character class syntax [%c %c] is reserved for future extensions", c, c);
} else {
/* Maternal grandfather:
STATIC void
S_checkposixcc(pTHX)
{
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) &&
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP) &&
(*PL_regcomp_parse == ':' ||
*PL_regcomp_parse == '=' ||
*PL_regcomp_parse == '.')) {
while(*s && isALNUM(*s))
s++;
if (*s && c == *s && s[1] == ']') {
- Perl_warner(aTHX_ WARN_UNSAFE,
+ Perl_warner(aTHX_ WARN_REGEXP,
"Character class syntax [%c %c] belongs inside character classes", c, c);
if (c == '=' || c == '.')
- Perl_warner(aTHX_ WARN_UNSAFE,
+ Perl_warner(aTHX_ WARN_REGEXP,
"Character class syntax [%c %c] is reserved for future extensions", c, c);
}
}
S_regclass(pTHX)
{
dTHR;
- register UV value;
+ register U32 value;
register I32 lastvalue = OOB_CHAR8;
register I32 range = 0;
register regnode *ret;
- register I32 def;
I32 numlen;
I32 namedclass;
char *rangebegin;
ANYOF_FLAGS(ret) |= ANYOF_INVERT;
}
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE))
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
checkposixcc();
if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
namedclass = regpposixcc(value);
else if (value == '\\') {
value = UCHARAT(PL_regcomp_parse++);
+ /* Some compilers cannot handle switching on 64-bit integer
+ * values, therefore value cannot be an UV. --jhi */
switch (value) {
case 'w': namedclass = ANYOF_ALNUM; break;
case 'W': namedclass = ANYOF_NALNUM; break;
case 'a': value = '\057'; break;
#endif
case 'x':
+ numlen = 0; /* disallow underscores */
value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen);
PL_regcomp_parse += numlen;
break;
break;
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
+ numlen = 0; /* disallow underscores */
value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen);
PL_regcomp_parse += numlen;
break;
default:
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(value))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: Unrecognized escape \\%c in character class passed through",
PL_regprecomp,
(int)value);
need_class = 1;
if (range) { /* a-\d, a-[:digit:] */
if (!SIZE_ONLY) {
- if (ckWARN(WARN_UNSAFE))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (ckWARN(WARN_REGEXP))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: false [] range \"%*.*s\" in regexp",
PL_regprecomp,
PL_regcomp_parse - rangebegin,
PL_regcomp_parse[1] != ']') {
PL_regcomp_parse++;
if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */
- if (ckWARN(WARN_UNSAFE))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (ckWARN(WARN_REGEXP))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: false [] range \"%*.*s\" in regexp",
PL_regprecomp,
PL_regcomp_parse - rangebegin,
{
dTHR;
register char *e;
- register UV value;
+ register U32 value;
register U32 lastvalue = OOB_UTF8;
register I32 range = 0;
register regnode *ret;
listsv = newSVpvn("# comment\n",10);
}
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE))
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP))
checkposixcc();
if (*PL_regcomp_parse == ']' || *PL_regcomp_parse == '-')
if (value == '[')
namedclass = regpposixcc(value);
else if (value == '\\') {
- value = utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
+ value = (U32)utf8_to_uv((U8*)PL_regcomp_parse, &numlen);
PL_regcomp_parse += numlen;
+ /* Some compilers cannot handle switching on 64-bit integer
+ * values, therefore value cannot be an UV. Yes, this will
+ * be a problem later if we want switch on Unicode. --jhi */
switch (value) {
case 'w': namedclass = ANYOF_ALNUM; break;
case 'W': namedclass = ANYOF_NALNUM; break;
e = strchr(PL_regcomp_parse++, '}');
if (!e)
FAIL("Missing right brace on \\x{}");
+ numlen = 1; /* allow underscores */
value = (UV)scan_hex(PL_regcomp_parse,
e - PL_regcomp_parse,
&numlen);
PL_regcomp_parse = e + 1;
}
else {
+ numlen = 0; /* disallow underscores */
value = (UV)scan_hex(PL_regcomp_parse, 2, &numlen);
PL_regcomp_parse += numlen;
}
break;
case '0': case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
+ numlen = 0; /* disallow underscores */
value = (UV)scan_oct(--PL_regcomp_parse, 3, &numlen);
PL_regcomp_parse += numlen;
break;
default:
- if (!SIZE_ONLY && ckWARN(WARN_UNSAFE) && isALPHA(value))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (!SIZE_ONLY && ckWARN(WARN_REGEXP) && isALPHA(value))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: Unrecognized escape \\%c in character class passed through",
PL_regprecomp,
(int)value);
if (namedclass > OOB_NAMEDCLASS) {
if (range) { /* a-\d, a-[:digit:] */
if (!SIZE_ONLY) {
- if (ckWARN(WARN_UNSAFE))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (ckWARN(WARN_REGEXP))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: false [] range \"%*.*s\" in regexp",
PL_regprecomp,
PL_regcomp_parse - rangebegin,
PL_regcomp_parse[1] != ']') {
PL_regcomp_parse++;
if (namedclass > OOB_NAMEDCLASS) { /* \w-, [:word:]- */
- if (ckWARN(WARN_UNSAFE))
- Perl_warner(aTHX_ WARN_UNSAFE,
+ if (ckWARN(WARN_REGEXP))
+ Perl_warner(aTHX_ WARN_REGEXP,
"/%.127s/: false [] range \"%*.*s\" in regexp",
PL_regprecomp,
PL_regcomp_parse - rangebegin,
{
dTHR;
if (SIZE_ONLY) {
- U8 tmpbuf[10];
+ U8 tmpbuf[UTF8_MAXLEN];
*lenp = uv_to_utf8(tmpbuf, uv) - tmpbuf;
}
else
dTHR;
register regnode *scan;
register regnode *temp;
- register I32 offset;
if (SIZE_ONLY)
return;
{
#ifdef DEBUGGING
register U8 op = EXACT; /* Arbitrary non-END op. */
- register regnode *next, *onode;
+ register regnode *next;
while (op != END && (!last || node < last)) {
/* While that wasn't END last time... */