Loading...
--- Libc/Libc-391/regex/FreeBSD/regcomp.c.patch
+++ Libc/Libc-763.12/regex/FreeBSD/regcomp.c.patch
@@ -1,15 +1,15 @@
---- regcomp.c.orig 2004-11-25 11:38:32.000000000 -0800
-+++ regcomp.c 2005-02-24 13:46:56.000000000 -0800
-@@ -43,6 +43,8 @@
+--- regcomp.c.orig 2010-06-21 14:05:04.000000000 -0700
++++ regcomp.c 2010-06-21 14:23:51.000000000 -0700
+@@ -39,6 +39,8 @@ static char sccsid[] = "@(#)regcomp.c 8.
#include <sys/cdefs.h>
- __FBSDID("$FreeBSD: src/lib/libc/regex/regcomp.c,v 1.34 2004/10/03 15:42:59 stefanf Exp $");
+ __FBSDID("$FreeBSD: src/lib/libc/regex/regcomp.c,v 1.36 2007/06/11 03:05:54 delphij Exp $");
+#include "xlocale_private.h"
+
#include <sys/types.h>
#include <stdio.h>
#include <string.h>
-@@ -73,6 +75,9 @@
+@@ -69,6 +71,9 @@ struct parse {
sopno ssize; /* malloced strip size (allocated) */
sopno slen; /* malloced strip length (used) */
int ncsalloc; /* number of csets allocated */
@@ -19,7 +19,7 @@
struct re_guts *g;
# define NPAREN 10 /* we need to remember () 1-9 for back refs */
sopno pbegin[NPAREN]; /* -> ( ([0] unused) */
-@@ -97,7 +102,7 @@
+@@ -93,7 +98,7 @@ static void p_b_cclass(struct parse *p,
static void p_b_eclass(struct parse *p, cset *cs);
static wint_t p_b_symbol(struct parse *p);
static wint_t p_b_coll_elem(struct parse *p, wint_t endc);
@@ -28,7 +28,7 @@
static void bothcases(struct parse *p, wint_t ch);
static void ordinary(struct parse *p, wint_t ch);
static void nonnewline(struct parse *p);
-@@ -108,7 +113,7 @@
+@@ -104,7 +109,7 @@ static void freeset(struct parse *p, cse
static void CHadd(struct parse *p, cset *cs, wint_t ch);
static void CHaddrange(struct parse *p, cset *cs, wint_t min, wint_t max);
static void CHaddtype(struct parse *p, cset *cs, wctype_t wct);
@@ -37,7 +37,7 @@
static sopno dupl(struct parse *p, sopno start, sopno finish);
static void doemit(struct parse *p, sop op, size_t opnd);
static void doinsert(struct parse *p, sop op, size_t opnd, sopno pos);
-@@ -227,10 +232,14 @@
+@@ -222,10 +227,14 @@ regcomp(regex_t * __restrict preg,
p->end = p->next + len;
p->error = 0;
p->ncsalloc = 0;
@@ -52,7 +52,7 @@
g->sets = NULL;
g->ncsets = 0;
g->cflags = cflags;
-@@ -308,8 +317,12 @@
+@@ -302,8 +311,12 @@ p_ere(struct parse *p,
conc = HERE();
while (MORE() && (c = PEEK()) != '|' && c != stop)
p_ere_exp(p);
@@ -66,7 +66,7 @@
if (!EAT('|'))
break; /* NOTE BREAK OUT */
-@@ -417,7 +430,7 @@
+@@ -410,7 +423,7 @@ p_ere_exp(struct parse *p)
ordinary(p, wc);
break;
case '{': /* okay as ordinary except if digit follows */
@@ -75,7 +75,7 @@
/* FALLTHROUGH */
default:
p->next--;
-@@ -431,7 +444,7 @@
+@@ -424,7 +437,7 @@ p_ere_exp(struct parse *p)
c = PEEK();
/* we call { a repetition if followed by a digit */
if (!( c == '*' || c == '+' || c == '?' ||
@@ -84,7 +84,7 @@
return; /* no repetition, we're done */
NEXT();
-@@ -460,7 +473,7 @@
+@@ -453,7 +466,7 @@ p_ere_exp(struct parse *p)
case '{':
count = p_count(p);
if (EAT(',')) {
@@ -93,7 +93,7 @@
count2 = p_count(p);
(void)REQUIRE(count <= count2, REG_BADBR);
} else /* single number with comma */
-@@ -481,7 +494,7 @@
+@@ -474,7 +487,7 @@ p_ere_exp(struct parse *p)
return;
c = PEEK();
if (!( c == '*' || c == '+' || c == '?' ||
@@ -102,9 +102,9 @@
return;
SETERROR(REG_BADRPT);
}
-@@ -494,7 +507,12 @@
- p_str(p)
- struct parse *p;
+@@ -486,7 +499,12 @@ p_ere_exp(struct parse *p)
+ static void
+ p_str(struct parse *p)
{
+#if __DARWIN_UNIX03
+ if (!p->zerorepeats) REQUIRE(MORE(), REG_EMPTY);
@@ -115,7 +115,7 @@
while (MORE())
ordinary(p, WGETNEXT());
}
-@@ -534,8 +552,12 @@
+@@ -525,8 +543,12 @@ p_bre(struct parse *p,
p->g->iflags |= USEEOL;
p->g->neol++;
}
@@ -129,8 +129,34 @@
}
/*
-@@ -639,7 +661,7 @@
+@@ -599,12 +621,22 @@ p_simp_re(struct parse *p,
+ i = (c&~BACKSL) - '0';
+ assert(i < NPAREN);
+ if (p->pend[i] != 0) {
++#if __DARWIN_UNIX03
++ int skip = 1;
++#endif /* __DARWIN_UNIX03 */
+ assert(i <= p->g->nsub);
+ EMIT(OBACK_, i);
+ assert(p->pbegin[i] != 0);
+ assert(OP(p->strip[p->pbegin[i]]) == OLPAREN);
+ assert(OP(p->strip[p->pend[i]]) == ORPAREN);
++#if __DARWIN_UNIX03
++ if (OP(p->strip[p->pbegin[i]+skip]) == OBOL) {
++ skip++; /* don't dup anchor in subexp */
++ }
++ (void) dupl(p, p->pbegin[i]+skip, p->pend[i]);
++#else /* !__DARWIN_UNIX03 */
+ (void) dupl(p, p->pbegin[i]+1, p->pend[i]);
++#endif /* __DARWIN_UNIX03 */
+ EMIT(O_BACK, i);
+ } else
+ SETERROR(REG_ESUBREG);
+@@ -627,9 +659,10 @@ p_simp_re(struct parse *p,
+ INSERT(OQUEST_, pos);
+ ASTERN(O_QUEST, pos);
} else if (EATTWO('\\', '{')) {
++ (void)REQUIRE(MORE(), REG_EBRACE);
count = p_count(p);
if (EAT(',')) {
- if (MORE() && isdigit((uch)PEEK())) {
@@ -138,7 +164,7 @@
count2 = p_count(p);
(void)REQUIRE(count <= count2, REG_BADBR);
} else /* single number with comma */
-@@ -670,7 +692,7 @@
+@@ -659,7 +692,7 @@ p_count(struct parse *p)
int count = 0;
int ndigits = 0;
@@ -147,29 +173,30 @@
count = count*10 + (GETNEXT() - '0');
ndigits++;
}
-@@ -709,10 +731,21 @@
+@@ -697,10 +730,22 @@ p_bracket(struct parse *p)
cs->icase = 1;
if (EAT('^'))
cs->invert = 1;
+#if __DARWIN_UNIX03
-+ if (PEEK2() != '-') { /* Don't eat '-' or ']' if they're part of ranges */
-+ if (EAT(']'))
-+ CHadd(p, cs, ']');
-+ else if (EAT('-'))
-+ CHadd(p, cs, '-');
++ if (PEEK2() != '-' && PEEK2() != ']') { /* Don't eat '-' or ']' if they're part of ranges
++ * but do process [^-] */
+ if (EAT(']'))
+ CHadd(p, cs, ']');
+ else if (EAT('-'))
+ CHadd(p, cs, '-');
+ }
+ if (MORE() && !SEETWO('-',']')) /* Parse RE []-'] */
+ p_b_term(p, cs);
+#else /* !__DARWIN_UNIX03 */
- if (EAT(']'))
- CHadd(p, cs, ']');
- else if (EAT('-'))
- CHadd(p, cs, '-');
++ if (EAT(']'))
++ CHadd(p, cs, ']');
++ else if (EAT('-'))
++ CHadd(p, cs, '-');
+#endif /* __DARWIN_UNIX03 */
while (MORE() && PEEK() != ']' && !SEETWO('-', ']'))
p_b_term(p, cs);
if (EAT('-'))
-@@ -725,7 +758,7 @@
+@@ -713,7 +758,7 @@ p_bracket(struct parse *p)
if (cs->invert && p->g->cflags®_NEWLINE)
cs->bmp['\n' >> 3] |= 1 << ('\n' & 7);
@@ -178,14 +205,14 @@
ordinary(p, ch);
freeset(p, cs);
} else
-@@ -751,8 +784,16 @@
+@@ -737,8 +782,16 @@ p_b_term(struct parse *p, cset *cs)
c = (MORE2()) ? PEEK2() : '\0';
break;
case '-':
+#if __DARWIN_UNIX03
+ if (PEEK2() != '-') { /* Allow [---] */
-+ SETERROR(REG_ERANGE);
-+ return; /* NOTE RETURN */
++ SETERROR(REG_ERANGE);
++ return; /* NOTE RETURN */
+ } else
+ c = '-';
+#else /* !__DARWIN_UNIX03 */
@@ -195,7 +222,7 @@
break;
default:
c = '\0';
-@@ -773,7 +814,11 @@
+@@ -759,7 +812,11 @@ p_b_term(struct parse *p, cset *cs)
NEXT2();
(void)REQUIRE(MORE(), REG_EBRACK);
c = PEEK();
@@ -207,7 +234,7 @@
p_b_eclass(p, cs);
(void)REQUIRE(MORE(), REG_EBRACK);
(void)REQUIRE(EATTWO('=', ']'), REG_ECOLLATE);
-@@ -792,14 +837,14 @@
+@@ -778,14 +835,14 @@ p_b_term(struct parse *p, cset *cs)
if (start == finish)
CHadd(p, cs, start);
else {
@@ -226,7 +253,7 @@
)
CHadd(p, cs, i);
}
-@@ -823,7 +868,7 @@
+@@ -807,7 +864,7 @@ p_b_cclass(struct parse *p, cset *cs)
wctype_t wct;
char clname[16];
@@ -235,7 +262,7 @@
NEXT();
len = p->next - sp;
if (len >= sizeof(clname) - 1) {
-@@ -832,7 +877,7 @@
+@@ -816,7 +873,7 @@ p_b_cclass(struct parse *p, cset *cs)
}
memcpy(clname, sp, len);
clname[len] = '\0';
@@ -244,27 +271,86 @@
SETERROR(REG_ECTYPE);
return;
}
-@@ -903,7 +948,7 @@
+@@ -826,14 +883,38 @@ p_b_cclass(struct parse *p, cset *cs)
+ /*
+ - p_b_eclass - parse an equivalence-class name and deal with it
+ == static void p_b_eclass(struct parse *p, cset *cs);
+- *
+- * This implementation is incomplete. xxx
+ */
+ static void
+ p_b_eclass(struct parse *p, cset *cs)
+ {
+- wint_t c;
+-
++ char *sp = p->next;
++ int len, ec;
++ mbstate_t mbs;
++ int *newequiv_classes;
++ wint_t c;
++
++ while (MORE() && !SEETWO('=', ']'))
++ NEXT();
++ if (!MORE()) {
++ SETERROR(REG_EBRACK);
++ return;
++ }
++ len = p->next - sp;
++ memset(&mbs, 0, sizeof(mbs));
++ ec = __collate_equiv_class(sp, len, &mbs, p->g->loc);
++ if (ec > 0) {
++ newequiv_classes = realloc(cs->equiv_classes,
++ (cs->nequiv_classes + 1) * sizeof(*cs->equiv_classes));
++ if (newequiv_classes == NULL) {
++ SETERROR(REG_ESPACE);
++ return;
++ }
++ cs->equiv_classes = newequiv_classes;
++ cs->equiv_classes[cs->nequiv_classes++] = ec;
++ return;
++ }
++ /* not an equivalence class, so fallback to a collating element */
++ p->next = sp;
+ c = p_b_coll_elem(p, '=');
+ CHadd(p, cs, c);
+ }
+@@ -866,10 +947,10 @@ p_b_coll_elem(struct parse *p,
+ wint_t endc) /* name ended by endc,']' */
+ {
+ char *sp = p->next;
+- struct cname *cp;
++ const struct cname *cp;
+ int len;
+ mbstate_t mbs;
+- wchar_t wc;
++ wchar_t wbuf[16];
+ size_t clen;
+
+ while (MORE() && !SEETWO(endc, ']'))
+@@ -883,9 +964,10 @@ p_b_coll_elem(struct parse *p,
if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0')
return(cp->code); /* known name */
memset(&mbs, 0, sizeof(mbs));
- if ((clen = mbrtowc(&wc, sp, len, &mbs)) == len)
-+ if ((clen = mbrtowc_l(&wc, sp, len, &mbs, p->g->loc)) == len)
- return (wc); /* single character */
- else if (clen == (size_t)-1 || clen == (size_t)-2)
+- return (wc); /* single character */
+- else if (clen == (size_t)-1 || clen == (size_t)-2)
++ clen = __collate_collating_symbol(wbuf, 16, sp, len, &mbs, p->g->loc);
++ if (clen == 1)
++ return (*wbuf); /* single character */
++ else if (clen == (size_t)-1)
SETERROR(REG_ILLSEQ);
-@@ -914,17 +959,18 @@
+ else
+ SETERROR(REG_ECOLLATE); /* neither */
+@@ -894,16 +976,16 @@ p_b_coll_elem(struct parse *p,
/*
- othercase - return the case counterpart of an alphabetic
- == static char othercase(int ch);
-+ == static char othercase(int ch, locale_t loc);
++ == static char othercase(wint_t ch, locale_t loc);
*/
static wint_t /* if no counterpart, return ch */
--othercase(ch)
-+othercase(ch, loc)
- wint_t ch;
-+locale_t loc;
+-othercase(wint_t ch)
++othercase(wint_t ch, locale_t loc)
{
- assert(iswalpha(ch));
- if (iswupper(ch))
@@ -279,7 +365,7 @@
else /* peculiar, but could happen */
return(ch);
}
-@@ -946,10 +992,10 @@
+@@ -923,10 +1005,10 @@ bothcases(struct parse *p, wint_t ch)
size_t n;
mbstate_t mbs;
@@ -292,7 +378,7 @@
assert(n != (size_t)-1);
bracket[n] = ']';
bracket[n + 1] = '\0';
-@@ -971,7 +1017,7 @@
+@@ -946,7 +1028,7 @@ ordinary(struct parse *p, wint_t ch)
{
cset *cs;
@@ -301,7 +387,7 @@
bothcases(p, ch);
else if ((ch & OPDMASK) == ch)
EMIT(OCHAR, ch);
-@@ -1039,6 +1085,9 @@
+@@ -1012,10 +1094,22 @@ repeat(struct parse *p,
switch (REP(MAP(from), MAP(to))) {
case REP(0, 0): /* must be user doing this */
DROP(finish-start); /* drop the operand */
@@ -309,9 +395,42 @@
+ p->zerorepeats++;
+#endif /* __DARWIN_UNIX03 */
break;
++ case REP(0, INF): /* as x{1,}? */
++#if __DARWIN_UNIX03
++ /* this case does not require the (y|) trick, noKLUDGE */
++ /* Just like * =+? */
++ INSERT(OPLUS_, start);
++ ASTERN(O_PLUS, start);
++ INSERT(OQUEST_, start);
++ ASTERN(O_QUEST, start);
++ break;
++#endif /* __DARWIN_UNIX03 */
case REP(0, 1): /* as x{1,1}? */
case REP(0, N): /* as x{1,n}? */
-@@ -1099,7 +1148,7 @@
+- case REP(0, INF): /* as x{1,}? */
+ /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
+ INSERT(OCH_, start); /* offset is wrong... */
+ repeat(p, start+1, 1, to);
+@@ -1029,6 +1123,10 @@ repeat(struct parse *p,
+ /* done */
+ break;
+ case REP(1, N): /* as x?x{1,n-1} */
++#if __DARWIN_UNIX03
++ INSERT(OQUEST_, start);
++ ASTERN(O_QUEST, start);
++#else /* !__DARWIN_UNIX03 */
+ /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */
+ INSERT(OCH_, start);
+ ASTERN(OOR1, start);
+@@ -1036,6 +1134,7 @@ repeat(struct parse *p,
+ EMIT(OOR2, 0); /* offset very wrong... */
+ AHEAD(THERE()); /* ...so fix it */
+ ASTERN(O_CH, THERETHERE());
++#endif /* __DARWIN_UNIX03 */
+ copy = dupl(p, start+1, finish+1);
+ assert(copy == finish+4);
+ repeat(p, copy, 1, to-1);
+@@ -1071,7 +1170,7 @@ wgetnext(struct parse *p)
size_t n;
memset(&mbs, 0, sizeof(mbs));
@@ -320,14 +439,12 @@
if (n == (size_t)-1 || n == (size_t)-2) {
SETERROR(REG_ILLSEQ);
return (0);
-@@ -1172,13 +1221,14 @@
+@@ -1139,12 +1238,12 @@ freeset(struct parse *p, cset *cs)
- returning it if so, otherwise returning OUT.
*/
static wint_t
--singleton(cs)
-+singleton(cs, loc)
- cset *cs;
-+locale_t loc;
+-singleton(cset *cs)
++singleton(cset *cs, locale_t loc)
{
wint_t i, s, n;
@@ -337,7 +454,7 @@
n++;
s = i;
}
-@@ -1215,9 +1265,9 @@
+@@ -1178,9 +1277,9 @@ CHadd(struct parse *p, cset *cs, wint_t
cs->wides[cs->nwides++] = ch;
}
if (cs->icase) {
@@ -349,7 +466,7 @@
cs->bmp[nch >> 3] |= 1 << (nch & 7);
}
}
-@@ -1262,7 +1312,7 @@
+@@ -1219,7 +1318,7 @@ CHaddtype(struct parse *p, cset *cs, wct
wctype_t *newtypes;
for (i = 0; i < NC; i++)
@@ -358,7 +475,7 @@
CHadd(p, cs, i);
newtypes = realloc(cs->types, (cs->ntypes + 1) *
sizeof(*cs->types));
-@@ -1451,6 +1501,7 @@
+@@ -1391,6 +1490,7 @@ findmust(struct parse *p, struct re_guts
char buf[MB_LEN_MAX];
size_t clen;
mbstate_t mbs;
@@ -366,7 +483,7 @@
/* avoid making error situations worse */
if (p->error != 0)
-@@ -1461,8 +1512,8 @@
+@@ -1401,8 +1501,8 @@ findmust(struct parse *p, struct re_guts
* multibyte character strings, but it's safe for at least
* UTF-8 (see RFC 3629).
*/
@@ -377,7 +494,7 @@
return;
/* find the longest OCHAR sequence in strip */
-@@ -1478,7 +1529,7 @@
+@@ -1418,7 +1518,7 @@ findmust(struct parse *p, struct re_guts
memset(&mbs, 0, sizeof(mbs));
newstart = scan - 1;
}
@@ -386,7 +503,7 @@
if (clen == (size_t)-1)
goto toohard;
newlen += clen;
-@@ -1597,7 +1648,7 @@
+@@ -1537,7 +1637,7 @@ findmust(struct parse *p, struct re_guts
while (cp < g->must + g->mlen) {
while (OP(s = *scan++) != OCHAR)
continue;