Bug#510219: Info received (Bug#510219: sed: cannot tell if some chars are letters, digits, etc.)
On Fri, Jan 09, 2009 at 09:33:45AM +0100, Paolo Bonzini wrote:
> This was bug 9697 in glibc's bugzilla, and is fixed by the patch at "git
> diff 4c2a6f3d 37bdc055ce^" in git://sources.redhat.com/git/glibc.git
> (gitweb interface at http://repo.or.cz/w/glibc.git).
>
> Another regex bug (bug 697 in glibc's bugzilla), by the way, is fixed by
> the patch at "git diff 37bdc055^ 9d565658".
Attaching those two for posterity.
diff --git a/ChangeLog b/ChangeLog
index 182bd26..8829b44 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2009-01-04 Paolo Bonzini <bonzini@gnu.org>
+
+ [BZ 9697]
+ * posix/bug-regex17.c: Add testcases.
+ * posix/regcomp.c (re_compile_fastmap_iter): Rewrite COMPLEX_BRACKET
+ handling.
+
2009-01-05 Martin Schwidefsky <schwidefsky@de.ibm.com>
* sysdeps/unix/sysv/linux/s390/bits/libc-vdso.h: New file.
diff --git a/posix/bug-regex17.c b/posix/bug-regex17.c
index b42f9b6..1c11a1d 100644
--- a/posix/bug-regex17.c
+++ b/posix/bug-regex17.c
@@ -1,5 +1,5 @@
-/* Turkish regular expression tests.
- Copyright (C) 2002, 2003 Free Software Foundation, Inc.
+/* German regular expression tests.
+ Copyright (C) 2002, 2003, 2009 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Jakub Jelinek <jakub@redhat.com>, 2002.
@@ -33,10 +33,10 @@ struct
int flags, nmatch;
regmatch_t rm[5];
} tests[] = {
- /* \xc3\x84 LATIN CAPITAL LETTER A WITH DIAERESIS
- \xc3\x96 LATIN CAPITAL LETTER O WITH DIAERESIS
- \xc3\xa4 LATIN SMALL LETTER A WITH DIAERESIS
- \xc3\xb6 LATIN SMALL LETTER O WITH DIAERESIS */
+ /* U+00C4 \xc3\x84 LATIN CAPITAL LETTER A WITH DIAERESIS
+ U+00D6 \xc3\x96 LATIN CAPITAL LETTER O WITH DIAERESIS
+ U+00E4 \xc3\xa4 LATIN SMALL LETTER A WITH DIAERESIS
+ U+00F6 \xc3\xb6 LATIN SMALL LETTER O WITH DIAERESIS */
{ "\xc3\x84\xc3\x96*\xc3\xb6$", "aB\xc3\xa4\xc3\xb6\xc3\xb6\xc3\x96", REG_ICASE, 2,
{ { 2, 10 }, { -1, -1 } } },
{ "[\xc3\x84x]\xc3\x96*\xc3\xb6$", "aB\xc3\x84\xc3\xb6\xc3\xb6\xc3\x96", REG_ICASE, 2,
@@ -45,10 +45,22 @@ struct
{ { 2, 10 }, { -1, -1 } } },
{ "[^x]\xc3\x96*\xc3\xb6$", "aB\xc3\xa4\xc3\xb6\xc3\xb6\xc3\x96", REG_ICASE, 2,
{ { 2, 10 }, { -1, -1 } } },
+
+ /* Tests for bug 9697:
+ U+00DF \xc3\x9f LATIN SMALL LETTER SHARP S
+ U+02DA \xcb\x9a RING ABOVE
+ U+02E2 \xcb\xa2 MODIFIER LETTER SMALL S */
+ { "[a-z]|[^a-z]", "\xcb\xa2", REG_EXTENDED, 2,
+ { { 0, 2 }, { -1, -1 } } },
+ { "[a-z]", "\xc3\x9f", REG_EXTENDED, 2,
+ { { 0, 2 }, { -1, -1 } } },
+ { "[^a-z]", "\xcb\x9a", REG_EXTENDED, 2,
+ { { 0, 2 }, { -1, -1 } } },
};
-int
-main (void)
+
+static int
+do_test (void)
{
regex_t re;
regmatch_t rm[5];
@@ -93,3 +105,6 @@ main (void)
return ret;
}
+
+#define TEST_FUNCTION do_test ()
+#include "../test-skeleton.c"
diff --git a/posix/regcomp.c b/posix/regcomp.c
index d5a0535..4843cfe 100644
--- a/posix/regcomp.c
+++ b/posix/regcomp.c
@@ -1,5 +1,6 @@
/* Extended regular expression matching and search library.
- Copyright (C) 2002,2003,2004,2005,2006,2007 Free Software Foundation, Inc.
+ Copyright (C) 2002,2003,2004,2005,2006,2007,2009
+ Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
@@ -350,47 +351,67 @@ re_compile_fastmap_iter (regex_t *bufp, const re_dfastate_t *init_state,
#ifdef RE_ENABLE_I18N
else if (type == COMPLEX_BRACKET)
{
- int i;
re_charset_t *cset = dfa->nodes[node].opr.mbcset;
- if (cset->non_match || cset->ncoll_syms || cset->nequiv_classes
- || cset->nranges || cset->nchar_classes)
- {
+ int i;
+
# ifdef _LIBC
- if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0)
+ /* See if we have to try all bytes which start multiple collation
+ elements.
+ e.g. In da_DK, we want to catch 'a' since "aa" is a valid
+ collation element, and don't catch 'b' since 'b' is
+ the only collation element which starts from 'b' (and
+ it is caught by SIMPLE_BRACKET). */
+ if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0
+ && (cset->ncoll_syms || cset->nranges))
{
- /* In this case we want to catch the bytes which are
- the first byte of any collation elements.
- e.g. In da_DK, we want to catch 'a' since "aa"
- is a valid collation element, and don't catch
- 'b' since 'b' is the only collation element
- which starts from 'b'. */
const int32_t *table = (const int32_t *)
_NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
for (i = 0; i < SBC_MAX; ++i)
if (table[i] < 0)
re_set_fastmap (fastmap, icase, i);
}
-# else
- if (dfa->mb_cur_max > 1)
- for (i = 0; i < SBC_MAX; ++i)
- if (__btowc (i) == WEOF)
- re_set_fastmap (fastmap, icase, i);
-# endif /* not _LIBC */
- }
- for (i = 0; i < cset->nmbchars; ++i)
+# endif /* _LIBC */
+
+ /* See if we have to start the match at all multibyte characters,
+ i.e. where we would not find an invalid sequence. This only
+ applies to multibyte character sets; for single byte character
+ sets, the SIMPLE_BRACKET again suffices. */
+ if (dfa->mb_cur_max > 1
+ && (cset->nchar_classes || cset->non_match
+# ifdef _LIBC
+ || cset->nequiv_classes
+# endif /* _LIBC */
+ ))
{
- char buf[256];
- mbstate_t state;
- memset (&state, '\0', sizeof (state));
- if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1)
- re_set_fastmap (fastmap, icase, *(unsigned char *) buf);
- if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
+ unsigned char c = 0;
+ do
{
- if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state)
- != (size_t) -1)
- re_set_fastmap (fastmap, 0, *(unsigned char *) buf);
+ mbstate_t mbs;
+ memset (&mbs, 0, sizeof (mbs));
+ if (__mbrtowc (NULL, (char *) &c, 1, &mbs) == (size_t) -2)
+ re_set_fastmap (fastmap, false, (int) c);
}
+ while (++c != 0);
}
+
+ else
+ {
+ /* ... Else catch all bytes which can start the mbchars. */
+ for (i = 0; i < cset->nmbchars; ++i)
+ {
+ char buf[256];
+ mbstate_t state;
+ memset (&state, '\0', sizeof (state));
+ if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1)
+ re_set_fastmap (fastmap, icase, *(unsigned char *) buf);
+ if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
+ {
+ if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state)
+ != (size_t) -1)
+ re_set_fastmap (fastmap, false, *(unsigned char *) buf);
+ }
+ }
+ }
}
#endif /* RE_ENABLE_I18N */
else if (type == OP_PERIOD
diff --git a/ChangeLog b/ChangeLog
index 8829b44..17e987b 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2009-01-05 Paolo Bonzini <bonzini@gnu.org>
+
+ [BZ 697]
+ * posix/regexec.c (prune_impossible_nodes): Handle sifted_states[0]
+ being NULL also if there are no backreferences.
+ * posix/rxspencer/tests: Add testcases.
+
2009-01-04 Paolo Bonzini <bonzini@gnu.org>
[BZ 9697]
diff --git a/posix/regexec.c b/posix/regexec.c
index 135efe7..7bf0c08 100644
--- a/posix/regexec.c
+++ b/posix/regexec.c
@@ -1,5 +1,5 @@
/* Extended regular expression matching and search library.
- Copyright (C) 2002, 2003, 2004, 2005, 2007 Free Software Foundation, Inc.
+ Copyright (C) 2002, 2003, 2004, 2005, 2007, 2009 Free Software Foundation, Inc.
This file is part of the GNU C Library.
Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
@@ -1004,6 +1004,11 @@ prune_impossible_nodes (mctx)
re_node_set_free (&sctx.limits);
if (BE (ret != REG_NOERROR, 0))
goto free_return;
+ if (sifted_states[0] == NULL)
+ {
+ ret = REG_NOMATCH;
+ goto free_return;
+ }
}
re_free (mctx->state_log);
mctx->state_log = sifted_states;
diff --git a/posix/rxspencer/tests b/posix/rxspencer/tests
index b84a270..3ad46e2 100644
--- a/posix/rxspencer/tests
+++ b/posix/rxspencer/tests
@@ -536,3 +536,7 @@ a.*\b & abT ab
\B & aSbTc
\B & SaT @SaT
\B & aSTSb @TSb
+
+o$($|.) - oN
+o$($|.) - op
+o$($|.) - o o
Reply to: