author | "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr> |
Tue Mar 03 17:41:59 2009 +0000 (2009-03-03) | |
changeset 1219 | 2b875ed306c2 |
permissions | -rw-r--r-- |
1 Original patch from: gentoo/src/patchsets/glibc/2.9/0052_all_glibc-2.9-regex-BZ9697.patch
3 -= BEGIN original header =-
4 http://sourceware.org/ml/libc-alpha/2009-01/msg00005.html
6 From ea8ca0dfcbf2721bcf2c08ce3c01d5764b827572 Mon Sep 17 00:00:00 2001
7 From: Ulrich Drepper <drepper@redhat.com>
8 Date: Thu, 8 Jan 2009 00:42:28 +0000
9 Subject: [PATCH] (re_compile_fastmap_iter): Rewrite COMPLEX_BRACKET handling.
11 -= END original header =-
13 diff -durN glibc-2_9.orig/posix/regcomp.c glibc-2_9/posix/regcomp.c
14 --- glibc-2_9.orig/posix/regcomp.c 2008-05-15 05:07:21.000000000 +0200
15 +++ glibc-2_9/posix/regcomp.c 2009-02-02 22:00:41.000000000 +0100
16 @@ -350,47 +350,67 @@
17 #ifdef RE_ENABLE_I18N
18 else if (type == COMPLEX_BRACKET)
19 {
20 - int i;
21 re_charset_t *cset = dfa->nodes[node].opr.mbcset;
22 - if (cset->non_match || cset->ncoll_syms || cset->nequiv_classes
23 - || cset->nranges || cset->nchar_classes)
24 - {
25 + int i;
26 +
27 # ifdef _LIBC
28 - if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0)
29 + /* See if we have to try all bytes which start multiple collation
30 + elements.
31 + e.g. In da_DK, we want to catch 'a' since "aa" is a valid
32 + collation element, and don't catch 'b' since 'b' is
33 + the only collation element which starts from 'b' (and
34 + it is caught by SIMPLE_BRACKET). */
35 + if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0
36 + && (cset->ncoll_syms || cset->nranges))
37 {
38 - /* In this case we want to catch the bytes which are
39 - the first byte of any collation elements.
40 - e.g. In da_DK, we want to catch 'a' since "aa"
41 - is a valid collation element, and don't catch
42 - 'b' since 'b' is the only collation element
43 - which starts from 'b'. */
44 const int32_t *table = (const int32_t *)
45 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
46 for (i = 0; i < SBC_MAX; ++i)
47 if (table[i] < 0)
48 re_set_fastmap (fastmap, icase, i);
49 }
50 -# else
51 - if (dfa->mb_cur_max > 1)
52 - for (i = 0; i < SBC_MAX; ++i)
53 - if (__btowc (i) == WEOF)
54 - re_set_fastmap (fastmap, icase, i);
55 -# endif /* not _LIBC */
56 - }
57 - for (i = 0; i < cset->nmbchars; ++i)
58 +# endif /* _LIBC */
59 +
60 + /* See if we have to start the match at all multibyte characters,
61 + i.e. where we would not find an invalid sequence. This only
62 + applies to multibyte character sets; for single byte character
63 + sets, the SIMPLE_BRACKET again suffices. */
64 + if (dfa->mb_cur_max > 1
65 + && (cset->nchar_classes || cset->non_match
66 +# ifdef _LIBC
67 + || cset->nequiv_classes
68 +# endif /* _LIBC */
69 + ))
70 {
71 - char buf[256];
72 - mbstate_t state;
73 - memset (&state, '\0', sizeof (state));
74 - if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1)
75 - re_set_fastmap (fastmap, icase, *(unsigned char *) buf);
76 - if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
77 + unsigned char c = 0;
78 + do
79 {
80 - if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state)
81 - != (size_t) -1)
82 - re_set_fastmap (fastmap, 0, *(unsigned char *) buf);
83 + mbstate_t mbs;
84 + memset (&mbs, 0, sizeof (mbs));
85 + if (__mbrtowc (NULL, (char *) &c, 1, &mbs) == (size_t) -2)
86 + re_set_fastmap (fastmap, false, (int) c);
87 }
88 + while (++c != 0);
89 }
90 +
91 + else
92 + {
93 + /* ... Else catch all bytes which can start the mbchars. */
94 + for (i = 0; i < cset->nmbchars; ++i)
95 + {
96 + char buf[256];
97 + mbstate_t state;
98 + memset (&state, '\0', sizeof (state));
99 + if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1)
100 + re_set_fastmap (fastmap, icase, *(unsigned char *) buf);
101 + if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1)
102 + {
103 + if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state)
104 + != (size_t) -1)
105 + re_set_fastmap (fastmap, false, *(unsigned char *) buf);
106 + }
107 + }
108 + }
109 }
110 #endif /* RE_ENABLE_I18N */
111 else if (type == OP_PERIOD