Mercurial > hg > octave-nkf > gnulib-hg
diff lib/regcomp.c @ 11000:683b379e6760
regex: fix glibc bug 9697
2009-01-09 Paolo Bonzini <bonzini@gnu.org>
* lib/regcomp.c (re_compile_fastmap_iter): Rewrite COMPLEX_BRACKET
handling.
author | Paolo Bonzini <bonzini@gnu.org> |
---|---|
date | Fri, 09 Jan 2009 09:10:36 +0100 (2009-01-09) |
parents | cc7a1af3872f |
children | 0b80ccdc9aa4 |
line wrap: on
line diff
--- a/lib/regcomp.c +++ b/lib/regcomp.c @@ -357,45 +357,65 @@ #ifdef RE_ENABLE_I18N else if (type == COMPLEX_BRACKET) { + re_charset_t *cset = dfa->nodes[node].opr.mbcset; Idx 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) - { + # 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, false, *(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); + } } } }