diff options
Diffstat (limited to 'posix')
| -rw-r--r-- | posix/bug-regex20.c | 22 | ||||
| -rw-r--r-- | posix/regcomp.c | 13 | ||||
| -rw-r--r-- | posix/regexec.c | 7 |
3 files changed, 31 insertions, 11 deletions
diff --git a/posix/bug-regex20.c b/posix/bug-regex20.c index 11b9484faf..74662e6246 100644 --- a/posix/bug-regex20.c +++ b/posix/bug-regex20.c @@ -43,15 +43,35 @@ static struct \xe2\x80\x94 EM DASH */ /* Should be optimized. */ {RE_SYNTAX_POSIX_BASIC, "foo", "b\xc3\xa4rfoob\xc3\xa4z", 4, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4*z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4*z", "b\xc3\xa4rfoobz", 7, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4\\+z", + "b\xc3\xa4rfoob\xc3\xa4\xc3\xa4z", 7, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4\\?z", "b\xc3\xa4rfoob\xc3\xa4z", 7, 1}, + {RE_SYNTAX_POSIX_BASIC, "b\xc3\xa4\\{1,2\\}z", + "b\xc3\xa4rfoob\xc3\xa4z", 7, 1}, {RE_SYNTAX_POSIX_BASIC, "^x\\|xy*z$", "\xc3\xb6xyyz", 2, 1}, {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{6\\}z\\+", "x\\yyyyyyzz\xc3\xb6", 0, 1}, {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{2,36\\}z\\+", "x\\yzz\xc3\xb6", -1, 1}, {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{,3\\}z\\+", "x\\yyyzz\xc3\xb6", 0, 1}, + {RE_SYNTAX_POSIX_BASIC, "^x\\|x\xc3\xa4*z$", + "\xc3\xb6x\xc3\xa4\xc3\xa4z", 2, 1}, + {RE_SYNTAX_POSIX_BASIC, "^x\\\\\xc3\x84\\{6\\}z\\+", + "x\\\xc3\x84\xc3\x84\xc3\x84\xc3\x84\xc3\x84\xc3\x84zz\xc3\xb6", 0, 1}, + {RE_SYNTAX_POSIX_BASIC, "^x\\\\\xc3\x84\\{2,36\\}z\\+", + "x\\\xc3\x84zz\xc3\xb6", -1, 1}, + {RE_SYNTAX_POSIX_BASIC, "^x\\\\\xc3\x84\\{,3\\}z\\+", + "x\\\xc3\x84\xc3\x84\xc3\x84zz\xc3\xb6", 0, 1}, {RE_SYNTAX_POSIX_BASIC, "x[C]y", "axCy", 1, 1}, {RE_SYNTAX_POSIX_BASIC, "x[ABC]y", "axCy", 1, 1}, {RE_SYNTAX_POSIX_BASIC, "\\`x\\|z\\'", "x\xe2\x80\x94", 0, 1}, {RE_SYNTAX_POSIX_BASIC, "\\(xy\\)z\\1a\\1", "\xe2\x80\x94xyzxyaxy\xc3\x84", 3, 1}, {RE_SYNTAX_POSIX_BASIC, "xy\\?z", "\xc3\x84xz\xc3\xb6", 2, 1}, + {RE_SYNTAX_POSIX_BASIC, "\\`\xc3\x84\\|z\\'", "\xc3\x84\xe2\x80\x94", 0, 1}, + {RE_SYNTAX_POSIX_BASIC, "\\(x\xc3\x84\\)z\\1\x61\\1", + "\xe2\x80\x94x\xc3\x84zx\xc3\x84\x61x\xc3\x84\xc3\x96", 3, 1}, + {RE_SYNTAX_POSIX_BASIC, "x\xc3\x96\\?z", "\xc3\x84xz\xc3\xb6", 2, 1}, {RE_SYNTAX_POSIX_EXTENDED, "foo", "b\xc3\xa4rfoob\xc3\xa4z", 4, 1}, {RE_SYNTAX_POSIX_EXTENDED, "^x|xy*z$", "\xc3\xb6xyyz", 2, 1}, {RE_SYNTAX_POSIX_EXTENDED, "^x\\\\y{6}z+", "x\\yyyyyyzz\xc3\xb6", 0, 1}, @@ -64,7 +84,6 @@ static struct {RE_SYNTAX_POSIX_EXTENDED, "xy?z", "\xc3\x84xz\xc3\xb6", 2, 1}, /* Should not be optimized. */ {RE_SYNTAX_POSIX_BASIC, "x.y", "ax\xe2\x80\x94yz", 1, 0}, - {RE_SYNTAX_POSIX_BASIC, "x\xc3\x96*y", "ax\xc3\x96\xc3\x96yz", 1, 0}, {RE_SYNTAX_POSIX_BASIC, "x[\xc3\x84\xc3\xa4]y", "ax\xc3\xa4y", 1, 0}, {RE_SYNTAX_POSIX_BASIC, "x[A-Z,]y", "axCy", 1, 0}, {RE_SYNTAX_POSIX_BASIC, "x[^y]z", "ax\xe2\x80\x94z", 1, 0}, @@ -77,7 +96,6 @@ static struct {RE_SYNTAX_POSIX_BASIC, "a\\wz", "a\xc3\x84z", 0, 0}, {RE_SYNTAX_POSIX_BASIC, "x\\Wz", "\xc3\x96x\xe2\x80\x94z", 2, 0}, {RE_SYNTAX_POSIX_EXTENDED, "x.y", "ax\xe2\x80\x94yz", 1, 0}, - {RE_SYNTAX_POSIX_EXTENDED, "x\xc3\x96*y", "ax\xc3\x96\xc3\x96yz", 1, 0}, {RE_SYNTAX_POSIX_EXTENDED, "x[\xc3\x84\xc3\xa4]y", "ax\xc3\xa4y", 1, 0}, {RE_SYNTAX_POSIX_EXTENDED, "x[A-Z,]y", "axCy", 1, 0}, {RE_SYNTAX_POSIX_EXTENDED, "x[^y]z", "ax\xe2\x80\x94z", 1, 0}, diff --git a/posix/regcomp.c b/posix/regcomp.c index 68ce551c3a..b5f0c92a3a 100644 --- a/posix/regcomp.c +++ b/posix/regcomp.c @@ -965,17 +965,14 @@ static void optimize_utf8 (dfa) re_dfa_t *dfa; { - int node, i; + int node, i, mb_chars = 0; for (node = 0; node < dfa->nodes_len; ++node) switch (dfa->nodes[node].type) { case CHARACTER: - /* Chars >= 0x80 are optimizable in some cases (e.g. when not - followed by DUP operator, not in bracket etc.). - For now punt on them all. */ if (dfa->nodes[node].opr.c >= 0x80) - return; + mb_chars = 1; break; case ANCHOR: switch (dfa->nodes[node].opr.idx) @@ -1010,6 +1007,12 @@ optimize_utf8 (dfa) return; } + if (mb_chars) + for (node = 0; node < dfa->nodes_len; ++node) + if (dfa->nodes[node].type == CHARACTER + && dfa->nodes[node].opr.c >= 0x80) + dfa->nodes[node].mb_partial = 0; + /* The search can be in single byte locale. */ dfa->mb_cur_max = 1; dfa->is_utf8 = 0; diff --git a/posix/regexec.c b/posix/regexec.c index 7470197506..09756b7691 100644 --- a/posix/regexec.c +++ b/posix/regexec.c @@ -3483,10 +3483,6 @@ check_node_accept_bytes (preg, node_idx, input, str_idx) int elem_len = re_string_elem_size_at (input, str_idx); int char_len = re_string_char_size_at (input, str_idx); int i; -# ifdef _LIBC - int j; - uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); -# endif /* _LIBC */ if (elem_len <= 1 && char_len <= 1) return 0; if (node->type == OP_PERIOD) @@ -3505,6 +3501,8 @@ check_node_accept_bytes (preg, node_idx, input, str_idx) # ifdef _LIBC const unsigned char *pin = ((char *) re_string_get_buffer (input) + str_idx); + int j; + uint32_t nrules; # endif /* _LIBC */ int match_len = 0; wchar_t wc = ((cset->nranges || cset->nchar_classes || cset->nmbchars) @@ -3529,6 +3527,7 @@ check_node_accept_bytes (preg, node_idx, input, str_idx) } # ifdef _LIBC + nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); if (nrules != 0) { unsigned int in_collseq = 0; |