]>
Commit | Line | Data |
---|---|---|
1 | struct locale_data { | |
2 | unsigned int nstrings; | |
3 | union locale_data_value { | |
4 | char *string; | |
5 | unsigned int word; | |
6 | } values[]; | |
7 | }; | |
8 | extern __thread struct locale_data **_nl_current_LC_CTYPE; | |
9 | typedef unsigned long int reg_syntax_t; | |
10 | extern reg_syntax_t re_syntax_options; | |
11 | typedef enum { | |
12 | REG_ENOSYS | |
13 | } reg_errcode_t; | |
14 | typedef struct { | |
15 | reg_syntax_t syntax; | |
16 | unsigned char *translate; | |
17 | unsigned int re_nsub; | |
18 | } regex_t; | |
19 | extern __thread void *__libc_tsd_CTYPE_TOLOWER; | |
20 | inline int **__ctype_tolower_loc(void) | |
21 | { | |
22 | union { | |
23 | void **ptr; | |
24 | int **tablep; | |
25 | } u; | |
26 | u.ptr = (&__libc_tsd_CTYPE_TOLOWER); | |
27 | if (*u.tablep == 0) | |
28 | *u.tablep = (*_nl_current_LC_CTYPE)->values[0].string; | |
29 | return u.tablep; | |
30 | } | |
31 | inline int tolower(int __c) | |
32 | { | |
33 | return *__ctype_tolower_loc()[__c]; | |
34 | } | |
35 | typedef struct { | |
36 | } mbstate_t; | |
37 | typedef struct { | |
38 | int nelem; | |
39 | } re_node_set; | |
40 | typedef enum { | |
41 | CHARACTER = 1, COMPLEX_BRACKET = 6 | |
42 | } re_token_type_t; | |
43 | typedef struct { | |
44 | long int *mbchars; | |
45 | int nmbchars; | |
46 | } re_charset_t; | |
47 | typedef struct { | |
48 | union { | |
49 | unsigned char c; | |
50 | unsigned long int *sbcset; | |
51 | } opr; | |
52 | re_token_type_t type; | |
53 | } re_token_t; | |
54 | typedef struct re_dfa_t re_dfa_t; | |
55 | typedef struct { | |
56 | re_node_set nodes; | |
57 | } re_dfastate_t; | |
58 | struct re_dfa_t { | |
59 | re_token_t *nodes; | |
60 | unsigned int nodes_len; | |
61 | re_dfastate_t *init_state; | |
62 | unsigned long int *sb_char; | |
63 | int nbackref; | |
64 | unsigned int has_mb_node:1; | |
65 | unsigned int is_utf8:1; | |
66 | unsigned int map_notascii:1; | |
67 | int mb_cur_max; | |
68 | int *subexp_map; | |
69 | }; | |
70 | static inline void re_set_fastmap(char *fastmap, int icase, int ch) | |
71 | { | |
72 | if (icase) { | |
73 | int __res; | |
74 | if (sizeof(ch) > 1) { | |
75 | if (ch) { | |
76 | } else | |
77 | __res = tolower(ch); | |
78 | } | |
79 | fastmap[__res] = 1; | |
80 | } | |
81 | } | |
82 | static void re_compile_fastmap_iter(regex_t * bufp, re_dfastate_t * init_state, | |
83 | char *fastmap) | |
84 | { | |
85 | re_dfa_t *dfa; | |
86 | int node_cnt; | |
87 | int icase; | |
88 | for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) { | |
89 | re_token_type_t type; | |
90 | if (type == CHARACTER) { | |
91 | if ((bufp->syntax & 0x40000) && dfa->mb_cur_max > 1) { | |
92 | void *buf = alloca(dfa->mb_cur_max); | |
93 | } | |
94 | int i; | |
95 | int ch; | |
96 | for (; i < (256 / (sizeof(unsigned long int) * 8));) { | |
97 | int j; | |
98 | int w; | |
99 | for (; j < 8;) | |
100 | if (w & 1) | |
101 | re_set_fastmap(fastmap, icase, | |
102 | ch); | |
103 | } | |
104 | } else if (type == COMPLEX_BRACKET) { | |
105 | int i; | |
106 | re_charset_t *cset; | |
107 | for (; i < cset->nmbchars;) { | |
108 | char buf[256]; | |
109 | mbstate_t state; | |
110 | if (__wcrtomb(buf, cset->mbchars[i], &state) != | |
111 | -1) | |
112 | re_set_fastmap(fastmap, icase, *buf); | |
113 | } | |
114 | } | |
115 | } | |
116 | } | |
117 | static unsigned long int *utf8_sb_map; | |
118 | static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern, | |
119 | unsigned int length, | |
120 | reg_syntax_t syntax) | |
121 | { | |
122 | re_dfa_t *dfa; | |
123 | if (dfa->is_utf8 && !(syntax & 0x400000) && preg->translate == 0) | |
124 | optimize_utf8(dfa); | |
125 | int result; | |
126 | if (strlen((*_nl_current_LC_CTYPE)->values[0].string) < 4) | |
127 | result = (*_nl_current_LC_CTYPE)->values[0].string[0] - 'U'; | |
128 | else | |
129 | result = | |
130 | strcmp((*_nl_current_LC_CTYPE)->values[0].string, "UTF-8"); | |
131 | if (dfa->mb_cur_max == 6 && result == 0) | |
132 | dfa->is_utf8 = 1; | |
133 | dfa->map_notascii = (*_nl_current_LC_CTYPE)->values[0].word != 0; | |
134 | if (dfa->mb_cur_max > 1) { | |
135 | if (dfa->is_utf8) | |
136 | dfa->sb_char = utf8_sb_map; | |
137 | else { | |
138 | int i; | |
139 | for (; i < (256 / (sizeof(unsigned long int) * 8));) ; | |
140 | } | |
141 | } | |
142 | int node; | |
143 | int i; | |
144 | int mb_chars; | |
145 | int has_period; | |
146 | for (; node < dfa->nodes_len;) | |
147 | switch (dfa->nodes[node].type) { | |
148 | case CHARACTER: | |
149 | if (dfa->nodes[node].opr.c >= 0x80) | |
150 | mb_chars = 1; | |
151 | for (; i < (256 / (sizeof(unsigned long int) * 8));) | |
152 | if (dfa->nodes[node].opr.sbcset[i]) | |
153 | return; | |
154 | } | |
155 | dfa->has_mb_node = dfa->nbackref > 0 || has_period; | |
156 | if (dfa->subexp_map != 0) { | |
157 | for (; i < preg->re_nsub; i++) | |
158 | if (dfa->subexp_map[i] != i) | |
159 | break; | |
160 | } | |
161 | } | |
162 | const char *__re_compile_pattern(const char *pattern, unsigned int length, | |
163 | regex_t * bufp) | |
164 | { | |
165 | reg_errcode_t ret = | |
166 | re_compile_internal(bufp, pattern, length, re_syntax_options); | |
167 | } | |
168 | ||
169 | void __re_compile_fastmap(regex_t * bufp) | |
170 | { | |
171 | re_dfa_t *dfa; | |
172 | char *fastmap; | |
173 | re_compile_fastmap_iter(bufp, dfa->init_state, fastmap); | |
174 | } |