]>
Commit | Line | Data |
---|---|---|
45516216 | 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 * | |
9 | *_nl_current_LC_CTYPE | |
10 | __attribute__ ((tls_model("initial-exec"))); | |
11 | typedef unsigned long int reg_syntax_t; | |
12 | extern reg_syntax_t re_syntax_options; | |
13 | typedef enum { | |
14 | REG_ENOSYS = -1, REG_NOERROR = | |
15 | 0, REG_NOMATCH, REG_BADPAT, REG_ECOLLATE, REG_ECTYPE, REG_EESCAPE, | |
16 | REG_ESUBREG, REG_EBRACK, REG_EPAREN, REG_EBRACE, REG_BADBR, | |
17 | REG_ERANGE, REG_ESPACE, REG_BADRPT, REG_EEND, REG_ESIZE, REG_ERPAREN | |
18 | } reg_errcode_t; | |
19 | typedef struct { | |
20 | unsigned char *buffer; | |
21 | reg_syntax_t syntax; | |
22 | char *fastmap; | |
23 | unsigned char *translate; | |
24 | unsigned int re_nsub; | |
25 | unsigned fastmap_accurate:1; | |
26 | unsigned not_bol:1; | |
27 | unsigned not_eol:1; | |
28 | } regex_t; | |
29 | extern __thread void *__libc_tsd_CTYPE_TOLOWER; | |
30 | inline | |
31 | int ** | |
32 | __ctype_tolower_loc(void) | |
33 | { | |
34 | union { | |
35 | void **ptr; | |
36 | int **tablep; | |
37 | } u; | |
38 | u.ptr = (&__libc_tsd_CTYPE_TOLOWER); | |
39 | if (*u.tablep == 0) | |
40 | *u.tablep = | |
41 | ((int *)((*_nl_current_LC_CTYPE)->values[0].string) + 128); | |
42 | return u.tablep; | |
43 | } | |
44 | inline int tolower(int __c) | |
45 | { | |
46 | return *__ctype_tolower_loc()[__c] | |
47 | ;} | |
48 | typedef struct { | |
49 | } mbstate_t; | |
50 | typedef unsigned long int *bitset_t; | |
51 | typedef unsigned long int *re_bitset_ptr_t; | |
52 | typedef struct { | |
53 | int nelem; | |
54 | int *elems; | |
55 | } re_node_set; | |
56 | typedef enum { | |
57 | NON_TYPE = 0, CHARACTER = 1, END_OF_RE = 2, SIMPLE_BRACKET = | |
58 | 3, OP_BACK_REF = 4, OP_PERIOD = 5, COMPLEX_BRACKET = | |
59 | 6, OP_UTF8_PERIOD = 7, OP_OPEN_SUBEXP = 8 | 0, OP_CLOSE_SUBEXP = | |
60 | 8 | 1, OP_ALT = 8 | 2, OP_DUP_ASTERISK = 8 | 3, ANCHOR = | |
61 | 8 | 4, CONCAT = 16, SUBEXP = 17, OP_DUP_PLUS = | |
62 | 18, OP_DUP_QUESTION, OP_OPEN_BRACKET, OP_CLOSE_BRACKET, | |
63 | OP_CHARSET_RANGE, OP_OPEN_DUP_NUM, OP_CLOSE_DUP_NUM, | |
64 | OP_NON_MATCH_LIST, OP_OPEN_COLL_ELEM, OP_CLOSE_COLL_ELEM, | |
65 | OP_OPEN_EQUIV_CLASS, OP_CLOSE_EQUIV_CLASS, OP_OPEN_CHAR_CLASS, | |
66 | OP_CLOSE_CHAR_CLASS, OP_WORD, OP_NOTWORD, OP_SPACE, OP_NOTSPACE, | |
67 | BACK_SLASH | |
68 | } re_token_type_t; | |
69 | typedef struct { | |
70 | long int *mbchars; | |
71 | int nmbchars; | |
72 | } re_charset_t; | |
73 | typedef struct { | |
74 | union { | |
75 | unsigned char c; | |
76 | re_bitset_ptr_t sbcset; | |
77 | re_charset_t *mbcset; | |
78 | int idx; | |
79 | } opr; | |
80 | re_token_type_t type:8; | |
81 | } re_token_t; | |
82 | typedef struct re_dfa_t re_dfa_t; | |
83 | typedef struct bin_tree_t bin_tree_t; | |
84 | struct re_dfastate_t { | |
85 | re_node_set nodes; | |
86 | }; | |
87 | typedef struct re_dfastate_t re_dfastate_t; | |
88 | struct re_dfa_t { | |
89 | re_token_t *nodes; | |
90 | unsigned int nodes_len; | |
91 | int *nexts; | |
92 | int *org_indices; | |
93 | re_node_set *edests; | |
94 | re_node_set *eclosures; | |
95 | re_dfastate_t *init_state; | |
96 | bin_tree_t *str_tree; | |
97 | re_bitset_ptr_t sb_char; | |
98 | int nbackref; | |
99 | unsigned int has_mb_node:1; | |
100 | unsigned int is_utf8:1; | |
101 | unsigned int map_notascii:1; | |
102 | int mb_cur_max; | |
103 | int *subexp_map; | |
104 | }; | |
105 | static reg_errcode_t re_compile_internal(regex_t * preg, | |
106 | const | |
107 | char *pattern, | |
108 | unsigned int length, | |
109 | reg_syntax_t syntax); | |
110 | static void re_compile_fastmap_iter(regex_t * bufp, | |
111 | const re_dfastate_t * init_state, | |
112 | char *fastmap); | |
113 | static reg_errcode_t analyze(regex_t * preg); | |
114 | static reg_errcode_t optimize_subexps(void *extra, bin_tree_t * node); | |
115 | const char *__re_compile_pattern(pattern, length, bufp) | |
116 | const char *pattern; | |
117 | { | |
118 | reg_errcode_t ret; | |
119 | ret = re_compile_internal(bufp, pattern, length, re_syntax_options); | |
120 | } | |
121 | ||
122 | int __re_compile_fastmap(regex_t *bufp) | |
123 | { | |
124 | re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; | |
125 | char *fastmap = bufp->fastmap; | |
126 | re_compile_fastmap_iter(bufp, dfa->init_state, fastmap); | |
127 | } | |
128 | static inline void | |
129 | re_set_fastmap(char *fastmap, int icase, | |
130 | int ch) | |
131 | { | |
132 | if (icase) { | |
133 | int __res; | |
134 | if (sizeof(ch) > 1) { | |
135 | if (ch) { | |
136 | } else | |
137 | __res = tolower(ch); | |
138 | } | |
139 | fastmap[__res] = 1; | |
140 | } | |
141 | } | |
142 | static void re_compile_fastmap_iter(regex_t * bufp, | |
143 | const re_dfastate_t * init_state, | |
144 | char *fastmap) | |
145 | { | |
146 | re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; | |
147 | int node_cnt; | |
148 | int icase = (dfa->mb_cur_max == 1 && (bufp->syntax & 0x40000)); | |
149 | for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) { | |
150 | int node = init_state->nodes.elems[node_cnt]; | |
151 | re_token_type_t type = dfa->nodes[node].type; | |
152 | if (type == CHARACTER) { | |
153 | if ((bufp->syntax & 0x40000) && dfa->mb_cur_max > 1) { | |
154 | unsigned char *buf | |
155 | = | |
156 | alloca(dfa->mb_cur_max); | |
157 | } | |
158 | int i; | |
159 | int ch; | |
160 | for (; | |
161 | i < (256 / (sizeof(unsigned long int) * 8)); ) { | |
162 | int j; | |
163 | unsigned long int w = | |
164 | dfa->nodes[node].opr.sbcset[i]; | |
165 | for (; j < (sizeof(unsigned long int) * 8);) | |
166 | if (w & ((unsigned long int)1 << j)) | |
167 | re_set_fastmap(fastmap, icase, | |
168 | ch); | |
169 | } | |
170 | } else if (type == COMPLEX_BRACKET) { | |
171 | int i; | |
172 | re_charset_t *cset = dfa->nodes[node].opr.mbcset; | |
173 | for (i = 0; i < cset->nmbchars; ++i) { | |
174 | char buf[256]; | |
175 | mbstate_t state; | |
176 | if (__wcrtomb(buf, cset->mbchars[i], &state) != | |
177 | (unsigned int)-1) | |
178 | re_set_fastmap(fastmap, icase, | |
179 | *(unsigned char *)buf); | |
180 | if ((bufp->syntax & 0x4000) | |
181 | && dfa->mb_cur_max > 1) { | |
182 | if (__wcrtomb | |
183 | (buf, towlower(cset->mbchars[i]), | |
184 | &state) != (unsigned int)-1) | |
185 | re_set_fastmap(fastmap, 0, | |
186 | *(unsigned char | |
187 | *)buf); | |
188 | } | |
189 | } | |
190 | } | |
191 | } | |
192 | } | |
193 | static const bitset_t utf8_sb_map; | |
194 | static reg_errcode_t re_compile_internal(regex_t * preg, const char *pattern, | |
195 | unsigned int length, | |
196 | reg_syntax_t syntax) | |
197 | { | |
198 | reg_errcode_t err; | |
199 | re_dfa_t *dfa; | |
200 | preg->fastmap_accurate = 0; | |
201 | preg->not_bol = preg->not_eol = 0; | |
202 | err = analyze(preg); | |
203 | if (dfa->is_utf8 && !(syntax & 0x400000) && preg->translate == 0) | |
204 | optimize_utf8(dfa); | |
205 | if (dfa->mb_cur_max == 6 && __extension__( { | |
206 | unsigned int __s1_len; | |
207 | unsigned int __s2_len; | |
208 | ( | |
209 | (__s1_len = | |
210 | strlen(((*_nl_current_LC_CTYPE)->values[0].string)), __s2_len = strlen("UTF-8"), (!((unsigned int)(const void *)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (unsigned int)(const void *)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) || __s1_len >= 4) && (!((unsigned int)(const void *)(("UTF-8") + 1) - (unsigned int)(const void *)("UTF-8") == 1) || __s2_len >= 4)) ? strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : ( | |
211 | ((unsigned int)(const void *)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (unsigned int)(const void *)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) && (__s1_len = strlen(((*_nl_current_LC_CTYPE)->values[0].string)), __s1_len < 4) ? ( | |
212 | ((unsigned int)(const void *)(("UTF-8") + 1) - (unsigned int)(const void *)("UTF-8") == 1) ? | |
213 | strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (__extension__( { | |
214 | __const | |
215 | unsigned | |
216 | char | |
217 | *__s2 | |
218 | = | |
219 | (__const | |
220 | unsigned | |
221 | char | |
222 | *) | |
223 | (__const | |
224 | char | |
225 | *) | |
226 | ("UTF-8"); | |
227 | register | |
228 | int | |
229 | __result | |
230 | = | |
231 | (((__const unsigned char *)(__const char *)(((*_nl_current_LC_CTYPE)->values[0].string)))[0] - __s2[0]); __result;} | |
232 | ))): ( | |
233 | ((unsigned int)(const void *)(("UTF-8") + 1) - (unsigned int)(const void *)("UTF-8") == 1) && (__s2_len = strlen("UTF-8"), __s2_len < 4) ? ( | |
234 | ((unsigned int)(const void *)((((*_nl_current_LC_CTYPE)->values[0].string)) + 1) - (unsigned int)(const void *)(((*_nl_current_LC_CTYPE)->values[0].string)) == 1) ? | |
235 | strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8") : (__extension__( { | |
236 | __const | |
237 | unsigned | |
238 | char | |
239 | *__s1 | |
240 | = | |
241 | (__const | |
242 | unsigned | |
243 | char | |
244 | *) | |
245 | (__const | |
246 | char | |
247 | *) | |
248 | (((*_nl_current_LC_CTYPE)->values[0].string)); register int __result = __s1[0] - ((__const unsigned char *)(__const char *)("UTF-8"))[0]; __result;} | |
249 | ))): strcmp(((*_nl_current_LC_CTYPE)->values[0].string), "UTF-8"))));} | |
250 | ) == 0) | |
251 | dfa->is_utf8 = 1; | |
252 | dfa->map_notascii = | |
253 | (((unsigned int)(*_nl_current_LC_CTYPE)->values[0].word) != 0); | |
254 | if (dfa->mb_cur_max > 1) { | |
255 | if (dfa->is_utf8) | |
256 | dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map; | |
257 | else { | |
258 | int i; | |
259 | int j; | |
260 | int ch; | |
261 | dfa->sb_char = | |
262 | (re_bitset_ptr_t) calloc(sizeof(bitset_t), 1); | |
263 | for (i = 0, ch = 0; | |
264 | i < (256 / (sizeof(unsigned long int) * 8)); ++i) | |
265 | for (j = 0; j < (sizeof(unsigned long int) * 8); | |
266 | ++j, ++ch) { | |
267 | unsigned int wch = __btowc(ch); | |
268 | if (wch != (0xffffffffu)) | |
269 | dfa->sb_char[i] |= | |
270 | (unsigned long int)1 << j; | |
271 | } | |
272 | } | |
273 | } | |
274 | int node; | |
275 | int i; | |
276 | int mb_chars; | |
277 | int has_period; | |
278 | for ( | |
279 | ; node < dfa->nodes_len;) | |
280 | switch (dfa->nodes[node].type) { | |
281 | case CHARACTER: | |
282 | if (dfa->nodes[node].opr.c >= 0x80) | |
283 | mb_chars = 1; | |
284 | case ANCHOR: | |
285 | switch (dfa->nodes[node].opr.idx) { | |
286 | } | |
287 | case OP_PERIOD: | |
288 | has_period = 1; | |
289 | for (i = 0x80 / (sizeof(unsigned long int) * 8); | |
290 | i < (256 / (sizeof(unsigned long int) * 8)); ++i) | |
291 | if (dfa->nodes[node].opr.sbcset[i]) | |
292 | return; | |
293 | } | |
294 | dfa->has_mb_node = dfa->nbackref > 0 || has_period; | |
295 | } | |
296 | static reg_errcode_t analyze(regex_t * preg) | |
297 | { | |
298 | re_dfa_t *dfa = (re_dfa_t *) preg->buffer; | |
299 | if ( | |
300 | (dfa->nexts == 0 || dfa->org_indices == 0 || dfa->edests == 0 | |
301 | || dfa->eclosures == 0)) | |
302 | return REG_ESPACE; | |
303 | if (dfa->subexp_map != 0) { | |
304 | int i; | |
305 | for ( | |
306 | ; i < preg->re_nsub; | |
307 | ) | |
308 | dfa->subexp_map[i] = i; | |
309 | preorder(dfa->str_tree, optimize_subexps, dfa); | |
310 | for (i = 0; i < preg->re_nsub; i++) | |
311 | if (dfa->subexp_map[i] != i) | |
312 | break; | |
313 | } | |
314 | } |