1 /***********************************************************************
2 * *
3 * This software is part of the ast package *
4 * Copyright (c) 1985-2011 AT&T Intellectual Property *
5 * and is licensed under the *
6 * Eclipse Public License, Version 1.0 *
7 * by AT&T Intellectual Property *
8 * *
9 * A copy of the License is available at *
10 * http://www.eclipse.org/org/documents/epl-v10.html *
11 * (with md5 checksum b35adb5213ca9657e911e9befb180842) *
12 * *
13 * Information and Software Systems Research *
14 * AT&T Research *
15 * Florham Park NJ *
16 * *
17 * Glenn Fowler <gsf@research.att.com> *
18 * David Korn <dgk@research.att.com> *
19 * Phong Vo <kpv@research.att.com> *
20 * *
21 ***********************************************************************/
22 #pragma prototyped
23
24 /*
25 * posix regex state and alloc
26 */
27
28 #include "reglib.h"
29
30 #if _PACKAGE_ast
31
32 #include <ccode.h>
33
34 #else
35
36 #define CC_bel '\a'
37 #define CC_esc '\033'
38 #define CC_vt '\v'
39
40 #endif
41
42 /*
43 * state shared by all threads
44 */
45
46 State_t state =
47 {
48 { -1, -1 },
49
50 /*
51 * escape code table
52 * the "funny" things get special treatment at ends of BRE
53 *
54 * BRE 0:normal 1:escaped 2:escaped-char-class
55 * ERE 3:normal 4:escaped 5:escaped-char-class
56 * ARE 6:normal 7:escaped 8:escaped-char-class
57 * SRE 9:normal 10:escaped 11:escaped-char-class
58 * KRE 12:normal 13:escaped 14:escaped-char-class
59 */
60
61 '\\',
62 '\\', '\\', '\\',
63 '\\', '\\', '\\',
64 '\\', '\\', '\\',
65 '\\', '\\', '\\',
66 '\\', '\\', '\\',
67 '^', /* funny */
68 '^', '^', '^',
69 T_CFLX, '^', '^',
70 T_CFLX, '^', '^',
71 '^', '^', '^',
72 '^', '^', '^',
73 '.',
74 T_DOT, '.', T_BAD,
75 T_DOT, '.', T_BAD,
76 T_DOT, '.', T_BAD,
77 '.', '.', T_BAD,
78 '.', '.', T_BAD,
79 '$', /* funny */
80 '$', '$', T_BAD,
81 T_DOLL, '$', T_BAD,
82 T_DOLL, '$', T_BAD,
83 '$', '$', T_BAD,
84 '$', '$', T_BAD,
85 '*',
86 T_STAR, '*', T_BAD,
87 T_STAR, '*', T_BAD,
88 T_STAR, '*', T_BAD,
89 T_STAR, '*', '*',
90 T_STAR, '*', '*',
91 '[',
92 T_BRA, '[', '[',
93 T_BRA, '[', '[',
94 T_BRA, '[', '[',
95 T_BRA, '[', '[',
96 T_BRA, '[', '[',
97 '|',
98 '|', T_BAD, T_BAD,
99 T_BAR, '|', T_BAD,
100 T_BAR, '|', T_BAD,
101 '|', '|', T_BAD,
102 T_BAR, '|', T_BAD,
103 '+',
104 '+', T_BAD, T_BAD,
105 T_PLUS, '+', T_BAD,
106 T_PLUS, '+', T_BAD,
107 '+', '+', T_BAD,
108 T_PLUS, '+', T_BAD,
109 '?',
110 '?', T_BAD, T_BAD,
111 T_QUES, '?', T_BAD,
112 T_QUES, '?', T_BAD,
113 T_QUES, '?', '?',
114 T_QUES, '?', '?',
115 '(',
116 '(', T_OPEN, T_BAD,
117 T_OPEN, '(', T_BAD,
118 T_OPEN, '(', T_BAD,
119 '(', '(', '(',
120 T_OPEN, '(', '(',
121 ')',
122 ')', T_CLOSE, T_BAD,
123 T_CLOSE, ')', T_BAD,
124 T_CLOSE, ')', T_BAD,
125 ')', ')', ')',
126 T_CLOSE, ')', ')',
127 '{',
128 '{', T_LEFT, T_BAD,
129 T_LEFT, '{', T_BAD,
130 T_LEFT, '{', T_BAD,
131 '{', '{', '{',
132 T_LEFT, '{', '{',
133 '}',
134 '}', T_RIGHT, T_BAD,
135 '}', T_BAD, T_BAD,
136 '}', T_BAD, T_BAD,
137 '}', '}', '}',
138 '}', '}', '}',
139 '&',
140 '&', T_BAD, T_BAD,
141 '&', T_AND, T_BAD,
142 T_AND, '&', T_BAD,
143 '&', '&', T_BAD,
144 T_AND, '&', T_BAD,
145 '!',
146 '!', T_BAD, T_BAD,
147 '!', T_BANG, T_BAD,
148 T_BANG, '!', T_BAD,
149 '!', '!', T_BAD,
150 T_BANG, '!', T_BAD,
151 '@',
152 '@', T_BAD, T_BAD,
153 '@', T_BAD, T_BAD,
154 '@', T_BAD, T_BAD,
155 '@', '@', T_BAD,
156 T_AT, '@', T_BAD,
157 '~',
158 '~', T_BAD, T_BAD,
159 '~', T_BAD, T_BAD,
160 '~', T_BAD, T_BAD,
161 '~', '~', T_BAD,
162 T_TILDE, '~', T_BAD,
163 '%',
164 '%', T_BAD, T_BAD,
165 '%', T_BAD, T_BAD,
166 '%', T_BAD, T_BAD,
167 '%', '%', T_BAD,
168 T_PERCENT, '%', T_BAD,
169 '<',
170 '<', T_LT, T_BAD,
171 '<', T_LT, T_BAD,
172 T_LT, '<', T_BAD,
173 '<', '<', T_BAD,
174 '<', '<', T_BAD,
175 '>',
176 '>', T_GT, T_BAD,
177 '>', T_GT, T_BAD,
178 T_GT, '>', T_BAD,
179 '>', '>', T_BAD,
180 '>', '>', T_BAD,
181
182 /* backrefs */
183
184 '0',
185 '0', T_BACK+0, T_ESCAPE,
186 '0', T_BACK+0, T_ESCAPE,
187 '0', T_BACK+0, T_ESCAPE,
188 '0', T_BACK+0, T_ESCAPE,
189 '0', T_BACK+0, T_ESCAPE,
190 '1',
191 '1', T_BACK+1, T_ESCAPE,
192 '1', T_BACK+1, T_ESCAPE,
193 '1', T_BACK+1, T_ESCAPE,
194 '1', T_BACK+1, T_ESCAPE,
195 '1', T_BACK+1, T_ESCAPE,
196 '2',
197 '2', T_BACK+2, T_ESCAPE,
198 '2', T_BACK+2, T_ESCAPE,
199 '2', T_BACK+2, T_ESCAPE,
200 '2', T_BACK+2, T_ESCAPE,
201 '2', T_BACK+2, T_ESCAPE,
202 '3',
203 '3', T_BACK+3, T_ESCAPE,
204 '3', T_BACK+3, T_ESCAPE,
205 '3', T_BACK+3, T_ESCAPE,
206 '3', T_BACK+3, T_ESCAPE,
207 '3', T_BACK+3, T_ESCAPE,
208 '4',
209 '4', T_BACK+4, T_ESCAPE,
210 '4', T_BACK+4, T_ESCAPE,
211 '4', T_BACK+4, T_ESCAPE,
212 '4', T_BACK+4, T_ESCAPE,
213 '4', T_BACK+4, T_ESCAPE,
214 '5',
215 '5', T_BACK+5, T_ESCAPE,
216 '5', T_BACK+5, T_ESCAPE,
217 '5', T_BACK+5, T_ESCAPE,
218 '5', T_BACK+5, T_ESCAPE,
219 '5', T_BACK+5, T_ESCAPE,
220 '6',
221 '6', T_BACK+6, T_ESCAPE,
222 '6', T_BACK+6, T_ESCAPE,
223 '6', T_BACK+6, T_ESCAPE,
224 '6', T_BACK+6, T_ESCAPE,
225 '6', T_BACK+6, T_ESCAPE,
226 '7',
227 '7', T_BACK+7, T_ESCAPE,
228 '7', T_BACK+7, T_ESCAPE,
229 '7', T_BACK+7, T_ESCAPE,
230 '7', T_BACK+7, T_ESCAPE,
231 '7', T_BACK+7, T_ESCAPE,
232 '8',
233 '8', T_BACK+8, T_ESCAPE,
234 '8', T_BACK+8, T_ESCAPE,
235 '8', T_BACK+8, T_ESCAPE,
236 '8', '8', T_ESCAPE,
237 '8', T_BACK+8, T_ESCAPE,
238 '9',
239 '9', T_BACK+9, T_ESCAPE,
240 '9', T_BACK+9, T_ESCAPE,
241 '9', T_BACK+9, T_ESCAPE,
242 '9', '9', T_ESCAPE,
243 '9', T_BACK+9, T_ESCAPE,
244
245 /* perl */
246
247 'A',
248 'A', T_BEG_STR, T_BAD,
249 'A', T_BEG_STR, T_BAD,
250 'A', T_BEG_STR, T_BAD,
251 'A', T_BEG_STR, T_BAD,
252 'A', T_BEG_STR, T_BAD,
253 'b',
254 'b', T_WORD, '\b',
255 'b', T_WORD, '\b',
256 'b', T_WORD, '\b',
257 'b', T_WORD, '\b',
258 'b', T_WORD, '\b',
259 'B',
260 'B', T_WORD_NOT, T_BAD,
261 'B', T_WORD_NOT, T_BAD,
262 'B', T_WORD_NOT, T_BAD,
263 'B', T_WORD_NOT, T_BAD,
264 'B', T_WORD_NOT, T_BAD,
265 'd',
266 'd', T_DIGIT, T_DIGIT,
267 'd', T_DIGIT, T_DIGIT,
268 'd', T_DIGIT, T_DIGIT,
269 'd', T_DIGIT, T_DIGIT,
270 'd', T_DIGIT, T_DIGIT,
271 'D',
272 'D', T_DIGIT_NOT, T_DIGIT_NOT,
273 'D', T_DIGIT_NOT, T_DIGIT_NOT,
274 'D', T_DIGIT_NOT, T_DIGIT_NOT,
275 'D', T_DIGIT_NOT, T_DIGIT_NOT,
276 'D', T_DIGIT_NOT, T_DIGIT_NOT,
277 's',
278 's', T_SPACE, T_SPACE,
279 's', T_SPACE, T_SPACE,
280 's', T_SPACE, T_SPACE,
281 's', T_SPACE, T_SPACE,
282 's', T_SPACE, T_SPACE,
283 'S',
284 'S', T_SPACE_NOT, T_SPACE_NOT,
285 'S', T_SPACE_NOT, T_SPACE_NOT,
286 'S', T_SPACE_NOT, T_SPACE_NOT,
287 'S', T_SPACE_NOT, T_SPACE_NOT,
288 'S', T_SPACE_NOT, T_SPACE_NOT,
289 'w',
290 'w', T_ALNUM, T_ALNUM,
291 'w', T_ALNUM, T_ALNUM,
292 'w', T_ALNUM, T_ALNUM,
293 'w', T_ALNUM, T_ALNUM,
294 'w', T_ALNUM, T_ALNUM,
295 'W',
296 'W', T_ALNUM_NOT, T_ALNUM_NOT,
297 'W', T_ALNUM_NOT, T_ALNUM_NOT,
298 'W', T_ALNUM_NOT, T_ALNUM_NOT,
299 'W', T_ALNUM_NOT, T_ALNUM_NOT,
300 'W', T_ALNUM_NOT, T_ALNUM_NOT,
301 'z',
302 'z', T_FIN_STR, T_BAD,
303 'z', T_FIN_STR, T_BAD,
304 'z', T_FIN_STR, T_BAD,
305 'z', T_FIN_STR, T_BAD,
306 'z', T_FIN_STR, T_BAD,
307 'Z',
308 'Z', T_END_STR, T_BAD,
309 'Z', T_END_STR, T_BAD,
310 'Z', T_END_STR, T_BAD,
311 'Z', T_END_STR, T_BAD,
312 'Z', T_END_STR, T_BAD,
313
314 /* C escapes */
315
316 'a',
317 'a', CC_bel, CC_bel,
318 'a', CC_bel, CC_bel,
319 'a', CC_bel, CC_bel,
320 'a', CC_bel, CC_bel,
321 'a', CC_bel, CC_bel,
322 'c',
323 'c', T_ESCAPE, T_ESCAPE,
324 'c', T_ESCAPE, T_ESCAPE,
325 'c', T_ESCAPE, T_ESCAPE,
326 'c', T_ESCAPE, T_ESCAPE,
327 'c', T_ESCAPE, T_ESCAPE,
328 'C',
329 'C', T_ESCAPE, T_ESCAPE,
330 'C', T_ESCAPE, T_ESCAPE,
331 'C', T_ESCAPE, T_ESCAPE,
332 'C', T_ESCAPE, T_ESCAPE,
333 'C', T_ESCAPE, T_ESCAPE,
334 'e',
335 'e', CC_esc, CC_esc,
336 'e', CC_esc, CC_esc,
337 'e', CC_esc, CC_esc,
338 'e', CC_esc, CC_esc,
339 'e', CC_esc, CC_esc,
340 'E',
341 'E', CC_esc, CC_esc,
342 'E', CC_esc, CC_esc,
343 'E', CC_esc, CC_esc,
344 'E', CC_esc, CC_esc,
345 'E', CC_esc, CC_esc,
346 'f',
347 'f', '\f', '\f',
348 'f', '\f', '\f',
349 'f', '\f', '\f',
350 'f', '\f', '\f',
351 'f', '\f', '\f',
352 'n',
353 'n', '\n', '\n',
354 'n', '\n', '\n',
355 'n', '\n', '\n',
356 'n', '\n', '\n',
357 'n', '\n', '\n',
358 'r',
359 'r', '\r', '\r',
360 'r', '\r', '\r',
361 'r', '\r', '\r',
362 'r', '\r', '\r',
363 'r', '\r', '\r',
364 't',
365 't', '\t', '\t',
366 't', '\t', '\t',
367 't', '\t', '\t',
368 't', '\t', '\t',
369 't', '\t', '\t',
370 'v',
371 'v', CC_vt, CC_vt,
372 'v', CC_vt, CC_vt,
373 'v', CC_vt, CC_vt,
374 'v', CC_vt, CC_vt,
375 'v', CC_vt, CC_vt,
376 'x',
377 'x', T_ESCAPE, T_ESCAPE,
378 'x', T_ESCAPE, T_ESCAPE,
379 'x', T_ESCAPE, T_ESCAPE,
380 'x', T_ESCAPE, T_ESCAPE,
381 'x', T_ESCAPE, T_ESCAPE,
382 };
383
384 /*
385 * all allocation/free done here
386 * interface compatible with vmresize()
387 *
388 * malloc(n) alloc(0,n)
389 * realloc(p,n) alloc(p,n)
390 * free(p) alloc(p,0)
391 */
392
393 void*
alloc(register regdisc_t * disc,void * p,size_t n)394 alloc(register regdisc_t* disc, void* p, size_t n)
395 {
396 if (disc->re_resizef)
397 {
398 if (!n && (disc->re_flags & REG_NOFREE))
399 return 0;
400 return (*disc->re_resizef)(disc->re_resizehandle, p, n);
401 }
402 else if (!n)
403 {
404 if (!(disc->re_flags & REG_NOFREE))
405 free(p);
406 return 0;
407 }
408 else if (p)
409 return realloc(p, n);
410 else
411 return malloc(n);
412 }
413