1 | /* Copyright (C) 1998-2019 Free Software Foundation, Inc. |
2 | This file is part of the GNU C Library. |
3 | Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998. |
4 | |
5 | This program is free software; you can redistribute it and/or modify |
6 | it under the terms of the GNU General Public License as published |
7 | by the Free Software Foundation; version 2 of the License, or |
8 | (at your option) any later version. |
9 | |
10 | This program is distributed in the hope that it will be useful, |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | GNU General Public License for more details. |
14 | |
15 | You should have received a copy of the GNU General Public License |
16 | along with this program; if not, see <http://www.gnu.org/licenses/>. */ |
17 | |
18 | #ifdef HAVE_CONFIG_H |
19 | # include <config.h> |
20 | #endif |
21 | |
22 | #include <errno.h> |
23 | #include <limits.h> |
24 | #include <obstack.h> |
25 | #include <search.h> |
26 | #include <stdlib.h> |
27 | #include <string.h> |
28 | #include <unistd.h> |
29 | #include <stdint.h> |
30 | |
31 | #include "localedef.h" |
32 | #include "linereader.h" |
33 | #include "charmap.h" |
34 | #include "repertoire.h" |
35 | #include "simple-hash.h" |
36 | |
37 | |
38 | /* Simple keyword hashing for the repertoiremap. */ |
39 | static const struct keyword_t *repertoiremap_hash (const char *str, |
40 | unsigned int len); |
41 | static void repertoire_new_char (struct linereader *lr, hash_table *ht, |
42 | hash_table *rt, struct obstack *ob, |
43 | uint32_t value, const char *from, |
44 | const char *to, int decimal_ellipsis); |
45 | static int repertoire_compare (const void *p1, const void *p2); |
46 | |
47 | /* Already known repertoire maps. */ |
48 | static void *known; |
49 | |
50 | /* List of repertoire maps which are not available and which have been |
51 | reported to not be. */ |
52 | static void *unavailable; |
53 | |
54 | |
55 | struct repertoire_t * |
56 | repertoire_read (const char *filename) |
57 | { |
58 | struct linereader *repfile; |
59 | struct repertoire_t *result; |
60 | struct repertoire_t **resultp; |
61 | struct repertoire_t search; |
62 | int state; |
63 | char *from_name = NULL; |
64 | char *to_name = NULL; |
65 | enum token_t ellipsis = tok_none; |
66 | |
67 | search.name = filename; |
68 | resultp = tfind (&search, &known, &repertoire_compare); |
69 | if (resultp != NULL) |
70 | return *resultp; |
71 | |
72 | /* Determine path. */ |
73 | repfile = lr_open (filename, repertoiremap_hash); |
74 | if (repfile == NULL) |
75 | { |
76 | if (strchr (filename, '/') == NULL) |
77 | { |
78 | char *i18npath = getenv ("I18NPATH" ); |
79 | if (i18npath != NULL && *i18npath != '\0') |
80 | { |
81 | const size_t pathlen = strlen (i18npath); |
82 | char i18npathbuf[pathlen + 1]; |
83 | char path[strlen (filename) + 1 + pathlen |
84 | + sizeof ("/repertoiremaps/" ) - 1]; |
85 | char *next; |
86 | i18npath = memcpy (i18npathbuf, i18npath, pathlen + 1); |
87 | |
88 | while (repfile == NULL |
89 | && (next = strsep (&i18npath, ":" )) != NULL) |
90 | { |
91 | stpcpy (stpcpy (stpcpy (path, next), "/repertoiremaps/" ), |
92 | filename); |
93 | |
94 | repfile = lr_open (path, repertoiremap_hash); |
95 | |
96 | if (repfile == NULL) |
97 | { |
98 | stpcpy (stpcpy (stpcpy (path, next), "/" ), filename); |
99 | |
100 | repfile = lr_open (path, repertoiremap_hash); |
101 | } |
102 | } |
103 | } |
104 | |
105 | if (repfile == NULL) |
106 | { |
107 | /* Look in the systems charmap directory. */ |
108 | char *buf = xmalloc (strlen (filename) + 1 |
109 | + sizeof (REPERTOIREMAP_PATH)); |
110 | |
111 | stpcpy (stpcpy (stpcpy (buf, REPERTOIREMAP_PATH), "/" ), |
112 | filename); |
113 | repfile = lr_open (buf, repertoiremap_hash); |
114 | |
115 | free (buf); |
116 | } |
117 | } |
118 | |
119 | if (repfile == NULL) |
120 | return NULL; |
121 | } |
122 | |
123 | /* We don't want symbolic names in string to be translated. */ |
124 | repfile->translate_strings = 0; |
125 | |
126 | /* Allocate room for result. */ |
127 | result = (struct repertoire_t *) xmalloc (sizeof (struct repertoire_t)); |
128 | memset (result, '\0', sizeof (struct repertoire_t)); |
129 | |
130 | result->name = xstrdup (filename); |
131 | |
132 | #define obstack_chunk_alloc malloc |
133 | #define obstack_chunk_free free |
134 | obstack_init (&result->mem_pool); |
135 | |
136 | if (init_hash (&result->char_table, 256) |
137 | || init_hash (&result->reverse_table, 256) |
138 | || init_hash (&result->seq_table, 256)) |
139 | { |
140 | free (result); |
141 | return NULL; |
142 | } |
143 | |
144 | /* We use a state machine to describe the charmap description file |
145 | format. */ |
146 | state = 1; |
147 | while (1) |
148 | { |
149 | /* What's on? */ |
150 | struct token *now = lr_token (repfile, NULL, NULL, NULL, verbose); |
151 | enum token_t nowtok = now->tok; |
152 | struct token *arg; |
153 | |
154 | if (nowtok == tok_eof) |
155 | break; |
156 | |
157 | switch (state) |
158 | { |
159 | case 1: |
160 | /* We haven't yet read any character definition. This is where |
161 | we accept escape_char and comment_char definitions. */ |
162 | if (nowtok == tok_eol) |
163 | /* Ignore empty lines. */ |
164 | continue; |
165 | |
166 | if (nowtok == tok_escape_char || nowtok == tok_comment_char) |
167 | { |
168 | /* We know that we need an argument. */ |
169 | arg = lr_token (repfile, NULL, NULL, NULL, verbose); |
170 | |
171 | if (arg->tok != tok_ident) |
172 | { |
173 | lr_error (repfile, _("syntax error in prolog: %s" ), |
174 | _("bad argument" )); |
175 | |
176 | lr_ignore_rest (repfile, 0); |
177 | continue; |
178 | } |
179 | |
180 | if (arg->val.str.lenmb != 1) |
181 | { |
182 | lr_error (repfile, _("\ |
183 | argument to <%s> must be a single character" ), |
184 | nowtok == tok_escape_char ? "escape_char" |
185 | : "comment_char" ); |
186 | |
187 | lr_ignore_rest (repfile, 0); |
188 | continue; |
189 | } |
190 | |
191 | if (nowtok == tok_escape_char) |
192 | repfile->escape_char = *arg->val.str.startmb; |
193 | else |
194 | repfile->comment_char = *arg->val.str.startmb; |
195 | |
196 | lr_ignore_rest (repfile, 1); |
197 | continue; |
198 | } |
199 | |
200 | if (nowtok == tok_charids) |
201 | { |
202 | lr_ignore_rest (repfile, 1); |
203 | |
204 | state = 2; |
205 | continue; |
206 | } |
207 | |
208 | /* Otherwise we start reading the character definitions. */ |
209 | state = 2; |
210 | /* FALLTHROUGH */ |
211 | |
212 | case 2: |
213 | /* We are now are in the body. Each line |
214 | must have the format "%s %s %s\n" or "%s...%s %s %s\n". */ |
215 | if (nowtok == tok_eol) |
216 | /* Ignore empty lines. */ |
217 | continue; |
218 | |
219 | if (nowtok == tok_end) |
220 | { |
221 | state = 90; |
222 | continue; |
223 | } |
224 | |
225 | if (nowtok != tok_bsymbol) |
226 | { |
227 | lr_error (repfile, |
228 | _("syntax error in repertoire map definition: %s" ), |
229 | _("no symbolic name given" )); |
230 | |
231 | lr_ignore_rest (repfile, 0); |
232 | continue; |
233 | } |
234 | |
235 | /* If the previous line was not completely correct free the |
236 | used memory. */ |
237 | if (from_name != NULL) |
238 | obstack_free (&result->mem_pool, from_name); |
239 | |
240 | from_name = (char *) obstack_copy0 (&result->mem_pool, |
241 | now->val.str.startmb, |
242 | now->val.str.lenmb); |
243 | to_name = NULL; |
244 | |
245 | state = 3; |
246 | continue; |
247 | |
248 | case 3: |
249 | /* We have two possibilities: We can see an ellipsis or an |
250 | encoding value. */ |
251 | if (nowtok == tok_ellipsis3 || nowtok == tok_ellipsis4 |
252 | || nowtok == tok_ellipsis2) |
253 | { |
254 | ellipsis = nowtok; |
255 | state = 4; |
256 | continue; |
257 | } |
258 | /* FALLTHROUGH */ |
259 | |
260 | case 5: |
261 | /* We expect a value of the form <Uxxxx> or <Uxxxxxxxx> where |
262 | the xxx mean a hexadecimal value. */ |
263 | state = 2; |
264 | |
265 | errno = 0; |
266 | if (nowtok != tok_ucs4) |
267 | { |
268 | lr_error (repfile, |
269 | _("syntax error in repertoire map definition: %s" ), |
270 | _("no <Uxxxx> or <Uxxxxxxxx> value given" )); |
271 | |
272 | lr_ignore_rest (repfile, 0); |
273 | continue; |
274 | } |
275 | |
276 | /* We've found a new valid definition. */ |
277 | repertoire_new_char (repfile, &result->char_table, |
278 | &result->reverse_table, &result->mem_pool, |
279 | now->val.ucs4, from_name, to_name, |
280 | ellipsis != tok_ellipsis2); |
281 | |
282 | /* Ignore the rest of the line. */ |
283 | lr_ignore_rest (repfile, 0); |
284 | |
285 | from_name = NULL; |
286 | to_name = NULL; |
287 | |
288 | continue; |
289 | |
290 | case 4: |
291 | if (nowtok != tok_bsymbol) |
292 | { |
293 | lr_error (repfile, |
294 | _("syntax error in repertoire map definition: %s" ), |
295 | _("no symbolic name given for end of range" )); |
296 | |
297 | lr_ignore_rest (repfile, 0); |
298 | state = 2; |
299 | continue; |
300 | } |
301 | |
302 | /* Copy the to-name in a safe place. */ |
303 | to_name = (char *) obstack_copy0 (&result->mem_pool, |
304 | repfile->token.val.str.startmb, |
305 | repfile->token.val.str.lenmb); |
306 | |
307 | state = 5; |
308 | continue; |
309 | |
310 | case 90: |
311 | if (nowtok != tok_charids) |
312 | lr_error (repfile, _("\ |
313 | %1$s: definition does not end with `END %1$s'" ), "CHARIDS" ); |
314 | |
315 | lr_ignore_rest (repfile, nowtok == tok_charids); |
316 | break; |
317 | } |
318 | |
319 | break; |
320 | } |
321 | |
322 | if (state != 2 && state != 90 && !be_quiet) |
323 | record_error (0, 0, _("%s: premature end of file" ), |
324 | repfile->fname); |
325 | |
326 | lr_close (repfile); |
327 | |
328 | if (tsearch (result, &known, &repertoire_compare) == NULL) |
329 | /* Something went wrong. */ |
330 | record_error (0, errno, _("cannot save new repertoire map" )); |
331 | |
332 | return result; |
333 | } |
334 | |
335 | |
336 | void |
337 | repertoire_complain (const char *name) |
338 | { |
339 | if (tfind (name, &unavailable, (__compar_fn_t) strcmp) == NULL) |
340 | { |
341 | record_error (0, errno, _("\ |
342 | repertoire map file `%s' not found" ), name); |
343 | |
344 | /* Remember that we reported this map. */ |
345 | tsearch (name, &unavailable, (__compar_fn_t) strcmp); |
346 | } |
347 | } |
348 | |
349 | |
350 | static int |
351 | repertoire_compare (const void *p1, const void *p2) |
352 | { |
353 | struct repertoire_t *r1 = (struct repertoire_t *) p1; |
354 | struct repertoire_t *r2 = (struct repertoire_t *) p2; |
355 | |
356 | return strcmp (r1->name, r2->name); |
357 | } |
358 | |
359 | |
360 | static const struct keyword_t * |
361 | repertoiremap_hash (const char *str, unsigned int len) |
362 | { |
363 | static const struct keyword_t wordlist[] = |
364 | { |
365 | {"escape_char" , tok_escape_char, 0}, |
366 | {"comment_char" , tok_comment_char, 0}, |
367 | {"CHARIDS" , tok_charids, 0}, |
368 | {"END" , tok_end, 0}, |
369 | }; |
370 | |
371 | if (len == 11 && memcmp (wordlist[0].name, str, 11) == 0) |
372 | return &wordlist[0]; |
373 | if (len == 12 && memcmp (wordlist[1].name, str, 12) == 0) |
374 | return &wordlist[1]; |
375 | if (len == 7 && memcmp (wordlist[2].name, str, 7) == 0) |
376 | return &wordlist[2]; |
377 | if (len == 3 && memcmp (wordlist[3].name, str, 3) == 0) |
378 | return &wordlist[3]; |
379 | |
380 | return NULL; |
381 | } |
382 | |
383 | |
384 | static void |
385 | repertoire_new_char (struct linereader *lr, hash_table *ht, hash_table *rt, |
386 | struct obstack *ob, uint32_t value, const char *from, |
387 | const char *to, int decimal_ellipsis) |
388 | { |
389 | char *from_end; |
390 | char *to_end; |
391 | const char *cp; |
392 | char *buf = NULL; |
393 | int prefix_len, len1, len2; |
394 | unsigned long int from_nr, to_nr, cnt; |
395 | |
396 | if (to == NULL) |
397 | { |
398 | insert_entry (ht, from, strlen (from), |
399 | (void *) (unsigned long int) value); |
400 | /* Please note that it isn't a bug if a symbol is defined more |
401 | than once. All later definitions are simply discarded. */ |
402 | |
403 | insert_entry (rt, obstack_copy (ob, &value, sizeof (value)), |
404 | sizeof (value), (void *) from); |
405 | |
406 | return; |
407 | } |
408 | |
409 | /* We have a range: the names must have names with equal prefixes |
410 | and an equal number of digits, where the second number is greater |
411 | or equal than the first. */ |
412 | len1 = strlen (from); |
413 | len2 = strlen (to); |
414 | |
415 | if (len1 != len2) |
416 | { |
417 | invalid_range: |
418 | lr_error (lr, _("invalid names for character range" )); |
419 | return; |
420 | } |
421 | |
422 | cp = &from[len1 - 1]; |
423 | if (decimal_ellipsis) |
424 | while (isdigit (*cp) && cp >= from) |
425 | --cp; |
426 | else |
427 | while (isxdigit (*cp) && cp >= from) |
428 | { |
429 | if (!isdigit (*cp) && !isupper (*cp)) |
430 | lr_error (lr, _("\ |
431 | hexadecimal range format should use only capital characters" )); |
432 | --cp; |
433 | } |
434 | |
435 | prefix_len = (cp - from) + 1; |
436 | |
437 | if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0) |
438 | goto invalid_range; |
439 | |
440 | errno = 0; |
441 | from_nr = strtoul (&from[prefix_len], &from_end, decimal_ellipsis ? 10 : 16); |
442 | if (*from_end != '\0' || (from_nr == ULONG_MAX && errno == ERANGE) |
443 | || ((to_nr = strtoul (&to[prefix_len], &to_end, |
444 | decimal_ellipsis ? 10 : 16)) == ULONG_MAX |
445 | && errno == ERANGE) |
446 | || *to_end != '\0') |
447 | { |
448 | lr_error (lr, _("<%s> and <%s> are invalid names for range" ), |
449 | from, to); |
450 | return; |
451 | } |
452 | |
453 | if (from_nr > to_nr) |
454 | { |
455 | lr_error (lr, _("upper limit in range is smaller than lower limit" )); |
456 | return; |
457 | } |
458 | |
459 | for (cnt = from_nr; cnt <= to_nr; ++cnt) |
460 | { |
461 | uint32_t this_value = value + (cnt - from_nr); |
462 | |
463 | obstack_printf (ob, decimal_ellipsis ? "%.*s%0*ld" : "%.*s%0*lX" , |
464 | prefix_len, from, len1 - prefix_len, cnt); |
465 | obstack_1grow (ob, '\0'); |
466 | |
467 | insert_entry (ht, buf, len1, |
468 | (void *) (unsigned long int) this_value); |
469 | /* Please note we don't examine the return value since it is no error |
470 | if we have two definitions for a symbol. */ |
471 | |
472 | insert_entry (rt, obstack_copy (ob, &this_value, sizeof (this_value)), |
473 | sizeof (this_value), (void *) from); |
474 | } |
475 | } |
476 | |
477 | |
478 | uint32_t |
479 | repertoire_find_value (const struct repertoire_t *rep, const char *name, |
480 | size_t len) |
481 | { |
482 | void *result; |
483 | |
484 | if (rep == NULL) |
485 | return ILLEGAL_CHAR_VALUE; |
486 | |
487 | if (find_entry ((hash_table *) &rep->char_table, name, len, &result) < 0) |
488 | return ILLEGAL_CHAR_VALUE; |
489 | |
490 | return (uint32_t) ((unsigned long int) result); |
491 | } |
492 | |
493 | |
494 | const char * |
495 | repertoire_find_symbol (const struct repertoire_t *rep, uint32_t ucs) |
496 | { |
497 | void *result; |
498 | |
499 | if (rep == NULL) |
500 | return NULL; |
501 | |
502 | if (find_entry ((hash_table *) &rep->reverse_table, &ucs, sizeof (ucs), |
503 | &result) < 0) |
504 | return NULL; |
505 | |
506 | return (const char *) result; |
507 | } |
508 | |
509 | |
510 | struct charseq * |
511 | repertoire_find_seq (const struct repertoire_t *rep, uint32_t ucs) |
512 | { |
513 | void *result; |
514 | |
515 | if (rep == NULL) |
516 | return NULL; |
517 | |
518 | if (find_entry ((hash_table *) &rep->seq_table, &ucs, sizeof (ucs), |
519 | &result) < 0) |
520 | return NULL; |
521 | |
522 | return (struct charseq *) result; |
523 | } |
524 | |