1 | /* Copyright (C) 1998-2017 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 <error.h> |
24 | #include <limits.h> |
25 | #include <obstack.h> |
26 | #include <search.h> |
27 | #include <stdlib.h> |
28 | #include <string.h> |
29 | #include <unistd.h> |
30 | #include <stdint.h> |
31 | |
32 | #include "localedef.h" |
33 | #include "linereader.h" |
34 | #include "charmap.h" |
35 | #include "repertoire.h" |
36 | #include "simple-hash.h" |
37 | |
38 | |
39 | /* Simple keyword hashing for the repertoiremap. */ |
40 | static const struct keyword_t *repertoiremap_hash (const char *str, |
41 | unsigned int len); |
42 | static void repertoire_new_char (struct linereader *lr, hash_table *ht, |
43 | hash_table *rt, struct obstack *ob, |
44 | uint32_t value, const char *from, |
45 | const char *to, int decimal_ellipsis); |
46 | static int repertoire_compare (const void *p1, const void *p2); |
47 | |
48 | /* Already known repertoire maps. */ |
49 | static void *known; |
50 | |
51 | /* List of repertoire maps which are not available and which have been |
52 | reported to not be. */ |
53 | static void *unavailable; |
54 | |
55 | |
56 | struct repertoire_t * |
57 | repertoire_read (const char *filename) |
58 | { |
59 | struct linereader *repfile; |
60 | struct repertoire_t *result; |
61 | struct repertoire_t **resultp; |
62 | struct repertoire_t search; |
63 | int state; |
64 | char *from_name = NULL; |
65 | char *to_name = NULL; |
66 | enum token_t ellipsis = tok_none; |
67 | |
68 | search.name = filename; |
69 | resultp = tfind (&search, &known, &repertoire_compare); |
70 | if (resultp != NULL) |
71 | return *resultp; |
72 | |
73 | /* Determine path. */ |
74 | repfile = lr_open (filename, repertoiremap_hash); |
75 | if (repfile == NULL) |
76 | { |
77 | if (strchr (filename, '/') == NULL) |
78 | { |
79 | char *i18npath = getenv ("I18NPATH" ); |
80 | if (i18npath != NULL && *i18npath != '\0') |
81 | { |
82 | const size_t pathlen = strlen (i18npath); |
83 | char i18npathbuf[pathlen + 1]; |
84 | char path[strlen (filename) + 1 + pathlen |
85 | + sizeof ("/repertoiremaps/" ) - 1]; |
86 | char *next; |
87 | i18npath = memcpy (i18npathbuf, i18npath, pathlen + 1); |
88 | |
89 | while (repfile == NULL |
90 | && (next = strsep (&i18npath, ":" )) != NULL) |
91 | { |
92 | stpcpy (stpcpy (stpcpy (path, next), "/repertoiremaps/" ), |
93 | filename); |
94 | |
95 | repfile = lr_open (path, repertoiremap_hash); |
96 | |
97 | if (repfile == NULL) |
98 | { |
99 | stpcpy (stpcpy (stpcpy (path, next), "/" ), filename); |
100 | |
101 | repfile = lr_open (path, repertoiremap_hash); |
102 | } |
103 | } |
104 | } |
105 | |
106 | if (repfile == NULL) |
107 | { |
108 | /* Look in the systems charmap directory. */ |
109 | char *buf = xmalloc (strlen (filename) + 1 |
110 | + sizeof (REPERTOIREMAP_PATH)); |
111 | |
112 | stpcpy (stpcpy (stpcpy (buf, REPERTOIREMAP_PATH), "/" ), |
113 | filename); |
114 | repfile = lr_open (buf, repertoiremap_hash); |
115 | |
116 | free (buf); |
117 | } |
118 | } |
119 | |
120 | if (repfile == NULL) |
121 | return NULL; |
122 | } |
123 | |
124 | /* We don't want symbolic names in string to be translated. */ |
125 | repfile->translate_strings = 0; |
126 | |
127 | /* Allocate room for result. */ |
128 | result = (struct repertoire_t *) xmalloc (sizeof (struct repertoire_t)); |
129 | memset (result, '\0', sizeof (struct repertoire_t)); |
130 | |
131 | result->name = xstrdup (filename); |
132 | |
133 | #define obstack_chunk_alloc malloc |
134 | #define obstack_chunk_free free |
135 | obstack_init (&result->mem_pool); |
136 | |
137 | if (init_hash (&result->char_table, 256) |
138 | || init_hash (&result->reverse_table, 256) |
139 | || init_hash (&result->seq_table, 256)) |
140 | { |
141 | free (result); |
142 | return NULL; |
143 | } |
144 | |
145 | /* We use a state machine to describe the charmap description file |
146 | format. */ |
147 | state = 1; |
148 | while (1) |
149 | { |
150 | /* What's on? */ |
151 | struct token *now = lr_token (repfile, NULL, NULL, NULL, verbose); |
152 | enum token_t nowtok = now->tok; |
153 | struct token *arg; |
154 | |
155 | if (nowtok == tok_eof) |
156 | break; |
157 | |
158 | switch (state) |
159 | { |
160 | case 1: |
161 | /* We haven't yet read any character definition. This is where |
162 | we accept escape_char and comment_char definitions. */ |
163 | if (nowtok == tok_eol) |
164 | /* Ignore empty lines. */ |
165 | continue; |
166 | |
167 | if (nowtok == tok_escape_char || nowtok == tok_comment_char) |
168 | { |
169 | /* We know that we need an argument. */ |
170 | arg = lr_token (repfile, NULL, NULL, NULL, verbose); |
171 | |
172 | if (arg->tok != tok_ident) |
173 | { |
174 | lr_error (repfile, _("syntax error in prolog: %s" ), |
175 | _("bad argument" )); |
176 | |
177 | lr_ignore_rest (repfile, 0); |
178 | continue; |
179 | } |
180 | |
181 | if (arg->val.str.lenmb != 1) |
182 | { |
183 | lr_error (repfile, _("\ |
184 | argument to <%s> must be a single character" ), |
185 | nowtok == tok_escape_char ? "escape_char" |
186 | : "comment_char" ); |
187 | |
188 | lr_ignore_rest (repfile, 0); |
189 | continue; |
190 | } |
191 | |
192 | if (nowtok == tok_escape_char) |
193 | repfile->escape_char = *arg->val.str.startmb; |
194 | else |
195 | repfile->comment_char = *arg->val.str.startmb; |
196 | |
197 | lr_ignore_rest (repfile, 1); |
198 | continue; |
199 | } |
200 | |
201 | if (nowtok == tok_charids) |
202 | { |
203 | lr_ignore_rest (repfile, 1); |
204 | |
205 | state = 2; |
206 | continue; |
207 | } |
208 | |
209 | /* Otherwise we start reading the character definitions. */ |
210 | state = 2; |
211 | /* FALLTHROUGH */ |
212 | |
213 | case 2: |
214 | /* We are now are in the body. Each line |
215 | must have the format "%s %s %s\n" or "%s...%s %s %s\n". */ |
216 | if (nowtok == tok_eol) |
217 | /* Ignore empty lines. */ |
218 | continue; |
219 | |
220 | if (nowtok == tok_end) |
221 | { |
222 | state = 90; |
223 | continue; |
224 | } |
225 | |
226 | if (nowtok != tok_bsymbol) |
227 | { |
228 | lr_error (repfile, |
229 | _("syntax error in repertoire map definition: %s" ), |
230 | _("no symbolic name given" )); |
231 | |
232 | lr_ignore_rest (repfile, 0); |
233 | continue; |
234 | } |
235 | |
236 | /* If the previous line was not completely correct free the |
237 | used memory. */ |
238 | if (from_name != NULL) |
239 | obstack_free (&result->mem_pool, from_name); |
240 | |
241 | from_name = (char *) obstack_copy0 (&result->mem_pool, |
242 | now->val.str.startmb, |
243 | now->val.str.lenmb); |
244 | to_name = NULL; |
245 | |
246 | state = 3; |
247 | continue; |
248 | |
249 | case 3: |
250 | /* We have two possibilities: We can see an ellipsis or an |
251 | encoding value. */ |
252 | if (nowtok == tok_ellipsis3 || nowtok == tok_ellipsis4 |
253 | || nowtok == tok_ellipsis2) |
254 | { |
255 | ellipsis = nowtok; |
256 | state = 4; |
257 | continue; |
258 | } |
259 | /* FALLTHROUGH */ |
260 | |
261 | case 5: |
262 | /* We expect a value of the form <Uxxxx> or <Uxxxxxxxx> where |
263 | the xxx mean a hexadecimal value. */ |
264 | state = 2; |
265 | |
266 | errno = 0; |
267 | if (nowtok != tok_ucs4) |
268 | { |
269 | lr_error (repfile, |
270 | _("syntax error in repertoire map definition: %s" ), |
271 | _("no <Uxxxx> or <Uxxxxxxxx> value given" )); |
272 | |
273 | lr_ignore_rest (repfile, 0); |
274 | continue; |
275 | } |
276 | |
277 | /* We've found a new valid definition. */ |
278 | repertoire_new_char (repfile, &result->char_table, |
279 | &result->reverse_table, &result->mem_pool, |
280 | now->val.ucs4, from_name, to_name, |
281 | ellipsis != tok_ellipsis2); |
282 | |
283 | /* Ignore the rest of the line. */ |
284 | lr_ignore_rest (repfile, 0); |
285 | |
286 | from_name = NULL; |
287 | to_name = NULL; |
288 | |
289 | continue; |
290 | |
291 | case 4: |
292 | if (nowtok != tok_bsymbol) |
293 | { |
294 | lr_error (repfile, |
295 | _("syntax error in repertoire map definition: %s" ), |
296 | _("no symbolic name given for end of range" )); |
297 | |
298 | lr_ignore_rest (repfile, 0); |
299 | state = 2; |
300 | continue; |
301 | } |
302 | |
303 | /* Copy the to-name in a safe place. */ |
304 | to_name = (char *) obstack_copy0 (&result->mem_pool, |
305 | repfile->token.val.str.startmb, |
306 | repfile->token.val.str.lenmb); |
307 | |
308 | state = 5; |
309 | continue; |
310 | |
311 | case 90: |
312 | if (nowtok != tok_charids) |
313 | lr_error (repfile, _("\ |
314 | %1$s: definition does not end with `END %1$s'" ), "CHARIDS" ); |
315 | |
316 | lr_ignore_rest (repfile, nowtok == tok_charids); |
317 | break; |
318 | } |
319 | |
320 | break; |
321 | } |
322 | |
323 | if (state != 2 && state != 90 && !be_quiet) |
324 | WITH_CUR_LOCALE (error (0, 0, _("%s: premature end of file" ), |
325 | repfile->fname)); |
326 | |
327 | lr_close (repfile); |
328 | |
329 | if (tsearch (result, &known, &repertoire_compare) == NULL) |
330 | /* Something went wrong. */ |
331 | WITH_CUR_LOCALE (error (0, errno, _("cannot save new repertoire map" ))); |
332 | |
333 | return result; |
334 | } |
335 | |
336 | |
337 | void |
338 | repertoire_complain (const char *name) |
339 | { |
340 | if (tfind (name, &unavailable, (__compar_fn_t) strcmp) == NULL) |
341 | { |
342 | WITH_CUR_LOCALE (error (0, errno, _("\ |
343 | repertoire map file `%s' not found" ), name)); |
344 | |
345 | /* Remember that we reported this map. */ |
346 | tsearch (name, &unavailable, (__compar_fn_t) strcmp); |
347 | } |
348 | } |
349 | |
350 | |
351 | static int |
352 | repertoire_compare (const void *p1, const void *p2) |
353 | { |
354 | struct repertoire_t *r1 = (struct repertoire_t *) p1; |
355 | struct repertoire_t *r2 = (struct repertoire_t *) p2; |
356 | |
357 | return strcmp (r1->name, r2->name); |
358 | } |
359 | |
360 | |
361 | static const struct keyword_t * |
362 | repertoiremap_hash (const char *str, unsigned int len) |
363 | { |
364 | static const struct keyword_t wordlist[] = |
365 | { |
366 | {"escape_char" , tok_escape_char, 0}, |
367 | {"comment_char" , tok_comment_char, 0}, |
368 | {"CHARIDS" , tok_charids, 0}, |
369 | {"END" , tok_end, 0}, |
370 | }; |
371 | |
372 | if (len == 11 && memcmp (wordlist[0].name, str, 11) == 0) |
373 | return &wordlist[0]; |
374 | if (len == 12 && memcmp (wordlist[1].name, str, 12) == 0) |
375 | return &wordlist[1]; |
376 | if (len == 7 && memcmp (wordlist[2].name, str, 7) == 0) |
377 | return &wordlist[2]; |
378 | if (len == 3 && memcmp (wordlist[3].name, str, 3) == 0) |
379 | return &wordlist[3]; |
380 | |
381 | return NULL; |
382 | } |
383 | |
384 | |
385 | static void |
386 | repertoire_new_char (struct linereader *lr, hash_table *ht, hash_table *rt, |
387 | struct obstack *ob, uint32_t value, const char *from, |
388 | const char *to, int decimal_ellipsis) |
389 | { |
390 | char *from_end; |
391 | char *to_end; |
392 | const char *cp; |
393 | char *buf = NULL; |
394 | int prefix_len, len1, len2; |
395 | unsigned long int from_nr, to_nr, cnt; |
396 | |
397 | if (to == NULL) |
398 | { |
399 | insert_entry (ht, from, strlen (from), |
400 | (void *) (unsigned long int) value); |
401 | /* Please note that it isn't a bug if a symbol is defined more |
402 | than once. All later definitions are simply discarded. */ |
403 | |
404 | insert_entry (rt, obstack_copy (ob, &value, sizeof (value)), |
405 | sizeof (value), (void *) from); |
406 | |
407 | return; |
408 | } |
409 | |
410 | /* We have a range: the names must have names with equal prefixes |
411 | and an equal number of digits, where the second number is greater |
412 | or equal than the first. */ |
413 | len1 = strlen (from); |
414 | len2 = strlen (to); |
415 | |
416 | if (len1 != len2) |
417 | { |
418 | invalid_range: |
419 | lr_error (lr, _("invalid names for character range" )); |
420 | return; |
421 | } |
422 | |
423 | cp = &from[len1 - 1]; |
424 | if (decimal_ellipsis) |
425 | while (isdigit (*cp) && cp >= from) |
426 | --cp; |
427 | else |
428 | while (isxdigit (*cp) && cp >= from) |
429 | { |
430 | if (!isdigit (*cp) && !isupper (*cp)) |
431 | lr_error (lr, _("\ |
432 | hexadecimal range format should use only capital characters" )); |
433 | --cp; |
434 | } |
435 | |
436 | prefix_len = (cp - from) + 1; |
437 | |
438 | if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0) |
439 | goto invalid_range; |
440 | |
441 | errno = 0; |
442 | from_nr = strtoul (&from[prefix_len], &from_end, decimal_ellipsis ? 10 : 16); |
443 | if (*from_end != '\0' || (from_nr == ULONG_MAX && errno == ERANGE) |
444 | || ((to_nr = strtoul (&to[prefix_len], &to_end, |
445 | decimal_ellipsis ? 10 : 16)) == ULONG_MAX |
446 | && errno == ERANGE) |
447 | || *to_end != '\0') |
448 | { |
449 | lr_error (lr, _("<%s> and <%s> are invalid names for range" ), |
450 | from, to); |
451 | return; |
452 | } |
453 | |
454 | if (from_nr > to_nr) |
455 | { |
456 | lr_error (lr, _("upper limit in range is smaller than lower limit" )); |
457 | return; |
458 | } |
459 | |
460 | for (cnt = from_nr; cnt <= to_nr; ++cnt) |
461 | { |
462 | uint32_t this_value = value + (cnt - from_nr); |
463 | |
464 | obstack_printf (ob, decimal_ellipsis ? "%.*s%0*ld" : "%.*s%0*lX" , |
465 | prefix_len, from, len1 - prefix_len, cnt); |
466 | obstack_1grow (ob, '\0'); |
467 | |
468 | insert_entry (ht, buf, len1, |
469 | (void *) (unsigned long int) this_value); |
470 | /* Please note we don't examine the return value since it is no error |
471 | if we have two definitions for a symbol. */ |
472 | |
473 | insert_entry (rt, obstack_copy (ob, &this_value, sizeof (this_value)), |
474 | sizeof (this_value), (void *) from); |
475 | } |
476 | } |
477 | |
478 | |
479 | uint32_t |
480 | repertoire_find_value (const struct repertoire_t *rep, const char *name, |
481 | size_t len) |
482 | { |
483 | void *result; |
484 | |
485 | if (rep == NULL) |
486 | return ILLEGAL_CHAR_VALUE; |
487 | |
488 | if (find_entry ((hash_table *) &rep->char_table, name, len, &result) < 0) |
489 | return ILLEGAL_CHAR_VALUE; |
490 | |
491 | return (uint32_t) ((unsigned long int) result); |
492 | } |
493 | |
494 | |
495 | const char * |
496 | repertoire_find_symbol (const struct repertoire_t *rep, uint32_t ucs) |
497 | { |
498 | void *result; |
499 | |
500 | if (rep == NULL) |
501 | return NULL; |
502 | |
503 | if (find_entry ((hash_table *) &rep->reverse_table, &ucs, sizeof (ucs), |
504 | &result) < 0) |
505 | return NULL; |
506 | |
507 | return (const char *) result; |
508 | } |
509 | |
510 | |
511 | struct charseq * |
512 | repertoire_find_seq (const struct repertoire_t *rep, uint32_t ucs) |
513 | { |
514 | void *result; |
515 | |
516 | if (rep == NULL) |
517 | return NULL; |
518 | |
519 | if (find_entry ((hash_table *) &rep->seq_table, &ucs, sizeof (ucs), |
520 | &result) < 0) |
521 | return NULL; |
522 | |
523 | return (struct charseq *) result; |
524 | } |
525 | |