1 | /* Copyright (C) 1998-2023 Free Software Foundation, Inc. |
2 | This file is part of the GNU C Library. |
3 | |
4 | This program is free software; you can redistribute it and/or modify |
5 | it under the terms of the GNU General Public License as published |
6 | by the Free Software Foundation; version 2 of the License, or |
7 | (at your option) any later version. |
8 | |
9 | This program is distributed in the hope that it will be useful, |
10 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
11 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
12 | GNU General Public License for more details. |
13 | |
14 | You should have received a copy of the GNU General Public License |
15 | along with this program; if not, see <https://www.gnu.org/licenses/>. */ |
16 | |
17 | #ifdef HAVE_CONFIG_H |
18 | # include <config.h> |
19 | #endif |
20 | |
21 | #include <errno.h> |
22 | #include <limits.h> |
23 | #include <obstack.h> |
24 | #include <search.h> |
25 | #include <stdlib.h> |
26 | #include <string.h> |
27 | #include <unistd.h> |
28 | #include <stdint.h> |
29 | |
30 | #include "localedef.h" |
31 | #include "linereader.h" |
32 | #include "charmap.h" |
33 | #include "repertoire.h" |
34 | #include "simple-hash.h" |
35 | |
36 | |
37 | /* Simple keyword hashing for the repertoiremap. */ |
38 | static const struct keyword_t *repertoiremap_hash (const char *str, |
39 | size_t len); |
40 | static void repertoire_new_char (struct linereader *lr, hash_table *ht, |
41 | hash_table *rt, struct obstack *ob, |
42 | uint32_t value, const char *from, |
43 | const char *to, int decimal_ellipsis); |
44 | static int repertoire_compare (const void *p1, const void *p2); |
45 | |
46 | /* Already known repertoire maps. */ |
47 | static void *known; |
48 | |
49 | /* List of repertoire maps which are not available and which have been |
50 | reported to not be. */ |
51 | static void *unavailable; |
52 | |
53 | |
54 | struct repertoire_t * |
55 | repertoire_read (const char *filename) |
56 | { |
57 | struct linereader *repfile; |
58 | struct repertoire_t *result; |
59 | struct repertoire_t **resultp; |
60 | struct repertoire_t search; |
61 | int state; |
62 | char *from_name = NULL; |
63 | char *to_name = NULL; |
64 | enum token_t ellipsis = tok_none; |
65 | |
66 | search.name = filename; |
67 | resultp = tfind (&search, &known, &repertoire_compare); |
68 | if (resultp != NULL) |
69 | return *resultp; |
70 | |
71 | /* Determine path. */ |
72 | repfile = lr_open (filename, repertoiremap_hash); |
73 | if (repfile == NULL) |
74 | { |
75 | if (strchr (filename, '/') == NULL) |
76 | { |
77 | char *i18npath = getenv ("I18NPATH" ); |
78 | if (i18npath != NULL && *i18npath != '\0') |
79 | { |
80 | const size_t pathlen = strlen (i18npath); |
81 | char i18npathbuf[pathlen + 1]; |
82 | char path[strlen (filename) + 1 + pathlen |
83 | + sizeof ("/repertoiremaps/" ) - 1]; |
84 | char *next; |
85 | i18npath = memcpy (i18npathbuf, i18npath, pathlen + 1); |
86 | |
87 | while (repfile == NULL |
88 | && (next = strsep (&i18npath, ":" )) != NULL) |
89 | { |
90 | stpcpy (stpcpy (stpcpy (path, next), "/repertoiremaps/" ), |
91 | filename); |
92 | |
93 | repfile = lr_open (path, repertoiremap_hash); |
94 | |
95 | if (repfile == NULL) |
96 | { |
97 | stpcpy (stpcpy (stpcpy (path, next), "/" ), filename); |
98 | |
99 | repfile = lr_open (path, repertoiremap_hash); |
100 | } |
101 | } |
102 | } |
103 | |
104 | if (repfile == NULL) |
105 | { |
106 | /* Look in the systems charmap directory. */ |
107 | char *buf = xmalloc (strlen (filename) + 1 |
108 | + sizeof (REPERTOIREMAP_PATH)); |
109 | |
110 | stpcpy (stpcpy (stpcpy (buf, REPERTOIREMAP_PATH), "/" ), |
111 | filename); |
112 | repfile = lr_open (buf, repertoiremap_hash); |
113 | |
114 | free (buf); |
115 | } |
116 | } |
117 | |
118 | if (repfile == NULL) |
119 | return NULL; |
120 | } |
121 | |
122 | /* We don't want symbolic names in string to be translated. */ |
123 | repfile->translate_strings = 0; |
124 | |
125 | /* Allocate room for result. */ |
126 | result = (struct repertoire_t *) xmalloc (sizeof (struct repertoire_t)); |
127 | memset (result, '\0', sizeof (struct repertoire_t)); |
128 | |
129 | result->name = xstrdup (filename); |
130 | |
131 | #define obstack_chunk_alloc malloc |
132 | #define obstack_chunk_free free |
133 | obstack_init (&result->mem_pool); |
134 | |
135 | if (init_hash (&result->char_table, 256) |
136 | || init_hash (&result->reverse_table, 256) |
137 | || init_hash (&result->seq_table, 256)) |
138 | { |
139 | free (result); |
140 | return NULL; |
141 | } |
142 | |
143 | /* We use a state machine to describe the charmap description file |
144 | format. */ |
145 | state = 1; |
146 | while (1) |
147 | { |
148 | /* What's on? */ |
149 | struct token *now = lr_token (repfile, NULL, NULL, NULL, verbose); |
150 | enum token_t nowtok = now->tok; |
151 | struct token *arg; |
152 | |
153 | if (nowtok == tok_eof) |
154 | break; |
155 | |
156 | switch (state) |
157 | { |
158 | case 1: |
159 | /* We haven't yet read any character definition. This is where |
160 | we accept escape_char and comment_char definitions. */ |
161 | if (nowtok == tok_eol) |
162 | /* Ignore empty lines. */ |
163 | continue; |
164 | |
165 | if (nowtok == tok_escape_char || nowtok == tok_comment_char) |
166 | { |
167 | /* We know that we need an argument. */ |
168 | arg = lr_token (repfile, NULL, NULL, NULL, verbose); |
169 | |
170 | if (arg->tok != tok_ident) |
171 | { |
172 | lr_error (repfile, _("syntax error in prolog: %s" ), |
173 | _("bad argument" )); |
174 | |
175 | lr_ignore_rest (repfile, 0); |
176 | continue; |
177 | } |
178 | |
179 | if (arg->val.str.lenmb != 1) |
180 | { |
181 | lr_error (repfile, _("\ |
182 | argument to <%s> must be a single character" ), |
183 | nowtok == tok_escape_char ? "escape_char" |
184 | : "comment_char" ); |
185 | |
186 | lr_ignore_rest (repfile, 0); |
187 | continue; |
188 | } |
189 | |
190 | if (nowtok == tok_escape_char) |
191 | repfile->escape_char = *arg->val.str.startmb; |
192 | else |
193 | repfile->comment_char = *arg->val.str.startmb; |
194 | |
195 | lr_ignore_rest (repfile, 1); |
196 | continue; |
197 | } |
198 | |
199 | if (nowtok == tok_charids) |
200 | { |
201 | lr_ignore_rest (repfile, 1); |
202 | |
203 | state = 2; |
204 | continue; |
205 | } |
206 | |
207 | /* Otherwise we start reading the character definitions. */ |
208 | state = 2; |
209 | /* FALLTHROUGH */ |
210 | |
211 | case 2: |
212 | /* We are now are in the body. Each line |
213 | must have the format "%s %s %s\n" or "%s...%s %s %s\n". */ |
214 | if (nowtok == tok_eol) |
215 | /* Ignore empty lines. */ |
216 | continue; |
217 | |
218 | if (nowtok == tok_end) |
219 | { |
220 | state = 90; |
221 | continue; |
222 | } |
223 | |
224 | if (nowtok != tok_bsymbol) |
225 | { |
226 | lr_error (repfile, |
227 | _("syntax error in repertoire map definition: %s" ), |
228 | _("no symbolic name given" )); |
229 | |
230 | lr_ignore_rest (repfile, 0); |
231 | continue; |
232 | } |
233 | |
234 | /* If the previous line was not completely correct free the |
235 | used memory. */ |
236 | if (from_name != NULL) |
237 | obstack_free (&result->mem_pool, from_name); |
238 | |
239 | from_name = (char *) obstack_copy0 (&result->mem_pool, |
240 | now->val.str.startmb, |
241 | now->val.str.lenmb); |
242 | to_name = NULL; |
243 | |
244 | state = 3; |
245 | continue; |
246 | |
247 | case 3: |
248 | /* We have two possibilities: We can see an ellipsis or an |
249 | encoding value. */ |
250 | if (nowtok == tok_ellipsis3 || nowtok == tok_ellipsis4 |
251 | || nowtok == tok_ellipsis2) |
252 | { |
253 | ellipsis = nowtok; |
254 | state = 4; |
255 | continue; |
256 | } |
257 | /* FALLTHROUGH */ |
258 | |
259 | case 5: |
260 | /* We expect a value of the form <Uxxxx> or <Uxxxxxxxx> where |
261 | the xxx mean a hexadecimal value. */ |
262 | state = 2; |
263 | |
264 | errno = 0; |
265 | if (nowtok != tok_ucs4) |
266 | { |
267 | lr_error (repfile, |
268 | _("syntax error in repertoire map definition: %s" ), |
269 | _("no <Uxxxx> or <Uxxxxxxxx> value given" )); |
270 | |
271 | lr_ignore_rest (repfile, 0); |
272 | continue; |
273 | } |
274 | |
275 | /* We've found a new valid definition. */ |
276 | repertoire_new_char (repfile, &result->char_table, |
277 | &result->reverse_table, &result->mem_pool, |
278 | now->val.ucs4, from_name, to_name, |
279 | ellipsis != tok_ellipsis2); |
280 | |
281 | /* Ignore the rest of the line. */ |
282 | lr_ignore_rest (repfile, 0); |
283 | |
284 | from_name = NULL; |
285 | to_name = NULL; |
286 | |
287 | continue; |
288 | |
289 | case 4: |
290 | if (nowtok != tok_bsymbol) |
291 | { |
292 | lr_error (repfile, |
293 | _("syntax error in repertoire map definition: %s" ), |
294 | _("no symbolic name given for end of range" )); |
295 | |
296 | lr_ignore_rest (repfile, 0); |
297 | state = 2; |
298 | continue; |
299 | } |
300 | |
301 | /* Copy the to-name in a safe place. */ |
302 | to_name = (char *) obstack_copy0 (&result->mem_pool, |
303 | repfile->token.val.str.startmb, |
304 | repfile->token.val.str.lenmb); |
305 | |
306 | state = 5; |
307 | continue; |
308 | |
309 | case 90: |
310 | if (nowtok != tok_charids) |
311 | lr_error (repfile, _("\ |
312 | %1$s: definition does not end with `END %1$s'" ), "CHARIDS" ); |
313 | |
314 | lr_ignore_rest (repfile, nowtok == tok_charids); |
315 | break; |
316 | } |
317 | |
318 | break; |
319 | } |
320 | |
321 | if (state != 2 && state != 90 && !be_quiet) |
322 | record_error (0, 0, _("%s: premature end of file" ), |
323 | repfile->fname); |
324 | |
325 | lr_close (repfile); |
326 | |
327 | if (tsearch (result, &known, &repertoire_compare) == NULL) |
328 | /* Something went wrong. */ |
329 | record_error (0, errno, _("cannot save new repertoire map" )); |
330 | |
331 | return result; |
332 | } |
333 | |
334 | |
335 | void |
336 | repertoire_complain (const char *name) |
337 | { |
338 | if (tfind (name, &unavailable, (__compar_fn_t) strcmp) == NULL) |
339 | { |
340 | record_error (0, errno, _("\ |
341 | repertoire map file `%s' not found" ), name); |
342 | |
343 | /* Remember that we reported this map. */ |
344 | tsearch (name, &unavailable, (__compar_fn_t) strcmp); |
345 | } |
346 | } |
347 | |
348 | |
349 | static int |
350 | repertoire_compare (const void *p1, const void *p2) |
351 | { |
352 | struct repertoire_t *r1 = (struct repertoire_t *) p1; |
353 | struct repertoire_t *r2 = (struct repertoire_t *) p2; |
354 | |
355 | return strcmp (r1->name, r2->name); |
356 | } |
357 | |
358 | |
359 | static const struct keyword_t * |
360 | repertoiremap_hash (const char *str, size_t len) |
361 | { |
362 | static const struct keyword_t wordlist[] = |
363 | { |
364 | {"escape_char" , tok_escape_char, 0}, |
365 | {"comment_char" , tok_comment_char, 0}, |
366 | {"CHARIDS" , tok_charids, 0}, |
367 | {"END" , tok_end, 0}, |
368 | }; |
369 | |
370 | if (len == 11 && memcmp (wordlist[0].name, str, 11) == 0) |
371 | return &wordlist[0]; |
372 | if (len == 12 && memcmp (wordlist[1].name, str, 12) == 0) |
373 | return &wordlist[1]; |
374 | if (len == 7 && memcmp (wordlist[2].name, str, 7) == 0) |
375 | return &wordlist[2]; |
376 | if (len == 3 && memcmp (wordlist[3].name, str, 3) == 0) |
377 | return &wordlist[3]; |
378 | |
379 | return NULL; |
380 | } |
381 | |
382 | |
383 | static void |
384 | repertoire_new_char (struct linereader *lr, hash_table *ht, hash_table *rt, |
385 | struct obstack *ob, uint32_t value, const char *from, |
386 | const char *to, int decimal_ellipsis) |
387 | { |
388 | char *from_end; |
389 | char *to_end; |
390 | const char *cp; |
391 | char *buf = NULL; |
392 | int prefix_len, len1, len2; |
393 | unsigned long int from_nr, to_nr, cnt; |
394 | |
395 | if (to == NULL) |
396 | { |
397 | insert_entry (ht, from, strlen (from), |
398 | (void *) (unsigned long int) value); |
399 | /* Please note that it isn't a bug if a symbol is defined more |
400 | than once. All later definitions are simply discarded. */ |
401 | |
402 | insert_entry (rt, obstack_copy (ob, &value, sizeof (value)), |
403 | sizeof (value), (void *) from); |
404 | |
405 | return; |
406 | } |
407 | |
408 | /* We have a range: the names must have names with equal prefixes |
409 | and an equal number of digits, where the second number is greater |
410 | or equal than the first. */ |
411 | len1 = strlen (from); |
412 | len2 = strlen (to); |
413 | |
414 | if (len1 != len2) |
415 | { |
416 | invalid_range: |
417 | lr_error (lr, _("invalid names for character range" )); |
418 | return; |
419 | } |
420 | |
421 | cp = &from[len1 - 1]; |
422 | if (decimal_ellipsis) |
423 | while (isdigit (*cp) && cp >= from) |
424 | --cp; |
425 | else |
426 | while (isxdigit (*cp) && cp >= from) |
427 | { |
428 | if (!isdigit (*cp) && !isupper (*cp)) |
429 | lr_error (lr, _("\ |
430 | hexadecimal range format should use only capital characters" )); |
431 | --cp; |
432 | } |
433 | |
434 | prefix_len = (cp - from) + 1; |
435 | |
436 | if (cp == &from[len1 - 1] || strncmp (from, to, prefix_len) != 0) |
437 | goto invalid_range; |
438 | |
439 | errno = 0; |
440 | from_nr = strtoul (&from[prefix_len], &from_end, decimal_ellipsis ? 10 : 16); |
441 | if (*from_end != '\0' || (from_nr == ULONG_MAX && errno == ERANGE) |
442 | || ((to_nr = strtoul (&to[prefix_len], &to_end, |
443 | decimal_ellipsis ? 10 : 16)) == ULONG_MAX |
444 | && errno == ERANGE) |
445 | || *to_end != '\0') |
446 | { |
447 | lr_error (lr, _("<%s> and <%s> are invalid names for range" ), |
448 | from, to); |
449 | return; |
450 | } |
451 | |
452 | if (from_nr > to_nr) |
453 | { |
454 | lr_error (lr, _("upper limit in range is smaller than lower limit" )); |
455 | return; |
456 | } |
457 | |
458 | for (cnt = from_nr; cnt <= to_nr; ++cnt) |
459 | { |
460 | uint32_t this_value = value + (cnt - from_nr); |
461 | |
462 | obstack_printf (ob, decimal_ellipsis ? "%.*s%0*ld" : "%.*s%0*lX" , |
463 | prefix_len, from, len1 - prefix_len, cnt); |
464 | obstack_1grow (ob, '\0'); |
465 | |
466 | insert_entry (ht, buf, len1, |
467 | (void *) (unsigned long int) this_value); |
468 | /* Please note we don't examine the return value since it is no error |
469 | if we have two definitions for a symbol. */ |
470 | |
471 | insert_entry (rt, obstack_copy (ob, &this_value, sizeof (this_value)), |
472 | sizeof (this_value), (void *) from); |
473 | } |
474 | } |
475 | |
476 | |
477 | uint32_t |
478 | repertoire_find_value (const struct repertoire_t *rep, const char *name, |
479 | size_t len) |
480 | { |
481 | void *result; |
482 | |
483 | if (rep == NULL) |
484 | return ILLEGAL_CHAR_VALUE; |
485 | |
486 | if (find_entry ((hash_table *) &rep->char_table, name, len, &result) < 0) |
487 | return ILLEGAL_CHAR_VALUE; |
488 | |
489 | return (uint32_t) ((unsigned long int) result); |
490 | } |
491 | |
492 | |
493 | const char * |
494 | repertoire_find_symbol (const struct repertoire_t *rep, uint32_t ucs) |
495 | { |
496 | void *result; |
497 | |
498 | if (rep == NULL) |
499 | return NULL; |
500 | |
501 | if (find_entry ((hash_table *) &rep->reverse_table, &ucs, sizeof (ucs), |
502 | &result) < 0) |
503 | return NULL; |
504 | |
505 | return (const char *) result; |
506 | } |
507 | |
508 | |
509 | struct charseq * |
510 | repertoire_find_seq (const struct repertoire_t *rep, uint32_t ucs) |
511 | { |
512 | void *result; |
513 | |
514 | if (rep == NULL) |
515 | return NULL; |
516 | |
517 | if (find_entry ((hash_table *) &rep->seq_table, &ucs, sizeof (ucs), |
518 | &result) < 0) |
519 | return NULL; |
520 | |
521 | return (struct charseq *) result; |
522 | } |
523 | |