]>
Commit | Line | Data |
---|---|---|
19bc17a9 RM |
1 | /* Copyright (C) 1996 Free Software Foundation, Inc. |
2 | This file is part of the GNU C Library. | |
3 | Contributed by Ulrich Drepper, <drepper@gnu.ai.mit.edu>. | |
4 | ||
5 | The GNU C Library is free software; you can redistribute it and/or | |
6 | modify it under the terms of the GNU Library General Public License as | |
7 | published by the Free Software Foundation; either version 2 of the | |
8 | License, or (at your option) any later version. | |
9 | ||
10 | The GNU C Library 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 GNU | |
13 | Library General Public License for more details. | |
14 | ||
15 | You should have received a copy of the GNU Library General Public | |
16 | License along with the GNU C Library; see the file COPYING.LIB. If | |
17 | not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
18 | Boston, MA 02111-1307, USA. */ | |
19 | ||
20 | #ifdef HAVE_CONFIG_H | |
21 | # include <config.h> | |
22 | #endif | |
23 | ||
24 | #include <errno.h> | |
25 | #include <fcntl.h> | |
26 | #include <locale.h> | |
27 | #include <malloc.h> | |
75cd5204 | 28 | #include <stdio.h> |
19bc17a9 RM |
29 | #include <string.h> |
30 | #include <unistd.h> | |
31 | #include <sys/stat.h> | |
32 | #include <sys/uio.h> | |
33 | ||
34 | #include "locfile.h" | |
35 | #include "linereader.h" | |
36 | #include "localeinfo.h" | |
37 | #include "locales.h" | |
38 | ||
39 | ||
40 | /* Uncomment the following line in the production version. */ | |
41 | /* #define NDEBUG 1 */ | |
42 | #include <assert.h> | |
43 | ||
44 | /* Define the lookup function. */ | |
45 | #include "locfile-kw.h" | |
46 | ||
47 | ||
48 | /* Some useful macros. */ | |
49 | #define MIN(a, b) (__extension__ ({ typeof (a) _a = (a); \ | |
50 | typeof (b) _b = (b); \ | |
51 | _a < _b ? _a : _b; })) | |
52 | ||
53 | ||
54 | void *xmalloc (size_t __n); | |
55 | char *xstrdup (const char *__str); | |
56 | ||
57 | struct localedef_t * | |
58 | locfile_read (const char *filename, struct charset_t *charset) | |
59 | { | |
60 | struct linereader *ldfile; | |
61 | struct localedef_t *result; | |
62 | int state; | |
63 | enum token_t expected_tok = tok_none; | |
64 | const char *expected_str = NULL; | |
65 | enum token_t ctype_tok_sym = tok_none; | |
66 | const char *ctype_tok_str = NULL; | |
67 | int copy_category = 0; | |
68 | int cnt; | |
69 | ||
70 | /* Allocate space for result. */ | |
71 | result = (struct localedef_t *) xmalloc (sizeof (struct localedef_t)); | |
72 | memset (result, '\0', sizeof (struct localedef_t)); | |
73 | ||
74 | ldfile = lr_open (filename, locfile_hash); | |
75 | if (ldfile == NULL) | |
76 | { | |
77 | if (filename[0] != '/') | |
78 | { | |
79 | char path[strlen (filename) + 1 + sizeof (LOCSRCDIR)]; | |
80 | ||
81 | stpcpy (stpcpy (stpcpy (path, LOCSRCDIR), "/"), filename); | |
82 | ldfile = lr_open (path, locfile_hash); | |
83 | } | |
84 | ||
85 | if (ldfile == NULL) | |
86 | { | |
87 | result->failed = 1; | |
88 | return result; | |
89 | } | |
90 | } | |
91 | ||
92 | #define HANDLE_COPY(category, token, string) \ | |
93 | if (nowtok == tok_copy) \ | |
94 | { \ | |
95 | copy_category = category; \ | |
96 | expected_tok = token; \ | |
97 | expected_str = string; \ | |
98 | state = 8; \ | |
99 | continue; \ | |
100 | } \ | |
101 | ++state | |
102 | ||
103 | #define LOCALE_PROLOG(token, string) \ | |
104 | if (nowtok == tok_eol) \ | |
105 | /* Ignore empty lines. */ \ | |
106 | continue; \ | |
107 | if (nowtok == tok_end) \ | |
108 | { \ | |
109 | expected_tok = token; \ | |
110 | expected_str = string; \ | |
111 | state = 4; \ | |
112 | continue; \ | |
113 | } \ | |
114 | if (nowtok == tok_copy) \ | |
115 | goto only_copy; | |
116 | ||
117 | ||
118 | #define READ_STRING(fn, errlabel) \ | |
119 | do \ | |
120 | { \ | |
121 | arg = lr_token (ldfile, charset); \ | |
122 | if (arg->tok != tok_string) \ | |
123 | goto errlabel; \ | |
124 | fn (ldfile, result, nowtok, arg, charset); \ | |
125 | lr_ignore_rest (ldfile, 1); \ | |
126 | } \ | |
127 | while (0) | |
128 | ||
129 | #define READ_STRING_LIST(fn, errlabel) \ | |
130 | do \ | |
131 | { \ | |
132 | arg = lr_token (ldfile, charset); \ | |
133 | while (arg->tok == tok_string) \ | |
134 | { \ | |
135 | fn (ldfile, result, nowtok, arg, charset); \ | |
136 | arg = lr_token (ldfile, charset); \ | |
137 | if (arg->tok != tok_semicolon) \ | |
138 | break; \ | |
139 | arg = lr_token (ldfile, charset); \ | |
140 | } \ | |
141 | if (arg->tok != tok_eol) \ | |
142 | goto errlabel; \ | |
143 | } \ | |
144 | while (0) | |
145 | ||
146 | #define READ_NUMBER(fn, errlabel) \ | |
147 | do \ | |
148 | { \ | |
149 | arg = lr_token (ldfile, charset); \ | |
150 | if (arg->tok != tok_minus1 && arg->tok != tok_number) \ | |
151 | goto errlabel; \ | |
152 | fn (ldfile, result, nowtok, arg, charset); \ | |
153 | lr_ignore_rest (ldfile, 1); \ | |
154 | } \ | |
155 | while (0) | |
156 | ||
157 | #define READ_NUMBER_LIST(fn, errlabel) \ | |
158 | do \ | |
159 | { \ | |
160 | arg = lr_token (ldfile, charset); \ | |
161 | while (arg->tok == tok_minus1 || arg->tok == tok_number) \ | |
162 | { \ | |
163 | fn (ldfile, result, nowtok, arg, charset); \ | |
164 | arg = lr_token (ldfile, charset); \ | |
165 | if (arg->tok != tok_semicolon) \ | |
166 | break; \ | |
167 | arg = lr_token (ldfile, charset); \ | |
168 | } \ | |
169 | if (arg->tok != tok_eol) \ | |
170 | goto errlabel; \ | |
171 | } \ | |
172 | while (0) | |
173 | ||
174 | #define SYNTAX_ERROR(string) \ | |
175 | lr_error (ldfile, string); \ | |
176 | lr_ignore_rest (ldfile, 0); | |
177 | ||
178 | ||
179 | /* Parse locale definition file and store result in RESULT. */ | |
180 | state = 1; | |
181 | while (1) | |
182 | { | |
183 | /* What's on? */ | |
184 | struct token *now = lr_token (ldfile, charset); | |
185 | enum token_t nowtok = now->tok; | |
186 | struct token *arg; | |
187 | ||
188 | if (nowtok == tok_eof) | |
189 | break; | |
190 | ||
191 | switch (state) | |
192 | { | |
193 | case 1: | |
194 | /* The beginning. We expect the special declarations, EOL or | |
195 | the start of any locale. */ | |
196 | if (nowtok == tok_eol) | |
197 | /* Ignore empty lines. */ | |
198 | continue; | |
199 | ||
200 | switch (nowtok) | |
201 | { | |
202 | case tok_escape_char: | |
203 | case tok_comment_char: | |
204 | /* We need an argument. */ | |
205 | arg = lr_token (ldfile, charset); | |
206 | ||
207 | if (arg->tok != tok_ident) | |
208 | { | |
209 | SYNTAX_ERROR (_("bad argument")); | |
210 | continue; | |
211 | } | |
212 | ||
213 | if (arg->val.str.len != 1) | |
214 | { | |
215 | lr_error (ldfile, _("\ | |
216 | argument to `%s' must be a single character"), | |
217 | nowtok == tok_escape_char ? "escape_char" | |
218 | : "comment_char"); | |
219 | ||
220 | lr_ignore_rest (ldfile, 0); | |
221 | continue; | |
222 | } | |
223 | ||
224 | if (nowtok == tok_escape_char) | |
225 | ldfile->escape_char = *arg->val.str.start; | |
226 | else | |
227 | ldfile->comment_char = *arg->val.str.start; | |
228 | break; | |
229 | ||
230 | case tok_lc_ctype: | |
231 | state = 2; | |
232 | break; | |
233 | ||
234 | case tok_lc_collate: | |
235 | state = 10; | |
236 | break; | |
237 | ||
238 | case tok_lc_monetary: | |
239 | state = 20; | |
240 | break; | |
241 | ||
242 | case tok_lc_numeric: | |
243 | state = 30; | |
244 | break; | |
245 | ||
246 | case tok_lc_time: | |
247 | state = 40; | |
248 | break; | |
249 | ||
250 | case tok_lc_messages: | |
251 | state = 50; | |
252 | break; | |
253 | ||
254 | default: | |
255 | SYNTAX_ERROR (_("\ | |
256 | syntax error: not inside a locale definition section")); | |
257 | continue; | |
258 | } | |
259 | lr_ignore_rest (ldfile, 1); | |
260 | continue; | |
261 | ||
262 | case 2: | |
263 | HANDLE_COPY (LC_CTYPE, tok_lc_ctype, "LC_CYTPE"); | |
264 | ||
265 | ctype_startup (ldfile, result, charset); | |
266 | /* FALLTHROUGH */ | |
267 | ||
268 | case 3: | |
269 | /* Here we accept all the character classes, tolower/toupper, | |
270 | and following ANSI C:1995 self-defined classes. */ | |
271 | LOCALE_PROLOG (tok_lc_ctype, "LC_CTYPE"); | |
272 | ||
273 | if (nowtok == tok_charclass) | |
274 | { | |
275 | READ_STRING_LIST (ctype_class_new, bad_new_charclass); | |
276 | continue; | |
277 | bad_new_charclass: | |
278 | SYNTAX_ERROR (_("\ | |
279 | syntax error in definition of new character class")); | |
280 | continue; | |
281 | } | |
282 | ||
503054c0 | 283 | if (nowtok == tok_charconv) |
19bc17a9 | 284 | { |
503054c0 | 285 | READ_STRING_LIST (ctype_map_new, bad_new_charconv); |
19bc17a9 | 286 | continue; |
503054c0 | 287 | bad_new_charconv: |
19bc17a9 RM |
288 | SYNTAX_ERROR (_("\ |
289 | syntax error in definition of new character map")); | |
290 | continue; | |
291 | } | |
292 | ||
293 | if (nowtok == tok_upper || nowtok == tok_lower | |
294 | || nowtok == tok_alpha || nowtok == tok_digit | |
295 | || nowtok == tok_alnum || nowtok == tok_space | |
296 | || nowtok == tok_cntrl || nowtok == tok_punct | |
297 | || nowtok == tok_graph || nowtok == tok_print | |
298 | || nowtok == tok_xdigit || nowtok == tok_blank) | |
299 | { | |
300 | ctype_tok_sym = nowtok; | |
301 | ctype_tok_str = NULL; | |
302 | state = 5; | |
303 | continue; | |
304 | } | |
305 | ||
306 | if (nowtok == tok_toupper|| nowtok == tok_tolower) | |
307 | { | |
308 | ctype_tok_sym = nowtok; | |
309 | ctype_tok_str = NULL; | |
310 | state = 6; | |
311 | continue; | |
312 | } | |
313 | ||
314 | if (nowtok != tok_ident) | |
315 | goto bad_charclass; | |
316 | ||
317 | /* We possibly have a self-defined character class. */ | |
318 | if (ctype_is_charclass (ldfile, result, now->val.str.start)) | |
319 | { | |
320 | ctype_tok_sym = nowtok; | |
321 | ctype_tok_str = now->val.str.start; | |
322 | state = 5; | |
323 | continue; | |
324 | } | |
325 | ||
326 | /* ...or a self-defined character map. */ | |
503054c0 | 327 | if (ctype_is_charconv (ldfile, result, now->val.str.start)) |
19bc17a9 RM |
328 | { |
329 | ctype_tok_sym = nowtok; | |
330 | ctype_tok_str = now->val.str.start; | |
331 | state = 6; | |
332 | continue; | |
333 | } | |
334 | ||
335 | SYNTAX_ERROR (_("syntax error in definition of LC_CTYPE category")); | |
336 | continue; | |
337 | ||
338 | case 4: | |
339 | /* Handle `END xxx'. */ | |
340 | if (nowtok != expected_tok) | |
341 | lr_error (ldfile, _("\ | |
342 | `%1$s' definition does not end with `END %1$s'"), expected_str); | |
343 | ||
344 | lr_ignore_rest (ldfile, nowtok == expected_tok); | |
345 | state = 1; | |
346 | continue; | |
347 | ||
348 | case 5: | |
349 | /* Here we expect a semicolon separated list of bsymbols. The | |
350 | bit to be set in the word is given in CHARCLASS_BIT. */ | |
351 | arg = now; | |
352 | ||
353 | ctype_class_start (ldfile, result, ctype_tok_sym, ctype_tok_str, | |
354 | charset); | |
355 | ||
356 | while (arg->tok != tok_eol) | |
357 | { | |
358 | /* Any token other than a bsymbol is an error. */ | |
359 | if (arg->tok != tok_bsymbol) | |
360 | { | |
361 | bad_charclass: | |
362 | SYNTAX_ERROR (_("\ | |
363 | syntax error in character class definition")); | |
364 | break; | |
365 | } | |
366 | ||
367 | /* Lookup value for token and write into array. */ | |
368 | ctype_class_from (ldfile, result, arg, charset); | |
369 | ||
370 | arg = lr_token (ldfile, charset); | |
371 | if (arg->tok == tok_semicolon) | |
372 | arg = lr_token (ldfile, charset); | |
373 | else if (arg->tok != tok_eol) | |
374 | goto bad_charclass; | |
375 | ||
376 | /* Look for ellipsis. */ | |
377 | if (arg->tok == tok_ellipsis) | |
378 | { | |
379 | arg = lr_token (ldfile, charset); | |
380 | if (arg->tok != tok_semicolon) | |
381 | goto bad_charclass; | |
382 | ||
383 | arg = lr_token (ldfile, charset); | |
384 | if (arg->tok != tok_bsymbol) | |
385 | goto bad_charclass; | |
386 | ||
387 | /* Write range starting at LAST to ARG->VAL. */ | |
388 | ctype_class_to (ldfile, result, arg, charset); | |
389 | ||
390 | arg = lr_token (ldfile, charset); | |
391 | if (arg->tok == tok_semicolon) | |
392 | arg = lr_token (ldfile, charset); | |
393 | else if (arg->tok != tok_eol) | |
394 | goto bad_charclass; | |
395 | } | |
396 | } | |
397 | ||
398 | /* Mark class as already seen. */ | |
399 | ctype_class_end (ldfile, result); | |
400 | state = 3; | |
401 | ||
402 | continue; | |
403 | ||
404 | case 6: | |
405 | /* Here we expect a list of character mappings. Note: the | |
406 | first opening brace is already matched. */ | |
407 | ctype_map_start (ldfile, result, ctype_tok_sym, ctype_tok_str, | |
408 | charset); | |
409 | ||
410 | while (1) | |
411 | { | |
412 | /* Match ( bsymbol , bsymbol ) */ | |
413 | if (now->tok != tok_open_brace) | |
503054c0 | 414 | goto bad_charconv; |
19bc17a9 RM |
415 | |
416 | now = lr_token (ldfile, charset); | |
417 | if (now->tok != tok_bsymbol) | |
418 | { | |
503054c0 | 419 | bad_charconv: |
19bc17a9 | 420 | SYNTAX_ERROR (_("\ |
503054c0 | 421 | syntax error in character conversion definition")); |
19bc17a9 RM |
422 | state = 3; |
423 | break; | |
424 | } | |
425 | ||
426 | /* Lookup arg and assign to FROM. */ | |
427 | ctype_map_from (ldfile, result, now, charset); | |
428 | ||
429 | now = lr_token (ldfile, charset); | |
430 | if (now->tok != tok_comma) | |
503054c0 | 431 | goto bad_charconv; |
19bc17a9 RM |
432 | |
433 | now = lr_token (ldfile, charset); | |
434 | if (now->tok != tok_bsymbol) | |
503054c0 | 435 | goto bad_charconv; |
19bc17a9 RM |
436 | |
437 | /* Lookup arg and assign to TO. */ | |
438 | ctype_map_to (ldfile, result, now, charset); | |
439 | ||
440 | now = lr_token (ldfile, charset); | |
441 | if (now->tok != tok_close_brace) | |
503054c0 | 442 | goto bad_charconv; |
19bc17a9 RM |
443 | |
444 | now = lr_token (ldfile, charset); | |
445 | if (now->tok == tok_eol) | |
446 | { | |
447 | state = 3; | |
448 | break; | |
449 | } | |
450 | if (now->tok != tok_semicolon) | |
503054c0 | 451 | goto bad_charconv; |
19bc17a9 RM |
452 | |
453 | now = lr_token (ldfile, charset); | |
454 | } | |
455 | ||
456 | ctype_map_end (ldfile, result); | |
457 | continue; | |
458 | ||
459 | case 8: | |
460 | { | |
461 | /* We have seen `copy'. First match the argument. */ | |
462 | int warned = 0; | |
463 | ||
464 | if (nowtok != tok_string) | |
465 | lr_error (ldfile, _("expect string argument for `copy'")); | |
466 | else | |
467 | def_to_process (now->val.str.start, 1 << copy_category); | |
468 | ||
469 | lr_ignore_rest (ldfile, nowtok == tok_string); | |
470 | ||
471 | /* The rest of the line must be empty | |
472 | and the next keyword must be `END xxx'. */ | |
473 | ||
474 | while (lr_token (ldfile, charset)->tok != tok_end) | |
475 | { | |
476 | if (warned == 0) | |
477 | { | |
478 | only_copy: | |
479 | lr_error (ldfile, _("\ | |
480 | no other keyword shall be specified when `copy' is used")); | |
481 | warned = 1; | |
482 | } | |
483 | ||
484 | lr_ignore_rest (ldfile, 0); | |
485 | } | |
486 | ||
487 | state = 4; | |
488 | } | |
489 | continue; | |
490 | ||
491 | case 10: | |
492 | HANDLE_COPY (LC_COLLATE, tok_lc_collate, "LC_COLLATE"); | |
493 | ||
494 | collate_startup (ldfile, result, charset); | |
495 | /* FALLTHROUGH */ | |
496 | ||
497 | case 11: | |
498 | /* Process the LC_COLLATE section. We expect `END LC_COLLATE' | |
499 | any of the collation specifications, or any bsymbol. */ | |
500 | LOCALE_PROLOG (tok_lc_collate, "LC_COLLATE"); | |
501 | ||
502 | if (nowtok == tok_order_start) | |
503 | { | |
504 | state = 12; | |
505 | continue; | |
506 | } | |
507 | ||
508 | if (nowtok != tok_collating_element | |
509 | && nowtok != tok_collating_symbol) | |
510 | { | |
511 | bad_collation: | |
512 | lr_error (ldfile, _("\ | |
513 | syntax error in collation definition")); | |
514 | lr_ignore_rest (ldfile, 0); | |
515 | continue; | |
516 | } | |
517 | ||
518 | /* Get argument. */ | |
519 | arg = lr_token (ldfile, charset); | |
520 | if (arg->tok != tok_bsymbol) | |
521 | { | |
522 | lr_error (ldfile, _("\ | |
523 | collation symbol expected after `%s'"), | |
524 | nowtok == tok_collating_element | |
525 | ? "collating-element" : "collating-symbol"); | |
526 | lr_ignore_rest (ldfile, 0); | |
527 | continue; | |
528 | } | |
529 | ||
530 | if (nowtok == tok_collating_element) | |
531 | { | |
532 | /* Save to-value as new name. */ | |
533 | collate_element_to (ldfile, result, arg, charset); | |
534 | ||
535 | arg = lr_token (ldfile, charset); | |
536 | if (arg->tok != tok_from) | |
537 | { | |
538 | lr_error (ldfile, _("\ | |
539 | `from' expected after first argument to `collating-element'")); | |
540 | lr_ignore_rest (ldfile, 0); | |
541 | continue; | |
542 | } | |
543 | ||
544 | arg = lr_token (ldfile, charset); | |
545 | if (arg->tok != tok_string) | |
546 | { | |
547 | lr_error (ldfile, _("\ | |
548 | from-value of `collating-element' must be a string")); | |
549 | lr_ignore_rest (ldfile, 0); | |
550 | continue; | |
551 | } | |
552 | ||
553 | /* Enter new collating element. */ | |
554 | collate_element_from (ldfile, result, arg, charset); | |
555 | } | |
556 | else | |
557 | /* Enter new collating symbol into table. */ | |
558 | collate_symbol (ldfile, result, arg, charset); | |
559 | ||
560 | lr_ignore_rest (ldfile, 1); | |
561 | continue; | |
562 | ||
563 | case 12: | |
564 | /* We parse the rest of the line containing `order_start'. | |
565 | In any case we continue with parsing the symbols. */ | |
566 | state = 13; | |
567 | ||
568 | cnt = 0; | |
569 | while (now->tok != tok_eol) | |
570 | { | |
571 | int collation_method = 0; | |
572 | ||
573 | ++cnt; | |
574 | ||
575 | do | |
576 | { | |
577 | if (now->tok == tok_forward) | |
578 | collation_method |= sort_forward; | |
579 | else if (now->tok == tok_backward) | |
580 | collation_method |= sort_backward; | |
581 | else if (now->tok == tok_position) | |
582 | collation_method |= sort_position; | |
583 | else | |
584 | { | |
585 | lr_error (ldfile, _("unknown collation directive")); | |
586 | lr_ignore_rest (ldfile, 0); | |
587 | continue; | |
588 | } | |
589 | ||
590 | now = lr_token (ldfile, charset); | |
591 | } | |
592 | while (now->tok == tok_comma | |
593 | && (now == lr_token (ldfile, charset) != tok_none)); | |
594 | ||
595 | /* Check for consistency: forward and backwards are | |
596 | mutually exclusive. */ | |
597 | if ((collation_method & sort_forward) != 0 | |
598 | && (collation_method & sort_backward) != 0) | |
599 | { | |
600 | lr_error (ldfile, _("\ | |
601 | sorting order `forward' and `backward' are mutually exclusive")); | |
602 | /* The recover clear the backward flag. */ | |
603 | collation_method &= ~sort_backward; | |
604 | } | |
605 | ||
606 | /* ??? I don't know whether this is correct but while | |
607 | thinking about the `strcoll' functions I found that I | |
608 | need a direction when performing position depended | |
609 | collation. So I assume here that implicitly the | |
610 | direction `forward' is given when `position' alone is | |
611 | written. --drepper */ | |
612 | if (collation_method == sort_position) | |
613 | collation_method |= sort_forward; | |
614 | ||
615 | /* Enter info about next collation order. */ | |
616 | collate_new_order (ldfile, result, collation_method); | |
617 | ||
618 | if (now->tok != tok_eol && now->tok != tok_semicolon) | |
619 | { | |
620 | lr_error (ldfile, _("\ | |
621 | syntax error in `order_start' directive")); | |
622 | lr_ignore_rest (ldfile, 0); | |
623 | break; | |
624 | } | |
625 | ||
626 | if (now->tok == tok_semicolon) | |
627 | now = lr_token (ldfile, charset); | |
628 | } | |
629 | ||
630 | /* If no argument to `order_start' is given, one `forward' | |
631 | argument is implicitely assumed. */ | |
632 | if (cnt == 0) | |
633 | collate_new_order (ldfile, result, sort_forward); | |
634 | ||
635 | ||
636 | /* We now know about all sorting rules. */ | |
637 | collate_build_arrays (ldfile, result); | |
638 | ||
639 | continue; | |
640 | ||
641 | case 13: | |
642 | /* We read one symbol a line until `order_end' is found. */ | |
643 | { | |
644 | static int last_correct = 1; | |
645 | ||
646 | if (nowtok == tok_order_end) | |
647 | { | |
648 | state = 14; | |
649 | lr_ignore_rest (ldfile, 1); | |
650 | continue; | |
651 | } | |
652 | ||
653 | /* Ignore empty lines. */ | |
654 | if (nowtok == tok_eol) | |
655 | continue; | |
656 | ||
657 | if (nowtok != tok_bsymbol && nowtok != tok_undefined | |
658 | && nowtok != tok_ellipsis) | |
659 | { | |
660 | if (last_correct == 1) | |
661 | { | |
662 | lr_error (ldfile, _("\ | |
663 | syntax error in collating order definition")); | |
664 | last_correct = 0; | |
665 | } | |
666 | lr_ignore_rest (ldfile, 0); | |
667 | continue; | |
668 | } | |
669 | else | |
670 | { | |
671 | last_correct = 1; | |
672 | ||
673 | /* Remember current token. */ | |
674 | if (collate_order_elem (ldfile, result, now, charset) < 0) | |
675 | continue; | |
676 | } | |
677 | ||
678 | /* Read optional arguments. */ | |
679 | arg = lr_token (ldfile, charset); | |
680 | while (arg->tok != tok_eol) | |
681 | { | |
682 | if (arg->tok != tok_ignore && arg->tok != tok_ellipsis | |
683 | && arg->tok != tok_bsymbol && arg->tok != tok_string) | |
684 | break; | |
685 | ||
686 | if (arg->tok == tok_ignore || arg->tok == tok_ellipsis | |
687 | || arg->tok == tok_string) | |
688 | { | |
689 | /* Call handler for simple weights. */ | |
690 | if (collate_simple_weight (ldfile, result, arg, charset) | |
691 | < 0) | |
692 | goto illegal_weight; | |
693 | ||
694 | arg = lr_token (ldfile, charset); | |
695 | } | |
696 | else | |
697 | do | |
698 | { | |
699 | /* Collect char. */ | |
700 | int ok = collate_weight_bsymbol (ldfile, result, arg, | |
701 | charset); | |
702 | if (ok < 0) | |
703 | goto illegal_weight; | |
704 | ||
705 | arg = lr_token (ldfile, charset); | |
706 | } | |
707 | while (arg->tok == tok_bsymbol); | |
708 | ||
709 | /* Are there more weights? */ | |
710 | if (arg->tok != tok_semicolon) | |
711 | break; | |
712 | ||
713 | /* Yes, prepare next weight. */ | |
714 | if (collate_next_weight (ldfile, result) < 0) | |
715 | goto illegal_weight; | |
716 | ||
717 | arg = lr_token (ldfile, charset); | |
718 | } | |
719 | ||
720 | if (arg->tok != tok_eol) | |
721 | { | |
722 | SYNTAX_ERROR (_("syntax error in order specification")); | |
723 | } | |
724 | ||
725 | collate_end_weight (ldfile, result); | |
726 | illegal_weight: | |
727 | } | |
728 | continue; | |
729 | ||
730 | case 14: | |
731 | /* Following to the `order_end' keyword we don't expect | |
732 | anything but the `END'. */ | |
733 | if (nowtok == tok_eol) | |
734 | continue; | |
735 | ||
736 | if (nowtok != tok_end) | |
737 | goto bad_collation; | |
738 | ||
739 | expected_tok = tok_lc_collate; | |
740 | expected_str = "LC_COLLATE"; | |
741 | state = 4; | |
742 | ||
743 | ldfile->translate_strings = 1; | |
744 | continue; | |
745 | ||
746 | case 20: | |
747 | HANDLE_COPY (LC_MONETARY, tok_lc_monetary, "LC_MONETARY"); | |
748 | ||
749 | monetary_startup (ldfile, result, charset); | |
750 | /* FALLTHROUGH */ | |
751 | ||
752 | case 21: | |
753 | LOCALE_PROLOG (tok_lc_monetary, "LC_MONETARY"); | |
754 | ||
755 | switch (nowtok) | |
756 | { | |
757 | case tok_int_curr_symbol: | |
758 | case tok_currency_symbol: | |
759 | case tok_mon_decimal_point: | |
760 | case tok_mon_thousands_sep: | |
761 | case tok_positive_sign: | |
762 | case tok_negative_sign: | |
763 | READ_STRING (monetary_add, bad_monetary); | |
764 | break; | |
765 | ||
766 | case tok_int_frac_digits: | |
767 | case tok_frac_digits: | |
768 | case tok_p_cs_precedes: | |
769 | case tok_p_sep_by_space: | |
770 | case tok_n_cs_precedes: | |
771 | case tok_n_sep_by_space: | |
772 | case tok_p_sign_posn: | |
773 | case tok_n_sign_posn: | |
774 | READ_NUMBER (monetary_add, bad_monetary); | |
775 | break; | |
776 | ||
777 | case tok_mon_grouping: | |
778 | /* We have a semicolon separated list of integers. */ | |
779 | READ_NUMBER_LIST (monetary_add, bad_monetary); | |
780 | break; | |
781 | ||
782 | default: | |
783 | bad_monetary: | |
784 | SYNTAX_ERROR (_("syntax error in monetary locale definition")); | |
785 | } | |
786 | continue; | |
787 | ||
788 | case 30: | |
789 | HANDLE_COPY (LC_NUMERIC, tok_lc_numeric, "LC_NUMERIC"); | |
790 | ||
791 | numeric_startup (ldfile, result, charset); | |
792 | /* FALLTHROUGH */ | |
793 | ||
794 | case 31: | |
795 | LOCALE_PROLOG (tok_lc_numeric, "LC_NUMERIC"); | |
796 | ||
797 | switch (nowtok) | |
798 | { | |
799 | case tok_decimal_point: | |
800 | case tok_thousands_sep: | |
801 | READ_STRING (numeric_add, bad_numeric); | |
802 | break; | |
803 | ||
804 | case tok_grouping: | |
805 | /* We have a semicolon separated list of integers. */ | |
806 | READ_NUMBER_LIST (numeric_add, bad_numeric); | |
807 | break; | |
808 | ||
809 | default: | |
810 | bad_numeric: | |
811 | SYNTAX_ERROR (_("syntax error in numeric locale definition")); | |
812 | } | |
813 | continue; | |
814 | ||
815 | case 40: | |
816 | HANDLE_COPY (LC_TIME, tok_lc_time, "LC_TIME"); | |
817 | ||
818 | time_startup (ldfile, result, charset); | |
819 | /* FALLTHROUGH */ | |
820 | ||
821 | case 41: | |
822 | LOCALE_PROLOG (tok_lc_time, "LC_TIME"); | |
823 | ||
824 | switch (nowtok) | |
825 | { | |
826 | case tok_abday: | |
827 | case tok_day: | |
828 | case tok_abmon: | |
829 | case tok_mon: | |
830 | case tok_am_pm: | |
831 | case tok_alt_digits: | |
c4029823 | 832 | case tok_era: |
19bc17a9 RM |
833 | READ_STRING_LIST (time_add, bad_time); |
834 | continue; | |
835 | ||
836 | case tok_d_t_fmt: | |
837 | case tok_d_fmt: | |
838 | case tok_t_fmt: | |
839 | case tok_t_fmt_ampm: | |
19bc17a9 RM |
840 | case tok_era_year: |
841 | case tok_era_d_t_fmt: | |
842 | case tok_era_d_fmt: | |
843 | case tok_era_t_fmt: | |
844 | READ_STRING (time_add, bad_time); | |
845 | break; | |
846 | ||
847 | default: | |
848 | bad_time: | |
849 | SYNTAX_ERROR (_("syntax error in time locale definition")); | |
850 | } | |
851 | continue; | |
852 | ||
853 | case 50: | |
854 | HANDLE_COPY (LC_MESSAGES, tok_lc_messages, "LC_MESSAGES"); | |
855 | ||
856 | messages_startup (ldfile, result, charset); | |
857 | /* FALLTHROUGH */ | |
858 | ||
859 | case 51: | |
860 | LOCALE_PROLOG (tok_lc_messages, "LC_MESSAGES"); | |
861 | ||
862 | switch (nowtok) | |
863 | { | |
864 | case tok_yesexpr: | |
865 | case tok_noexpr: | |
866 | case tok_yesstr: | |
867 | case tok_nostr: | |
868 | READ_STRING (messages_add, bad_message); | |
869 | break; | |
870 | ||
871 | default: | |
872 | bad_message: | |
873 | SYNTAX_ERROR (_("syntax error in message locale definition")); | |
874 | } | |
875 | continue; | |
876 | ||
877 | default: | |
878 | error (5, 0, _("%s: error in state machine"), __FILE__); | |
879 | /* NOTREACHED */ | |
880 | } | |
881 | ||
882 | break; | |
883 | } | |
884 | ||
885 | /* We read all of the file. */ | |
886 | lr_close (ldfile); | |
887 | ||
888 | /* Let's see what information is available. */ | |
889 | for (cnt = LC_CTYPE; cnt <= LC_MESSAGES; ++cnt) | |
890 | if (result->categories[cnt].generic != NULL) | |
891 | result->avail |= 1 << cnt; | |
892 | ||
893 | return result; | |
894 | } | |
895 | ||
896 | ||
897 | void | |
898 | check_all_categories (struct localedef_t *locale, struct charset_t *charset) | |
899 | { | |
900 | /* Call the finishing functions for all locales. */ | |
901 | if ((locale->binary & (1 << LC_CTYPE)) == 0) | |
902 | ctype_finish (locale, charset); | |
903 | if ((locale->binary & (1 << LC_COLLATE)) == 0) | |
904 | collate_finish (locale, charset); | |
905 | if ((locale->binary & (1 << LC_MONETARY)) == 0) | |
906 | monetary_finish (locale); | |
907 | if ((locale->binary & (1 << LC_NUMERIC)) == 0) | |
908 | numeric_finish (locale); | |
909 | if ((locale->binary & (1 << LC_TIME)) == 0) | |
910 | time_finish (locale); | |
911 | if ((locale->binary & (1 << LC_MESSAGES)) == 0) | |
912 | messages_finish (locale); | |
913 | } | |
914 | ||
915 | ||
916 | void | |
75cd5204 RM |
917 | write_all_categories (struct localedef_t *locale, struct charset_t *charset, |
918 | const char *output_path) | |
19bc17a9 RM |
919 | { |
920 | /* Call all functions to write locale data. */ | |
75cd5204 | 921 | ctype_output (locale, charset, output_path); |
ce7a5ef4 | 922 | collate_output (locale, charset, output_path); |
19bc17a9 RM |
923 | monetary_output (locale, output_path); |
924 | numeric_output (locale, output_path); | |
925 | time_output (locale, output_path); | |
926 | messages_output (locale, output_path); | |
927 | } | |
928 | ||
929 | ||
930 | void | |
931 | write_locale_data (const char *output_path, const char *category, | |
932 | size_t n_elem, struct iovec *vec) | |
933 | { | |
c199a24f | 934 | size_t cnt, step, maxiov; |
19bc17a9 RM |
935 | int fd; |
936 | char *fname; | |
937 | ||
c4029823 | 938 | fname = malloc (strlen (output_path) + 2 * strlen (category) + 6); |
036cc82f RM |
939 | if (fname == NULL) |
940 | error (5, errno, _("memory exhausted")); | |
941 | ||
942 | /* Normally we write to the directory pointed to by the OUTPUT_PATH. | |
943 | But for LC_MESSAGES we have to take care for the translation | |
944 | data. This means we need to have a directory LC_MESSAGES in | |
945 | which we place the file under the name SYS_LC_MESSAGES. */ | |
ce7a5ef4 | 946 | sprintf (fname, "%s%s", output_path, category); |
036cc82f | 947 | if (strcmp (category, "LC_MESSAGES") == 0) |
036cc82f | 948 | { |
ce7a5ef4 RM |
949 | struct stat st; |
950 | ||
951 | if (stat (fname, &st) < 0) | |
952 | { | |
953 | if (mkdir (fname, 0777) < 0) | |
954 | fd = creat (fname, 0666); | |
955 | else | |
956 | { | |
957 | fd = -1; | |
958 | errno = EISDIR; | |
959 | } | |
960 | } | |
961 | else if (S_ISREG (st.st_mode)) | |
962 | fd = creat (fname, 0666); | |
963 | else | |
964 | { | |
965 | fd = -1; | |
966 | errno = EISDIR; | |
967 | } | |
036cc82f | 968 | } |
ce7a5ef4 RM |
969 | else |
970 | fd = creat (fname, 0666); | |
036cc82f | 971 | |
19bc17a9 RM |
972 | if (fd == -1) |
973 | { | |
974 | int save_err = errno; | |
975 | ||
976 | if (errno == EISDIR) | |
977 | { | |
036cc82f | 978 | sprintf (fname, "%1$s%2$s/SYS_%2$s", output_path, category); |
19bc17a9 RM |
979 | fd = creat (fname, 0666); |
980 | if (fd == -1) | |
981 | save_err = errno; | |
982 | } | |
983 | ||
984 | if (fd == -1) | |
985 | { | |
ce7a5ef4 RM |
986 | error (0, save_err, _("\ |
987 | cannot open output file `%s' for category `%s'"), | |
988 | fname, category); | |
19bc17a9 RM |
989 | return; |
990 | } | |
991 | } | |
992 | free (fname); | |
993 | ||
c199a24f MB |
994 | #ifdef UIO_MAXIOV |
995 | maxiov = UIO_MAXIOV; | |
996 | #else | |
997 | maxiov = sysconf (_SC_UIO_MAXIOV); | |
998 | #endif | |
999 | ||
19bc17a9 RM |
1000 | /* Write the data using writev. But we must take care for the |
1001 | limitation of the implementation. */ | |
1002 | for (cnt = 0; cnt < n_elem; cnt += step) | |
1003 | { | |
c199a24f MB |
1004 | step = n_elem - cnt; |
1005 | if (maxiov > 0) | |
1006 | step = MIN (maxiov, step); | |
19bc17a9 RM |
1007 | |
1008 | if (writev (fd, &vec[cnt], step) < 0) | |
1009 | { | |
1010 | error (0, errno, _("failure while writing data for category `%s'"), | |
1011 | category); | |
1012 | break; | |
1013 | } | |
1014 | } | |
1015 | ||
1016 | close (fd); | |
1017 | } |