]>
Commit | Line | Data |
---|---|---|
b0c9067d | 1 | /* Copyright (C) 1996-2002, 2003 Free Software Foundation, Inc. |
df4ef2ab | 2 | This file is part of the GNU C Library. |
b6aa34eb | 3 | Contributed by Ulrich Drepper <drepper@redhat.com>, 1996. |
a641835a | 4 | |
df4ef2ab | 5 | The GNU C Library is free software; you can redistribute it and/or |
41bdb6e2 AJ |
6 | modify it under the terms of the GNU Lesser General Public |
7 | License as published by the Free Software Foundation; either | |
8 | version 2.1 of the License, or (at your option) any later version. | |
a641835a | 9 | |
df4ef2ab UD |
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 | |
41bdb6e2 | 13 | Lesser General Public License for more details. |
a641835a | 14 | |
41bdb6e2 AJ |
15 | You should have received a copy of the GNU Lesser General Public |
16 | License along with the GNU C Library; if not, write to the Free | |
17 | Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA | |
18 | 02111-1307 USA. */ | |
a641835a RM |
19 | |
20 | #ifdef HAVE_CONFIG_H | |
21 | # include <config.h> | |
22 | #endif | |
23 | ||
5a97622d | 24 | #include <argp.h> |
d2defdc4 | 25 | #include <assert.h> |
a641835a RM |
26 | #include <ctype.h> |
27 | #include <endian.h> | |
28 | #include <errno.h> | |
29 | #include <error.h> | |
30 | #include <fcntl.h> | |
d2defdc4 UD |
31 | #include <iconv.h> |
32 | #include <langinfo.h> | |
e75154a6 | 33 | #include <locale.h> |
a641835a RM |
34 | #include <libintl.h> |
35 | #include <limits.h> | |
36 | #include <nl_types.h> | |
37 | #include <obstack.h> | |
b6aa34eb | 38 | #include <stdint.h> |
a641835a RM |
39 | #include <stdio.h> |
40 | #include <stdlib.h> | |
41 | #include <string.h> | |
42 | #include <unistd.h> | |
d2defdc4 | 43 | #include <wchar.h> |
a641835a RM |
44 | |
45 | #include "version.h" | |
46 | ||
47 | #include "catgetsinfo.h" | |
48 | ||
49 | ||
50 | #define SWAPU32(w) \ | |
51 | (((w) << 24) | (((w) & 0xff00) << 8) | (((w) >> 8) & 0xff00) | ((w) >> 24)) | |
52 | ||
53 | struct message_list | |
54 | { | |
55 | int number; | |
56 | const char *message; | |
57 | ||
58 | const char *fname; | |
59 | size_t line; | |
60 | const char *symbol; | |
61 | ||
62 | struct message_list *next; | |
63 | }; | |
64 | ||
65 | ||
66 | struct set_list | |
67 | { | |
68 | int number; | |
69 | int deleted; | |
70 | struct message_list *messages; | |
71 | int last_message; | |
72 | ||
73 | const char *fname; | |
74 | size_t line; | |
75 | const char *symbol; | |
76 | ||
77 | struct set_list *next; | |
78 | }; | |
79 | ||
80 | ||
81 | struct catalog | |
82 | { | |
83 | struct set_list *all_sets; | |
84 | struct set_list *current_set; | |
85 | size_t total_messages; | |
d2defdc4 | 86 | wint_t quote_char; |
a641835a RM |
87 | int last_set; |
88 | ||
89 | struct obstack mem_pool; | |
90 | }; | |
91 | ||
92 | ||
93 | /* If non-zero force creation of new file, not using existing one. */ | |
94 | static int force_new; | |
95 | ||
5a97622d UD |
96 | /* Name of output file. */ |
97 | static const char *output_name; | |
98 | ||
99 | /* Name of generated C header file. */ | |
100 | static const char *header_name; | |
101 | ||
102 | /* Name and version of program. */ | |
103 | static void print_version (FILE *stream, struct argp_state *state); | |
104 | void (*argp_program_version_hook) (FILE *, struct argp_state *) = print_version; | |
105 | ||
106 | #define OPT_NEW 1 | |
107 | ||
108 | /* Definitions of arguments for argp functions. */ | |
109 | static const struct argp_option options[] = | |
110 | { | |
111 | { "header", 'H', N_("NAME"), 0, | |
112 | N_("Create C header file NAME containing symbol definitions") }, | |
113 | { "new", OPT_NEW, NULL, 0, | |
114 | N_("Do not use existing catalog, force new output file") }, | |
115 | { "output", 'o', N_("NAME"), 0, N_("Write output to file NAME") }, | |
116 | { NULL, 0, NULL, 0, NULL } | |
117 | }; | |
118 | ||
119 | /* Short description of program. */ | |
120 | static const char doc[] = N_("Generate message catalog.\ | |
121 | \vIf INPUT-FILE is -, input is read from standard input. If OUTPUT-FILE\n\ | |
122 | is -, output is written to standard output.\n"); | |
123 | ||
124 | /* Strings for arguments in help texts. */ | |
125 | static const char args_doc[] = N_("\ | |
126 | -o OUTPUT-FILE [INPUT-FILE]...\n[OUTPUT-FILE [INPUT-FILE]...]"); | |
127 | ||
128 | /* Prototype for option handler. */ | |
129 | static error_t parse_opt (int key, char *arg, struct argp_state *state); | |
130 | ||
131 | /* Function to print some extra text in the help message. */ | |
132 | static char *more_help (int key, const char *text, void *input); | |
133 | ||
134 | /* Data structure to communicate with argp functions. */ | |
135 | static struct argp argp = | |
a641835a | 136 | { |
5a97622d | 137 | options, parse_opt, args_doc, doc, NULL, more_help |
a641835a RM |
138 | }; |
139 | ||
5a97622d | 140 | |
a641835a RM |
141 | /* Wrapper functions with error checking for standard functions. */ |
142 | extern void *xmalloc (size_t n); | |
a3b2008a | 143 | extern void *xcalloc (size_t n, size_t s); |
d2defdc4 UD |
144 | extern void *xrealloc (void *o, size_t n); |
145 | extern char *xstrdup (const char *); | |
a641835a RM |
146 | |
147 | /* Prototypes for local functions. */ | |
a641835a RM |
148 | static void error_print (void); |
149 | static struct catalog *read_input_file (struct catalog *current, | |
150 | const char *fname); | |
151 | static void write_out (struct catalog *result, const char *output_name, | |
152 | const char *header_name); | |
153 | static struct set_list *find_set (struct catalog *current, int number); | |
d2defdc4 | 154 | static void normalize_line (const char *fname, size_t line, iconv_t cd, |
ee25ee65 UD |
155 | wchar_t *string, wchar_t quote_char, |
156 | wchar_t escape_char); | |
a641835a | 157 | static void read_old (struct catalog *catalog, const char *file_name); |
d2defdc4 | 158 | static int open_conversion (const char *codesetp, iconv_t *cd_towcp, |
ee25ee65 | 159 | iconv_t *cd_tombp, wchar_t *escape_charp); |
a641835a RM |
160 | |
161 | ||
162 | int | |
163 | main (int argc, char *argv[]) | |
164 | { | |
165 | struct catalog *result; | |
2f6d1f1b | 166 | int remaining; |
a641835a RM |
167 | |
168 | /* Set program name for messages. */ | |
169 | error_print_progname = error_print; | |
170 | ||
171 | /* Set locale via LC_ALL. */ | |
172 | setlocale (LC_ALL, ""); | |
173 | ||
174 | /* Set the text message domain. */ | |
175 | textdomain (PACKAGE); | |
176 | ||
177 | /* Initialize local variables. */ | |
a641835a RM |
178 | result = NULL; |
179 | ||
5a97622d | 180 | /* Parse and process arguments. */ |
2f6d1f1b | 181 | argp_parse (&argp, argc, argv, 0, &remaining, NULL); |
a641835a RM |
182 | |
183 | /* Determine output file. */ | |
184 | if (output_name == NULL) | |
2f6d1f1b | 185 | output_name = remaining < argc ? argv[remaining++] : "-"; |
a641835a RM |
186 | |
187 | /* Process all input files. */ | |
188 | setlocale (LC_CTYPE, "C"); | |
2f6d1f1b | 189 | if (remaining < argc) |
a641835a | 190 | do |
2f6d1f1b UD |
191 | result = read_input_file (result, argv[remaining]); |
192 | while (++remaining < argc); | |
a641835a RM |
193 | else |
194 | result = read_input_file (NULL, "-"); | |
195 | ||
196 | /* Write out the result. */ | |
197 | if (result != NULL) | |
198 | write_out (result, output_name, header_name); | |
199 | ||
819c56e7 | 200 | return error_message_count != 0; |
a641835a RM |
201 | } |
202 | ||
203 | ||
5a97622d UD |
204 | /* Handle program arguments. */ |
205 | static error_t | |
206 | parse_opt (int key, char *arg, struct argp_state *state) | |
a641835a | 207 | { |
5a97622d | 208 | switch (key) |
fafaa44e | 209 | { |
5a97622d UD |
210 | case 'H': |
211 | header_name = arg; | |
212 | break; | |
213 | case OPT_NEW: | |
214 | force_new = 1; | |
215 | break; | |
216 | case 'o': | |
217 | output_name = arg; | |
218 | break; | |
219 | default: | |
220 | return ARGP_ERR_UNKNOWN; | |
fafaa44e | 221 | } |
5a97622d UD |
222 | return 0; |
223 | } | |
a641835a | 224 | |
5a97622d UD |
225 | |
226 | static char * | |
227 | more_help (int key, const char *text, void *input) | |
228 | { | |
229 | switch (key) | |
230 | { | |
231 | case ARGP_KEY_HELP_EXTRA: | |
232 | /* We print some extra information. */ | |
233 | return strdup (gettext ("\ | |
f2ea0f5b | 234 | Report bugs using the `glibcbug' script to <bugs@gnu.org>.\n")); |
5a97622d UD |
235 | default: |
236 | break; | |
237 | } | |
238 | return (char *) text; | |
239 | } | |
240 | ||
241 | /* Print the version information. */ | |
242 | static void | |
243 | print_version (FILE *stream, struct argp_state *state) | |
244 | { | |
245 | fprintf (stream, "gencat (GNU %s) %s\n", PACKAGE, VERSION); | |
246 | fprintf (stream, gettext ("\ | |
247 | Copyright (C) %s Free Software Foundation, Inc.\n\ | |
248 | This is free software; see the source for copying conditions. There is NO\n\ | |
249 | warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n\ | |
b0c9067d | 250 | "), "2003"); |
5a97622d | 251 | fprintf (stream, gettext ("Written by %s.\n"), "Ulrich Drepper"); |
a641835a RM |
252 | } |
253 | ||
254 | ||
255 | /* The address of this function will be assigned to the hook in the | |
256 | error functions. */ | |
257 | static void | |
258 | error_print () | |
259 | { | |
260 | /* We don't want the program name to be printed in messages. Emacs' | |
261 | compile.el does not like this. */ | |
262 | } | |
263 | ||
264 | ||
265 | static struct catalog * | |
266 | read_input_file (struct catalog *current, const char *fname) | |
267 | { | |
268 | FILE *fp; | |
269 | char *buf; | |
270 | size_t len; | |
271 | size_t line_number; | |
d2defdc4 UD |
272 | wchar_t *wbuf; |
273 | size_t wbufsize; | |
274 | iconv_t cd_towc = (iconv_t) -1; | |
275 | iconv_t cd_tomb = (iconv_t) -1; | |
ee25ee65 | 276 | wchar_t escape_char = L'\\'; |
d2defdc4 | 277 | char *codeset = NULL; |
a641835a RM |
278 | |
279 | if (strcmp (fname, "-") == 0 || strcmp (fname, "/dev/stdin") == 0) | |
280 | { | |
281 | fp = stdin; | |
282 | fname = gettext ("*standard input*"); | |
283 | } | |
284 | else | |
285 | fp = fopen (fname, "r"); | |
286 | if (fp == NULL) | |
287 | { | |
288 | error (0, errno, gettext ("cannot open input file `%s'"), fname); | |
289 | return current; | |
290 | } | |
291 | ||
292 | /* If we haven't seen anything yet, allocate result structure. */ | |
293 | if (current == NULL) | |
294 | { | |
a3b2008a | 295 | current = (struct catalog *) xcalloc (1, sizeof (*current)); |
a641835a | 296 | |
df4ef2ab | 297 | #define obstack_chunk_alloc malloc |
a641835a RM |
298 | #define obstack_chunk_free free |
299 | obstack_init (¤t->mem_pool); | |
a3b2008a UD |
300 | |
301 | current->current_set = find_set (current, NL_SETD); | |
a641835a RM |
302 | } |
303 | ||
304 | buf = NULL; | |
305 | len = 0; | |
306 | line_number = 0; | |
d2defdc4 UD |
307 | |
308 | wbufsize = 1024; | |
309 | wbuf = (wchar_t *) xmalloc (wbufsize); | |
310 | ||
a641835a RM |
311 | while (!feof (fp)) |
312 | { | |
313 | int continued; | |
314 | int used; | |
315 | size_t start_line = line_number + 1; | |
316 | char *this_line; | |
317 | ||
318 | do | |
319 | { | |
320 | int act_len; | |
321 | ||
322 | act_len = getline (&buf, &len, fp); | |
323 | if (act_len <= 0) | |
324 | break; | |
325 | ++line_number; | |
326 | ||
327 | /* It the line continued? */ | |
eabea972 | 328 | continued = 0; |
a641835a RM |
329 | if (buf[act_len - 1] == '\n') |
330 | { | |
331 | --act_len; | |
eabea972 UD |
332 | |
333 | /* There might be more than one backslash at the end of | |
334 | the line. Only if there is an odd number of them is | |
335 | the line continued. */ | |
a5392bed | 336 | if (act_len > 0 && buf[act_len - 1] == '\\') |
eabea972 UD |
337 | { |
338 | int temp_act_len = act_len; | |
339 | ||
340 | do | |
341 | { | |
342 | --temp_act_len; | |
343 | continued = !continued; | |
344 | } | |
345 | while (temp_act_len > 0 && buf[temp_act_len - 1] == '\\'); | |
eabea972 | 346 | |
c70ad7d7 UD |
347 | if (continued) |
348 | --act_len; | |
349 | } | |
a641835a | 350 | } |
a641835a RM |
351 | |
352 | /* Append to currently selected line. */ | |
353 | obstack_grow (¤t->mem_pool, buf, act_len); | |
354 | } | |
355 | while (continued); | |
356 | ||
357 | obstack_1grow (¤t->mem_pool, '\0'); | |
358 | this_line = (char *) obstack_finish (¤t->mem_pool); | |
359 | ||
360 | used = 0; | |
361 | if (this_line[0] == '$') | |
362 | { | |
c3880fbd | 363 | if (isblank (this_line[1])) |
d2defdc4 UD |
364 | { |
365 | int cnt = 1; | |
366 | while (isblank (this_line[cnt])) | |
367 | ++cnt; | |
368 | if (strncmp (&this_line[cnt], "codeset=", 8) != 0) | |
369 | /* This is a comment line. Do nothing. */; | |
370 | else if (codeset != NULL) | |
371 | /* Ignore multiple codeset. */; | |
372 | else | |
373 | { | |
374 | int start = cnt + 8; | |
375 | cnt = start; | |
376 | while (this_line[cnt] != '\0' && !isspace (this_line[cnt])) | |
377 | ++cnt; | |
378 | if (cnt != start) | |
379 | { | |
380 | int len = cnt - start; | |
381 | codeset = xmalloc (len + 1); | |
382 | *((char *) mempcpy (codeset, &this_line[start], len)) | |
383 | = '\0'; | |
384 | } | |
385 | } | |
386 | } | |
a641835a RM |
387 | else if (strncmp (&this_line[1], "set", 3) == 0) |
388 | { | |
40a55d20 | 389 | int cnt = sizeof ("set"); |
6dbe2837 | 390 | int set_number; |
a641835a RM |
391 | const char *symbol = NULL; |
392 | while (isspace (this_line[cnt])) | |
393 | ++cnt; | |
394 | ||
395 | if (isdigit (this_line[cnt])) | |
396 | { | |
397 | set_number = atol (&this_line[cnt]); | |
398 | ||
399 | /* If the given number for the character set is | |
400 | higher than any we used for symbolic set names | |
401 | avoid clashing by using only higher numbers for | |
402 | the following symbolic definitions. */ | |
403 | if (set_number > current->last_set) | |
404 | current->last_set = set_number; | |
405 | } | |
406 | else | |
407 | { | |
408 | /* See whether it is a reasonable identifier. */ | |
409 | int start = cnt; | |
410 | while (isalnum (this_line[cnt]) || this_line[cnt] == '_') | |
411 | ++cnt; | |
412 | ||
413 | if (cnt == start) | |
414 | { | |
415 | /* No correct character found. */ | |
416 | error_at_line (0, 0, fname, start_line, | |
417 | gettext ("illegal set number")); | |
418 | set_number = 0; | |
419 | } | |
420 | else | |
421 | { | |
6d52618b | 422 | /* We have found seomthing that looks like a |
a641835a RM |
423 | correct identifier. */ |
424 | struct set_list *runp; | |
425 | ||
426 | this_line[cnt] = '\0'; | |
427 | used = 1; | |
428 | symbol = &this_line[start]; | |
429 | ||
430 | /* Test whether the identifier was already used. */ | |
431 | runp = current->all_sets; | |
432 | while (runp != 0) | |
433 | if (runp->symbol != NULL | |
434 | && strcmp (runp->symbol, symbol) == 0) | |
435 | break; | |
436 | else | |
437 | runp = runp->next; | |
438 | ||
439 | if (runp != NULL) | |
440 | { | |
441 | /* We cannot allow duplicate identifiers for | |
442 | message sets. */ | |
443 | error_at_line (0, 0, fname, start_line, | |
444 | gettext ("duplicate set definition")); | |
445 | error_at_line (0, 0, runp->fname, runp->line, | |
446 | gettext ("\ | |
447 | this is the first definition")); | |
448 | set_number = 0; | |
449 | } | |
450 | else | |
451 | /* Allocate next free message set for identifier. */ | |
452 | set_number = ++current->last_set; | |
453 | } | |
454 | } | |
455 | ||
456 | if (set_number != 0) | |
457 | { | |
458 | /* We found a legal set number. */ | |
459 | current->current_set = find_set (current, set_number); | |
460 | if (symbol != NULL) | |
461 | used = 1; | |
462 | current->current_set->symbol = symbol; | |
463 | current->current_set->fname = fname; | |
464 | current->current_set->line = start_line; | |
465 | } | |
466 | } | |
467 | else if (strncmp (&this_line[1], "delset", 6) == 0) | |
468 | { | |
469 | int cnt = sizeof ("delset"); | |
470 | size_t set_number; | |
471 | while (isspace (this_line[cnt])) | |
472 | ++cnt; | |
473 | ||
474 | if (isdigit (this_line[cnt])) | |
475 | { | |
476 | size_t set_number = atol (&this_line[cnt]); | |
477 | struct set_list *set; | |
478 | ||
479 | /* Mark the message set with the given number as | |
480 | deleted. */ | |
481 | set = find_set (current, set_number); | |
482 | set->deleted = 1; | |
483 | } | |
484 | else | |
485 | { | |
486 | /* See whether it is a reasonable identifier. */ | |
487 | int start = cnt; | |
488 | while (isalnum (this_line[cnt]) || this_line[cnt] == '_') | |
489 | ++cnt; | |
490 | ||
491 | if (cnt == start) | |
492 | { | |
493 | error_at_line (0, 0, fname, start_line, | |
494 | gettext ("illegal set number")); | |
495 | set_number = 0; | |
496 | } | |
497 | else | |
498 | { | |
499 | const char *symbol; | |
500 | struct set_list *runp; | |
501 | ||
502 | this_line[cnt] = '\0'; | |
503 | used = 1; | |
504 | symbol = &this_line[start]; | |
505 | ||
506 | /* We have a symbolic set name. This name must | |
507 | appear somewhere else in the catalogs read so | |
508 | far. */ | |
509 | set_number = 0; | |
510 | for (runp = current->all_sets; runp != NULL; | |
511 | runp = runp->next) | |
512 | { | |
513 | if (strcmp (runp->symbol, symbol) == 0) | |
514 | { | |
515 | runp->deleted = 1; | |
516 | break; | |
517 | } | |
518 | } | |
519 | if (runp == NULL) | |
520 | /* Name does not exist before. */ | |
521 | error_at_line (0, 0, fname, start_line, | |
522 | gettext ("unknown set `%s'"), symbol); | |
523 | } | |
524 | } | |
525 | } | |
526 | else if (strncmp (&this_line[1], "quote", 5) == 0) | |
527 | { | |
d2defdc4 UD |
528 | char buf[2]; |
529 | char *bufptr; | |
530 | size_t buflen; | |
531 | char *wbufptr; | |
532 | size_t wbuflen; | |
533 | int cnt; | |
534 | ||
535 | cnt = sizeof ("quote"); | |
a641835a RM |
536 | while (isspace (this_line[cnt])) |
537 | ++cnt; | |
d2defdc4 UD |
538 | |
539 | /* We need the conversion. */ | |
540 | if (cd_towc == (iconv_t) -1 | |
ee25ee65 UD |
541 | && open_conversion (codeset, &cd_towc, &cd_tomb, |
542 | &escape_char) != 0) | |
d2defdc4 UD |
543 | /* Something is wrong. */ |
544 | goto out; | |
545 | ||
a641835a | 546 | /* Yes, the quote char can be '\0'; this means no quote |
d2defdc4 UD |
547 | char. The function using the information works on |
548 | wide characters so we have to convert it here. */ | |
549 | buf[0] = this_line[cnt]; | |
550 | buf[1] = '\0'; | |
551 | bufptr = buf; | |
552 | buflen = 2; | |
553 | ||
554 | wbufptr = (char *) wbuf; | |
555 | wbuflen = wbufsize; | |
556 | ||
557 | /* Flush the state. */ | |
558 | iconv (cd_towc, NULL, NULL, NULL, NULL); | |
559 | ||
560 | iconv (cd_towc, &bufptr, &buflen, &wbufptr, &wbuflen); | |
561 | if (buflen != 0 || (wchar_t *) wbufptr != &wbuf[2]) | |
562 | error_at_line (0, 0, fname, start_line, | |
563 | gettext ("invalid quote character")); | |
564 | else | |
565 | /* Use the converted wide character. */ | |
566 | current->quote_char = wbuf[0]; | |
a641835a RM |
567 | } |
568 | else | |
569 | { | |
570 | int cnt; | |
571 | cnt = 2; | |
572 | while (this_line[cnt] != '\0' && !isspace (this_line[cnt])) | |
573 | ++cnt; | |
574 | this_line[cnt] = '\0'; | |
575 | error_at_line (0, 0, fname, start_line, | |
576 | gettext ("unknown directive `%s': line ignored"), | |
577 | &this_line[1]); | |
578 | } | |
579 | } | |
580 | else if (isalnum (this_line[0]) || this_line[0] == '_') | |
581 | { | |
582 | const char *ident = this_line; | |
0a70515e | 583 | char *line = this_line; |
a641835a RM |
584 | int message_number; |
585 | ||
586 | do | |
0a70515e UD |
587 | ++line; |
588 | while (line[0] != '\0' && !isspace (line[0])); | |
589 | if (line[0] != '\0') | |
590 | *line++ = '\0'; /* Terminate the identifier. */ | |
a641835a | 591 | |
a641835a RM |
592 | /* Now we found the beginning of the message itself. */ |
593 | ||
594 | if (isdigit (ident[0])) | |
595 | { | |
596 | struct message_list *runp; | |
96eaef36 | 597 | struct message_list *lastp; |
a641835a RM |
598 | |
599 | message_number = atoi (ident); | |
600 | ||
601 | /* Find location to insert the new message. */ | |
602 | runp = current->current_set->messages; | |
96eaef36 | 603 | lastp = NULL; |
a641835a RM |
604 | while (runp != NULL) |
605 | if (runp->number == message_number) | |
606 | break; | |
607 | else | |
96eaef36 UD |
608 | { |
609 | lastp = runp; | |
610 | runp = runp->next; | |
611 | } | |
a641835a RM |
612 | if (runp != NULL) |
613 | { | |
b6aa34eb UD |
614 | /* Oh, oh. There is already a message with this |
615 | number in the message set. */ | |
819c56e7 UD |
616 | if (runp->symbol == NULL) |
617 | { | |
618 | /* The existing message had its number specified | |
619 | by the user. Fatal collision type uh, oh. */ | |
620 | error_at_line (0, 0, fname, start_line, | |
621 | gettext ("duplicated message number")); | |
622 | error_at_line (0, 0, runp->fname, runp->line, | |
623 | gettext ("this is the first definition")); | |
624 | message_number = 0; | |
625 | } | |
626 | else | |
627 | { | |
628 | /* Collision was with number auto-assigned to a | |
629 | symbolic. Change existing symbolic number | |
630 | and move to end the list (if not already there). */ | |
631 | runp->number = ++current->current_set->last_message; | |
632 | ||
633 | if (runp->next != NULL) | |
634 | { | |
635 | struct message_list *endp; | |
636 | ||
637 | if (lastp == NULL) | |
638 | current->current_set->messages=runp->next; | |
639 | else | |
640 | lastp->next=runp->next; | |
641 | ||
642 | endp = runp->next; | |
643 | while (endp->next != NULL) | |
644 | endp = endp->next; | |
645 | ||
646 | endp->next = runp; | |
647 | runp->next = NULL; | |
648 | } | |
649 | } | |
a641835a RM |
650 | } |
651 | ident = NULL; /* We don't have a symbol. */ | |
652 | ||
653 | if (message_number != 0 | |
654 | && message_number > current->current_set->last_message) | |
655 | current->current_set->last_message = message_number; | |
656 | } | |
657 | else if (ident[0] != '\0') | |
658 | { | |
659 | struct message_list *runp; | |
96eaef36 | 660 | struct message_list *lastp; |
a641835a RM |
661 | |
662 | /* Test whether the symbolic name was not used for | |
663 | another message in this message set. */ | |
96eaef36 UD |
664 | runp = current->current_set->messages; |
665 | lastp = NULL; | |
a641835a RM |
666 | while (runp != NULL) |
667 | if (runp->symbol != NULL && strcmp (ident, runp->symbol) == 0) | |
668 | break; | |
669 | else | |
670 | runp = runp->next; | |
671 | if (runp != NULL) | |
672 | { | |
b6aa34eb UD |
673 | /* The name is already used. */ |
674 | error_at_line (0, 0, fname, start_line, gettext ("\ | |
96eaef36 | 675 | duplicated message identifier")); |
b6aa34eb UD |
676 | error_at_line (0, 0, runp->fname, runp->line, |
677 | gettext ("this is the first definition")); | |
a641835a RM |
678 | message_number = 0; |
679 | } | |
680 | else | |
681 | /* Give the message the next unused number. */ | |
682 | message_number = ++current->current_set->last_message; | |
683 | } | |
684 | else | |
685 | message_number = 0; | |
686 | ||
687 | if (message_number != 0) | |
688 | { | |
d2defdc4 UD |
689 | char *inbuf; |
690 | size_t inlen; | |
691 | char *outbuf; | |
692 | size_t outlen; | |
a641835a | 693 | struct message_list *newp; |
0a70515e | 694 | size_t line_len = strlen (line) + 1; |
9d37acc4 | 695 | size_t ident_len = 0; |
d2defdc4 UD |
696 | |
697 | /* We need the conversion. */ | |
698 | if (cd_towc == (iconv_t) -1 | |
ee25ee65 UD |
699 | && open_conversion (codeset, &cd_towc, &cd_tomb, |
700 | &escape_char) != 0) | |
d2defdc4 UD |
701 | /* Something is wrong. */ |
702 | goto out; | |
703 | ||
704 | /* Convert to a wide character string. We have to | |
705 | interpret escape sequences which will be impossible | |
706 | without doing the conversion if the codeset of the | |
707 | message is stateful. */ | |
708 | while (1) | |
709 | { | |
0a70515e UD |
710 | inbuf = line; |
711 | inlen = line_len; | |
d2defdc4 UD |
712 | outbuf = (char *) wbuf; |
713 | outlen = wbufsize; | |
714 | ||
715 | /* Flush the state. */ | |
716 | iconv (cd_towc, NULL, NULL, NULL, NULL); | |
717 | ||
718 | iconv (cd_towc, &inbuf, &inlen, &outbuf, &outlen); | |
719 | if (inlen == 0) | |
720 | { | |
721 | /* The string is converted. */ | |
722 | assert (outlen < wbufsize); | |
723 | assert (wbuf[(wbufsize - outlen) / sizeof (wchar_t) - 1] | |
724 | == L'\0'); | |
725 | break; | |
726 | } | |
727 | ||
728 | if (outlen != 0) | |
729 | { | |
730 | /* Something is wrong with this string, we ignore it. */ | |
731 | error_at_line (0, 0, fname, start_line, gettext ("\ | |
732 | invalid character: message ignored")); | |
733 | goto ignore; | |
734 | } | |
735 | ||
736 | /* The output buffer is too small. */ | |
737 | wbufsize *= 2; | |
738 | wbuf = (wchar_t *) xrealloc (wbuf, wbufsize); | |
739 | } | |
a641835a | 740 | |
a641835a RM |
741 | /* Strip quote characters, change escape sequences into |
742 | correct characters etc. */ | |
d2defdc4 | 743 | normalize_line (fname, start_line, cd_towc, wbuf, |
ee25ee65 | 744 | current->quote_char, escape_char); |
a641835a | 745 | |
9d37acc4 UD |
746 | if (ident) |
747 | ident_len = line - this_line; | |
748 | ||
d2defdc4 UD |
749 | /* Now the string is free of escape sequences. Convert it |
750 | back into a multibyte character string. First free the | |
751 | memory allocated for the original string. */ | |
752 | obstack_free (¤t->mem_pool, this_line); | |
753 | ||
0a70515e UD |
754 | used = 1; /* Yes, we use the line. */ |
755 | ||
d2defdc4 UD |
756 | /* Now fill in the new string. It should never happen that |
757 | the replaced string is longer than the original. */ | |
758 | inbuf = (char *) wbuf; | |
759 | inlen = (wcslen (wbuf) + 1) * sizeof (wchar_t); | |
760 | ||
761 | outlen = obstack_room (¤t->mem_pool); | |
0a70515e UD |
762 | obstack_blank (¤t->mem_pool, outlen); |
763 | this_line = (char *) obstack_base (¤t->mem_pool); | |
9d37acc4 UD |
764 | outbuf = this_line + ident_len; |
765 | outlen -= ident_len; | |
d2defdc4 UD |
766 | |
767 | /* Flush the state. */ | |
768 | iconv (cd_tomb, NULL, NULL, NULL, NULL); | |
769 | ||
770 | iconv (cd_tomb, &inbuf, &inlen, &outbuf, &outlen); | |
771 | if (inlen != 0) | |
772 | { | |
773 | error_at_line (0, 0, fname, start_line, | |
774 | gettext ("invalid line")); | |
775 | goto ignore; | |
776 | } | |
777 | assert (outbuf[-1] == '\0'); | |
778 | ||
779 | /* Free the memory in the obstack we don't use. */ | |
0a70515e UD |
780 | obstack_blank (¤t->mem_pool, -(int) outlen); |
781 | line = obstack_finish (¤t->mem_pool); | |
d2defdc4 | 782 | |
a641835a RM |
783 | newp = (struct message_list *) xmalloc (sizeof (*newp)); |
784 | newp->number = message_number; | |
9d37acc4 | 785 | newp->message = line + ident_len; |
a641835a | 786 | /* Remember symbolic name; is NULL if no is given. */ |
9d37acc4 | 787 | newp->symbol = ident ? line : NULL; |
a641835a RM |
788 | /* Remember where we found the character. */ |
789 | newp->fname = fname; | |
790 | newp->line = start_line; | |
791 | ||
792 | /* Find place to insert to message. We keep them in a | |
793 | sorted single linked list. */ | |
794 | if (current->current_set->messages == NULL | |
795 | || current->current_set->messages->number > message_number) | |
796 | { | |
797 | newp->next = current->current_set->messages; | |
798 | current->current_set->messages = newp; | |
799 | } | |
800 | else | |
801 | { | |
802 | struct message_list *runp; | |
803 | runp = current->current_set->messages; | |
804 | while (runp->next != NULL) | |
805 | if (runp->next->number > message_number) | |
806 | break; | |
807 | else | |
808 | runp = runp->next; | |
809 | newp->next = runp->next; | |
810 | runp->next = newp; | |
811 | } | |
812 | } | |
813 | ++current->total_messages; | |
814 | } | |
815 | else | |
816 | { | |
817 | size_t cnt; | |
818 | ||
819 | cnt = 0; | |
820 | /* See whether we have any non-white space character in this | |
821 | line. */ | |
822 | while (this_line[cnt] != '\0' && isspace (this_line[cnt])) | |
823 | ++cnt; | |
824 | ||
825 | if (this_line[cnt] != '\0') | |
826 | /* Yes, some unknown characters found. */ | |
827 | error_at_line (0, 0, fname, start_line, | |
828 | gettext ("malformed line ignored")); | |
829 | } | |
830 | ||
d2defdc4 | 831 | ignore: |
a641835a RM |
832 | /* We can save the memory for the line if it was not used. */ |
833 | if (!used) | |
834 | obstack_free (¤t->mem_pool, this_line); | |
835 | } | |
836 | ||
d2defdc4 UD |
837 | /* Close the conversion modules. */ |
838 | iconv_close (cd_towc); | |
839 | iconv_close (cd_tomb); | |
840 | free (codeset); | |
841 | ||
842 | out: | |
843 | free (wbuf); | |
844 | ||
a641835a RM |
845 | if (fp != stdin) |
846 | fclose (fp); | |
847 | return current; | |
848 | } | |
849 | ||
850 | ||
851 | static void | |
852 | write_out (struct catalog *catalog, const char *output_name, | |
853 | const char *header_name) | |
854 | { | |
855 | /* Computing the "optimal" size. */ | |
856 | struct set_list *set_run; | |
857 | size_t best_total, best_size, best_depth; | |
858 | size_t act_size, act_depth; | |
859 | struct catalog_obj obj; | |
860 | struct obstack string_pool; | |
861 | const char *strings; | |
862 | size_t strings_size; | |
b6aa34eb | 863 | uint32_t *array1, *array2; |
a641835a RM |
864 | size_t cnt; |
865 | int fd; | |
866 | ||
867 | /* If not otherwise told try to read file with existing | |
868 | translations. */ | |
869 | if (!force_new) | |
870 | read_old (catalog, output_name); | |
871 | ||
872 | /* Initialize best_size with a very high value. */ | |
873 | best_total = best_size = best_depth = UINT_MAX; | |
874 | ||
875 | /* We need some start size for testing. Let's start with | |
876 | TOTAL_MESSAGES / 5, which theoretically provides a mean depth of | |
877 | 5. */ | |
878 | act_size = 1 + catalog->total_messages / 5; | |
879 | ||
880 | /* We determine the size of a hash table here. Because the message | |
881 | numbers can be chosen arbitrary by the programmer we cannot use | |
882 | the simple method of accessing the array using the message | |
883 | number. The algorithm is based on the trivial hash function | |
884 | NUMBER % TABLE_SIZE, where collisions are stored in a second | |
885 | dimension up to TABLE_DEPTH. We here compute TABLE_SIZE so that | |
886 | the needed space (= TABLE_SIZE * TABLE_DEPTH) is minimal. */ | |
887 | while (act_size <= best_total) | |
888 | { | |
889 | size_t deep[act_size]; | |
890 | ||
891 | act_depth = 1; | |
892 | memset (deep, '\0', act_size * sizeof (size_t)); | |
893 | set_run = catalog->all_sets; | |
894 | while (set_run != NULL) | |
895 | { | |
896 | struct message_list *message_run; | |
897 | ||
898 | message_run = set_run->messages; | |
899 | while (message_run != NULL) | |
900 | { | |
901 | size_t idx = (message_run->number * set_run->number) % act_size; | |
902 | ||
903 | ++deep[idx]; | |
904 | if (deep[idx] > act_depth) | |
905 | { | |
906 | act_depth = deep[idx]; | |
907 | if (act_depth * act_size > best_total) | |
908 | break; | |
909 | } | |
910 | message_run = message_run->next; | |
911 | } | |
a641835a RM |
912 | set_run = set_run->next; |
913 | } | |
914 | ||
adc6ff7f RM |
915 | if (act_depth * act_size <= best_total) |
916 | { | |
917 | /* We have found a better solution. */ | |
918 | best_total = act_depth * act_size; | |
919 | best_size = act_size; | |
920 | best_depth = act_depth; | |
921 | } | |
922 | ||
a641835a RM |
923 | ++act_size; |
924 | } | |
925 | ||
926 | /* let's be prepared for an empty message file. */ | |
927 | if (best_size == UINT_MAX) | |
928 | { | |
929 | best_size = 1; | |
930 | best_depth = 1; | |
931 | } | |
932 | ||
933 | /* OK, now we have the size we will use. Fill in the header, build | |
934 | the table and the second one with swapped byte order. */ | |
935 | obj.magic = CATGETS_MAGIC; | |
936 | obj.plane_size = best_size; | |
937 | obj.plane_depth = best_depth; | |
938 | ||
939 | /* Allocate room for all needed arrays. */ | |
940 | array1 = | |
b6aa34eb UD |
941 | (uint32_t *) alloca (best_size * best_depth * sizeof (uint32_t) * 3); |
942 | memset (array1, '\0', best_size * best_depth * sizeof (uint32_t) * 3); | |
a641835a | 943 | array2 |
b6aa34eb | 944 | = (uint32_t *) alloca (best_size * best_depth * sizeof (uint32_t) * 3); |
a641835a RM |
945 | obstack_init (&string_pool); |
946 | ||
947 | set_run = catalog->all_sets; | |
948 | while (set_run != NULL) | |
949 | { | |
950 | struct message_list *message_run; | |
951 | ||
952 | message_run = set_run->messages; | |
953 | while (message_run != NULL) | |
954 | { | |
955 | size_t idx = (((message_run->number * set_run->number) % best_size) | |
956 | * 3); | |
957 | /* Determine collision depth. */ | |
958 | while (array1[idx] != 0) | |
959 | idx += best_size * 3; | |
960 | ||
961 | /* Store set number, message number and pointer into string | |
962 | space, relative to the first string. */ | |
963 | array1[idx + 0] = set_run->number; | |
964 | array1[idx + 1] = message_run->number; | |
965 | array1[idx + 2] = obstack_object_size (&string_pool); | |
966 | ||
967 | /* Add current string to the continuous space containing all | |
968 | strings. */ | |
969 | obstack_grow0 (&string_pool, message_run->message, | |
970 | strlen (message_run->message)); | |
971 | ||
972 | message_run = message_run->next; | |
973 | } | |
974 | ||
975 | set_run = set_run->next; | |
976 | } | |
977 | strings_size = obstack_object_size (&string_pool); | |
978 | strings = obstack_finish (&string_pool); | |
979 | ||
980 | /* Compute ARRAY2 by changing the byte order. */ | |
981 | for (cnt = 0; cnt < best_size * best_depth * 3; ++cnt) | |
982 | array2[cnt] = SWAPU32 (array1[cnt]); | |
983 | ||
984 | /* Now we can write out the whole data. */ | |
985 | if (strcmp (output_name, "-") == 0 | |
986 | || strcmp (output_name, "/dev/stdout") == 0) | |
987 | fd = STDOUT_FILENO; | |
988 | else | |
989 | { | |
990 | fd = creat (output_name, 0666); | |
991 | if (fd < 0) | |
992 | error (EXIT_FAILURE, errno, gettext ("cannot open output file `%s'"), | |
993 | output_name); | |
994 | } | |
995 | ||
996 | /* Write out header. */ | |
997 | write (fd, &obj, sizeof (obj)); | |
998 | ||
999 | /* We always write out the little endian version of the index | |
1000 | arrays. */ | |
1001 | #if __BYTE_ORDER == __LITTLE_ENDIAN | |
b6aa34eb UD |
1002 | write (fd, array1, best_size * best_depth * sizeof (uint32_t) * 3); |
1003 | write (fd, array2, best_size * best_depth * sizeof (uint32_t) * 3); | |
a641835a | 1004 | #elif __BYTE_ORDER == __BIG_ENDIAN |
b6aa34eb UD |
1005 | write (fd, array2, best_size * best_depth * sizeof (uint32_t) * 3); |
1006 | write (fd, array1, best_size * best_depth * sizeof (uint32_t) * 3); | |
a641835a RM |
1007 | #else |
1008 | # error Cannot handle __BYTE_ORDER byte order | |
1009 | #endif | |
1010 | ||
1011 | /* Finally write the strings. */ | |
1012 | write (fd, strings, strings_size); | |
1013 | ||
1014 | if (fd != STDOUT_FILENO) | |
1015 | close (fd); | |
1016 | ||
1017 | /* If requested now write out the header file. */ | |
1018 | if (header_name != NULL) | |
1019 | { | |
1020 | int first = 1; | |
1021 | FILE *fp; | |
1022 | ||
1023 | /* Open output file. "-" or "/dev/stdout" means write to | |
1024 | standard output. */ | |
1025 | if (strcmp (header_name, "-") == 0 | |
1026 | || strcmp (header_name, "/dev/stdout") == 0) | |
1027 | fp = stdout; | |
1028 | else | |
1029 | { | |
1030 | fp = fopen (header_name, "w"); | |
1031 | if (fp == NULL) | |
1032 | error (EXIT_FAILURE, errno, | |
1033 | gettext ("cannot open output file `%s'"), header_name); | |
1034 | } | |
1035 | ||
1036 | /* Iterate over all sets and all messages. */ | |
1037 | set_run = catalog->all_sets; | |
1038 | while (set_run != NULL) | |
1039 | { | |
1040 | struct message_list *message_run; | |
1041 | ||
1042 | /* If the current message set has a symbolic name write this | |
1043 | out first. */ | |
1044 | if (set_run->symbol != NULL) | |
a4242e25 | 1045 | fprintf (fp, "%s#define %sSet %#x\t/* %s:%Zu */\n", |
a641835a RM |
1046 | first ? "" : "\n", set_run->symbol, set_run->number - 1, |
1047 | set_run->fname, set_run->line); | |
1048 | first = 0; | |
1049 | ||
1050 | message_run = set_run->messages; | |
1051 | while (message_run != NULL) | |
1052 | { | |
1053 | /* If the current message has a symbolic name write | |
1054 | #define out. But we have to take care for the set | |
1055 | not having a symbolic name. */ | |
1056 | if (message_run->symbol != NULL) | |
6e4c40ba UD |
1057 | { |
1058 | if (set_run->symbol == NULL) | |
1059 | fprintf (fp, "#define AutomaticSet%d%s %#x\t/* %s:%Zu */\n", | |
1060 | set_run->number, message_run->symbol, | |
1061 | message_run->number, message_run->fname, | |
1062 | message_run->line); | |
1063 | else | |
1064 | fprintf (fp, "#define %s%s %#x\t/* %s:%Zu */\n", | |
1065 | set_run->symbol, message_run->symbol, | |
1066 | message_run->number, message_run->fname, | |
1067 | message_run->line); | |
1068 | } | |
a641835a RM |
1069 | |
1070 | message_run = message_run->next; | |
1071 | } | |
1072 | ||
1073 | set_run = set_run->next; | |
1074 | } | |
1075 | ||
1076 | if (fp != stdout) | |
1077 | fclose (fp); | |
1078 | } | |
1079 | } | |
1080 | ||
1081 | ||
1082 | static struct set_list * | |
1083 | find_set (struct catalog *current, int number) | |
1084 | { | |
1085 | struct set_list *result = current->all_sets; | |
1086 | ||
1087 | /* We must avoid set number 0 because a set of this number signals | |
1088 | in the tables that the entry is not occupied. */ | |
1089 | ++number; | |
1090 | ||
1091 | while (result != NULL) | |
1092 | if (result->number == number) | |
1093 | return result; | |
1094 | else | |
1095 | result = result->next; | |
1096 | ||
1097 | /* Prepare new message set. */ | |
a3b2008a | 1098 | result = (struct set_list *) xcalloc (1, sizeof (*result)); |
a641835a | 1099 | result->number = number; |
a641835a RM |
1100 | result->next = current->all_sets; |
1101 | current->all_sets = result; | |
1102 | ||
1103 | return result; | |
1104 | } | |
1105 | ||
1106 | ||
1107 | /* Normalize given string *in*place* by processing escape sequences | |
1108 | and quote characters. */ | |
1109 | static void | |
d2defdc4 | 1110 | normalize_line (const char *fname, size_t line, iconv_t cd, wchar_t *string, |
ee25ee65 | 1111 | wchar_t quote_char, wchar_t escape_char) |
a641835a RM |
1112 | { |
1113 | int is_quoted; | |
d2defdc4 UD |
1114 | wchar_t *rp = string; |
1115 | wchar_t *wp = string; | |
a641835a | 1116 | |
d2defdc4 | 1117 | if (quote_char != L'\0' && *rp == quote_char) |
a641835a RM |
1118 | { |
1119 | is_quoted = 1; | |
1120 | ++rp; | |
1121 | } | |
1122 | else | |
1123 | is_quoted = 0; | |
1124 | ||
d2defdc4 | 1125 | while (*rp != L'\0') |
a641835a RM |
1126 | if (*rp == quote_char) |
1127 | /* We simply end the string when we find the first time an | |
1128 | not-escaped quote character. */ | |
1129 | break; | |
ee25ee65 | 1130 | else if (*rp == escape_char) |
a641835a RM |
1131 | { |
1132 | ++rp; | |
d2defdc4 | 1133 | if (quote_char != L'\0' && *rp == quote_char) |
a641835a RM |
1134 | /* This is an extension to XPG. */ |
1135 | *wp++ = *rp++; | |
1136 | else | |
1137 | /* Recognize escape sequences. */ | |
1138 | switch (*rp) | |
1139 | { | |
d2defdc4 UD |
1140 | case L'n': |
1141 | *wp++ = L'\n'; | |
a641835a RM |
1142 | ++rp; |
1143 | break; | |
d2defdc4 UD |
1144 | case L't': |
1145 | *wp++ = L'\t'; | |
a641835a RM |
1146 | ++rp; |
1147 | break; | |
d2defdc4 UD |
1148 | case L'v': |
1149 | *wp++ = L'\v'; | |
a641835a RM |
1150 | ++rp; |
1151 | break; | |
d2defdc4 UD |
1152 | case L'b': |
1153 | *wp++ = L'\b'; | |
a641835a RM |
1154 | ++rp; |
1155 | break; | |
d2defdc4 UD |
1156 | case L'r': |
1157 | *wp++ = L'\r'; | |
a641835a RM |
1158 | ++rp; |
1159 | break; | |
d2defdc4 UD |
1160 | case L'f': |
1161 | *wp++ = L'\f'; | |
a641835a RM |
1162 | ++rp; |
1163 | break; | |
d2defdc4 | 1164 | case L'0' ... L'7': |
a641835a | 1165 | { |
d2defdc4 UD |
1166 | int number; |
1167 | char cbuf[2]; | |
1168 | char *cbufptr; | |
1169 | size_t cbufin; | |
1170 | wchar_t wcbuf[2]; | |
1171 | char *wcbufptr; | |
1172 | size_t wcbufin; | |
1173 | ||
1174 | number = *rp++ - L'0'; | |
1175 | while (number <= (255 / 8) && *rp >= L'0' && *rp <= L'7') | |
a641835a RM |
1176 | { |
1177 | number *= 8; | |
d2defdc4 | 1178 | number += *rp++ - L'0'; |
a641835a | 1179 | } |
d2defdc4 UD |
1180 | |
1181 | cbuf[0] = (char) number; | |
1182 | cbuf[1] = '\0'; | |
1183 | cbufptr = cbuf; | |
1184 | cbufin = 2; | |
1185 | ||
1186 | wcbufptr = (char *) wcbuf; | |
1187 | wcbufin = sizeof (wcbuf); | |
1188 | ||
1189 | /* Flush the state. */ | |
1190 | iconv (cd, NULL, NULL, NULL, NULL); | |
1191 | ||
1192 | iconv (cd, &cbufptr, &cbufin, &wcbufptr, &wcbufin); | |
1193 | if (cbufptr != &cbuf[2] || (wchar_t *) wcbufptr != &wcbuf[2]) | |
1194 | error_at_line (0, 0, fname, line, | |
1195 | gettext ("invalid escape sequence")); | |
1196 | else | |
1197 | *wp++ = wcbuf[0]; | |
a641835a RM |
1198 | } |
1199 | break; | |
1200 | default: | |
ee25ee65 UD |
1201 | if (*rp == escape_char) |
1202 | { | |
1203 | *wp++ = escape_char; | |
1204 | ++rp; | |
1205 | } | |
1206 | else | |
1207 | /* Simply ignore the backslash character. */; | |
a641835a RM |
1208 | break; |
1209 | } | |
1210 | } | |
1211 | else | |
1212 | *wp++ = *rp++; | |
1213 | ||
1214 | /* If we saw a quote character at the beginning we expect another | |
1215 | one at the end. */ | |
1216 | if (is_quoted && *rp != quote_char) | |
d2defdc4 | 1217 | error_at_line (0, 0, fname, line, gettext ("unterminated message")); |
a641835a RM |
1218 | |
1219 | /* Terminate string. */ | |
d2defdc4 | 1220 | *wp = L'\0'; |
a641835a RM |
1221 | return; |
1222 | } | |
1223 | ||
1224 | ||
1225 | static void | |
1226 | read_old (struct catalog *catalog, const char *file_name) | |
1227 | { | |
1228 | struct catalog_info old_cat_obj; | |
1229 | struct set_list *set = NULL; | |
1230 | int last_set = -1; | |
1231 | size_t cnt; | |
1232 | ||
a641835a | 1233 | /* Try to open catalog, but don't look through the NLSPATH. */ |
ca130fe4 | 1234 | if (__open_catalog (file_name, NULL, NULL, &old_cat_obj) != 0) |
6e4c40ba UD |
1235 | { |
1236 | if (errno == ENOENT) | |
1237 | /* No problem, the catalog simply does not exist. */ | |
1238 | return; | |
1239 | else | |
ca130fe4 UD |
1240 | error (EXIT_FAILURE, errno, |
1241 | gettext ("while opening old catalog file")); | |
6e4c40ba | 1242 | } |
a641835a RM |
1243 | |
1244 | /* OK, we have the catalog loaded. Now read all messages and merge | |
1245 | them. When set and message number clash for any message the new | |
b6aa34eb UD |
1246 | one is used. If the new one is empty it indicates that the |
1247 | message should be deleted. */ | |
a641835a RM |
1248 | for (cnt = 0; cnt < old_cat_obj.plane_size * old_cat_obj.plane_depth; ++cnt) |
1249 | { | |
1250 | struct message_list *message, *last; | |
1251 | ||
1252 | if (old_cat_obj.name_ptr[cnt * 3 + 0] == 0) | |
1253 | /* No message in this slot. */ | |
1254 | continue; | |
1255 | ||
b6aa34eb | 1256 | if (old_cat_obj.name_ptr[cnt * 3 + 0] - 1 != (uint32_t) last_set) |
a641835a RM |
1257 | { |
1258 | last_set = old_cat_obj.name_ptr[cnt * 3 + 0] - 1; | |
1259 | set = find_set (catalog, old_cat_obj.name_ptr[cnt * 3 + 0] - 1); | |
1260 | } | |
1261 | ||
1262 | last = NULL; | |
1263 | message = set->messages; | |
1264 | while (message != NULL) | |
1265 | { | |
b6aa34eb | 1266 | if ((uint32_t) message->number >= old_cat_obj.name_ptr[cnt * 3 + 1]) |
a641835a RM |
1267 | break; |
1268 | last = message; | |
1269 | message = message->next; | |
1270 | } | |
1271 | ||
1272 | if (message == NULL | |
b6aa34eb | 1273 | || (uint32_t) message->number > old_cat_obj.name_ptr[cnt * 3 + 1]) |
a641835a RM |
1274 | { |
1275 | /* We have found a message which is not yet in the catalog. | |
1276 | Insert it at the right position. */ | |
1277 | struct message_list *newp; | |
1278 | ||
1279 | newp = (struct message_list *) xmalloc (sizeof(*newp)); | |
1280 | newp->number = old_cat_obj.name_ptr[cnt * 3 + 1]; | |
1281 | newp->message = | |
1282 | &old_cat_obj.strings[old_cat_obj.name_ptr[cnt * 3 + 2]]; | |
1283 | newp->fname = NULL; | |
1284 | newp->line = 0; | |
1285 | newp->symbol = NULL; | |
1286 | newp->next = message; | |
1287 | ||
1288 | if (last == NULL) | |
1289 | set->messages = newp; | |
1290 | else | |
1291 | last->next = newp; | |
1292 | ||
1293 | ++catalog->total_messages; | |
1294 | } | |
b6aa34eb UD |
1295 | else if (*message->message == '\0') |
1296 | { | |
1297 | /* The new empty message has overridden the old one thus | |
1298 | "deleting" it as required. Now remove the empty remains. */ | |
1299 | if (last == NULL) | |
1300 | set->messages = message->next; | |
1301 | else | |
1302 | last->next = message->next; | |
1303 | } | |
a641835a RM |
1304 | } |
1305 | } | |
d2defdc4 UD |
1306 | |
1307 | ||
1308 | static int | |
ee25ee65 UD |
1309 | open_conversion (const char *codeset, iconv_t *cd_towcp, iconv_t *cd_tombp, |
1310 | wchar_t *escape_charp) | |
d2defdc4 | 1311 | { |
ee25ee65 UD |
1312 | char buf[2]; |
1313 | char *bufptr; | |
1314 | size_t bufsize; | |
1315 | wchar_t wbuf[2]; | |
1316 | char *wbufptr; | |
1317 | size_t wbufsize; | |
1318 | ||
d2defdc4 UD |
1319 | /* If the input file does not specify the codeset use the locale's. */ |
1320 | if (codeset == NULL) | |
1321 | { | |
1322 | setlocale (LC_ALL, ""); | |
1323 | codeset = nl_langinfo (CODESET); | |
1324 | setlocale (LC_ALL, "C"); | |
1325 | } | |
1326 | ||
1327 | /* Get the conversion modules. */ | |
1328 | *cd_towcp = iconv_open ("WCHAR_T", codeset); | |
1329 | *cd_tombp = iconv_open (codeset, "WCHAR_T"); | |
1330 | if (*cd_towcp == (iconv_t) -1 || *cd_tombp == (iconv_t) -1) | |
1331 | { | |
1332 | error (0, 0, gettext ("conversion modules not available")); | |
1333 | if (*cd_towcp != (iconv_t) -1) | |
1334 | iconv_close (*cd_towcp); | |
1335 | ||
1336 | return 1; | |
1337 | } | |
1338 | ||
ee25ee65 UD |
1339 | /* One special case for historical reasons is the backslash |
1340 | character. In some codesets the byte value 0x5c is not mapped to | |
1341 | U005c in Unicode. These charsets then don't have a backslash | |
1342 | character at all. Therefore we have to live with whatever the | |
1343 | codeset provides and recognize, instead of the U005c, the character | |
1344 | the byte value 0x5c is mapped to. */ | |
1345 | buf[0] = '\\'; | |
1346 | buf[1] = '\0'; | |
1347 | bufptr = buf; | |
1348 | bufsize = 2; | |
1349 | ||
1350 | wbufptr = (char *) wbuf; | |
1351 | wbufsize = sizeof (wbuf); | |
1352 | ||
1353 | iconv (*cd_towcp, &bufptr, &bufsize, &wbufptr, &wbufsize); | |
1354 | if (bufsize != 0 || wbufsize != 0) | |
1355 | { | |
1356 | /* Something went wrong, we couldn't convert the byte 0x5c. Go | |
1357 | on with using U005c. */ | |
1358 | error (0, 0, gettext ("cannot determine escape character")); | |
1359 | *escape_charp = L'\\'; | |
1360 | } | |
1361 | else | |
1362 | *escape_charp = wbuf[0]; | |
1363 | ||
d2defdc4 UD |
1364 | return 0; |
1365 | } |