]>
Commit | Line | Data |
---|---|---|
01ff9d0b | 1 | /* Copyright (C) 1995-1999, 2000 Free Software Foundation, Inc. |
c84142e8 | 2 | This file is part of the GNU C Library. |
4b10dd6c | 3 | Contributed by Ulrich Drepper <drepper@gnu.org>, 1995. |
19bc17a9 | 4 | |
c84142e8 UD |
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. | |
19bc17a9 | 9 | |
c84142e8 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 | |
13 | Library General Public License for more details. | |
19bc17a9 | 14 | |
c84142e8 UD |
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 not, | |
17 | write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
18 | Boston, MA 02111-1307, USA. */ | |
19bc17a9 RM |
19 | |
20 | #ifdef HAVE_CONFIG_H | |
21 | # include <config.h> | |
22 | #endif | |
23 | ||
a68b0d31 | 24 | #include <alloca.h> |
4b10dd6c | 25 | #include <byteswap.h> |
19bc17a9 | 26 | #include <endian.h> |
4b10dd6c | 27 | #include <errno.h> |
19bc17a9 | 28 | #include <limits.h> |
4b10dd6c UD |
29 | #include <obstack.h> |
30 | #include <stdlib.h> | |
19bc17a9 | 31 | #include <string.h> |
4b10dd6c UD |
32 | #include <wchar.h> |
33 | #include <wctype.h> | |
34 | #include <sys/uio.h> | |
19bc17a9 | 35 | |
4b10dd6c | 36 | #include "charmap.h" |
19bc17a9 RM |
37 | #include "localeinfo.h" |
38 | #include "langinfo.h" | |
4b10dd6c | 39 | #include "linereader.h" |
19bc17a9 | 40 | #include "locfile-token.h" |
4b10dd6c UD |
41 | #include "locfile.h" |
42 | #include "localedef.h" | |
19bc17a9 | 43 | |
19bc17a9 RM |
44 | #include <assert.h> |
45 | ||
46 | ||
011ebfab | 47 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
48 | /* These are the extra bits not in wctype.h since these are not preallocated |
49 | classes. */ | |
011ebfab UD |
50 | # define _ISwspecial1 (1 << 29) |
51 | # define _ISwspecial2 (1 << 30) | |
52 | # define _ISwspecial3 (1 << 31) | |
53 | #endif | |
19bc17a9 RM |
54 | |
55 | ||
56 | /* The bit used for representing a special class. */ | |
57 | #define BITPOS(class) ((class) - tok_upper) | |
4b10dd6c UD |
58 | #define BIT(class) (_ISbit (BITPOS (class))) |
59 | #define BITw(class) (_ISwbit (BITPOS (class))) | |
19bc17a9 RM |
60 | |
61 | #define ELEM(ctype, collection, idx, value) \ | |
62 | *find_idx (ctype, &ctype->collection idx, &ctype->collection##_max idx, \ | |
63 | &ctype->collection##_act idx, value) | |
64 | ||
19bc17a9 RM |
65 | |
66 | /* To be compatible with former implementations we for now restrict | |
67 | the number of bits for character classes to 16. When compatibility | |
68 | is not necessary anymore increase the number to 32. */ | |
4b10dd6c | 69 | #define char_class_t uint16_t |
4b10dd6c | 70 | #define char_class32_t uint32_t |
4b10dd6c UD |
71 | |
72 | ||
73 | /* Type to describe a transliteration action. We have a possibly | |
74 | multiple character from-string and a set of multiple character | |
75 | to-strings. All are 32bit values since this is what is used in | |
76 | the gconv functions. */ | |
77 | struct translit_to_t | |
78 | { | |
79 | uint32_t *str; | |
80 | ||
81 | struct translit_to_t *next; | |
82 | }; | |
83 | ||
84 | struct translit_t | |
85 | { | |
86 | uint32_t *from; | |
87 | ||
a673fbcb UD |
88 | const char *fname; |
89 | size_t lineno; | |
90 | ||
4b10dd6c UD |
91 | struct translit_to_t *to; |
92 | ||
93 | struct translit_t *next; | |
94 | }; | |
19bc17a9 | 95 | |
a673fbcb UD |
96 | struct translit_ignore_t |
97 | { | |
98 | uint32_t from; | |
99 | uint32_t to; | |
a0dc5206 | 100 | uint32_t step; |
a673fbcb UD |
101 | |
102 | const char *fname; | |
103 | size_t lineno; | |
104 | ||
105 | struct translit_ignore_t *next; | |
106 | }; | |
107 | ||
19bc17a9 RM |
108 | |
109 | /* The real definition of the struct for the LC_CTYPE locale. */ | |
110 | struct locale_ctype_t | |
111 | { | |
4b10dd6c | 112 | uint32_t *charnames; |
19bc17a9 RM |
113 | size_t charnames_max; |
114 | size_t charnames_act; | |
115 | ||
4b10dd6c UD |
116 | struct repertoire_t *repertoire; |
117 | ||
118 | /* We will allow up to 8 * sizeof (uint32_t) character classes. */ | |
119 | #define MAX_NR_CHARCLASS (8 * sizeof (uint32_t)) | |
ba1ffaa1 | 120 | size_t nr_charclass; |
19bc17a9 | 121 | const char *classnames[MAX_NR_CHARCLASS]; |
4b10dd6c UD |
122 | uint32_t last_class_char; |
123 | uint32_t class256_collection[256]; | |
124 | uint32_t *class_collection; | |
19bc17a9 RM |
125 | size_t class_collection_max; |
126 | size_t class_collection_act; | |
4b10dd6c UD |
127 | uint32_t class_done; |
128 | ||
129 | struct charseq **mbdigits; | |
130 | size_t mbdigits_act; | |
131 | size_t mbdigits_max; | |
132 | uint32_t *wcdigits; | |
133 | size_t wcdigits_act; | |
134 | size_t wcdigits_max; | |
135 | ||
136 | struct charseq *mboutdigits[10]; | |
137 | uint32_t wcoutdigits[10]; | |
138 | size_t outdigits_act; | |
19bc17a9 RM |
139 | |
140 | /* If the following number ever turns out to be too small simply | |
141 | increase it. But I doubt it will. --drepper@gnu */ | |
142 | #define MAX_NR_CHARMAP 16 | |
143 | const char *mapnames[MAX_NR_CHARMAP]; | |
4b10dd6c UD |
144 | uint32_t *map_collection[MAX_NR_CHARMAP]; |
145 | uint32_t map256_collection[2][256]; | |
9a0a462c UD |
146 | size_t map_collection_max[MAX_NR_CHARMAP]; |
147 | size_t map_collection_act[MAX_NR_CHARMAP]; | |
19bc17a9 RM |
148 | size_t map_collection_nr; |
149 | size_t last_map_idx; | |
4b10dd6c UD |
150 | int tomap_done[MAX_NR_CHARMAP]; |
151 | ||
152 | /* Transliteration information. */ | |
153 | const char *translit_copy_locale; | |
154 | const char *translit_copy_repertoire; | |
155 | struct translit_t *translit; | |
a673fbcb | 156 | struct translit_ignore_t *translit_ignore; |
a8e4c924 | 157 | uint32_t ntranslit_ignore; |
a673fbcb UD |
158 | |
159 | uint32_t *default_missing; | |
160 | const char *default_missing_file; | |
161 | size_t default_missing_lineno; | |
19bc17a9 RM |
162 | |
163 | /* The arrays for the binary representation. */ | |
4b10dd6c UD |
164 | uint32_t plane_size; |
165 | uint32_t plane_cnt; | |
19bc17a9 RM |
166 | char_class_t *ctype_b; |
167 | char_class32_t *ctype32_b; | |
4a33c2f5 UD |
168 | uint32_t *names; |
169 | uint32_t **map; | |
49f2be5b | 170 | uint32_t **map32; |
4b10dd6c UD |
171 | uint32_t *class_name_ptr; |
172 | uint32_t *map_name_ptr; | |
75cd5204 | 173 | unsigned char *width; |
4b10dd6c | 174 | uint32_t mb_cur_max; |
6990326c | 175 | const char *codeset_name; |
4a33c2f5 UD |
176 | uint32_t *translit_from_idx; |
177 | uint32_t *translit_from_tbl; | |
178 | uint32_t *translit_to_idx; | |
179 | uint32_t *translit_to_tbl; | |
04fbc779 | 180 | uint32_t translit_idx_size; |
4b10dd6c UD |
181 | size_t translit_from_tbl_size; |
182 | size_t translit_to_tbl_size; | |
183 | ||
a673fbcb | 184 | struct obstack mempool; |
19bc17a9 RM |
185 | }; |
186 | ||
187 | ||
4b10dd6c UD |
188 | #define obstack_chunk_alloc xmalloc |
189 | #define obstack_chunk_free free | |
190 | ||
191 | ||
19bc17a9 | 192 | /* Prototypes for local functions. */ |
4b10dd6c UD |
193 | static void ctype_startup (struct linereader *lr, struct localedef_t *locale, |
194 | struct charmap_t *charmap, int ignore_content); | |
195 | static void ctype_class_new (struct linereader *lr, | |
196 | struct locale_ctype_t *ctype, const char *name); | |
197 | static void ctype_map_new (struct linereader *lr, | |
198 | struct locale_ctype_t *ctype, | |
199 | const char *name, struct charmap_t *charmap); | |
200 | static uint32_t *find_idx (struct locale_ctype_t *ctype, uint32_t **table, | |
201 | size_t *max, size_t *act, unsigned int idx); | |
19bc17a9 | 202 | static void set_class_defaults (struct locale_ctype_t *ctype, |
4b10dd6c UD |
203 | struct charmap_t *charmap, |
204 | struct repertoire_t *repertoire); | |
75cd5204 | 205 | static void allocate_arrays (struct locale_ctype_t *ctype, |
4b10dd6c UD |
206 | struct charmap_t *charmap, |
207 | struct repertoire_t *repertoire); | |
19bc17a9 RM |
208 | |
209 | ||
4b10dd6c UD |
210 | static const char *longnames[] = |
211 | { | |
212 | "zero", "one", "two", "three", "four", | |
213 | "five", "six", "seven", "eight", "nine" | |
214 | }; | |
1b97149d UD |
215 | static const char *uninames[] = |
216 | { | |
217 | "U00000030", "U00000031", "U00000032", "U00000033", "U00000034", | |
218 | "U00000035", "U00000036", "U00000037", "U00000038", "U00000039" | |
219 | }; | |
4b10dd6c UD |
220 | static const unsigned char digits[] = "0123456789"; |
221 | ||
222 | ||
223 | static void | |
19bc17a9 | 224 | ctype_startup (struct linereader *lr, struct localedef_t *locale, |
4b10dd6c | 225 | struct charmap_t *charmap, int ignore_content) |
19bc17a9 RM |
226 | { |
227 | unsigned int cnt; | |
228 | struct locale_ctype_t *ctype; | |
229 | ||
4b10dd6c | 230 | if (!ignore_content) |
19bc17a9 | 231 | { |
4b10dd6c UD |
232 | /* Allocate the needed room. */ |
233 | locale->categories[LC_CTYPE].ctype = ctype = | |
234 | (struct locale_ctype_t *) xcalloc (1, sizeof (struct locale_ctype_t)); | |
235 | ||
236 | /* We have seen no names yet. */ | |
237 | ctype->charnames_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
238 | ctype->charnames = | |
239 | (unsigned int *) xmalloc (ctype->charnames_max | |
240 | * sizeof (unsigned int)); | |
241 | for (cnt = 0; cnt < 256; ++cnt) | |
242 | ctype->charnames[cnt] = cnt; | |
243 | ctype->charnames_act = 256; | |
244 | ||
245 | /* Fill character class information. */ | |
246 | ctype->last_class_char = ILLEGAL_CHAR_VALUE; | |
247 | /* The order of the following instructions determines the bit | |
248 | positions! */ | |
249 | ctype_class_new (lr, ctype, "upper"); | |
250 | ctype_class_new (lr, ctype, "lower"); | |
251 | ctype_class_new (lr, ctype, "alpha"); | |
252 | ctype_class_new (lr, ctype, "digit"); | |
253 | ctype_class_new (lr, ctype, "xdigit"); | |
254 | ctype_class_new (lr, ctype, "space"); | |
255 | ctype_class_new (lr, ctype, "print"); | |
256 | ctype_class_new (lr, ctype, "graph"); | |
257 | ctype_class_new (lr, ctype, "blank"); | |
258 | ctype_class_new (lr, ctype, "cntrl"); | |
259 | ctype_class_new (lr, ctype, "punct"); | |
260 | ctype_class_new (lr, ctype, "alnum"); | |
011ebfab | 261 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
262 | /* The following are extensions from ISO 14652. */ |
263 | ctype_class_new (lr, ctype, "left_to_right"); | |
264 | ctype_class_new (lr, ctype, "right_to_left"); | |
265 | ctype_class_new (lr, ctype, "num_terminator"); | |
266 | ctype_class_new (lr, ctype, "num_separator"); | |
267 | ctype_class_new (lr, ctype, "segment_separator"); | |
268 | ctype_class_new (lr, ctype, "block_separator"); | |
269 | ctype_class_new (lr, ctype, "direction_control"); | |
270 | ctype_class_new (lr, ctype, "sym_swap_layout"); | |
271 | ctype_class_new (lr, ctype, "char_shape_selector"); | |
272 | ctype_class_new (lr, ctype, "num_shape_selector"); | |
273 | ctype_class_new (lr, ctype, "non_spacing"); | |
274 | ctype_class_new (lr, ctype, "non_spacing_level3"); | |
275 | ctype_class_new (lr, ctype, "normal_connect"); | |
276 | ctype_class_new (lr, ctype, "r_connect"); | |
277 | ctype_class_new (lr, ctype, "no_connect"); | |
278 | ctype_class_new (lr, ctype, "no_connect-space"); | |
279 | ctype_class_new (lr, ctype, "vowel_connect"); | |
011ebfab | 280 | #endif |
4b10dd6c UD |
281 | |
282 | ctype->class_collection_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
283 | ctype->class_collection | |
284 | = (uint32_t *) xcalloc (sizeof (unsigned long int), | |
285 | ctype->class_collection_max); | |
286 | ctype->class_collection_act = 256; | |
287 | ||
288 | /* Fill character map information. */ | |
4b10dd6c UD |
289 | ctype->last_map_idx = MAX_NR_CHARMAP; |
290 | ctype_map_new (lr, ctype, "toupper", charmap); | |
291 | ctype_map_new (lr, ctype, "tolower", charmap); | |
011ebfab | 292 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 293 | ctype_map_new (lr, ctype, "tosymmetric", charmap); |
011ebfab | 294 | #endif |
4b10dd6c UD |
295 | |
296 | /* Fill first 256 entries in `toXXX' arrays. */ | |
297 | for (cnt = 0; cnt < 256; ++cnt) | |
298 | { | |
299 | ctype->map_collection[0][cnt] = cnt; | |
300 | ctype->map_collection[1][cnt] = cnt; | |
9e2b7438 | 301 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 302 | ctype->map_collection[2][cnt] = cnt; |
9e2b7438 | 303 | #endif |
4b10dd6c UD |
304 | ctype->map256_collection[0][cnt] = cnt; |
305 | ctype->map256_collection[1][cnt] = cnt; | |
306 | } | |
307 | ||
a673fbcb | 308 | obstack_init (&ctype->mempool); |
19bc17a9 RM |
309 | } |
310 | } | |
311 | ||
312 | ||
313 | void | |
4b10dd6c | 314 | ctype_finish (struct localedef_t *locale, struct charmap_t *charmap) |
19bc17a9 RM |
315 | { |
316 | /* See POSIX.2, table 2-6 for the meaning of the following table. */ | |
317 | #define NCLASS 12 | |
318 | static const struct | |
319 | { | |
320 | const char *name; | |
321 | const char allow[NCLASS]; | |
322 | } | |
323 | valid_table[NCLASS] = | |
324 | { | |
325 | /* The order is important. See token.h for more information. | |
326 | M = Always, D = Default, - = Permitted, X = Mutually exclusive */ | |
327 | { "upper", "--MX-XDDXXX-" }, | |
328 | { "lower", "--MX-XDDXXX-" }, | |
329 | { "alpha", "---X-XDDXXX-" }, | |
330 | { "digit", "XXX--XDDXXX-" }, | |
331 | { "xdigit", "-----XDDXXX-" }, | |
332 | { "space", "XXXXX------X" }, | |
333 | { "print", "---------X--" }, | |
334 | { "graph", "---------X--" }, | |
335 | { "blank", "XXXXXM-----X" }, | |
336 | { "cntrl", "XXXXX-XX--XX" }, | |
337 | { "punct", "XXXXX-DD-X-X" }, | |
338 | { "alnum", "-----XDDXXX-" } | |
339 | }; | |
340 | size_t cnt; | |
341 | int cls1, cls2; | |
4b10dd6c UD |
342 | uint32_t space_value; |
343 | struct charseq *space_seq; | |
19bc17a9 | 344 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
4b10dd6c | 345 | int warned; |
0e16ecfa UD |
346 | const void *key; |
347 | size_t len; | |
348 | void *vdata; | |
349 | void *curs; | |
19bc17a9 | 350 | |
b9eb05d6 UD |
351 | /* Now resolve copying and also handle completely missing definitions. */ |
352 | if (ctype == NULL) | |
353 | { | |
70e51ab9 UD |
354 | const char *repertoire_name; |
355 | ||
b9eb05d6 UD |
356 | /* First see whether we were supposed to copy. If yes, find the |
357 | actual definition. */ | |
358 | if (locale->copy_name[LC_CTYPE] != NULL) | |
359 | { | |
360 | /* Find the copying locale. This has to happen transitively since | |
361 | the locale we are copying from might also copying another one. */ | |
362 | struct localedef_t *from = locale; | |
363 | ||
364 | do | |
365 | from = find_locale (LC_CTYPE, from->copy_name[LC_CTYPE], | |
366 | from->repertoire_name, charmap); | |
367 | while (from->categories[LC_CTYPE].ctype == NULL | |
368 | && from->copy_name[LC_CTYPE] != NULL); | |
369 | ||
370 | ctype = locale->categories[LC_CTYPE].ctype | |
371 | = from->categories[LC_CTYPE].ctype; | |
372 | } | |
373 | ||
374 | /* If there is still no definition issue an warning and create an | |
375 | empty one. */ | |
376 | if (ctype == NULL) | |
377 | { | |
f6ada7ad UD |
378 | if (! be_quiet) |
379 | error (0, 0, _("No definition for %s category found"), "LC_CTYPE"); | |
b9eb05d6 UD |
380 | ctype_startup (NULL, locale, charmap, 0); |
381 | ctype = locale->categories[LC_CTYPE].ctype; | |
382 | } | |
70e51ab9 UD |
383 | |
384 | /* Get the repertoire we have to use. */ | |
385 | repertoire_name = locale->repertoire_name ?: repertoire_global; | |
386 | if (repertoire_name != NULL) | |
387 | ctype->repertoire = repertoire_read (repertoire_name); | |
b9eb05d6 UD |
388 | } |
389 | ||
db76d943 UD |
390 | /* We need the name of the currently used 8-bit character set to |
391 | make correct conversion between this 8-bit representation and the | |
392 | ISO 10646 character set used internally for wide characters. */ | |
393 | ctype->codeset_name = charmap->code_set_name; | |
394 | if (ctype->codeset_name == NULL) | |
395 | { | |
396 | if (! be_quiet) | |
397 | error (0, 0, "no character set name specified in charmap"); | |
398 | ctype->codeset_name = "//UNKNOWN//"; | |
399 | } | |
400 | ||
19bc17a9 | 401 | /* Set default value for classes not specified. */ |
4b10dd6c | 402 | set_class_defaults (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
403 | |
404 | /* Check according to table. */ | |
42d7c593 | 405 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) |
19bc17a9 | 406 | { |
4b10dd6c | 407 | uint32_t tmp = ctype->class_collection[cnt]; |
19bc17a9 | 408 | |
4b10dd6c UD |
409 | if (tmp != 0) |
410 | { | |
411 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
412 | if ((tmp & _ISwbit (cls1)) != 0) | |
413 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
414 | if (valid_table[cls1].allow[cls2] != '-') | |
19bc17a9 | 415 | { |
4b10dd6c UD |
416 | int eq = (tmp & _ISwbit (cls2)) != 0; |
417 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 418 | { |
4b10dd6c UD |
419 | case 'M': |
420 | if (!eq) | |
421 | { | |
422 | uint32_t value = ctype->charnames[cnt]; | |
423 | ||
424 | if (!be_quiet) | |
425 | error (0, 0, _("\ | |
426 | character L'\\u%0*x' in class `%s' must be in class `%s'"), | |
427 | value > 0xffff ? 8 : 4, value, | |
428 | valid_table[cls1].name, | |
429 | valid_table[cls2].name); | |
430 | } | |
431 | break; | |
432 | ||
433 | case 'X': | |
434 | if (eq) | |
435 | { | |
436 | uint32_t value = ctype->charnames[cnt]; | |
437 | ||
438 | if (!be_quiet) | |
439 | error (0, 0, _("\ | |
440 | character L'\\u%0*x' in class `%s' must not be in class `%s'"), | |
441 | value > 0xffff ? 8 : 4, value, | |
442 | valid_table[cls1].name, | |
443 | valid_table[cls2].name); | |
444 | } | |
445 | break; | |
446 | ||
447 | case 'D': | |
448 | ctype->class_collection[cnt] |= _ISwbit (cls2); | |
449 | break; | |
450 | ||
451 | default: | |
452 | error (5, 0, _("internal error in %s, line %u"), | |
453 | __FUNCTION__, __LINE__); | |
19bc17a9 | 454 | } |
4b10dd6c UD |
455 | } |
456 | } | |
457 | } | |
458 | ||
459 | for (cnt = 0; cnt < 256; ++cnt) | |
460 | { | |
461 | uint32_t tmp = ctype->class256_collection[cnt]; | |
19bc17a9 | 462 | |
4b10dd6c UD |
463 | if (tmp != 0) |
464 | { | |
465 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
466 | if ((tmp & _ISbit (cls1)) != 0) | |
467 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
468 | if (valid_table[cls1].allow[cls2] != '-') | |
469 | { | |
470 | int eq = (tmp & _ISbit (cls2)) != 0; | |
471 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 472 | { |
4b10dd6c UD |
473 | case 'M': |
474 | if (!eq) | |
475 | { | |
476 | char buf[17]; | |
477 | ||
5d431a3e | 478 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
479 | |
480 | if (!be_quiet) | |
481 | error (0, 0, _("\ | |
482 | character '%s' in class `%s' must be in class `%s'"), | |
483 | buf, valid_table[cls1].name, | |
484 | valid_table[cls2].name); | |
485 | } | |
486 | break; | |
487 | ||
488 | case 'X': | |
489 | if (eq) | |
490 | { | |
491 | char buf[17]; | |
492 | ||
5d431a3e | 493 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
494 | |
495 | if (!be_quiet) | |
496 | error (0, 0, _("\ | |
497 | character '%s' in class `%s' must not be in class `%s'"), | |
498 | buf, valid_table[cls1].name, | |
499 | valid_table[cls2].name); | |
500 | } | |
501 | break; | |
502 | ||
503 | case 'D': | |
504 | ctype->class256_collection[cnt] |= _ISbit (cls2); | |
505 | break; | |
506 | ||
507 | default: | |
508 | error (5, 0, _("internal error in %s, line %u"), | |
509 | __FUNCTION__, __LINE__); | |
19bc17a9 | 510 | } |
4b10dd6c UD |
511 | } |
512 | } | |
19bc17a9 RM |
513 | } |
514 | ||
515 | /* ... and now test <SP> as a special case. */ | |
a0dc5206 UD |
516 | space_value = 32; |
517 | if (((cnt = BITPOS (tok_space), | |
518 | (ELEM (ctype, class_collection, , space_value) | |
519 | & BITw (tok_space)) == 0) | |
520 | || (cnt = BITPOS (tok_blank), | |
521 | (ELEM (ctype, class_collection, , space_value) | |
522 | & BITw (tok_blank)) == 0))) | |
880f421f UD |
523 | { |
524 | if (!be_quiet) | |
525 | error (0, 0, _("<SP> character not in class `%s'"), | |
526 | valid_table[cnt].name); | |
527 | } | |
c84142e8 UD |
528 | else if (((cnt = BITPOS (tok_punct), |
529 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 530 | & BITw (tok_punct)) != 0) |
c84142e8 UD |
531 | || (cnt = BITPOS (tok_graph), |
532 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 533 | & BITw (tok_graph)) |
880f421f UD |
534 | != 0))) |
535 | { | |
536 | if (!be_quiet) | |
537 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
538 | valid_table[cnt].name); | |
539 | } | |
19bc17a9 | 540 | else |
4b10dd6c UD |
541 | ELEM (ctype, class_collection, , space_value) |= BITw (tok_print); |
542 | ||
543 | space_seq = charmap_find_value (charmap, "SP", 2); | |
ce177a84 | 544 | if (space_seq == NULL) |
45c95239 UD |
545 | space_seq = charmap_find_value (charmap, "space", 5); |
546 | if (space_seq == NULL) | |
1b97149d | 547 | space_seq = charmap_find_value (charmap, "U00000020", 9); |
4b10dd6c UD |
548 | if (space_seq == NULL || space_seq->nbytes != 1) |
549 | { | |
550 | if (!be_quiet) | |
551 | error (0, 0, _("character <SP> not defined in character map")); | |
552 | } | |
553 | else if (((cnt = BITPOS (tok_space), | |
554 | (ctype->class256_collection[space_seq->bytes[0]] | |
555 | & BIT (tok_space)) == 0) | |
556 | || (cnt = BITPOS (tok_blank), | |
557 | (ctype->class256_collection[space_seq->bytes[0]] | |
558 | & BIT (tok_blank)) == 0))) | |
559 | { | |
560 | if (!be_quiet) | |
561 | error (0, 0, _("<SP> character not in class `%s'"), | |
562 | valid_table[cnt].name); | |
563 | } | |
564 | else if (((cnt = BITPOS (tok_punct), | |
565 | (ctype->class256_collection[space_seq->bytes[0]] | |
566 | & BIT (tok_punct)) != 0) | |
567 | || (cnt = BITPOS (tok_graph), | |
568 | (ctype->class256_collection[space_seq->bytes[0]] | |
569 | & BIT (tok_graph)) != 0))) | |
570 | { | |
571 | if (!be_quiet) | |
572 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
573 | valid_table[cnt].name); | |
574 | } | |
575 | else | |
576 | ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print); | |
75cd5204 RM |
577 | |
578 | /* Now that the tests are done make sure the name array contains all | |
579 | characters which are handled in the WIDTH section of the | |
580 | character set definition file. */ | |
4b10dd6c UD |
581 | if (charmap->width_rules != NULL) |
582 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) | |
75cd5204 | 583 | { |
827ff758 UD |
584 | unsigned char bytes[charmap->mb_cur_max]; |
585 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
586 | ||
587 | /* We have the range of character for which the width is | |
588 | specified described using byte sequences of the multibyte | |
589 | charset. We have to convert this to UCS4 now. And we | |
590 | cannot simply convert the beginning and the end of the | |
591 | sequence, we have to iterate over the byte sequence and | |
592 | convert it for every single character. */ | |
593 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
594 | ||
595 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
596 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
597 | nbytes) <= 0) | |
598 | { | |
599 | /* Find the UCS value for `bytes'. */ | |
827ff758 | 600 | int inner; |
76e680a8 UD |
601 | uint32_t wch; |
602 | struct charseq *seq = charmap_find_symbol (charmap, bytes, nbytes); | |
603 | ||
604 | if (seq == NULL) | |
605 | wch = ILLEGAL_CHAR_VALUE; | |
606 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
607 | wch = seq->ucs4; | |
608 | else | |
609 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
610 | strlen (seq->name)); | |
827ff758 UD |
611 | |
612 | if (wch != ILLEGAL_CHAR_VALUE) | |
613 | /* We are only interested in the side-effects of the | |
614 | `find_idx' call. It will add appropriate entries in | |
615 | the name array if this is necessary. */ | |
616 | (void) find_idx (ctype, NULL, NULL, NULL, wch); | |
617 | ||
618 | /* "Increment" the bytes sequence. */ | |
619 | inner = nbytes - 1; | |
620 | while (inner >= 0 && bytes[inner] == 0xff) | |
621 | --inner; | |
622 | ||
623 | if (inner < 0) | |
624 | { | |
625 | /* We have to extend the byte sequence. */ | |
626 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
627 | break; | |
628 | ||
629 | bytes[0] = 1; | |
630 | memset (&bytes[1], 0, nbytes); | |
631 | ++nbytes; | |
632 | } | |
633 | else | |
634 | { | |
635 | ++bytes[inner]; | |
636 | while (++inner < nbytes) | |
637 | bytes[inner] = 0; | |
638 | } | |
639 | } | |
4b10dd6c UD |
640 | } |
641 | ||
0e16ecfa UD |
642 | /* Now set all the other characters of the character set to the |
643 | default width. */ | |
644 | curs = NULL; | |
645 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
646 | { | |
647 | struct charseq *data = (struct charseq *) vdata; | |
648 | ||
649 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
650 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
651 | data->name, len); | |
652 | ||
653 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) | |
654 | (void) find_idx (ctype, NULL, NULL, NULL, data->ucs4); | |
655 | } | |
656 | ||
4b10dd6c UD |
657 | /* There must be a multiple of 10 digits. */ |
658 | if (ctype->mbdigits_act % 10 != 0) | |
659 | { | |
660 | assert (ctype->mbdigits_act == ctype->wcdigits_act); | |
661 | ctype->wcdigits_act -= ctype->mbdigits_act % 10; | |
662 | ctype->mbdigits_act -= ctype->mbdigits_act % 10; | |
663 | error (0, 0, _("`digit' category has not entries in groups of ten")); | |
664 | } | |
665 | ||
666 | /* Check the input digits. There must be a multiple of ten available. | |
42d7c593 | 667 | In each group it could be that one or the other character is missing. |
4b10dd6c UD |
668 | In this case the whole group must be removed. */ |
669 | cnt = 0; | |
670 | while (cnt < ctype->mbdigits_act) | |
671 | { | |
672 | size_t inner; | |
673 | for (inner = 0; inner < 10; ++inner) | |
674 | if (ctype->mbdigits[cnt + inner] == NULL) | |
675 | break; | |
676 | ||
677 | if (inner == 10) | |
678 | cnt += 10; | |
679 | else | |
680 | { | |
681 | /* Remove the group. */ | |
682 | memmove (&ctype->mbdigits[cnt], &ctype->mbdigits[cnt + 10], | |
683 | ((ctype->wcdigits_act - cnt - 10) | |
684 | * sizeof (ctype->mbdigits[0]))); | |
685 | ctype->mbdigits_act -= 10; | |
686 | } | |
687 | } | |
688 | ||
689 | /* If no input digits are given use the default. */ | |
690 | if (ctype->mbdigits_act == 0) | |
691 | { | |
692 | if (ctype->mbdigits_max == 0) | |
693 | { | |
694 | ctype->mbdigits = obstack_alloc (&charmap->mem_pool, | |
695 | 10 * sizeof (struct charseq *)); | |
696 | ctype->mbdigits_max = 10; | |
697 | } | |
698 | ||
699 | for (cnt = 0; cnt < 10; ++cnt) | |
700 | { | |
701 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
702 | digits + cnt, 1); | |
703 | if (ctype->mbdigits[cnt] == NULL) | |
704 | { | |
705 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
706 | longnames[cnt], | |
707 | strlen (longnames[cnt])); | |
708 | if (ctype->mbdigits[cnt] == NULL) | |
709 | { | |
710 | /* Hum, this ain't good. */ | |
711 | error (0, 0, _("\ | |
712 | no input digits defined and none of the standard names in the charmap")); | |
713 | ||
714 | ctype->mbdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
715 | sizeof (struct charseq) + 1); | |
716 | ||
717 | /* This is better than nothing. */ | |
718 | ctype->mbdigits[cnt]->bytes[0] = digits[cnt]; | |
719 | ctype->mbdigits[cnt]->nbytes = 1; | |
720 | } | |
721 | } | |
722 | } | |
723 | ||
724 | ctype->mbdigits_act = 10; | |
725 | } | |
726 | ||
727 | /* Check the wide character input digits. There must be a multiple | |
42d7c593 | 728 | of ten available. In each group it could be that one or the other |
4b10dd6c UD |
729 | character is missing. In this case the whole group must be |
730 | removed. */ | |
731 | cnt = 0; | |
732 | while (cnt < ctype->wcdigits_act) | |
733 | { | |
734 | size_t inner; | |
735 | for (inner = 0; inner < 10; ++inner) | |
736 | if (ctype->wcdigits[cnt + inner] == ILLEGAL_CHAR_VALUE) | |
737 | break; | |
738 | ||
739 | if (inner == 10) | |
740 | cnt += 10; | |
741 | else | |
742 | { | |
743 | /* Remove the group. */ | |
744 | memmove (&ctype->wcdigits[cnt], &ctype->wcdigits[cnt + 10], | |
745 | ((ctype->wcdigits_act - cnt - 10) | |
746 | * sizeof (ctype->wcdigits[0]))); | |
747 | ctype->wcdigits_act -= 10; | |
748 | } | |
749 | } | |
750 | ||
751 | /* If no input digits are given use the default. */ | |
752 | if (ctype->wcdigits_act == 0) | |
753 | { | |
754 | if (ctype->wcdigits_max == 0) | |
755 | { | |
756 | ctype->wcdigits = obstack_alloc (&charmap->mem_pool, | |
757 | 10 * sizeof (uint32_t)); | |
758 | ctype->wcdigits_max = 10; | |
759 | } | |
760 | ||
761 | for (cnt = 0; cnt < 10; ++cnt) | |
762 | ctype->wcdigits[cnt] = L'0' + cnt; | |
763 | ||
764 | ctype->mbdigits_act = 10; | |
765 | } | |
766 | ||
767 | /* Check the outdigits. */ | |
768 | warned = 0; | |
769 | for (cnt = 0; cnt < 10; ++cnt) | |
770 | if (ctype->mboutdigits[cnt] == NULL) | |
771 | { | |
772 | static struct charseq replace[2]; | |
773 | ||
774 | if (!warned) | |
775 | { | |
776 | error (0, 0, _("\ | |
777 | not all characters used in `outdigit' are available in the charmap")); | |
778 | warned = 1; | |
779 | } | |
780 | ||
781 | replace[0].nbytes = 1; | |
782 | replace[0].bytes[0] = '?'; | |
783 | replace[0].bytes[1] = '\0'; | |
784 | ctype->mboutdigits[cnt] = &replace[0]; | |
785 | } | |
786 | ||
787 | warned = 0; | |
788 | for (cnt = 0; cnt < 10; ++cnt) | |
789 | if (ctype->wcoutdigits[cnt] == 0) | |
790 | { | |
791 | if (!warned) | |
792 | { | |
793 | error (0, 0, _("\ | |
794 | not all characters used in `outdigit' are available in the repertoire")); | |
795 | warned = 1; | |
796 | } | |
797 | ||
798 | ctype->wcoutdigits[cnt] = L'?'; | |
75cd5204 | 799 | } |
a8e4c924 UD |
800 | |
801 | /* Sort the entries in the translit_ignore list. */ | |
802 | if (ctype->translit_ignore != NULL) | |
803 | { | |
804 | struct translit_ignore_t *firstp = ctype->translit_ignore; | |
805 | struct translit_ignore_t *runp; | |
806 | ||
807 | ctype->ntranslit_ignore = 1; | |
808 | ||
809 | for (runp = firstp->next; runp != NULL; runp = runp->next) | |
810 | { | |
811 | struct translit_ignore_t *lastp = NULL; | |
812 | struct translit_ignore_t *cmpp; | |
813 | ||
814 | ++ctype->ntranslit_ignore; | |
815 | ||
816 | for (cmpp = firstp; cmpp != NULL; lastp = cmpp, cmpp = cmpp->next) | |
817 | if (runp->from < cmpp->from) | |
818 | break; | |
819 | ||
820 | runp->next = lastp; | |
821 | if (lastp == NULL) | |
822 | firstp = runp; | |
823 | } | |
824 | ||
825 | ctype->translit_ignore = firstp; | |
826 | } | |
19bc17a9 RM |
827 | } |
828 | ||
829 | ||
830 | void | |
4b10dd6c | 831 | ctype_output (struct localedef_t *locale, struct charmap_t *charmap, |
75cd5204 | 832 | const char *output_path) |
19bc17a9 | 833 | { |
e43e0dd6 | 834 | static const char nulbytes[4] = { 0, 0, 0, 0 }; |
19bc17a9 RM |
835 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
836 | const size_t nelems = (_NL_ITEM_INDEX (_NL_NUM_LC_CTYPE) | |
5491da0d | 837 | + (ctype->map_collection_nr - 2)); |
75cd5204 | 838 | struct iovec iov[2 + nelems + ctype->nr_charclass |
f175216d | 839 | + ctype->map_collection_nr + 2]; |
19bc17a9 | 840 | struct locale_file data; |
4b10dd6c | 841 | uint32_t idx[nelems + 1]; |
1d96d74d | 842 | uint32_t default_missing_len; |
75cd5204 | 843 | size_t elem, cnt, offset, total; |
4b10dd6c | 844 | char *cp; |
19bc17a9 RM |
845 | |
846 | /* Now prepare the output: Find the sizes of the table we can use. */ | |
4b10dd6c | 847 | allocate_arrays (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
848 | |
849 | data.magic = LIMAGIC (LC_CTYPE); | |
850 | data.n = nelems; | |
851 | iov[0].iov_base = (void *) &data; | |
852 | iov[0].iov_len = sizeof (data); | |
853 | ||
854 | iov[1].iov_base = (void *) idx; | |
a0edd63e | 855 | iov[1].iov_len = nelems * sizeof (uint32_t); |
19bc17a9 RM |
856 | |
857 | idx[0] = iov[0].iov_len + iov[1].iov_len; | |
858 | offset = 0; | |
859 | ||
860 | for (elem = 0; elem < nelems; ++elem) | |
861 | { | |
862 | if (elem < _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) | |
863 | switch (elem) | |
864 | { | |
c6df09ad UD |
865 | #define CTYPE_EMPTY(name) \ |
866 | case name: \ | |
04fbc779 | 867 | iov[2 + elem + offset].iov_base = (void *) ""; \ |
c6df09ad UD |
868 | iov[2 + elem + offset].iov_len = 0; \ |
869 | idx[elem + 1] = idx[elem]; \ | |
870 | break | |
871 | ||
872 | CTYPE_EMPTY(_NL_CTYPE_GAP1); | |
873 | CTYPE_EMPTY(_NL_CTYPE_GAP2); | |
874 | CTYPE_EMPTY(_NL_CTYPE_GAP3); | |
875 | ||
19bc17a9 RM |
876 | #define CTYPE_DATA(name, base, len) \ |
877 | case _NL_ITEM_INDEX (name): \ | |
ce7a5ef4 RM |
878 | iov[2 + elem + offset].iov_base = (base); \ |
879 | iov[2 + elem + offset].iov_len = (len); \ | |
1d96d74d | 880 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; \ |
19bc17a9 RM |
881 | break |
882 | ||
883 | CTYPE_DATA (_NL_CTYPE_CLASS, | |
884 | ctype->ctype_b, | |
885 | (256 + 128) * sizeof (char_class_t)); | |
886 | ||
4a33c2f5 UD |
887 | CTYPE_DATA (_NL_CTYPE_TOUPPER, |
888 | ctype->map[0], | |
f1d8b804 | 889 | (256 + 128) * sizeof (uint32_t)); |
4a33c2f5 UD |
890 | CTYPE_DATA (_NL_CTYPE_TOLOWER, |
891 | ctype->map[1], | |
f1d8b804 | 892 | (256 + 128) * sizeof (uint32_t)); |
19bc17a9 | 893 | |
49f2be5b UD |
894 | CTYPE_DATA (_NL_CTYPE_TOUPPER32, |
895 | ctype->map32[0], | |
f1d8b804 | 896 | (ctype->plane_size * ctype->plane_cnt) |
49f2be5b UD |
897 | * sizeof (uint32_t)); |
898 | CTYPE_DATA (_NL_CTYPE_TOLOWER32, | |
899 | ctype->map32[1], | |
f1d8b804 | 900 | (ctype->plane_size * ctype->plane_cnt) |
49f2be5b UD |
901 | * sizeof (uint32_t)); |
902 | ||
19bc17a9 RM |
903 | CTYPE_DATA (_NL_CTYPE_CLASS32, |
904 | ctype->ctype32_b, | |
905 | (ctype->plane_size * ctype->plane_cnt | |
906 | * sizeof (char_class32_t))); | |
907 | ||
4a33c2f5 UD |
908 | CTYPE_DATA (_NL_CTYPE_NAMES, |
909 | ctype->names, (ctype->plane_size * ctype->plane_cnt | |
910 | * sizeof (uint32_t))); | |
911 | ||
04fbc779 UD |
912 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TAB_SIZE, |
913 | &ctype->translit_idx_size, sizeof (uint32_t)); | |
4a33c2f5 UD |
914 | |
915 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_IDX, | |
916 | ctype->translit_from_idx, | |
04fbc779 | 917 | ctype->translit_idx_size * sizeof (uint32_t)); |
4b10dd6c | 918 | |
4a33c2f5 UD |
919 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_TBL, |
920 | ctype->translit_from_tbl, | |
4b10dd6c UD |
921 | ctype->translit_from_tbl_size); |
922 | ||
4a33c2f5 UD |
923 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_IDX, |
924 | ctype->translit_to_idx, | |
04fbc779 | 925 | ctype->translit_idx_size * sizeof (uint32_t)); |
4b10dd6c | 926 | |
4a33c2f5 UD |
927 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_TBL, |
928 | ctype->translit_to_tbl, ctype->translit_to_tbl_size); | |
4b10dd6c | 929 | |
4a33c2f5 | 930 | CTYPE_DATA (_NL_CTYPE_HASH_SIZE, |
4b10dd6c | 931 | &ctype->plane_size, sizeof (uint32_t)); |
4a33c2f5 | 932 | CTYPE_DATA (_NL_CTYPE_HASH_LAYERS, |
4b10dd6c | 933 | &ctype->plane_cnt, sizeof (uint32_t)); |
19bc17a9 | 934 | |
75cd5204 RM |
935 | case _NL_ITEM_INDEX (_NL_CTYPE_CLASS_NAMES): |
936 | /* The class name array. */ | |
937 | total = 0; | |
938 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt, ++offset) | |
939 | { | |
940 | iov[2 + elem + offset].iov_base | |
941 | = (void *) ctype->classnames[cnt]; | |
942 | iov[2 + elem + offset].iov_len | |
943 | = strlen (ctype->classnames[cnt]) + 1; | |
944 | total += iov[2 + elem + offset].iov_len; | |
945 | } | |
e43e0dd6 | 946 | iov[2 + elem + offset].iov_base = (void *) nulbytes; |
ce7a5ef4 RM |
947 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); |
948 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 949 | |
4b10dd6c | 950 | idx[elem + 1] = idx[elem] + total; |
75cd5204 RM |
951 | break; |
952 | ||
953 | case _NL_ITEM_INDEX (_NL_CTYPE_MAP_NAMES): | |
954 | /* The class name array. */ | |
955 | total = 0; | |
956 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt, ++offset) | |
957 | { | |
958 | iov[2 + elem + offset].iov_base | |
959 | = (void *) ctype->mapnames[cnt]; | |
960 | iov[2 + elem + offset].iov_len | |
961 | = strlen (ctype->mapnames[cnt]) + 1; | |
962 | total += iov[2 + elem + offset].iov_len; | |
963 | } | |
e43e0dd6 | 964 | iov[2 + elem + offset].iov_base = (void *) nulbytes; |
ce7a5ef4 RM |
965 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); |
966 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 967 | |
4b10dd6c | 968 | idx[elem + 1] = idx[elem] + total; |
75cd5204 | 969 | break; |
19bc17a9 RM |
970 | |
971 | CTYPE_DATA (_NL_CTYPE_WIDTH, | |
5866b131 UD |
972 | ctype->width, |
973 | (ctype->plane_size * ctype->plane_cnt + 3) & ~3ul); | |
19bc17a9 | 974 | |
0200214b | 975 | CTYPE_DATA (_NL_CTYPE_MB_CUR_MAX, |
4b10dd6c | 976 | &ctype->mb_cur_max, sizeof (uint32_t)); |
0200214b | 977 | |
ce7a5ef4 RM |
978 | case _NL_ITEM_INDEX (_NL_CTYPE_CODESET_NAME): |
979 | total = strlen (ctype->codeset_name) + 1; | |
980 | if (total % 4 == 0) | |
981 | iov[2 + elem + offset].iov_base = (char *) ctype->codeset_name; | |
982 | else | |
983 | { | |
984 | iov[2 + elem + offset].iov_base = alloca ((total + 3) & ~3); | |
9756dfe1 UD |
985 | memset (mempcpy (iov[2 + elem + offset].iov_base, |
986 | ctype->codeset_name, total), | |
987 | '\0', 4 - (total & 3)); | |
ce7a5ef4 RM |
988 | total = (total + 3) & ~3; |
989 | } | |
990 | iov[2 + elem + offset].iov_len = total; | |
4b10dd6c UD |
991 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
992 | break; | |
993 | ||
4a33c2f5 | 994 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN): |
4b10dd6c UD |
995 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
996 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
997 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
998 | ctype->mbdigits_act / 10; | |
a9c27b3e | 999 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
1000 | break; |
1001 | ||
4a33c2f5 | 1002 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN): |
e43e0dd6 UD |
1003 | /* Align entries. */ |
1004 | iov[2 + elem + offset].iov_base = (void *) nulbytes; | |
1005 | iov[2 + elem + offset].iov_len = (4 - idx[elem] % 4) % 4; | |
1006 | idx[elem] += iov[2 + elem + offset].iov_len; | |
1007 | ++offset; | |
1008 | ||
4b10dd6c UD |
1009 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
1010 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
1011 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
1012 | ctype->wcdigits_act / 10; | |
a9c27b3e | 1013 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
1014 | break; |
1015 | ||
e43e0dd6 | 1016 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_MB): |
4b10dd6c UD |
1017 | /* Compute the length of all possible characters. For INDIGITS |
1018 | there might be more than one. We simply concatenate all of | |
1019 | them with a NUL byte following. The NUL byte wouldn't be | |
1020 | necessary but it makes it easier for the user. */ | |
1021 | total = 0; | |
f175216d | 1022 | |
498b733e | 1023 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB); |
4b10dd6c UD |
1024 | cnt < ctype->mbdigits_act; cnt += 10) |
1025 | total += ctype->mbdigits[cnt]->nbytes + 1; | |
1026 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
1027 | iov[2 + elem + offset].iov_len = total; | |
1028 | ||
1029 | cp = iov[2 + elem + offset].iov_base; | |
498b733e | 1030 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB); |
4b10dd6c UD |
1031 | cnt < ctype->mbdigits_act; cnt += 10) |
1032 | { | |
1033 | cp = mempcpy (cp, ctype->mbdigits[cnt]->bytes, | |
1034 | ctype->mbdigits[cnt]->nbytes); | |
1035 | *cp++ = '\0'; | |
1036 | } | |
a9c27b3e | 1037 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1038 | break; |
1039 | ||
1040 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_MB): | |
1041 | /* Compute the length of all possible characters. For INDIGITS | |
1042 | there might be more than one. We simply concatenate all of | |
1043 | them with a NUL byte following. The NUL byte wouldn't be | |
1044 | necessary but it makes it easier for the user. */ | |
498b733e | 1045 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB); |
4b10dd6c UD |
1046 | total = ctype->mboutdigits[cnt]->nbytes + 1; |
1047 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
1048 | iov[2 + elem + offset].iov_len = total; | |
1049 | ||
1050 | *(char *) mempcpy (iov[2 + elem + offset].iov_base, | |
498b733e UD |
1051 | ctype->mboutdigits[cnt]->bytes, |
1052 | ctype->mboutdigits[cnt]->nbytes) = '\0'; | |
a9c27b3e | 1053 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1054 | break; |
1055 | ||
e43e0dd6 | 1056 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC): |
4b10dd6c UD |
1057 | total = ctype->wcdigits_act / 10; |
1058 | ||
1059 | iov[2 + elem + offset].iov_base = | |
1060 | (uint32_t *) alloca (total * sizeof (uint32_t)); | |
1061 | iov[2 + elem + offset].iov_len = total * sizeof (uint32_t); | |
1062 | ||
498b733e | 1063 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC); |
4b10dd6c UD |
1064 | cnt < ctype->wcdigits_act; cnt += 10) |
1065 | ((uint32_t *) iov[2 + elem + offset].iov_base)[cnt / 10] | |
4a33c2f5 | 1066 | = ctype->wcdigits[cnt]; |
a9c27b3e | 1067 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1068 | break; |
1069 | ||
e43e0dd6 UD |
1070 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC): |
1071 | /* Align entries. */ | |
1072 | iov[2 + elem + offset].iov_base = (void *) nulbytes; | |
1073 | iov[2 + elem + offset].iov_len = (4 - idx[elem] % 4) % 4; | |
1074 | idx[elem] += iov[2 + elem + offset].iov_len; | |
1075 | ++offset; | |
1076 | /* FALLTRHOUGH */ | |
1077 | ||
1078 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT1_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC): | |
498b733e | 1079 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC); |
4b10dd6c UD |
1080 | iov[2 + elem + offset].iov_base = &ctype->wcoutdigits[cnt]; |
1081 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
a9c27b3e | 1082 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1083 | break; |
1084 | ||
a8e4c924 UD |
1085 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING_LEN): |
1086 | default_missing_len = (ctype->default_missing | |
1087 | ? wcslen ((wchar_t *)ctype->default_missing) | |
7f455351 | 1088 | : 0); |
a8e4c924 UD |
1089 | iov[2 + elem + offset].iov_base = &default_missing_len; |
1090 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
1091 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1092 | break; | |
1093 | ||
1d96d74d UD |
1094 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING): |
1095 | iov[2 + elem + offset].iov_base = | |
1096 | ctype->default_missing ?: (uint32_t *) L""; | |
1097 | iov[2 + elem + offset].iov_len = | |
1098 | wcslen (iov[2 + elem + offset].iov_base); | |
1099 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1100 | break; | |
1101 | ||
a8e4c924 UD |
1102 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE_LEN): |
1103 | iov[2 + elem + offset].iov_base = &ctype->ntranslit_ignore; | |
1d96d74d | 1104 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); |
a8e4c924 UD |
1105 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
1106 | break; | |
1107 | ||
1108 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE): | |
1109 | { | |
1110 | uint32_t *ranges = (uint32_t *) alloca (ctype->ntranslit_ignore | |
1111 | * 3 * sizeof (uint32_t)); | |
1112 | struct translit_ignore_t *runp; | |
1113 | ||
1114 | iov[2 + elem + offset].iov_base = ranges; | |
1115 | iov[2 + elem + offset].iov_len = (ctype->ntranslit_ignore | |
1116 | * 3 * sizeof (uint32_t)); | |
1117 | ||
1118 | for (runp = ctype->translit_ignore; runp != NULL; | |
1119 | runp = runp->next) | |
1120 | { | |
1121 | *ranges++ = runp->from; | |
1122 | *ranges++ = runp->to; | |
1123 | *ranges++ = runp->step; | |
1124 | } | |
1125 | } | |
1d96d74d UD |
1126 | /* Remove the following line in case a new entry is added |
1127 | after _NL_CTYPE_TRANSLIT_DEFAULT_MISSING_LEN. */ | |
1128 | if (elem < nelems) | |
1129 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1130 | break; | |
1131 | ||
19bc17a9 RM |
1132 | default: |
1133 | assert (! "unknown CTYPE element"); | |
1134 | } | |
1135 | else | |
1136 | { | |
1137 | /* Handle extra maps. */ | |
5491da0d | 1138 | size_t nr = (elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) + 2; |
19bc17a9 | 1139 | |
49f2be5b | 1140 | iov[2 + elem + offset].iov_base = ctype->map32[nr]; |
75cd5204 | 1141 | iov[2 + elem + offset].iov_len = ((ctype->plane_size |
f1d8b804 | 1142 | * ctype->plane_cnt) |
4b10dd6c | 1143 | * sizeof (uint32_t)); |
19bc17a9 | 1144 | |
4b10dd6c | 1145 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
75cd5204 | 1146 | } |
19bc17a9 | 1147 | } |
19bc17a9 | 1148 | |
75cd5204 | 1149 | assert (2 + elem + offset == (nelems + ctype->nr_charclass |
f175216d | 1150 | + ctype->map_collection_nr + 2 + 2)); |
19bc17a9 | 1151 | |
83b1b6d8 | 1152 | write_locale_data (output_path, "LC_CTYPE", 2 + elem + offset, iov); |
19bc17a9 RM |
1153 | } |
1154 | ||
1155 | ||
4b10dd6c UD |
1156 | /* Local functions. */ |
1157 | static void | |
1158 | ctype_class_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1159 | const char *name) | |
19bc17a9 | 1160 | { |
4b10dd6c | 1161 | size_t cnt; |
19bc17a9 | 1162 | |
4b10dd6c UD |
1163 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1164 | if (strcmp (ctype->classnames[cnt], name) == 0) | |
1165 | break; | |
19bc17a9 | 1166 | |
4b10dd6c UD |
1167 | if (cnt < ctype->nr_charclass) |
1168 | { | |
1169 | lr_error (lr, _("character class `%s' already defined"), name); | |
1170 | return; | |
1171 | } | |
19bc17a9 | 1172 | |
4b10dd6c UD |
1173 | if (ctype->nr_charclass == MAX_NR_CHARCLASS) |
1174 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1175 | error (2, 0, _("\ | |
5d431a3e | 1176 | implementation limit: no more than %Zd character classes allowed"), |
4b10dd6c | 1177 | MAX_NR_CHARCLASS); |
19bc17a9 | 1178 | |
4b10dd6c | 1179 | ctype->classnames[ctype->nr_charclass++] = name; |
19bc17a9 RM |
1180 | } |
1181 | ||
1182 | ||
4b10dd6c UD |
1183 | static void |
1184 | ctype_map_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1185 | const char *name, struct charmap_t *charmap) | |
19bc17a9 | 1186 | { |
4b10dd6c | 1187 | size_t max_chars = 0; |
ba1ffaa1 | 1188 | size_t cnt; |
19bc17a9 | 1189 | |
4b10dd6c | 1190 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
19bc17a9 | 1191 | { |
4b10dd6c UD |
1192 | if (strcmp (ctype->mapnames[cnt], name) == 0) |
1193 | break; | |
1194 | ||
1195 | if (max_chars < ctype->map_collection_max[cnt]) | |
1196 | max_chars = ctype->map_collection_max[cnt]; | |
19bc17a9 RM |
1197 | } |
1198 | ||
4b10dd6c UD |
1199 | if (cnt < ctype->map_collection_nr) |
1200 | { | |
1201 | lr_error (lr, _("character map `%s' already defined"), name); | |
1202 | return; | |
1203 | } | |
19bc17a9 | 1204 | |
4b10dd6c UD |
1205 | if (ctype->map_collection_nr == MAX_NR_CHARMAP) |
1206 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1207 | error (2, 0, _("\ | |
1208 | implementation limit: no more than %d character maps allowed"), | |
1209 | MAX_NR_CHARMAP); | |
19bc17a9 | 1210 | |
4b10dd6c UD |
1211 | ctype->mapnames[cnt] = name; |
1212 | ||
1213 | if (max_chars == 0) | |
1214 | ctype->map_collection_max[cnt] = charmap->mb_cur_max == 1 ? 256 : 512; | |
1215 | else | |
1216 | ctype->map_collection_max[cnt] = max_chars; | |
1217 | ||
1218 | ctype->map_collection[cnt] = (uint32_t *) | |
5866b131 | 1219 | xcalloc (sizeof (uint32_t), ctype->map_collection_max[cnt]); |
4b10dd6c | 1220 | ctype->map_collection_act[cnt] = 256; |
19bc17a9 | 1221 | |
4b10dd6c | 1222 | ++ctype->map_collection_nr; |
19bc17a9 RM |
1223 | } |
1224 | ||
1225 | ||
4b10dd6c | 1226 | /* We have to be prepared that TABLE, MAX, and ACT can be NULL. This |
42d7c593 | 1227 | is possible if we only want to extend the name array. */ |
4b10dd6c UD |
1228 | static uint32_t * |
1229 | find_idx (struct locale_ctype_t *ctype, uint32_t **table, size_t *max, | |
1230 | size_t *act, uint32_t idx) | |
19bc17a9 | 1231 | { |
4b10dd6c | 1232 | size_t cnt; |
19bc17a9 | 1233 | |
4b10dd6c UD |
1234 | if (idx < 256) |
1235 | return table == NULL ? NULL : &(*table)[idx]; | |
19bc17a9 | 1236 | |
4b10dd6c UD |
1237 | for (cnt = 256; cnt < ctype->charnames_act; ++cnt) |
1238 | if (ctype->charnames[cnt] == idx) | |
1239 | break; | |
19bc17a9 | 1240 | |
4b10dd6c UD |
1241 | /* We have to distinguish two cases: the name is found or not. */ |
1242 | if (cnt == ctype->charnames_act) | |
1243 | { | |
1244 | /* Extend the name array. */ | |
1245 | if (ctype->charnames_act == ctype->charnames_max) | |
1246 | { | |
1247 | ctype->charnames_max *= 2; | |
5866b131 | 1248 | ctype->charnames = (uint32_t *) |
4b10dd6c | 1249 | xrealloc (ctype->charnames, |
5866b131 | 1250 | sizeof (uint32_t) * ctype->charnames_max); |
4b10dd6c UD |
1251 | } |
1252 | ctype->charnames[ctype->charnames_act++] = idx; | |
1253 | } | |
19bc17a9 | 1254 | |
4b10dd6c UD |
1255 | if (table == NULL) |
1256 | /* We have done everything we are asked to do. */ | |
1257 | return NULL; | |
19bc17a9 | 1258 | |
4b10dd6c UD |
1259 | if (cnt >= *act) |
1260 | { | |
1261 | if (cnt >= *max) | |
1262 | { | |
1263 | size_t old_max = *max; | |
1264 | do | |
1265 | *max *= 2; | |
1266 | while (*max <= cnt); | |
19bc17a9 | 1267 | |
4b10dd6c | 1268 | *table = |
5866b131 | 1269 | (uint32_t *) xrealloc (*table, *max * sizeof (uint32_t)); |
4b10dd6c UD |
1270 | memset (&(*table)[old_max], '\0', |
1271 | (*max - old_max) * sizeof (uint32_t)); | |
1272 | } | |
19bc17a9 | 1273 | |
76e680a8 | 1274 | *act = cnt + 1; |
4b10dd6c | 1275 | } |
19bc17a9 | 1276 | |
4b10dd6c | 1277 | return &(*table)[cnt]; |
19bc17a9 RM |
1278 | } |
1279 | ||
1280 | ||
4b10dd6c UD |
1281 | static int |
1282 | get_character (struct token *now, struct charmap_t *charmap, | |
1283 | struct repertoire_t *repertoire, | |
1284 | struct charseq **seqp, uint32_t *wchp) | |
19bc17a9 | 1285 | { |
4b10dd6c UD |
1286 | if (now->tok == tok_bsymbol) |
1287 | { | |
1288 | /* This will hopefully be the normal case. */ | |
1289 | *wchp = repertoire_find_value (repertoire, now->val.str.startmb, | |
1290 | now->val.str.lenmb); | |
1291 | *seqp = charmap_find_value (charmap, now->val.str.startmb, | |
1292 | now->val.str.lenmb); | |
1293 | } | |
1294 | else if (now->tok == tok_ucs4) | |
1295 | { | |
f0a4b6b1 UD |
1296 | char utmp[10]; |
1297 | ||
1298 | snprintf (utmp, sizeof (utmp), "U%08X", now->val.ucs4); | |
1299 | *seqp = charmap_find_value (charmap, utmp, 9); | |
1300 | ||
1301 | if (*seqp == NULL) | |
1302 | *seqp = repertoire_find_seq (repertoire, now->val.ucs4); | |
19bc17a9 | 1303 | |
4b10dd6c UD |
1304 | if (*seqp == NULL) |
1305 | { | |
1306 | /* Compute the value in the charmap from the UCS value. */ | |
1307 | const char *symbol = repertoire_find_symbol (repertoire, | |
1308 | now->val.ucs4); | |
19bc17a9 | 1309 | |
4b10dd6c UD |
1310 | if (symbol == NULL) |
1311 | *seqp = NULL; | |
1312 | else | |
1313 | *seqp = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1314 | |
4b10dd6c UD |
1315 | if (*seqp == NULL) |
1316 | { | |
723faa38 UD |
1317 | if (repertoire != NULL) |
1318 | { | |
1319 | /* Insert a negative entry. */ | |
1320 | static const struct charseq negative | |
1321 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1322 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, | |
1323 | sizeof (uint32_t)); | |
1324 | *newp = now->val.ucs4; | |
1325 | ||
1326 | insert_entry (&repertoire->seq_table, newp, | |
1327 | sizeof (uint32_t), (void *) &negative); | |
1328 | } | |
4b10dd6c UD |
1329 | } |
1330 | else | |
1331 | (*seqp)->ucs4 = now->val.ucs4; | |
1332 | } | |
1333 | else if ((*seqp)->ucs4 != now->val.ucs4) | |
1334 | *seqp = NULL; | |
19bc17a9 | 1335 | |
4b10dd6c UD |
1336 | *wchp = now->val.ucs4; |
1337 | } | |
1338 | else if (now->tok == tok_charcode) | |
1339 | { | |
1340 | /* We must map from the byte code to UCS4. */ | |
1341 | *seqp = charmap_find_symbol (charmap, now->val.str.startmb, | |
1342 | now->val.str.lenmb); | |
19bc17a9 | 1343 | |
4b10dd6c UD |
1344 | if (*seqp == NULL) |
1345 | *wchp = ILLEGAL_CHAR_VALUE; | |
1346 | else | |
1347 | { | |
1348 | if ((*seqp)->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1349 | (*seqp)->ucs4 = repertoire_find_value (repertoire, (*seqp)->name, | |
1350 | strlen ((*seqp)->name)); | |
1351 | *wchp = (*seqp)->ucs4; | |
1352 | } | |
1353 | } | |
1354 | else | |
1355 | return 1; | |
19bc17a9 RM |
1356 | |
1357 | return 0; | |
1358 | } | |
1359 | ||
1360 | ||
a0dc5206 UD |
1361 | /* Ellipsis like in `<foo123>..<foo12a>' or `<j1234>....<j1245>' and |
1362 | the .(2). counterparts. */ | |
4b10dd6c UD |
1363 | static void |
1364 | charclass_symbolic_ellipsis (struct linereader *ldfile, | |
1365 | struct locale_ctype_t *ctype, | |
1366 | struct charmap_t *charmap, | |
1367 | struct repertoire_t *repertoire, | |
1368 | struct token *now, | |
1369 | const char *last_str, | |
1370 | unsigned long int class256_bit, | |
1371 | unsigned long int class_bit, int base, | |
a0dc5206 | 1372 | int ignore_content, int handle_digits, int step) |
19bc17a9 | 1373 | { |
4b10dd6c UD |
1374 | const char *nowstr = now->val.str.startmb; |
1375 | char tmp[now->val.str.lenmb + 1]; | |
1376 | const char *cp; | |
1377 | char *endp; | |
1378 | unsigned long int from; | |
1379 | unsigned long int to; | |
19bc17a9 | 1380 | |
4b10dd6c UD |
1381 | /* We have to compute the ellipsis values using the symbolic names. */ |
1382 | assert (last_str != NULL); | |
1383 | ||
1384 | if (strlen (last_str) != now->val.str.lenmb) | |
19bc17a9 | 1385 | { |
4b10dd6c UD |
1386 | invalid_range: |
1387 | lr_error (ldfile, | |
549b3c3a | 1388 | _("`%s' and `%.*s' are no valid names for symbolic range"), |
f6ada7ad | 1389 | last_str, (int) now->val.str.lenmb, nowstr); |
4b10dd6c | 1390 | return; |
19bc17a9 RM |
1391 | } |
1392 | ||
4b10dd6c UD |
1393 | if (memcmp (last_str, nowstr, now->val.str.lenmb) == 0) |
1394 | /* Nothing to do, the names are the same. */ | |
1395 | return; | |
19bc17a9 | 1396 | |
4b10dd6c UD |
1397 | for (cp = last_str; *cp == *(nowstr + (cp - last_str)); ++cp) |
1398 | ; | |
19bc17a9 | 1399 | |
4b10dd6c UD |
1400 | errno = 0; |
1401 | from = strtoul (cp, &endp, base); | |
1402 | if ((from == UINT_MAX && errno == ERANGE) || *endp != '\0') | |
1403 | goto invalid_range; | |
19bc17a9 | 1404 | |
4b10dd6c | 1405 | to = strtoul (nowstr + (cp - last_str), &endp, base); |
549b3c3a UD |
1406 | if ((to == UINT_MAX && errno == ERANGE) |
1407 | || (endp - nowstr) != now->val.str.lenmb || from >= to) | |
4b10dd6c | 1408 | goto invalid_range; |
19bc17a9 | 1409 | |
4b10dd6c UD |
1410 | /* OK, we have a range FROM - TO. Now we can create the symbolic names. */ |
1411 | if (!ignore_content) | |
1412 | { | |
1413 | now->val.str.startmb = tmp; | |
a0dc5206 | 1414 | while ((from += step) <= to) |
4b10dd6c UD |
1415 | { |
1416 | struct charseq *seq; | |
1417 | uint32_t wch; | |
19bc17a9 | 1418 | |
4b10dd6c UD |
1419 | sprintf (tmp, (base == 10 ? "%.*s%0*d" : "%.*s%0*X"), cp - last_str, |
1420 | last_str, now->val.str.lenmb - (cp - last_str), from); | |
19bc17a9 | 1421 | |
4b10dd6c UD |
1422 | get_character (now, charmap, repertoire, &seq, &wch); |
1423 | ||
1424 | if (seq != NULL && seq->nbytes == 1) | |
1425 | /* Yep, we can store information about this byte sequence. */ | |
1426 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
19bc17a9 | 1427 | |
4b10dd6c UD |
1428 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) |
1429 | /* We have the UCS4 position. */ | |
1430 | *find_idx (ctype, &ctype->class_collection, | |
1431 | &ctype->class_collection_max, | |
1432 | &ctype->class_collection_act, wch) |= class_bit; | |
19bc17a9 | 1433 | |
4b10dd6c UD |
1434 | if (handle_digits == 1) |
1435 | { | |
1436 | /* We must store the digit values. */ | |
1437 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1438 | { | |
1439 | ctype->mbdigits_max *= 2; | |
1440 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1441 | (ctype->mbdigits_max | |
1442 | * sizeof (char *))); | |
1443 | ctype->wcdigits_max *= 2; | |
1444 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1445 | (ctype->wcdigits_max | |
1446 | * sizeof (uint32_t))); | |
1447 | } | |
1448 | ||
1449 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1450 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1451 | } | |
1452 | else if (handle_digits == 2) | |
1453 | { | |
1454 | /* We must store the digit values. */ | |
1455 | if (ctype->outdigits_act >= 10) | |
1456 | { | |
1457 | lr_error (ldfile, _("\ | |
1458 | %s: field `%s' does not contain exactly ten entries"), | |
1459 | "LC_CTYPE", "outdigit"); | |
1460 | return; | |
1461 | } | |
1462 | ||
1463 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1464 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1465 | ++ctype->outdigits_act; | |
1466 | } | |
1467 | } | |
1468 | } | |
19bc17a9 RM |
1469 | } |
1470 | ||
1471 | ||
a0dc5206 | 1472 | /* Ellipsis like in `<U1234>..<U2345>' or `<U1234>..(2)..<U2345>'. */ |
4b10dd6c UD |
1473 | static void |
1474 | charclass_ucs4_ellipsis (struct linereader *ldfile, | |
1475 | struct locale_ctype_t *ctype, | |
1476 | struct charmap_t *charmap, | |
1477 | struct repertoire_t *repertoire, | |
1478 | struct token *now, uint32_t last_wch, | |
1479 | unsigned long int class256_bit, | |
1480 | unsigned long int class_bit, int ignore_content, | |
a0dc5206 | 1481 | int handle_digits, int step) |
19bc17a9 | 1482 | { |
4b10dd6c | 1483 | if (last_wch > now->val.ucs4) |
19bc17a9 | 1484 | { |
4b10dd6c UD |
1485 | lr_error (ldfile, _("\ |
1486 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1487 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, now->val.ucs4, | |
1488 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, last_wch); | |
19bc17a9 RM |
1489 | return; |
1490 | } | |
1491 | ||
4b10dd6c | 1492 | if (!ignore_content) |
a0dc5206 | 1493 | while ((last_wch += step) <= now->val.ucs4) |
4b10dd6c UD |
1494 | { |
1495 | /* We have to find out whether there is a byte sequence corresponding | |
1496 | to this UCS4 value. */ | |
f0a4b6b1 UD |
1497 | struct charseq *seq; |
1498 | char utmp[10]; | |
1499 | ||
1500 | snprintf (utmp, sizeof (utmp), "U%08X", last_wch); | |
1501 | seq = charmap_find_value (charmap, utmp, 9); | |
a0dc5206 UD |
1502 | if (seq == NULL) |
1503 | { | |
1504 | snprintf (utmp, sizeof (utmp), "U%04X", last_wch); | |
1505 | seq = charmap_find_value (charmap, utmp, 5); | |
1506 | } | |
f0a4b6b1 UD |
1507 | |
1508 | if (seq == NULL) | |
1509 | /* Try looking in the repertoire map. */ | |
1510 | seq = repertoire_find_seq (repertoire, last_wch); | |
19bc17a9 | 1511 | |
4b10dd6c UD |
1512 | /* If this is the first time we look for this sequence create a new |
1513 | entry. */ | |
1514 | if (seq == NULL) | |
1515 | { | |
f0a4b6b1 UD |
1516 | static const struct charseq negative |
1517 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
19bc17a9 | 1518 | |
f0a4b6b1 UD |
1519 | /* Find the symbolic name for this UCS4 value. */ |
1520 | if (repertoire != NULL) | |
4b10dd6c | 1521 | { |
f0a4b6b1 UD |
1522 | const char *symbol = repertoire_find_symbol (repertoire, |
1523 | last_wch); | |
5866b131 UD |
1524 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, |
1525 | sizeof (uint32_t)); | |
f0a4b6b1 UD |
1526 | *newp = last_wch; |
1527 | ||
1528 | if (symbol != NULL) | |
1529 | /* We have a name, now search the multibyte value. */ | |
1530 | seq = charmap_find_value (charmap, symbol, strlen (symbol)); | |
1531 | ||
1532 | if (seq == NULL) | |
1533 | /* We have to create a fake entry. */ | |
1534 | seq = (struct charseq *) &negative; | |
1535 | else | |
1536 | seq->ucs4 = last_wch; | |
1537 | ||
5866b131 UD |
1538 | insert_entry (&repertoire->seq_table, newp, sizeof (uint32_t), |
1539 | seq); | |
4b10dd6c UD |
1540 | } |
1541 | else | |
f0a4b6b1 UD |
1542 | /* We have to create a fake entry. */ |
1543 | seq = (struct charseq *) &negative; | |
4b10dd6c UD |
1544 | } |
1545 | ||
1546 | /* We have a name, now search the multibyte value. */ | |
1547 | if (seq->ucs4 == last_wch && seq->nbytes == 1) | |
1548 | /* Yep, we can store information about this byte sequence. */ | |
1549 | ctype->class256_collection[(size_t) seq->bytes[0]] | |
1550 | |= class256_bit; | |
1551 | ||
1552 | /* And of course we have the UCS4 position. */ | |
5866b131 | 1553 | if (class_bit != 0) |
4b10dd6c UD |
1554 | *find_idx (ctype, &ctype->class_collection, |
1555 | &ctype->class_collection_max, | |
1556 | &ctype->class_collection_act, last_wch) |= class_bit; | |
1557 | ||
1558 | if (handle_digits == 1) | |
1559 | { | |
1560 | /* We must store the digit values. */ | |
1561 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1562 | { | |
1563 | ctype->mbdigits_max *= 2; | |
1564 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1565 | (ctype->mbdigits_max | |
1566 | * sizeof (char *))); | |
1567 | ctype->wcdigits_max *= 2; | |
1568 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1569 | (ctype->wcdigits_max | |
1570 | * sizeof (uint32_t))); | |
1571 | } | |
1572 | ||
1573 | ctype->mbdigits[ctype->mbdigits_act++] = (seq->ucs4 == last_wch | |
1574 | ? seq : NULL); | |
1575 | ctype->wcdigits[ctype->wcdigits_act++] = last_wch; | |
1576 | } | |
1577 | else if (handle_digits == 2) | |
1578 | { | |
1579 | /* We must store the digit values. */ | |
1580 | if (ctype->outdigits_act >= 10) | |
1581 | { | |
1582 | lr_error (ldfile, _("\ | |
1583 | %s: field `%s' does not contain exactly ten entries"), | |
1584 | "LC_CTYPE", "outdigit"); | |
1585 | return; | |
1586 | } | |
19bc17a9 | 1587 | |
4b10dd6c UD |
1588 | ctype->mboutdigits[ctype->outdigits_act] = (seq->ucs4 == last_wch |
1589 | ? seq : NULL); | |
1590 | ctype->wcoutdigits[ctype->outdigits_act] = last_wch; | |
1591 | ++ctype->outdigits_act; | |
1592 | } | |
1593 | } | |
19bc17a9 RM |
1594 | } |
1595 | ||
1596 | ||
4b10dd6c | 1597 | /* Ellipsis as in `/xea/x12.../xea/x34'. */ |
19bc17a9 | 1598 | static void |
4b10dd6c UD |
1599 | charclass_charcode_ellipsis (struct linereader *ldfile, |
1600 | struct locale_ctype_t *ctype, | |
1601 | struct charmap_t *charmap, | |
1602 | struct repertoire_t *repertoire, | |
1603 | struct token *now, char *last_charcode, | |
1604 | uint32_t last_charcode_len, | |
1605 | unsigned long int class256_bit, | |
1606 | unsigned long int class_bit, int ignore_content, | |
1607 | int handle_digits) | |
19bc17a9 | 1608 | { |
4b10dd6c UD |
1609 | /* First check whether the to-value is larger. */ |
1610 | if (now->val.charcode.nbytes != last_charcode_len) | |
1611 | { | |
1612 | lr_error (ldfile, _("\ | |
1613 | start end end character sequence of range must have the same length")); | |
1614 | return; | |
1615 | } | |
19bc17a9 | 1616 | |
4b10dd6c | 1617 | if (memcmp (last_charcode, now->val.charcode.bytes, last_charcode_len) > 0) |
19bc17a9 | 1618 | { |
4b10dd6c UD |
1619 | lr_error (ldfile, _("\ |
1620 | to-value character sequence is smaller than from-value sequence")); | |
19bc17a9 RM |
1621 | return; |
1622 | } | |
1623 | ||
4b10dd6c UD |
1624 | if (!ignore_content) |
1625 | { | |
1626 | do | |
1627 | { | |
1628 | /* Increment the byte sequence value. */ | |
1629 | struct charseq *seq; | |
1630 | uint32_t wch; | |
1631 | int i; | |
1632 | ||
1633 | for (i = last_charcode_len - 1; i >= 0; --i) | |
1634 | if (++last_charcode[i] != 0) | |
1635 | break; | |
1636 | ||
1637 | if (last_charcode_len == 1) | |
1638 | /* Of course we have the charcode value. */ | |
1639 | ctype->class256_collection[(size_t) last_charcode[0]] | |
1640 | |= class256_bit; | |
1641 | ||
1642 | /* Find the symbolic name. */ | |
1643 | seq = charmap_find_symbol (charmap, last_charcode, | |
1644 | last_charcode_len); | |
1645 | if (seq != NULL) | |
1646 | { | |
1647 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1648 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1649 | strlen (seq->name)); | |
f0a4b6b1 | 1650 | wch = seq == NULL ? ILLEGAL_CHAR_VALUE : seq->ucs4; |
4b10dd6c UD |
1651 | |
1652 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) | |
1653 | *find_idx (ctype, &ctype->class_collection, | |
1654 | &ctype->class_collection_max, | |
1655 | &ctype->class_collection_act, wch) |= class_bit; | |
1656 | } | |
1657 | else | |
1658 | wch = ILLEGAL_CHAR_VALUE; | |
19bc17a9 | 1659 | |
4b10dd6c UD |
1660 | if (handle_digits == 1) |
1661 | { | |
1662 | /* We must store the digit values. */ | |
1663 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1664 | { | |
1665 | ctype->mbdigits_max *= 2; | |
1666 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1667 | (ctype->mbdigits_max | |
1668 | * sizeof (char *))); | |
1669 | ctype->wcdigits_max *= 2; | |
1670 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1671 | (ctype->wcdigits_max | |
1672 | * sizeof (uint32_t))); | |
1673 | } | |
1674 | ||
1675 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1676 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1677 | seq->nbytes = last_charcode_len; | |
1678 | ||
1679 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1680 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1681 | } | |
1682 | else if (handle_digits == 2) | |
1683 | { | |
1684 | struct charseq *seq; | |
1685 | /* We must store the digit values. */ | |
1686 | if (ctype->outdigits_act >= 10) | |
1687 | { | |
1688 | lr_error (ldfile, _("\ | |
1689 | %s: field `%s' does not contain exactly ten entries"), | |
1690 | "LC_CTYPE", "outdigit"); | |
1691 | return; | |
1692 | } | |
1693 | ||
1694 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1695 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1696 | seq->nbytes = last_charcode_len; | |
1697 | ||
1698 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1699 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1700 | ++ctype->outdigits_act; | |
1701 | } | |
1702 | } | |
1703 | while (memcmp (last_charcode, now->val.charcode.bytes, | |
1704 | last_charcode_len) != 0); | |
1705 | } | |
19bc17a9 RM |
1706 | } |
1707 | ||
1708 | ||
4b10dd6c UD |
1709 | /* Read one transliteration entry. */ |
1710 | static uint32_t * | |
1711 | read_widestring (struct linereader *ldfile, struct token *now, | |
1712 | struct charmap_t *charmap, struct repertoire_t *repertoire) | |
19bc17a9 | 1713 | { |
4b10dd6c | 1714 | uint32_t *wstr; |
19bc17a9 | 1715 | |
4b10dd6c UD |
1716 | if (now->tok == tok_default_missing) |
1717 | /* The special name "" will denote this case. */ | |
5866b131 | 1718 | wstr = ((uint32_t *) { 0 }); |
4b10dd6c | 1719 | else if (now->tok == tok_bsymbol) |
19bc17a9 | 1720 | { |
4b10dd6c | 1721 | /* Get the value from the repertoire. */ |
a673fbcb | 1722 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1723 | wstr[0] = repertoire_find_value (repertoire, now->val.str.startmb, |
1724 | now->val.str.lenmb); | |
1725 | if (wstr[0] == ILLEGAL_CHAR_VALUE) | |
f0a4b6b1 UD |
1726 | { |
1727 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1728 | free (wstr); | |
1729 | return NULL; | |
1730 | } | |
4b10dd6c UD |
1731 | |
1732 | wstr[1] = 0; | |
19bc17a9 | 1733 | } |
4b10dd6c | 1734 | else if (now->tok == tok_ucs4) |
19bc17a9 | 1735 | { |
a673fbcb | 1736 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1737 | wstr[0] = now->val.ucs4; |
1738 | wstr[1] = 0; | |
1739 | } | |
1740 | else if (now->tok == tok_charcode) | |
1741 | { | |
1742 | /* Argh, we have to convert to the symbol name first and then to the | |
1743 | UCS4 value. */ | |
1744 | struct charseq *seq = charmap_find_symbol (charmap, | |
1745 | now->val.str.startmb, | |
1746 | now->val.str.lenmb); | |
1747 | if (seq == NULL) | |
1748 | /* Cannot find the UCS4 value. */ | |
1749 | return NULL; | |
1750 | ||
1751 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1752 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1753 | strlen (seq->name)); | |
1754 | if (seq->ucs4 == ILLEGAL_CHAR_VALUE) | |
1755 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1756 | return NULL; | |
1757 | ||
a673fbcb | 1758 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1759 | wstr[0] = seq->ucs4; |
1760 | wstr[1] = 0; | |
1761 | } | |
1762 | else if (now->tok == tok_string) | |
1763 | { | |
1764 | wstr = now->val.str.startwc; | |
a673fbcb | 1765 | if (wstr == NULL || wstr[0] == 0) |
4b10dd6c UD |
1766 | return NULL; |
1767 | } | |
1768 | else | |
1769 | { | |
1770 | if (now->tok != tok_eol && now->tok != tok_eof) | |
1771 | lr_ignore_rest (ldfile, 0); | |
1772 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
1773 | return (uint32_t *) -1l; | |
19bc17a9 RM |
1774 | } |
1775 | ||
4b10dd6c UD |
1776 | return wstr; |
1777 | } | |
19bc17a9 | 1778 | |
19bc17a9 | 1779 | |
4b10dd6c UD |
1780 | static void |
1781 | read_translit_entry (struct linereader *ldfile, struct locale_ctype_t *ctype, | |
1782 | struct token *now, struct charmap_t *charmap, | |
1783 | struct repertoire_t *repertoire) | |
1784 | { | |
1785 | uint32_t *from_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1786 | struct translit_t *result; | |
1787 | struct translit_to_t **top; | |
a673fbcb | 1788 | struct obstack *ob = &ctype->mempool; |
4b10dd6c UD |
1789 | int first; |
1790 | int ignore; | |
1791 | ||
1792 | if (from_wstr == NULL) | |
1793 | /* There is no valid from string. */ | |
1794 | return; | |
19bc17a9 | 1795 | |
4b10dd6c UD |
1796 | result = (struct translit_t *) obstack_alloc (ob, |
1797 | sizeof (struct translit_t)); | |
1798 | result->from = from_wstr; | |
a673fbcb UD |
1799 | result->fname = ldfile->fname; |
1800 | result->lineno = ldfile->lineno; | |
4b10dd6c UD |
1801 | result->next = NULL; |
1802 | result->to = NULL; | |
1803 | top = &result->to; | |
1804 | first = 1; | |
1805 | ignore = 0; | |
1806 | ||
1807 | while (1) | |
1808 | { | |
1809 | uint32_t *to_wstr; | |
1810 | ||
1811 | /* Next we have one or more transliterations. They are | |
1812 | separated by semicolons. */ | |
1813 | now = lr_token (ldfile, charmap, repertoire); | |
1814 | ||
1815 | if (!first && (now->tok == tok_semicolon || now->tok == tok_eol)) | |
1816 | { | |
1817 | /* One string read. */ | |
1818 | const uint32_t zero = 0; | |
1819 | ||
1820 | if (!ignore) | |
1821 | { | |
1822 | obstack_grow (ob, &zero, 4); | |
1823 | to_wstr = obstack_finish (ob); | |
1824 | ||
1825 | *top = obstack_alloc (ob, sizeof (struct translit_to_t)); | |
1826 | (*top)->str = to_wstr; | |
1827 | (*top)->next = NULL; | |
1828 | } | |
1829 | ||
1830 | if (now->tok == tok_eol) | |
1831 | { | |
1832 | result->next = ctype->translit; | |
1833 | ctype->translit = result; | |
1834 | return; | |
1835 | } | |
1836 | ||
1837 | if (!ignore) | |
1838 | top = &(*top)->next; | |
1839 | ignore = 0; | |
1840 | } | |
1841 | else | |
1842 | { | |
1843 | to_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1844 | if (to_wstr == (uint32_t *) -1l) | |
1845 | { | |
1846 | /* An error occurred. */ | |
1847 | obstack_free (ob, result); | |
1848 | return; | |
1849 | } | |
1850 | ||
1851 | if (to_wstr == NULL) | |
1852 | ignore = 1; | |
1853 | else | |
1854 | /* This value is usable. */ | |
1855 | obstack_grow (ob, to_wstr, wcslen ((wchar_t *) to_wstr) * 4); | |
19bc17a9 | 1856 | |
4b10dd6c UD |
1857 | first = 0; |
1858 | } | |
1859 | } | |
19bc17a9 RM |
1860 | } |
1861 | ||
1862 | ||
a673fbcb UD |
1863 | static void |
1864 | read_translit_ignore_entry (struct linereader *ldfile, | |
1865 | struct locale_ctype_t *ctype, | |
1866 | struct charmap_t *charmap, | |
1867 | struct repertoire_t *repertoire) | |
1868 | { | |
1869 | /* We expect a semicolon-separated list of characters we ignore. We are | |
1870 | only interested in the wide character definitions. These must be | |
1871 | single characters, possibly defining a range when an ellipsis is used. */ | |
1872 | while (1) | |
1873 | { | |
1874 | struct token *now = lr_token (ldfile, charmap, repertoire); | |
1875 | struct translit_ignore_t *newp; | |
1876 | uint32_t from; | |
1877 | ||
1878 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1879 | { | |
1880 | lr_error (ldfile, | |
1881 | _("premature end of `translit_ignore' definition")); | |
1882 | return; | |
1883 | } | |
1884 | ||
1885 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
1886 | { | |
1887 | lr_error (ldfile, _("syntax error")); | |
1888 | lr_ignore_rest (ldfile, 0); | |
1889 | return; | |
1890 | } | |
1891 | ||
1892 | if (now->tok == tok_ucs4) | |
1893 | from = now->val.ucs4; | |
1894 | else | |
f0a4b6b1 UD |
1895 | /* Try to get the value. */ |
1896 | from = repertoire_find_value (repertoire, now->val.str.startmb, | |
1897 | now->val.str.lenmb); | |
a673fbcb UD |
1898 | |
1899 | if (from == ILLEGAL_CHAR_VALUE) | |
1900 | { | |
1901 | lr_error (ldfile, "invalid character name"); | |
1902 | newp = NULL; | |
1903 | } | |
1904 | else | |
1905 | { | |
1906 | newp = (struct translit_ignore_t *) | |
1907 | obstack_alloc (&ctype->mempool, sizeof (struct translit_ignore_t)); | |
1908 | newp->from = from; | |
1909 | newp->to = from; | |
a0dc5206 | 1910 | newp->step = 1; |
a673fbcb UD |
1911 | |
1912 | newp->next = ctype->translit_ignore; | |
1913 | ctype->translit_ignore = newp; | |
1914 | } | |
1915 | ||
1916 | /* Now we expect either a semicolon, an ellipsis, or the end of the | |
1917 | line. */ | |
1918 | now = lr_token (ldfile, charmap, repertoire); | |
1919 | ||
a0dc5206 | 1920 | if (now->tok == tok_ellipsis2 || now->tok == tok_ellipsis2_2) |
a673fbcb UD |
1921 | { |
1922 | /* XXX Should we bother implementing `....'? `...' certainly | |
1923 | will not be implemented. */ | |
1924 | uint32_t to; | |
a0dc5206 | 1925 | int step = now->tok == tok_ellipsis2_2 ? 2 : 1; |
a673fbcb UD |
1926 | |
1927 | now = lr_token (ldfile, charmap, repertoire); | |
1928 | ||
1929 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1930 | { | |
1931 | lr_error (ldfile, | |
1932 | _("premature end of `translit_ignore' definition")); | |
1933 | return; | |
1934 | } | |
1935 | ||
1936 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
1937 | { | |
1938 | lr_error (ldfile, _("syntax error")); | |
1939 | lr_ignore_rest (ldfile, 0); | |
1940 | return; | |
1941 | } | |
1942 | ||
1943 | if (now->tok == tok_ucs4) | |
1944 | to = now->val.ucs4; | |
1945 | else | |
f0a4b6b1 UD |
1946 | /* Try to get the value. */ |
1947 | to = repertoire_find_value (repertoire, now->val.str.startmb, | |
1948 | now->val.str.lenmb); | |
a673fbcb UD |
1949 | |
1950 | if (to == ILLEGAL_CHAR_VALUE) | |
1951 | lr_error (ldfile, "invalid character name"); | |
1952 | else | |
1953 | { | |
1954 | /* Make sure the `to'-value is larger. */ | |
1955 | if (to >= from) | |
a0dc5206 UD |
1956 | { |
1957 | newp->to = to; | |
1958 | newp->step = step; | |
1959 | } | |
a673fbcb UD |
1960 | else |
1961 | lr_error (ldfile, _("\ | |
1962 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1963 | (to | from) < 65536 ? 4 : 8, to, | |
1964 | (to | from) < 65536 ? 4 : 8, from); | |
1965 | } | |
1966 | ||
1967 | /* And the next token. */ | |
1968 | now = lr_token (ldfile, charmap, repertoire); | |
1969 | } | |
1970 | ||
1971 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1972 | /* We are done. */ | |
1973 | return; | |
1974 | ||
1975 | if (now->tok == tok_semicolon) | |
1976 | /* Next round. */ | |
1977 | continue; | |
1978 | ||
1979 | /* If we come here something is wrong. */ | |
1980 | lr_error (ldfile, _("syntax error")); | |
1981 | lr_ignore_rest (ldfile, 0); | |
1982 | return; | |
1983 | } | |
1984 | } | |
1985 | ||
1986 | ||
4b10dd6c UD |
1987 | /* The parser for the LC_CTYPE section of the locale definition. */ |
1988 | void | |
1989 | ctype_read (struct linereader *ldfile, struct localedef_t *result, | |
1990 | struct charmap_t *charmap, const char *repertoire_name, | |
1991 | int ignore_content) | |
19bc17a9 | 1992 | { |
4b10dd6c UD |
1993 | struct repertoire_t *repertoire = NULL; |
1994 | struct locale_ctype_t *ctype; | |
1995 | struct token *now; | |
1996 | enum token_t nowtok; | |
19bc17a9 | 1997 | size_t cnt; |
4b10dd6c UD |
1998 | struct charseq *last_seq; |
1999 | uint32_t last_wch = 0; | |
2000 | enum token_t last_token; | |
2001 | enum token_t ellipsis_token; | |
a0dc5206 | 2002 | int step; |
4b10dd6c UD |
2003 | char last_charcode[16]; |
2004 | size_t last_charcode_len = 0; | |
2005 | const char *last_str = NULL; | |
2006 | int mapidx; | |
19bc17a9 | 2007 | |
4b10dd6c UD |
2008 | /* Get the repertoire we have to use. */ |
2009 | if (repertoire_name != NULL) | |
2010 | repertoire = repertoire_read (repertoire_name); | |
19bc17a9 | 2011 | |
4b10dd6c UD |
2012 | /* The rest of the line containing `LC_CTYPE' must be free. */ |
2013 | lr_ignore_rest (ldfile, 1); | |
19bc17a9 | 2014 | |
4b10dd6c UD |
2015 | |
2016 | do | |
19bc17a9 | 2017 | { |
4b10dd6c UD |
2018 | now = lr_token (ldfile, charmap, NULL); |
2019 | nowtok = now->tok; | |
19bc17a9 | 2020 | } |
4b10dd6c | 2021 | while (nowtok == tok_eol); |
19bc17a9 | 2022 | |
4b10dd6c UD |
2023 | /* If we see `copy' now we are almost done. */ |
2024 | if (nowtok == tok_copy) | |
2025 | { | |
01ff9d0b UD |
2026 | handle_copy (ldfile, charmap, repertoire_name, result, tok_lc_ctype, |
2027 | LC_CTYPE, "LC_CTYPE", ignore_content); | |
4b10dd6c UD |
2028 | return; |
2029 | } | |
75cd5204 | 2030 | |
4b10dd6c UD |
2031 | /* Prepare the data structures. */ |
2032 | ctype_startup (ldfile, result, charmap, ignore_content); | |
2033 | ctype = result->categories[LC_CTYPE].ctype; | |
2034 | ||
2035 | /* Remember the repertoire we use. */ | |
2036 | if (!ignore_content) | |
2037 | ctype->repertoire = repertoire; | |
2038 | ||
2039 | while (1) | |
19bc17a9 | 2040 | { |
4b10dd6c UD |
2041 | unsigned long int class_bit = 0; |
2042 | unsigned long int class256_bit = 0; | |
2043 | int handle_digits = 0; | |
2044 | ||
2045 | /* Of course we don't proceed beyond the end of file. */ | |
2046 | if (nowtok == tok_eof) | |
2047 | break; | |
2048 | ||
2049 | /* Ingore empty lines. */ | |
2050 | if (nowtok == tok_eol) | |
19bc17a9 | 2051 | { |
4b10dd6c UD |
2052 | now = lr_token (ldfile, charmap, NULL); |
2053 | nowtok = now->tok; | |
2054 | continue; | |
2055 | } | |
19bc17a9 | 2056 | |
4b10dd6c UD |
2057 | switch (nowtok) |
2058 | { | |
5491da0d UD |
2059 | case tok_charclass: |
2060 | now = lr_token (ldfile, charmap, NULL); | |
2061 | while (now->tok == tok_ident || now->tok == tok_string) | |
2062 | { | |
2063 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
2064 | now = lr_token (ldfile, charmap, NULL); | |
2065 | if (now->tok != tok_semicolon) | |
2066 | break; | |
2067 | now = lr_token (ldfile, charmap, NULL); | |
2068 | } | |
2069 | if (now->tok != tok_eol) | |
2070 | SYNTAX_ERROR (_("\ | |
2071 | %s: syntax error in definition of new character class"), "LC_CTYPE"); | |
2072 | break; | |
2073 | ||
2074 | case tok_charconv: | |
2075 | now = lr_token (ldfile, charmap, NULL); | |
2076 | while (now->tok == tok_ident || now->tok == tok_string) | |
2077 | { | |
2078 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2079 | now = lr_token (ldfile, charmap, NULL); | |
2080 | if (now->tok != tok_semicolon) | |
2081 | break; | |
2082 | now = lr_token (ldfile, charmap, NULL); | |
2083 | } | |
2084 | if (now->tok != tok_eol) | |
2085 | SYNTAX_ERROR (_("\ | |
2086 | %s: syntax error in definition of new character map"), "LC_CTYPE"); | |
2087 | break; | |
2088 | ||
4b10dd6c | 2089 | case tok_class: |
b9eb05d6 UD |
2090 | /* Ignore the rest of the line if we don't need the input of |
2091 | this line. */ | |
2092 | if (ignore_content) | |
2093 | { | |
2094 | lr_ignore_rest (ldfile, 0); | |
2095 | break; | |
2096 | } | |
2097 | ||
4b10dd6c UD |
2098 | /* We simply forget the `class' keyword and use the following |
2099 | operand to determine the bit. */ | |
2100 | now = lr_token (ldfile, charmap, NULL); | |
2101 | if (now->tok == tok_ident || now->tok == tok_string) | |
2102 | { | |
87372aa9 | 2103 | /* Must can be one of the predefined class names. */ |
4b10dd6c UD |
2104 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
2105 | if (strcmp (ctype->classnames[cnt], now->val.str.startmb) == 0) | |
2106 | break; | |
2107 | if (cnt >= ctype->nr_charclass) | |
2108 | { | |
011ebfab | 2109 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2110 | if (now->val.str.lenmb == 8 |
2111 | && memcmp ("special1", now->val.str.startmb, 8) == 0) | |
2112 | class_bit = _ISwspecial1; | |
2113 | else if (now->val.str.lenmb == 8 | |
2114 | && memcmp ("special2", now->val.str.startmb, 8) == 0) | |
2115 | class_bit = _ISwspecial2; | |
2116 | else if (now->val.str.lenmb == 8 | |
2117 | && memcmp ("special3", now->val.str.startmb, 8) == 0) | |
2118 | class_bit = _ISwspecial3; | |
2119 | else | |
011ebfab | 2120 | #endif |
4b10dd6c | 2121 | { |
87372aa9 UD |
2122 | /* OK, it's a new class. */ |
2123 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
4b10dd6c | 2124 | |
87372aa9 | 2125 | class_bit = _ISwbit (ctype->nr_charclass - 1); |
4b10dd6c UD |
2126 | } |
2127 | } | |
2128 | else | |
7f653277 UD |
2129 | { |
2130 | class_bit = _ISwbit (cnt); | |
4b10dd6c | 2131 | |
7f653277 UD |
2132 | free (now->val.str.startmb); |
2133 | } | |
4b10dd6c UD |
2134 | } |
2135 | else if (now->tok == tok_digit) | |
2136 | goto handle_tok_digit; | |
2137 | else if (now->tok < tok_upper || now->tok > tok_blank) | |
2138 | goto err_label; | |
2139 | else | |
2140 | { | |
2141 | class_bit = BITw (now->tok); | |
2142 | class256_bit = BIT (now->tok); | |
2143 | } | |
2144 | ||
2145 | /* The next character must be a semicolon. */ | |
2146 | now = lr_token (ldfile, charmap, NULL); | |
2147 | if (now->tok != tok_semicolon) | |
2148 | goto err_label; | |
2149 | goto read_charclass; | |
2150 | ||
2151 | case tok_upper: | |
2152 | case tok_lower: | |
2153 | case tok_alpha: | |
2154 | case tok_alnum: | |
2155 | case tok_space: | |
2156 | case tok_cntrl: | |
2157 | case tok_punct: | |
2158 | case tok_graph: | |
2159 | case tok_print: | |
2160 | case tok_xdigit: | |
2161 | case tok_blank: | |
b9eb05d6 UD |
2162 | /* Ignore the rest of the line if we don't need the input of |
2163 | this line. */ | |
2164 | if (ignore_content) | |
2165 | { | |
2166 | lr_ignore_rest (ldfile, 0); | |
2167 | break; | |
2168 | } | |
2169 | ||
4b10dd6c UD |
2170 | class_bit = BITw (now->tok); |
2171 | class256_bit = BIT (now->tok); | |
2172 | handle_digits = 0; | |
2173 | read_charclass: | |
2174 | ctype->class_done |= class_bit; | |
2175 | last_token = tok_none; | |
2176 | ellipsis_token = tok_none; | |
a0dc5206 | 2177 | step = 1; |
4b10dd6c UD |
2178 | now = lr_token (ldfile, charmap, NULL); |
2179 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2180 | { | |
2181 | uint32_t wch; | |
2182 | struct charseq *seq; | |
2183 | ||
2184 | if (ellipsis_token == tok_none) | |
2185 | { | |
2186 | if (get_character (now, charmap, repertoire, &seq, &wch)) | |
2187 | goto err_label; | |
2188 | ||
2189 | if (!ignore_content && seq != NULL && seq->nbytes == 1) | |
2190 | /* Yep, we can store information about this byte | |
2191 | sequence. */ | |
2192 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
2193 | ||
2194 | if (!ignore_content && wch != ILLEGAL_CHAR_VALUE | |
2195 | && class_bit != 0) | |
2196 | /* We have the UCS4 position. */ | |
2197 | *find_idx (ctype, &ctype->class_collection, | |
2198 | &ctype->class_collection_max, | |
2199 | &ctype->class_collection_act, wch) |= class_bit; | |
2200 | ||
2201 | last_token = now->tok; | |
549b3c3a | 2202 | /* Terminate the string. */ |
9e2b7438 UD |
2203 | if (last_token == tok_bsymbol) |
2204 | { | |
2205 | now->val.str.startmb[now->val.str.lenmb] = '\0'; | |
2206 | last_str = now->val.str.startmb; | |
2207 | } | |
2208 | else | |
2209 | last_str = NULL; | |
4b10dd6c UD |
2210 | last_seq = seq; |
2211 | last_wch = wch; | |
2212 | memcpy (last_charcode, now->val.charcode.bytes, 16); | |
2213 | last_charcode_len = now->val.charcode.nbytes; | |
2214 | ||
2215 | if (!ignore_content && handle_digits == 1) | |
2216 | { | |
2217 | /* We must store the digit values. */ | |
2218 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
2219 | { | |
b9eb05d6 | 2220 | ctype->mbdigits_max += 10; |
4b10dd6c UD |
2221 | ctype->mbdigits = xrealloc (ctype->mbdigits, |
2222 | (ctype->mbdigits_max | |
2223 | * sizeof (char *))); | |
b9eb05d6 | 2224 | ctype->wcdigits_max += 10; |
4b10dd6c UD |
2225 | ctype->wcdigits = xrealloc (ctype->wcdigits, |
2226 | (ctype->wcdigits_max | |
2227 | * sizeof (uint32_t))); | |
2228 | } | |
2229 | ||
2230 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
2231 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
2232 | } | |
2233 | else if (!ignore_content && handle_digits == 2) | |
2234 | { | |
2235 | /* We must store the digit values. */ | |
2236 | if (ctype->outdigits_act >= 10) | |
2237 | { | |
2238 | lr_error (ldfile, _("\ | |
2239 | %s: field `%s' does not contain exactly ten entries"), | |
2240 | "LC_CTYPE", "outdigit"); | |
2241 | goto err_label; | |
2242 | } | |
2243 | ||
2244 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
2245 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
2246 | ++ctype->outdigits_act; | |
2247 | } | |
2248 | } | |
2249 | else | |
2250 | { | |
2251 | /* Now it gets complicated. We have to resolve the | |
2252 | ellipsis problem. First we must distinguish between | |
2253 | the different kind of ellipsis and this must match the | |
2254 | tokens we have seen. */ | |
2255 | assert (last_token != tok_none); | |
2256 | ||
2257 | if (last_token != now->tok) | |
2258 | { | |
2259 | lr_error (ldfile, _("\ | |
2260 | ellipsis range must be marked by two operands of same type")); | |
2261 | lr_ignore_rest (ldfile, 0); | |
2262 | break; | |
2263 | } | |
2264 | ||
2265 | if (last_token == tok_bsymbol) | |
2266 | { | |
2267 | if (ellipsis_token == tok_ellipsis3) | |
2268 | lr_error (ldfile, _("with symbolic name range values \ | |
2269 | the absolute ellipsis `...' must not be used")); | |
2270 | ||
2271 | charclass_symbolic_ellipsis (ldfile, ctype, charmap, | |
2272 | repertoire, now, last_str, | |
2273 | class256_bit, class_bit, | |
2274 | (ellipsis_token | |
2275 | == tok_ellipsis4 | |
2276 | ? 10 : 16), | |
2277 | ignore_content, | |
a0dc5206 | 2278 | handle_digits, step); |
4b10dd6c UD |
2279 | } |
2280 | else if (last_token == tok_ucs4) | |
2281 | { | |
2282 | if (ellipsis_token != tok_ellipsis2) | |
2283 | lr_error (ldfile, _("\ | |
2284 | with UCS range values one must use the hexadecimal symbolic ellipsis `..'")); | |
2285 | ||
2286 | charclass_ucs4_ellipsis (ldfile, ctype, charmap, | |
2287 | repertoire, now, last_wch, | |
2288 | class256_bit, class_bit, | |
a0dc5206 UD |
2289 | ignore_content, handle_digits, |
2290 | step); | |
4b10dd6c UD |
2291 | } |
2292 | else | |
2293 | { | |
2294 | assert (last_token == tok_charcode); | |
2295 | ||
2296 | if (ellipsis_token != tok_ellipsis3) | |
2297 | lr_error (ldfile, _("\ | |
2298 | with character code range values one must use the absolute ellipsis `...'")); | |
2299 | ||
2300 | charclass_charcode_ellipsis (ldfile, ctype, charmap, | |
2301 | repertoire, now, | |
2302 | last_charcode, | |
2303 | last_charcode_len, | |
2304 | class256_bit, class_bit, | |
2305 | ignore_content, | |
2306 | handle_digits); | |
2307 | } | |
2308 | ||
2309 | /* Now we have used the last value. */ | |
2310 | last_token = tok_none; | |
2311 | } | |
2312 | ||
2313 | /* Next we expect a semicolon or the end of the line. */ | |
2314 | now = lr_token (ldfile, charmap, NULL); | |
2315 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2316 | break; | |
2317 | ||
2318 | if (last_token != tok_none | |
a0dc5206 | 2319 | && now->tok >= tok_ellipsis2 && now->tok <= tok_ellipsis4_2) |
4b10dd6c | 2320 | { |
a0dc5206 UD |
2321 | if (now->tok == tok_ellipsis2_2) |
2322 | { | |
2323 | now->tok = tok_ellipsis2; | |
2324 | step = 2; | |
2325 | } | |
2326 | else if (now->tok == tok_ellipsis4_2) | |
2327 | { | |
2328 | now->tok = tok_ellipsis4; | |
2329 | step = 2; | |
2330 | } | |
2331 | ||
4b10dd6c | 2332 | ellipsis_token = now->tok; |
a0dc5206 | 2333 | |
4b10dd6c UD |
2334 | now = lr_token (ldfile, charmap, NULL); |
2335 | continue; | |
2336 | } | |
2337 | ||
2338 | if (now->tok != tok_semicolon) | |
2339 | goto err_label; | |
2340 | ||
2341 | /* And get the next character. */ | |
2342 | now = lr_token (ldfile, charmap, NULL); | |
2343 | ||
2344 | ellipsis_token = tok_none; | |
a0dc5206 | 2345 | step = 1; |
4b10dd6c UD |
2346 | } |
2347 | break; | |
2348 | ||
2349 | case tok_digit: | |
b9eb05d6 UD |
2350 | /* Ignore the rest of the line if we don't need the input of |
2351 | this line. */ | |
2352 | if (ignore_content) | |
42d7c593 UD |
2353 | { |
2354 | lr_ignore_rest (ldfile, 0); | |
2355 | break; | |
2356 | } | |
b9eb05d6 | 2357 | |
4b10dd6c UD |
2358 | handle_tok_digit: |
2359 | class_bit = _ISwdigit; | |
2360 | class256_bit = _ISdigit; | |
2361 | handle_digits = 1; | |
2362 | goto read_charclass; | |
2363 | ||
2364 | case tok_outdigit: | |
b9eb05d6 UD |
2365 | /* Ignore the rest of the line if we don't need the input of |
2366 | this line. */ | |
2367 | if (ignore_content) | |
2368 | { | |
2369 | lr_ignore_rest (ldfile, 0); | |
2370 | break; | |
2371 | } | |
2372 | ||
4b10dd6c UD |
2373 | if (ctype->outdigits_act != 0) |
2374 | lr_error (ldfile, _("\ | |
2375 | %s: field `%s' declared more than once"), | |
2376 | "LC_CTYPE", "outdigit"); | |
2377 | class_bit = 0; | |
2378 | class256_bit = 0; | |
2379 | handle_digits = 2; | |
2380 | goto read_charclass; | |
2381 | ||
2382 | case tok_toupper: | |
b9eb05d6 UD |
2383 | /* Ignore the rest of the line if we don't need the input of |
2384 | this line. */ | |
2385 | if (ignore_content) | |
2386 | { | |
2387 | lr_ignore_rest (ldfile, 0); | |
2388 | break; | |
2389 | } | |
2390 | ||
4b10dd6c UD |
2391 | mapidx = 0; |
2392 | goto read_mapping; | |
2393 | ||
2394 | case tok_tolower: | |
b9eb05d6 UD |
2395 | /* Ignore the rest of the line if we don't need the input of |
2396 | this line. */ | |
2397 | if (ignore_content) | |
2398 | { | |
2399 | lr_ignore_rest (ldfile, 0); | |
2400 | break; | |
2401 | } | |
2402 | ||
4b10dd6c UD |
2403 | mapidx = 1; |
2404 | goto read_mapping; | |
2405 | ||
2406 | case tok_map: | |
b9eb05d6 UD |
2407 | /* Ignore the rest of the line if we don't need the input of |
2408 | this line. */ | |
2409 | if (ignore_content) | |
2410 | { | |
2411 | lr_ignore_rest (ldfile, 0); | |
2412 | break; | |
2413 | } | |
2414 | ||
4b10dd6c UD |
2415 | /* We simply forget the `map' keyword and use the following |
2416 | operand to determine the mapping. */ | |
2417 | now = lr_token (ldfile, charmap, NULL); | |
2418 | if (now->tok == tok_ident || now->tok == tok_string) | |
2419 | { | |
2420 | size_t cnt; | |
2421 | ||
2422 | for (cnt = 2; cnt < ctype->map_collection_nr; ++cnt) | |
2423 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2424 | break; | |
2425 | ||
7f653277 UD |
2426 | if (cnt < ctype->map_collection_nr) |
2427 | free (now->val.str.startmb); | |
2428 | else | |
87372aa9 UD |
2429 | /* OK, it's a new map. */ |
2430 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2431 | ||
2432 | mapidx = cnt; | |
4b10dd6c UD |
2433 | } |
2434 | else if (now->tok < tok_toupper || now->tok > tok_tolower) | |
2435 | goto err_label; | |
2436 | else | |
2437 | mapidx = now->tok - tok_toupper; | |
2438 | ||
2439 | now = lr_token (ldfile, charmap, NULL); | |
2440 | /* This better should be a semicolon. */ | |
2441 | if (now->tok != tok_semicolon) | |
2442 | goto err_label; | |
2443 | ||
2444 | read_mapping: | |
2445 | /* Test whether this mapping was already defined. */ | |
2446 | if (ctype->tomap_done[mapidx]) | |
2447 | { | |
2448 | lr_error (ldfile, _("duplicated definition for mapping `%s'"), | |
2449 | ctype->mapnames[mapidx]); | |
2450 | lr_ignore_rest (ldfile, 0); | |
2451 | break; | |
2452 | } | |
2453 | ctype->tomap_done[mapidx] = 1; | |
2454 | ||
2455 | now = lr_token (ldfile, charmap, NULL); | |
2456 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2457 | { | |
2458 | struct charseq *from_seq; | |
2459 | uint32_t from_wch; | |
2460 | struct charseq *to_seq; | |
2461 | uint32_t to_wch; | |
2462 | ||
2463 | /* Every pair starts with an opening brace. */ | |
2464 | if (now->tok != tok_open_brace) | |
2465 | goto err_label; | |
2466 | ||
2467 | /* Next comes the from-value. */ | |
2468 | now = lr_token (ldfile, charmap, NULL); | |
2469 | if (get_character (now, charmap, repertoire, &from_seq, | |
2470 | &from_wch) != 0) | |
2471 | goto err_label; | |
2472 | ||
2473 | /* The next is a comma. */ | |
2474 | now = lr_token (ldfile, charmap, NULL); | |
2475 | if (now->tok != tok_comma) | |
2476 | goto err_label; | |
2477 | ||
2478 | /* And the other value. */ | |
2479 | now = lr_token (ldfile, charmap, NULL); | |
2480 | if (get_character (now, charmap, repertoire, &to_seq, | |
2481 | &to_wch) != 0) | |
2482 | goto err_label; | |
2483 | ||
2484 | /* And the last thing is the closing brace. */ | |
2485 | now = lr_token (ldfile, charmap, NULL); | |
2486 | if (now->tok != tok_close_brace) | |
2487 | goto err_label; | |
2488 | ||
2489 | if (!ignore_content) | |
2490 | { | |
2491 | if (mapidx < 2 && from_seq != NULL && to_seq != NULL | |
2492 | && from_seq->nbytes == 1 && to_seq->nbytes == 1) | |
2493 | /* We can use this value. */ | |
2494 | ctype->map256_collection[mapidx][from_seq->bytes[0]] | |
2495 | = to_seq->bytes[0]; | |
2496 | ||
2497 | if (from_wch != ILLEGAL_CHAR_VALUE | |
2498 | && to_wch != ILLEGAL_CHAR_VALUE) | |
2499 | /* Both correct values. */ | |
2500 | *find_idx (ctype, &ctype->map_collection[mapidx], | |
2501 | &ctype->map_collection_max[mapidx], | |
2502 | &ctype->map_collection_act[mapidx], | |
2503 | from_wch) = to_wch; | |
2504 | } | |
2505 | ||
2506 | /* Now comes a semicolon or the end of the line/file. */ | |
2507 | now = lr_token (ldfile, charmap, NULL); | |
2508 | if (now->tok == tok_semicolon) | |
2509 | now = lr_token (ldfile, charmap, NULL); | |
2510 | } | |
2511 | break; | |
2512 | ||
2513 | case tok_translit_start: | |
b9eb05d6 UD |
2514 | /* Ignore the rest of the line if we don't need the input of |
2515 | this line. */ | |
2516 | if (ignore_content) | |
2517 | { | |
2518 | lr_ignore_rest (ldfile, 0); | |
2519 | break; | |
2520 | } | |
2521 | ||
4b10dd6c UD |
2522 | /* The rest of the line better should be empty. */ |
2523 | lr_ignore_rest (ldfile, 1); | |
2524 | ||
2525 | /* We count here the number of allocated entries in the `translit' | |
2526 | array. */ | |
2527 | cnt = 0; | |
2528 | ||
2529 | /* We proceed until we see the `translit_end' token. */ | |
2530 | while (now = lr_token (ldfile, charmap, repertoire), | |
2531 | now->tok != tok_translit_end && now->tok != tok_eof) | |
2532 | { | |
2533 | if (now->tok == tok_eol) | |
2534 | /* Ignore empty lines. */ | |
2535 | continue; | |
2536 | ||
2537 | if (now->tok == tok_translit_end) | |
2538 | { | |
2539 | lr_ignore_rest (ldfile, 0); | |
2540 | break; | |
2541 | } | |
2542 | ||
2543 | if (now->tok == tok_include) | |
2544 | { | |
2545 | /* We have to include locale. */ | |
2546 | const char *locale_name; | |
2547 | const char *repertoire_name; | |
2548 | ||
2549 | now = lr_token (ldfile, charmap, NULL); | |
2550 | /* This should be a string or an identifier. In any | |
2551 | case something to name a locale. */ | |
2552 | if (now->tok != tok_string && now->tok != tok_ident) | |
2553 | { | |
2554 | translit_syntax: | |
2555 | lr_error (ldfile, _("%s: syntax error"), "LC_CTYPE"); | |
2556 | lr_ignore_rest (ldfile, 0); | |
2557 | continue; | |
2558 | } | |
2559 | locale_name = now->val.str.startmb; | |
2560 | ||
2561 | /* Next should be a semicolon. */ | |
2562 | now = lr_token (ldfile, charmap, NULL); | |
2563 | if (now->tok != tok_semicolon) | |
2564 | goto translit_syntax; | |
2565 | ||
2566 | /* Now the repertoire name. */ | |
2567 | now = lr_token (ldfile, charmap, NULL); | |
2568 | if ((now->tok != tok_string && now->tok != tok_ident) | |
2569 | || now->val.str.startmb == NULL) | |
2570 | goto translit_syntax; | |
2571 | repertoire_name = now->val.str.startmb; | |
2572 | ||
2573 | /* We must not have more than one `include'. */ | |
2574 | if (ctype->translit_copy_locale != NULL) | |
2575 | { | |
2576 | lr_error (ldfile, _("\ | |
2577 | %s: only one `include' instruction allowed"), "LC_CTYPE"); | |
2578 | lr_ignore_rest (ldfile, 0); | |
2579 | continue; | |
2580 | } | |
2581 | ||
2582 | ctype->translit_copy_locale = locale_name; | |
2583 | ctype->translit_copy_repertoire = repertoire_name; | |
2584 | ||
2585 | /* The rest of the line must be empty. */ | |
2586 | lr_ignore_rest (ldfile, 1); | |
a673fbcb UD |
2587 | |
2588 | /* Make sure the locale is read. */ | |
2589 | add_to_readlist (LC_CTYPE, ctype->translit_copy_locale, | |
07dab0c3 | 2590 | repertoire_name, 1, NULL); |
a673fbcb UD |
2591 | continue; |
2592 | } | |
2593 | else if (now->tok == tok_default_missing) | |
2594 | { | |
2595 | uint32_t *wstr; | |
2596 | ||
2597 | /* We expect a single character or string as the | |
2598 | argument. */ | |
2599 | now = lr_token (ldfile, charmap, NULL); | |
2600 | wstr = read_widestring (ldfile, now, charmap, repertoire); | |
2601 | ||
2602 | if (wstr != NULL) | |
2603 | { | |
2604 | if (ctype->default_missing != NULL) | |
2605 | { | |
2606 | lr_error (ldfile, _("\ | |
2607 | %s: duplicate `default_missing' definition"), "LC_CTYPE"); | |
2608 | error_at_line (0, 0, ctype->default_missing_file, | |
2609 | ctype->default_missing_lineno, | |
2610 | _("previous definition was here")); | |
2611 | } | |
2612 | else | |
2613 | { | |
2614 | ctype->default_missing = wstr; | |
2615 | ctype->default_missing_file = ldfile->fname; | |
2616 | ctype->default_missing_lineno = ldfile->lineno; | |
2617 | } | |
2618 | } | |
2619 | lr_ignore_rest (ldfile, 1); | |
2620 | continue; | |
2621 | } | |
2622 | else if (now->tok == tok_translit_ignore) | |
2623 | { | |
2624 | read_translit_ignore_entry (ldfile, ctype, charmap, | |
2625 | repertoire); | |
4b10dd6c UD |
2626 | continue; |
2627 | } | |
2628 | ||
2629 | read_translit_entry (ldfile, ctype, now, charmap, repertoire); | |
2630 | } | |
2631 | break; | |
2632 | ||
2633 | case tok_ident: | |
b9eb05d6 UD |
2634 | /* Ignore the rest of the line if we don't need the input of |
2635 | this line. */ | |
2636 | if (ignore_content) | |
2637 | { | |
2638 | lr_ignore_rest (ldfile, 0); | |
2639 | break; | |
2640 | } | |
2641 | ||
4b10dd6c UD |
2642 | /* This could mean one of several things. First test whether |
2643 | it's a character class name. */ | |
2644 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
2645 | if (strcmp (now->val.str.startmb, ctype->classnames[cnt]) == 0) | |
2646 | break; | |
2647 | if (cnt < ctype->nr_charclass) | |
2648 | { | |
2649 | class_bit = _ISwbit (cnt); | |
2650 | class256_bit = cnt <= 11 ? _ISbit (cnt) : 0; | |
2651 | free (now->val.str.startmb); | |
2652 | goto read_charclass; | |
2653 | } | |
5491da0d UD |
2654 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
2655 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2656 | break; | |
2657 | if (cnt < ctype->map_collection_nr) | |
2658 | { | |
2659 | mapidx = cnt; | |
2660 | free (now->val.str.startmb); | |
2661 | goto read_mapping; | |
2662 | } | |
011ebfab | 2663 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2664 | if (strcmp (now->val.str.startmb, "special1") == 0) |
2665 | { | |
2666 | class_bit = _ISwspecial1; | |
2667 | free (now->val.str.startmb); | |
2668 | goto read_charclass; | |
2669 | } | |
2670 | if (strcmp (now->val.str.startmb, "special2") == 0) | |
2671 | { | |
2672 | class_bit = _ISwspecial2; | |
2673 | free (now->val.str.startmb); | |
2674 | goto read_charclass; | |
2675 | } | |
2676 | if (strcmp (now->val.str.startmb, "special3") == 0) | |
2677 | { | |
2678 | class_bit = _ISwspecial3; | |
2679 | free (now->val.str.startmb); | |
2680 | goto read_charclass; | |
2681 | } | |
2682 | if (strcmp (now->val.str.startmb, "tosymmetric") == 0) | |
2683 | { | |
2684 | mapidx = 2; | |
2685 | goto read_mapping; | |
2686 | } | |
011ebfab | 2687 | #endif |
4b10dd6c UD |
2688 | break; |
2689 | ||
2690 | case tok_end: | |
2691 | /* Next we assume `LC_CTYPE'. */ | |
2692 | now = lr_token (ldfile, charmap, NULL); | |
2693 | if (now->tok == tok_eof) | |
2694 | break; | |
2695 | if (now->tok == tok_eol) | |
2696 | lr_error (ldfile, _("%s: incomplete `END' line"), | |
2697 | "LC_CTYPE"); | |
2698 | else if (now->tok != tok_lc_ctype) | |
2699 | lr_error (ldfile, _("\ | |
2700 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2701 | lr_ignore_rest (ldfile, now->tok == tok_lc_ctype); | |
2702 | return; | |
2703 | ||
2704 | default: | |
2705 | err_label: | |
2706 | if (now->tok != tok_eof) | |
2707 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
19bc17a9 RM |
2708 | } |
2709 | ||
4b10dd6c UD |
2710 | /* Prepare for the next round. */ |
2711 | now = lr_token (ldfile, charmap, NULL); | |
2712 | nowtok = now->tok; | |
19bc17a9 RM |
2713 | } |
2714 | ||
4b10dd6c UD |
2715 | /* When we come here we reached the end of the file. */ |
2716 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); | |
19bc17a9 RM |
2717 | } |
2718 | ||
2719 | ||
2720 | static void | |
4b10dd6c UD |
2721 | set_class_defaults (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2722 | struct repertoire_t *repertoire) | |
19bc17a9 | 2723 | { |
4b10dd6c UD |
2724 | size_t cnt; |
2725 | ||
19bc17a9 RM |
2726 | /* These function defines the default values for the classes and conversions |
2727 | according to POSIX.2 2.5.2.1. | |
2728 | It may seem that the order of these if-blocks is arbitrary but it is NOT. | |
2729 | Don't move them unless you know what you do! */ | |
2730 | ||
4b10dd6c | 2731 | void set_default (int bitpos, int from, int to) |
19bc17a9 RM |
2732 | { |
2733 | char tmp[2]; | |
2734 | int ch; | |
4b10dd6c UD |
2735 | int bit = _ISbit (bitpos); |
2736 | int bitw = _ISwbit (bitpos); | |
19bc17a9 RM |
2737 | /* Define string. */ |
2738 | strcpy (tmp, "?"); | |
2739 | ||
2740 | for (ch = from; ch <= to; ++ch) | |
2741 | { | |
4b10dd6c | 2742 | struct charseq *seq; |
19bc17a9 RM |
2743 | tmp[0] = ch; |
2744 | ||
4b10dd6c UD |
2745 | seq = charmap_find_value (charmap, tmp, 1); |
2746 | if (seq == NULL) | |
2747 | { | |
2748 | if (!be_quiet) | |
2749 | error (0, 0, _("\ | |
2750 | %s: character `%s' not defined in charmap while needed as default value"), | |
2751 | "LC_CTYPE", tmp); | |
19bc17a9 | 2752 | } |
4b10dd6c UD |
2753 | else if (seq->nbytes != 1) |
2754 | error (0, 0, _("\ | |
2755 | %s: character `%s' in charmap not representable with one byte"), | |
2756 | "LC_CTYPE", tmp); | |
19bc17a9 | 2757 | else |
4b10dd6c | 2758 | ctype->class256_collection[seq->bytes[0]] |= bit; |
f0a4b6b1 UD |
2759 | |
2760 | /* No need to search here, the ASCII value is also the Unicode | |
2761 | value. */ | |
2762 | ELEM (ctype, class_collection, , ch) |= bitw; | |
19bc17a9 RM |
2763 | } |
2764 | } | |
2765 | ||
2766 | /* Set default values if keyword was not present. */ | |
4b10dd6c | 2767 | if ((ctype->class_done & BITw (tok_upper)) == 0) |
19bc17a9 RM |
2768 | /* "If this keyword [lower] is not specified, the lowercase letters |
2769 | `A' through `Z', ..., shall automatically belong to this class, | |
2770 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2771 | set_default (BITPOS (tok_upper), 'A', 'Z'); |
19bc17a9 | 2772 | |
4b10dd6c | 2773 | if ((ctype->class_done & BITw (tok_lower)) == 0) |
19bc17a9 RM |
2774 | /* "If this keyword [lower] is not specified, the lowercase letters |
2775 | `a' through `z', ..., shall automatically belong to this class, | |
2776 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2777 | set_default (BITPOS (tok_lower), 'a', 'z'); |
19bc17a9 | 2778 | |
4b10dd6c | 2779 | if ((ctype->class_done & BITw (tok_alpha)) == 0) |
19bc17a9 RM |
2780 | { |
2781 | /* Table 2-6 in P1003.2 says that characters in class `upper' or | |
2782 | class `lower' *must* be in class `alpha'. */ | |
2783 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower); | |
96f0d1f5 UD |
2784 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower); |
2785 | ||
2786 | for (cnt = 0; cnt < 256; ++cnt) | |
2787 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2788 | ctype->class256_collection[cnt] |= BIT (tok_alpha); | |
19bc17a9 RM |
2789 | |
2790 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2791 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2792 | ctype->class_collection[cnt] |= BITw (tok_alpha); | |
19bc17a9 RM |
2793 | } |
2794 | ||
4b10dd6c | 2795 | if ((ctype->class_done & BITw (tok_digit)) == 0) |
19bc17a9 RM |
2796 | /* "If this keyword [digit] is not specified, the digits `0' through |
2797 | `9', ..., shall automatically belong to this class, with | |
2798 | implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2799 | set_default (BITPOS (tok_digit), '0', '9'); |
19bc17a9 RM |
2800 | |
2801 | /* "Only characters specified for the `alpha' and `digit' keyword | |
2802 | shall be specified. Characters specified for the keyword `alpha' | |
2803 | and `digit' are automatically included in this class. */ | |
2804 | { | |
2805 | unsigned long int mask = BIT (tok_alpha) | BIT (tok_digit); | |
96f0d1f5 UD |
2806 | unsigned long int maskw = BITw (tok_alpha) | BITw (tok_digit); |
2807 | ||
2808 | for (cnt = 0; cnt < 256; ++cnt) | |
2809 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2810 | ctype->class256_collection[cnt] |= BIT (tok_alnum); | |
19bc17a9 RM |
2811 | |
2812 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2813 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2814 | ctype->class_collection[cnt] |= BITw (tok_alnum); | |
19bc17a9 RM |
2815 | } |
2816 | ||
4b10dd6c | 2817 | if ((ctype->class_done & BITw (tok_space)) == 0) |
19bc17a9 RM |
2818 | /* "If this keyword [space] is not specified, the characters <space>, |
2819 | <form-feed>, <newline>, <carriage-return>, <tab>, and | |
2820 | <vertical-tab>, ..., shall automatically belong to this class, | |
2821 | with implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
2822 | { | |
4b10dd6c | 2823 | struct charseq *seq; |
19bc17a9 | 2824 | |
4b10dd6c | 2825 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
2826 | if (seq == NULL) |
2827 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
2828 | if (seq == NULL) |
2829 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 2830 | if (seq == NULL) |
880f421f UD |
2831 | { |
2832 | if (!be_quiet) | |
2833 | error (0, 0, _("\ | |
4b10dd6c UD |
2834 | %s: character `%s' not defined while needed as default value"), |
2835 | "LC_CTYPE", "<space>"); | |
2836 | } | |
2837 | else if (seq->nbytes != 1) | |
2838 | error (0, 0, _("\ | |
2839 | %s: character `%s' in charmap not representable with one byte"), | |
2840 | "LC_CTYPE", "<space>"); | |
2841 | else | |
2842 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2843 | ||
f0a4b6b1 | 2844 | /* No need to search. */ |
ce177a84 | 2845 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_space); |
19bc17a9 | 2846 | |
4b10dd6c | 2847 | seq = charmap_find_value (charmap, "form-feed", 9); |
f0a4b6b1 UD |
2848 | if (seq == NULL) |
2849 | seq = charmap_find_value (charmap, "U0000000C", 9); | |
4b10dd6c | 2850 | if (seq == NULL) |
880f421f UD |
2851 | { |
2852 | if (!be_quiet) | |
2853 | error (0, 0, _("\ | |
4b10dd6c UD |
2854 | %s: character `%s' not defined while needed as default value"), |
2855 | "LC_CTYPE", "<form-feed>"); | |
2856 | } | |
2857 | else if (seq->nbytes != 1) | |
2858 | error (0, 0, _("\ | |
2859 | %s: character `%s' in charmap not representable with one byte"), | |
2860 | "LC_CTYPE", "<form-feed>"); | |
2861 | else | |
2862 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2863 | ||
f0a4b6b1 | 2864 | /* No need to search. */ |
ce177a84 | 2865 | ELEM (ctype, class_collection, , L'\f') |= BITw (tok_space); |
4b10dd6c | 2866 | |
19bc17a9 | 2867 | |
4b10dd6c | 2868 | seq = charmap_find_value (charmap, "newline", 7); |
f0a4b6b1 UD |
2869 | if (seq == NULL) |
2870 | seq = charmap_find_value (charmap, "U0000000A", 9); | |
4b10dd6c | 2871 | if (seq == NULL) |
880f421f UD |
2872 | { |
2873 | if (!be_quiet) | |
2874 | error (0, 0, _("\ | |
19bc17a9 | 2875 | character `%s' not defined while needed as default value"), |
4b10dd6c UD |
2876 | "<newline>"); |
2877 | } | |
2878 | else if (seq->nbytes != 1) | |
2879 | error (0, 0, _("\ | |
2880 | %s: character `%s' in charmap not representable with one byte"), | |
2881 | "LC_CTYPE", "<newline>"); | |
2882 | else | |
2883 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2884 | ||
f0a4b6b1 | 2885 | /* No need to search. */ |
ce177a84 | 2886 | ELEM (ctype, class_collection, , L'\n') |= BITw (tok_space); |
4b10dd6c | 2887 | |
19bc17a9 | 2888 | |
4b10dd6c | 2889 | seq = charmap_find_value (charmap, "carriage-return", 15); |
f0a4b6b1 UD |
2890 | if (seq == NULL) |
2891 | seq = charmap_find_value (charmap, "U0000000D", 9); | |
4b10dd6c | 2892 | if (seq == NULL) |
880f421f UD |
2893 | { |
2894 | if (!be_quiet) | |
2895 | error (0, 0, _("\ | |
4b10dd6c UD |
2896 | %s: character `%s' not defined while needed as default value"), |
2897 | "LC_CTYPE", "<carriage-return>"); | |
2898 | } | |
2899 | else if (seq->nbytes != 1) | |
2900 | error (0, 0, _("\ | |
2901 | %s: character `%s' in charmap not representable with one byte"), | |
2902 | "LC_CTYPE", "<carriage-return>"); | |
2903 | else | |
2904 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2905 | ||
f0a4b6b1 | 2906 | /* No need to search. */ |
ce177a84 | 2907 | ELEM (ctype, class_collection, , L'\r') |= BITw (tok_space); |
4b10dd6c | 2908 | |
19bc17a9 | 2909 | |
4b10dd6c | 2910 | seq = charmap_find_value (charmap, "tab", 3); |
f0a4b6b1 UD |
2911 | if (seq == NULL) |
2912 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c | 2913 | if (seq == NULL) |
880f421f UD |
2914 | { |
2915 | if (!be_quiet) | |
2916 | error (0, 0, _("\ | |
4b10dd6c UD |
2917 | %s: character `%s' not defined while needed as default value"), |
2918 | "LC_CTYPE", "<tab>"); | |
2919 | } | |
2920 | else if (seq->nbytes != 1) | |
2921 | error (0, 0, _("\ | |
2922 | %s: character `%s' in charmap not representable with one byte"), | |
2923 | "LC_CTYPE", "<tab>"); | |
2924 | else | |
2925 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2926 | ||
f0a4b6b1 | 2927 | /* No need to search. */ |
ce177a84 | 2928 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_space); |
4b10dd6c | 2929 | |
4b10dd6c UD |
2930 | |
2931 | seq = charmap_find_value (charmap, "vertical-tab", 12); | |
f0a4b6b1 UD |
2932 | if (seq == NULL) |
2933 | seq = charmap_find_value (charmap, "U0000000B", 9); | |
4b10dd6c UD |
2934 | if (seq == NULL) |
2935 | { | |
2936 | if (!be_quiet) | |
2937 | error (0, 0, _("\ | |
2938 | %s: character `%s' not defined while needed as default value"), | |
2939 | "LC_CTYPE", "<vertical-tab>"); | |
2940 | } | |
2941 | else if (seq->nbytes != 1) | |
2942 | error (0, 0, _("\ | |
2943 | %s: character `%s' in charmap not representable with one byte"), | |
2944 | "LC_CTYPE", "<vertical-tab>"); | |
2945 | else | |
2946 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
f0a4b6b1 UD |
2947 | |
2948 | /* No need to search. */ | |
ce177a84 | 2949 | ELEM (ctype, class_collection, , L'\v') |= BITw (tok_space); |
19bc17a9 RM |
2950 | } |
2951 | ||
4b10dd6c | 2952 | if ((ctype->class_done & BITw (tok_xdigit)) == 0) |
19bc17a9 RM |
2953 | /* "If this keyword is not specified, the digits `0' to `9', the |
2954 | uppercase letters `A' through `F', and the lowercase letters `a' | |
2955 | through `f', ..., shell automatically belong to this class, with | |
2956 | implementation defined character values." [P1003.2, 2.5.2.1] */ | |
2957 | { | |
4b10dd6c UD |
2958 | set_default (BITPOS (tok_xdigit), '0', '9'); |
2959 | set_default (BITPOS (tok_xdigit), 'A', 'F'); | |
2960 | set_default (BITPOS (tok_xdigit), 'a', 'f'); | |
19bc17a9 RM |
2961 | } |
2962 | ||
4b10dd6c | 2963 | if ((ctype->class_done & BITw (tok_blank)) == 0) |
19bc17a9 RM |
2964 | /* "If this keyword [blank] is unspecified, the characters <space> and |
2965 | <tab> shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
2966 | { | |
4b10dd6c | 2967 | struct charseq *seq; |
19bc17a9 | 2968 | |
4b10dd6c | 2969 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
2970 | if (seq == NULL) |
2971 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
2972 | if (seq == NULL) |
2973 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 2974 | if (seq == NULL) |
880f421f UD |
2975 | { |
2976 | if (!be_quiet) | |
2977 | error (0, 0, _("\ | |
4b10dd6c UD |
2978 | %s: character `%s' not defined while needed as default value"), |
2979 | "LC_CTYPE", "<space>"); | |
2980 | } | |
2981 | else if (seq->nbytes != 1) | |
2982 | error (0, 0, _("\ | |
2983 | %s: character `%s' in charmap not representable with one byte"), | |
2984 | "LC_CTYPE", "<space>"); | |
2985 | else | |
2986 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
2987 | ||
f0a4b6b1 | 2988 | /* No need to search. */ |
ce177a84 | 2989 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_blank); |
4b10dd6c | 2990 | |
4b10dd6c UD |
2991 | |
2992 | seq = charmap_find_value (charmap, "tab", 3); | |
f0a4b6b1 UD |
2993 | if (seq == NULL) |
2994 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c UD |
2995 | if (seq == NULL) |
2996 | { | |
2997 | if (!be_quiet) | |
2998 | error (0, 0, _("\ | |
2999 | %s: character `%s' not defined while needed as default value"), | |
3000 | "LC_CTYPE", "<tab>"); | |
3001 | } | |
3002 | else if (seq->nbytes != 1) | |
3003 | error (0, 0, _("\ | |
3004 | %s: character `%s' in charmap not representable with one byte"), | |
3005 | "LC_CTYPE", "<tab>"); | |
3006 | else | |
3007 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
f0a4b6b1 UD |
3008 | |
3009 | /* No need to search. */ | |
ce177a84 | 3010 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_blank); |
19bc17a9 RM |
3011 | } |
3012 | ||
4b10dd6c | 3013 | if ((ctype->class_done & BITw (tok_graph)) == 0) |
19bc17a9 RM |
3014 | /* "If this keyword [graph] is not specified, characters specified for |
3015 | the keywords `upper', `lower', `alpha', `digit', `xdigit' and `punct', | |
3016 | shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
3017 | { | |
3018 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3019 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3020 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3021 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3022 | BITw (tok_punct); | |
19bc17a9 RM |
3023 | size_t cnt; |
3024 | ||
3025 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3026 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3027 | ctype->class_collection[cnt] |= BITw (tok_graph); | |
4b10dd6c UD |
3028 | |
3029 | for (cnt = 0; cnt < 256; ++cnt) | |
3030 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3031 | ctype->class256_collection[cnt] |= BIT (tok_graph); | |
19bc17a9 RM |
3032 | } |
3033 | ||
4b10dd6c | 3034 | if ((ctype->class_done & BITw (tok_print)) == 0) |
19bc17a9 RM |
3035 | /* "If this keyword [print] is not provided, characters specified for |
3036 | the keywords `upper', `lower', `alpha', `digit', `xdigit', `punct', | |
3037 | and the <space> character shall belong to this character class." | |
3038 | [P1003.2, 2.5.2.1] */ | |
3039 | { | |
3040 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3041 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3042 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3043 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3044 | BITw (tok_punct); | |
19bc17a9 | 3045 | size_t cnt; |
4b10dd6c | 3046 | struct charseq *seq; |
19bc17a9 RM |
3047 | |
3048 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3049 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3050 | ctype->class_collection[cnt] |= BITw (tok_print); | |
19bc17a9 | 3051 | |
4b10dd6c UD |
3052 | for (cnt = 0; cnt < 256; ++cnt) |
3053 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3054 | ctype->class256_collection[cnt] |= BIT (tok_print); | |
3055 | ||
3056 | ||
4b10dd6c | 3057 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3058 | if (seq == NULL) |
3059 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3060 | if (seq == NULL) |
3061 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c UD |
3062 | if (seq == NULL) |
3063 | { | |
3064 | if (!be_quiet) | |
3065 | error (0, 0, _("\ | |
3066 | %s: character `%s' not defined while needed as default value"), | |
3067 | "LC_CTYPE", "<space>"); | |
3068 | } | |
3069 | else if (seq->nbytes != 1) | |
3070 | error (0, 0, _("\ | |
3071 | %s: character `%s' in charmap not representable with one byte"), | |
3072 | "LC_CTYPE", "<space>"); | |
3073 | else | |
3074 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_print); | |
f0a4b6b1 UD |
3075 | |
3076 | /* No need to search. */ | |
ce177a84 | 3077 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_print); |
19bc17a9 RM |
3078 | } |
3079 | ||
4b10dd6c | 3080 | if (ctype->tomap_done[0] == 0) |
6d52618b | 3081 | /* "If this keyword [toupper] is not specified, the lowercase letters |
19bc17a9 RM |
3082 | `a' through `z', and their corresponding uppercase letters `A' to |
3083 | `Z', ..., shall automatically be included, with implementation- | |
3084 | defined character values." [P1003.2, 2.5.2.1] */ | |
3085 | { | |
3086 | char tmp[4]; | |
3087 | int ch; | |
3088 | ||
3089 | strcpy (tmp, "<?>"); | |
3090 | ||
3091 | for (ch = 'a'; ch <= 'z'; ++ch) | |
3092 | { | |
4b10dd6c | 3093 | struct charseq *seq_from, *seq_to; |
19bc17a9 RM |
3094 | |
3095 | tmp[1] = (char) ch; | |
3096 | ||
4b10dd6c UD |
3097 | seq_from = charmap_find_value (charmap, &tmp[1], 1); |
3098 | if (seq_from == NULL) | |
19bc17a9 | 3099 | { |
880f421f UD |
3100 | if (!be_quiet) |
3101 | error (0, 0, _("\ | |
4b10dd6c UD |
3102 | %s: character `%s' not defined while needed as default value"), |
3103 | "LC_CTYPE", tmp); | |
3104 | } | |
3105 | else if (seq_from->nbytes != 1) | |
3106 | { | |
3107 | if (!be_quiet) | |
3108 | error (0, 0, _("\ | |
3109 | %s: character `%s' needed as default value not representable with one byte"), | |
3110 | "LC_CTYPE", tmp); | |
3111 | } | |
3112 | else | |
3113 | { | |
3114 | /* This conversion is implementation defined. */ | |
3115 | tmp[1] = (char) (ch + ('A' - 'a')); | |
3116 | seq_to = charmap_find_value (charmap, &tmp[1], 1); | |
3117 | if (seq_to == NULL) | |
3118 | { | |
3119 | if (!be_quiet) | |
3120 | error (0, 0, _("\ | |
3121 | %s: character `%s' not defined while needed as default value"), | |
3122 | "LC_CTYPE", tmp); | |
3123 | } | |
3124 | else if (seq_to->nbytes != 1) | |
3125 | { | |
3126 | if (!be_quiet) | |
3127 | error (0, 0, _("\ | |
3128 | %s: character `%s' needed as default value not representable with one byte"), | |
3129 | "LC_CTYPE", tmp); | |
3130 | } | |
3131 | else | |
3132 | /* The index [0] is determined by the order of the | |
3133 | `ctype_map_newP' calls in `ctype_startup'. */ | |
3134 | ctype->map256_collection[0][seq_from->bytes[0]] | |
3135 | = seq_to->bytes[0]; | |
19bc17a9 | 3136 | } |
f0a4b6b1 UD |
3137 | |
3138 | /* No need to search. */ | |
3139 | ELEM (ctype, map_collection, [0], ch) = ch + ('A' - 'a'); | |
19bc17a9 RM |
3140 | } |
3141 | } | |
3142 | ||
4b10dd6c | 3143 | if (ctype->tomap_done[1] == 0) |
19bc17a9 RM |
3144 | /* "If this keyword [tolower] is not specified, the mapping shall be |
3145 | the reverse mapping of the one specified to `toupper'." [P1003.2] */ | |
3146 | { | |
19bc17a9 RM |
3147 | for (cnt = 0; cnt < ctype->map_collection_act[0]; ++cnt) |
3148 | if (ctype->map_collection[0][cnt] != 0) | |
3149 | ELEM (ctype, map_collection, [1], | |
3150 | ctype->map_collection[0][cnt]) | |
3151 | = ctype->charnames[cnt]; | |
4b10dd6c UD |
3152 | |
3153 | for (cnt = 0; cnt < 256; ++cnt) | |
3154 | if (ctype->map256_collection[0][cnt] != 0) | |
85cb60ff | 3155 | ctype->map256_collection[1][ctype->map256_collection[0][cnt]] = cnt; |
4b10dd6c UD |
3156 | } |
3157 | ||
3158 | if (ctype->outdigits_act == 0) | |
3159 | { | |
3160 | for (cnt = 0; cnt < 10; ++cnt) | |
3161 | { | |
3162 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
3163 | digits + cnt, 1); | |
3164 | ||
3165 | if (ctype->mboutdigits[cnt] == NULL) | |
1b97149d UD |
3166 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, |
3167 | longnames[cnt], | |
3168 | strlen (longnames[cnt])); | |
b9eb05d6 | 3169 | |
1b97149d UD |
3170 | if (ctype->mboutdigits[cnt] == NULL) |
3171 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
3172 | uninames[cnt], 9); | |
b9eb05d6 | 3173 | |
1b97149d | 3174 | if (ctype->mboutdigits[cnt] == NULL) |
b9eb05d6 | 3175 | { |
1b97149d UD |
3176 | /* Provide a replacement. */ |
3177 | error (0, 0, _("\ | |
3178 | no output digits defined and none of the standard names in the charmap")); | |
b9eb05d6 | 3179 | |
1b97149d UD |
3180 | ctype->mboutdigits[cnt] = obstack_alloc (&charmap->mem_pool, |
3181 | sizeof (struct charseq) | |
3182 | + 1); | |
b9eb05d6 | 3183 | |
1b97149d UD |
3184 | /* This is better than nothing. */ |
3185 | ctype->mboutdigits[cnt]->bytes[0] = digits[cnt]; | |
3186 | ctype->mboutdigits[cnt]->nbytes = 1; | |
b9eb05d6 | 3187 | } |
1b97149d UD |
3188 | |
3189 | ctype->wcoutdigits[cnt] = L'0' + cnt; | |
4b10dd6c UD |
3190 | } |
3191 | ||
3192 | ctype->outdigits_act = 10; | |
19bc17a9 RM |
3193 | } |
3194 | } | |
3195 | ||
3196 | ||
3197 | static void | |
4b10dd6c UD |
3198 | allocate_arrays (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
3199 | struct repertoire_t *repertoire) | |
19bc17a9 RM |
3200 | { |
3201 | size_t idx; | |
a53e3292 | 3202 | size_t width_table_size; |
0e16ecfa UD |
3203 | const void *key; |
3204 | size_t len; | |
3205 | void *vdata; | |
3206 | void *curs; | |
5d431a3e | 3207 | |
6d52618b UD |
3208 | /* First we have to decide how we organize the arrays. It is easy |
3209 | for a one-byte character set. But multi-byte character set | |
3210 | cannot be stored flat because the chars might be sparsely used. | |
3211 | So we determine an optimal hashing function for the used | |
3212 | characters. | |
3213 | ||
3214 | We use a very trivial hashing function to store the sparse | |
3215 | table. CH % TABSIZE is used as an index. To solve multiple hits | |
3216 | we have N planes. This guarantees a fixed search time for a | |
42d7c593 | 3217 | character [N / 2]. In the following code we determine the minimum |
66ac0abe UD |
3218 | value for TABSIZE * N, where TABSIZE >= 256. |
3219 | ||
3220 | Some people complained that this algorithm takes too long. Well, | |
3221 | go on, improve it. But changing the step size is *not* an | |
3222 | option. Some people changed this to use only sizes of prime | |
3223 | numbers. Think again, do some math. We are looking for the | |
3224 | optimal solution, not something which works in general. Unless | |
3225 | somebody can provide a dynamic programming solution I think this | |
3226 | implementation is as good as it can get. */ | |
19bc17a9 RM |
3227 | size_t min_total = UINT_MAX; |
3228 | size_t act_size = 256; | |
3229 | ||
66ac0abe | 3230 | if (!be_quiet && ctype->charnames_act > 512) |
c84142e8 | 3231 | fputs (_("\ |
19bc17a9 | 3232 | Computing table size for character classes might take a while..."), |
c84142e8 | 3233 | stderr); |
19bc17a9 | 3234 | |
66ac0abe UD |
3235 | /* While we want to have a small total size we are willing to use a |
3236 | little bit larger table if this reduces the number of layers. | |
3237 | Therefore we add a little penalty to the number of planes. | |
3238 | Maybe this constant has to be adjusted a bit. */ | |
3239 | #define PENALTY 128 | |
3240 | do | |
19bc17a9 RM |
3241 | { |
3242 | size_t cnt[act_size]; | |
3243 | size_t act_planes = 1; | |
3244 | ||
3245 | memset (cnt, '\0', sizeof cnt); | |
3246 | ||
3247 | for (idx = 0; idx < 256; ++idx) | |
3248 | cnt[idx] = 1; | |
3249 | ||
3250 | for (idx = 0; idx < ctype->charnames_act; ++idx) | |
3251 | if (ctype->charnames[idx] >= 256) | |
3252 | { | |
3253 | size_t nr = ctype->charnames[idx] % act_size; | |
3254 | ||
3255 | if (++cnt[nr] > act_planes) | |
3256 | { | |
3257 | act_planes = cnt[nr]; | |
66ac0abe | 3258 | if ((act_size + PENALTY) * act_planes >= min_total) |
19bc17a9 RM |
3259 | break; |
3260 | } | |
3261 | } | |
3262 | ||
66ac0abe | 3263 | if ((act_size + PENALTY) * act_planes < min_total) |
19bc17a9 | 3264 | { |
66ac0abe | 3265 | min_total = (act_size + PENALTY) * act_planes; |
19bc17a9 RM |
3266 | ctype->plane_size = act_size; |
3267 | ctype->plane_cnt = act_planes; | |
3268 | } | |
3269 | ||
3270 | ++act_size; | |
3271 | } | |
66ac0abe | 3272 | while (act_size < min_total); |
19bc17a9 | 3273 | |
66ac0abe | 3274 | if (!be_quiet && ctype->charnames_act > 512) |
c84142e8 | 3275 | fputs (_(" done\n"), stderr); |
19bc17a9 | 3276 | |
75cd5204 | 3277 | |
4a33c2f5 UD |
3278 | ctype->names = (uint32_t *) xcalloc (ctype->plane_size |
3279 | * ctype->plane_cnt, | |
3280 | sizeof (uint32_t)); | |
19bc17a9 RM |
3281 | |
3282 | for (idx = 1; idx < 256; ++idx) | |
4a33c2f5 | 3283 | ctype->names[idx] = idx; |
19bc17a9 RM |
3284 | |
3285 | /* Trick: change the 0th entry's name to 1 to mark the cell occupied. */ | |
4a33c2f5 | 3286 | ctype->names[0] = 1; |
19bc17a9 RM |
3287 | |
3288 | for (idx = 256; idx < ctype->charnames_act; ++idx) | |
3289 | { | |
3290 | size_t nr = (ctype->charnames[idx] % ctype->plane_size); | |
3291 | size_t depth = 0; | |
3292 | ||
4a33c2f5 | 3293 | while (ctype->names[nr + depth * ctype->plane_size]) |
19bc17a9 RM |
3294 | ++depth; |
3295 | assert (depth < ctype->plane_cnt); | |
3296 | ||
4a33c2f5 | 3297 | ctype->names[nr + depth * ctype->plane_size] = ctype->charnames[idx]; |
19bc17a9 RM |
3298 | |
3299 | /* Now for faster access remember the index in the NAMES_B array. */ | |
3300 | ctype->charnames[idx] = nr + depth * ctype->plane_size; | |
3301 | } | |
4a33c2f5 | 3302 | ctype->names[0] = 0; |
19bc17a9 RM |
3303 | |
3304 | ||
3305 | /* You wonder about this amount of memory? This is only because some | |
3306 | users do not manage to address the array with unsigned values or | |
3307 | data types with range >= 256. '\200' would result in the array | |
3308 | index -128. To help these poor people we duplicate the entries for | |
3309 | 128 up to 255 below the entry for \0. */ | |
3310 | ctype->ctype_b = (char_class_t *) xcalloc (256 + 128, | |
3311 | sizeof (char_class_t)); | |
3312 | ctype->ctype32_b = (char_class32_t *) xcalloc (ctype->plane_size | |
3313 | * ctype->plane_cnt, | |
3314 | sizeof (char_class32_t)); | |
3315 | ||
4a33c2f5 | 3316 | /* This is the array accessed using the multibyte string elements. */ |
4b10dd6c | 3317 | for (idx = 0; idx < 256; ++idx) |
4a33c2f5 | 3318 | ctype->ctype_b[128 + idx] = ctype->class256_collection[idx]; |
19bc17a9 | 3319 | |
75cd5204 RM |
3320 | /* Mirror first 127 entries. We must take care that entry -1 is not |
3321 | mirrored because EOF == -1. */ | |
3322 | for (idx = 0; idx < 127; ++idx) | |
19bc17a9 RM |
3323 | ctype->ctype_b[idx] = ctype->ctype_b[256 + idx]; |
3324 | ||
3325 | /* The 32 bit array contains all characters. */ | |
3326 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
4a33c2f5 | 3327 | ctype->ctype32_b[ctype->charnames[idx]] = ctype->class_collection[idx]; |
19bc17a9 RM |
3328 | |
3329 | /* Room for table of mappings. */ | |
49f2be5b UD |
3330 | ctype->map = (uint32_t **) xmalloc (2 * sizeof (uint32_t *)); |
3331 | ctype->map32 = (uint32_t **) xmalloc (ctype->map_collection_nr | |
4a33c2f5 | 3332 | * sizeof (uint32_t *)); |
19bc17a9 RM |
3333 | |
3334 | /* Fill in all mappings. */ | |
49f2be5b | 3335 | for (idx = 0; idx < 2; ++idx) |
19bc17a9 RM |
3336 | { |
3337 | unsigned int idx2; | |
3338 | ||
3339 | /* Allocate table. */ | |
49f2be5b | 3340 | ctype->map[idx] = (uint32_t *) xmalloc ((256 + 128) * sizeof (uint32_t)); |
19bc17a9 RM |
3341 | |
3342 | /* Copy values from collection. */ | |
4b10dd6c | 3343 | for (idx2 = 0; idx2 < 256; ++idx2) |
4a33c2f5 | 3344 | ctype->map[idx][128 + idx2] = ctype->map256_collection[idx][idx2]; |
19bc17a9 | 3345 | |
75cd5204 RM |
3346 | /* Mirror first 127 entries. We must take care not to map entry |
3347 | -1 because EOF == -1. */ | |
3348 | for (idx2 = 0; idx2 < 127; ++idx2) | |
4a33c2f5 | 3349 | ctype->map[idx][idx2] = ctype->map[idx][256 + idx2]; |
19bc17a9 | 3350 | |
75cd5204 | 3351 | /* EOF must map to EOF. */ |
4a33c2f5 | 3352 | ctype->map[idx][127] = EOF; |
49f2be5b | 3353 | } |
a9c27b3e | 3354 | |
49f2be5b UD |
3355 | for (idx = 0; idx < ctype->map_collection_nr; ++idx) |
3356 | { | |
3357 | unsigned int idx2; | |
3358 | ||
3359 | /* Allocate table. */ | |
f1d8b804 UD |
3360 | ctype->map32[idx] = (uint32_t *) xmalloc (ctype->plane_size |
3361 | * ctype->plane_cnt | |
3362 | * sizeof (uint32_t)); | |
49f2be5b UD |
3363 | |
3364 | /* Copy default value (identity mapping). */ | |
f1d8b804 | 3365 | memcpy (ctype->map32[idx], ctype->names, |
49f2be5b UD |
3366 | ctype->plane_size * ctype->plane_cnt * sizeof (uint32_t)); |
3367 | ||
3368 | /* Copy values from collection. */ | |
3369 | for (idx2 = 0; idx2 < 256; ++idx2) | |
a9c27b3e | 3370 | if (ctype->map_collection[idx][idx2] != 0) |
f1d8b804 UD |
3371 | ctype->map32[idx][idx2] = ctype->map_collection[idx][idx2]; |
3372 | ||
3373 | while (idx2 < ctype->map_collection_act[idx]) | |
b06c53e7 UD |
3374 | { |
3375 | if (ctype->map_collection[idx][idx2] != 0) | |
450bf66e UD |
3376 | ctype->map32[idx][ctype->charnames[idx2]] = |
3377 | ctype->map_collection[idx][idx2]; | |
b06c53e7 UD |
3378 | ++idx2; |
3379 | } | |
19bc17a9 RM |
3380 | } |
3381 | ||
3382 | /* Extra array for class and map names. */ | |
4b10dd6c UD |
3383 | ctype->class_name_ptr = (uint32_t *) xmalloc (ctype->nr_charclass |
3384 | * sizeof (uint32_t)); | |
3385 | ctype->map_name_ptr = (uint32_t *) xmalloc (ctype->map_collection_nr | |
3386 | * sizeof (uint32_t)); | |
75cd5204 RM |
3387 | |
3388 | /* Array for width information. Because the expected width are very | |
3389 | small we use only one single byte. This save space and we need | |
3390 | not provide the information twice with both endianesses. */ | |
5866b131 UD |
3391 | width_table_size = (ctype->plane_size * ctype->plane_cnt + 3) & ~3ul; |
3392 | ctype->width = (unsigned char *) xmalloc (width_table_size); | |
3393 | ||
0e16ecfa UD |
3394 | /* Initialize with -1. */ |
3395 | memset (ctype->width, '\xff', width_table_size); | |
4b10dd6c | 3396 | if (charmap->width_rules != NULL) |
75cd5204 RM |
3397 | { |
3398 | size_t cnt; | |
3399 | ||
4b10dd6c | 3400 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
827ff758 UD |
3401 | { |
3402 | unsigned char bytes[charmap->mb_cur_max]; | |
3403 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
3404 | ||
3405 | /* We have the range of character for which the width is | |
3406 | specified described using byte sequences of the multibyte | |
3407 | charset. We have to convert this to UCS4 now. And we | |
3408 | cannot simply convert the beginning and the end of the | |
3409 | sequence, we have to iterate over the byte sequence and | |
3410 | convert it for every single character. */ | |
3411 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
3412 | ||
3413 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
3414 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
3415 | nbytes) <= 0) | |
75cd5204 | 3416 | { |
827ff758 | 3417 | /* Find the UCS value for `bytes'. */ |
827ff758 | 3418 | int inner; |
76e680a8 UD |
3419 | uint32_t wch; |
3420 | struct charseq *seq = | |
3421 | charmap_find_symbol (charmap, bytes, nbytes); | |
3422 | ||
3423 | if (seq == NULL) | |
3424 | wch = ILLEGAL_CHAR_VALUE; | |
3425 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
3426 | wch = seq->ucs4; | |
3427 | else | |
3428 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
3429 | strlen (seq->name)); | |
827ff758 UD |
3430 | |
3431 | if (wch != ILLEGAL_CHAR_VALUE) | |
3432 | { | |
3433 | /* Store the value. */ | |
b1c9ad82 | 3434 | size_t nr = wch % ctype->plane_size; |
827ff758 UD |
3435 | size_t depth = 0; |
3436 | ||
b1c9ad82 | 3437 | while (ctype->names[nr + depth * ctype->plane_size] != wch) |
0e16ecfa UD |
3438 | { |
3439 | ++depth; | |
3440 | assert (depth < ctype->plane_cnt); | |
3441 | } | |
827ff758 UD |
3442 | |
3443 | ctype->width[nr + depth * ctype->plane_size] | |
3444 | = charmap->width_rules[cnt].width; | |
3445 | } | |
3446 | ||
3447 | /* "Increment" the bytes sequence. */ | |
3448 | inner = nbytes - 1; | |
3449 | while (inner >= 0 && bytes[inner] == 0xff) | |
3450 | --inner; | |
75cd5204 | 3451 | |
827ff758 UD |
3452 | if (inner < 0) |
3453 | { | |
3454 | /* We have to extend the byte sequence. */ | |
3455 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
3456 | break; | |
75cd5204 | 3457 | |
827ff758 UD |
3458 | bytes[0] = 1; |
3459 | memset (&bytes[1], 0, nbytes); | |
3460 | ++nbytes; | |
3461 | } | |
3462 | else | |
3463 | { | |
3464 | ++bytes[inner]; | |
3465 | while (++inner < nbytes) | |
3466 | bytes[inner] = 0; | |
3467 | } | |
75cd5204 | 3468 | } |
827ff758 | 3469 | } |
75cd5204 | 3470 | } |
0200214b | 3471 | |
0e16ecfa UD |
3472 | /* Now set all the other characters of the character set to the |
3473 | default width. */ | |
3474 | curs = NULL; | |
3475 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
3476 | { | |
3477 | struct charseq *data = (struct charseq *) vdata; | |
3478 | size_t nr; | |
3479 | size_t depth; | |
3480 | ||
3481 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
3482 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
3483 | data->name, len); | |
3484 | ||
3485 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) | |
3486 | { | |
3487 | nr = data->ucs4 % ctype->plane_size; | |
3488 | depth = 0; | |
3489 | ||
3490 | while (ctype->names[nr + depth * ctype->plane_size] != data->ucs4) | |
3491 | { | |
3492 | ++depth; | |
3493 | assert (depth < ctype->plane_cnt); | |
3494 | } | |
3495 | ||
3496 | if (ctype->width[nr + depth * ctype->plane_size] | |
3497 | == (unsigned char) '\xff') | |
3498 | ctype->width[nr + depth * ctype->plane_size] = | |
3499 | charmap->width_default; | |
3500 | } | |
3501 | } | |
3502 | ||
4b10dd6c UD |
3503 | /* Set MB_CUR_MAX. */ |
3504 | ctype->mb_cur_max = charmap->mb_cur_max; | |
6990326c | 3505 | |
4b10dd6c UD |
3506 | /* Now determine the table for the transliteration information. |
3507 | ||
3508 | XXX It is not yet clear to me whether it is worth implementing a | |
3509 | complicated algorithm which uses a hash table to locate the entries. | |
3510 | For now I'll use a simple array which can be searching using binary | |
3511 | search. */ | |
3512 | if (ctype->translit_copy_locale != NULL) | |
3513 | { | |
3514 | /* Fold in the transliteration information from the locale mentioned | |
3515 | in the `include' statement. */ | |
3516 | struct locale_ctype_t *here = ctype; | |
3517 | ||
3518 | do | |
3519 | { | |
3520 | struct localedef_t *other = find_locale (LC_CTYPE, | |
3521 | here->translit_copy_locale, | |
3522 | repertoire->name, charmap); | |
3523 | ||
3524 | if (other == NULL) | |
3525 | { | |
3526 | error (0, 0, _("\ | |
3527 | %s: transliteration data from locale `%s' not available"), | |
3528 | "LC_CTYPE", here->translit_copy_locale); | |
3529 | break; | |
3530 | } | |
3531 | ||
3532 | here = other->categories[LC_CTYPE].ctype; | |
3533 | ||
3534 | /* Enqueue the information if necessary. */ | |
3535 | if (here->translit != NULL) | |
3536 | { | |
3537 | struct translit_t *endp = here->translit; | |
3538 | while (endp->next != NULL) | |
3539 | endp = endp->next; | |
3540 | ||
3541 | endp->next = ctype->translit; | |
3542 | ctype->translit = here->translit; | |
3543 | } | |
3544 | } | |
3545 | while (here->translit_copy_locale != NULL); | |
3546 | } | |
3547 | ||
3548 | if (ctype->translit != NULL) | |
3549 | { | |
3550 | /* First count how many entries we have. This is the upper limit | |
3551 | since some entries from the included files might be overwritten. */ | |
3552 | size_t number = 0; | |
3553 | size_t cnt; | |
3554 | struct translit_t *runp = ctype->translit; | |
3555 | struct translit_t **sorted; | |
3556 | size_t from_len, to_len; | |
3557 | ||
3558 | while (runp != NULL) | |
3559 | { | |
3560 | ++number; | |
3561 | runp = runp->next; | |
3562 | } | |
3563 | ||
3564 | /* Next we allocate an array large enough and fill in the values. */ | |
a9c27b3e UD |
3565 | sorted = (struct translit_t **) alloca (number |
3566 | * sizeof (struct translit_t **)); | |
4b10dd6c UD |
3567 | runp = ctype->translit; |
3568 | number = 0; | |
3569 | do | |
3570 | { | |
3571 | /* Search for the place where to insert this string. | |
3572 | XXX Better use a real sorting algorithm later. */ | |
3573 | size_t idx = 0; | |
3574 | int replace = 0; | |
3575 | ||
3576 | while (idx < number) | |
3577 | { | |
3578 | int res = wcscmp ((const wchar_t *) sorted[idx]->from, | |
3579 | (const wchar_t *) runp->from); | |
3580 | if (res == 0) | |
3581 | { | |
3582 | replace = 1; | |
3583 | break; | |
3584 | } | |
3585 | if (res > 0) | |
3586 | break; | |
3587 | ++idx; | |
3588 | } | |
3589 | ||
3590 | if (replace) | |
3591 | sorted[idx] = runp; | |
3592 | else | |
3593 | { | |
3594 | memmove (&sorted[idx + 1], &sorted[idx], | |
3595 | (number - idx) * sizeof (struct translit_t *)); | |
3596 | sorted[idx] = runp; | |
3597 | ++number; | |
3598 | } | |
3599 | ||
3600 | runp = runp->next; | |
3601 | } | |
3602 | while (runp != NULL); | |
3603 | ||
3604 | /* The next step is putting all the possible transliteration | |
3605 | strings in one memory block so that we can write it out. | |
3606 | We need several different blocks: | |
9ca23765 | 3607 | - index to the from-string array |
4b10dd6c UD |
3608 | - from-string array |
3609 | - index to the to-string array | |
3610 | - to-string array. | |
4b10dd6c UD |
3611 | */ |
3612 | from_len = to_len = 0; | |
3613 | for (cnt = 0; cnt < number; ++cnt) | |
3614 | { | |
3615 | struct translit_to_t *srunp; | |
3616 | from_len += wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
3617 | srunp = sorted[cnt]->to; | |
3618 | while (srunp != NULL) | |
3619 | { | |
3620 | to_len += wcslen ((const wchar_t *) srunp->str) + 1; | |
3621 | srunp = srunp->next; | |
3622 | } | |
3623 | /* Plus one for the extra NUL character marking the end of | |
3624 | the list for the current entry. */ | |
3625 | ++to_len; | |
3626 | } | |
3627 | ||
3628 | /* We can allocate the arrays for the results. */ | |
4a33c2f5 UD |
3629 | ctype->translit_from_idx = xmalloc (number * sizeof (uint32_t)); |
3630 | ctype->translit_from_tbl = xmalloc (from_len * sizeof (uint32_t)); | |
3631 | ctype->translit_to_idx = xmalloc (number * sizeof (uint32_t)); | |
3632 | ctype->translit_to_tbl = xmalloc (to_len * sizeof (uint32_t)); | |
4b10dd6c UD |
3633 | |
3634 | from_len = 0; | |
3635 | to_len = 0; | |
3636 | for (cnt = 0; cnt < number; ++cnt) | |
3637 | { | |
3638 | size_t len; | |
3639 | struct translit_to_t *srunp; | |
3640 | ||
4a33c2f5 UD |
3641 | ctype->translit_from_idx[cnt] = from_len; |
3642 | ctype->translit_to_idx[cnt] = to_len; | |
4b10dd6c UD |
3643 | |
3644 | len = wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4a33c2f5 | 3645 | wmemcpy ((wchar_t *) &ctype->translit_from_tbl[from_len], |
4b10dd6c UD |
3646 | (const wchar_t *) sorted[cnt]->from, len); |
3647 | from_len += len; | |
3648 | ||
4a33c2f5 | 3649 | ctype->translit_to_idx[cnt] = to_len; |
4b10dd6c UD |
3650 | srunp = sorted[cnt]->to; |
3651 | while (srunp != NULL) | |
3652 | { | |
3653 | len = wcslen ((const wchar_t *) srunp->str) + 1; | |
4a33c2f5 | 3654 | wmemcpy ((wchar_t *) &ctype->translit_to_tbl[to_len], |
4b10dd6c UD |
3655 | (const wchar_t *) srunp->str, len); |
3656 | to_len += len; | |
3657 | srunp = srunp->next; | |
3658 | } | |
4a33c2f5 | 3659 | ctype->translit_to_tbl[to_len++] = L'\0'; |
4b10dd6c | 3660 | } |
4b10dd6c UD |
3661 | |
3662 | /* Store the information about the length. */ | |
04fbc779 | 3663 | ctype->translit_idx_size = number; |
4b10dd6c UD |
3664 | ctype->translit_from_tbl_size = from_len * sizeof (uint32_t); |
3665 | ctype->translit_to_tbl_size = to_len * sizeof (uint32_t); | |
3666 | } | |
3667 | else | |
3668 | { | |
3669 | /* Provide some dummy pointers since we have nothing to write out. */ | |
3670 | static uint32_t no_str = { 0 }; | |
3671 | ||
4a33c2f5 UD |
3672 | ctype->translit_from_idx = &no_str; |
3673 | ctype->translit_from_tbl = &no_str; | |
3674 | ctype->translit_to_tbl = &no_str; | |
4b10dd6c UD |
3675 | ctype->translit_idx_size = 0; |
3676 | ctype->translit_from_tbl_size = 0; | |
3677 | ctype->translit_to_tbl_size = 0; | |
3678 | } | |
19bc17a9 | 3679 | } |