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