]>
Commit | Line | Data |
---|---|---|
f76d7052 | 1 | /* Copyright (C) 1995, 1996, 1997, 1998, 1999 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 | ||
88 | struct translit_to_t *to; | |
89 | ||
90 | struct translit_t *next; | |
91 | }; | |
19bc17a9 RM |
92 | |
93 | ||
94 | /* The real definition of the struct for the LC_CTYPE locale. */ | |
95 | struct locale_ctype_t | |
96 | { | |
4b10dd6c | 97 | uint32_t *charnames; |
19bc17a9 RM |
98 | size_t charnames_max; |
99 | size_t charnames_act; | |
100 | ||
4b10dd6c UD |
101 | struct repertoire_t *repertoire; |
102 | ||
103 | /* We will allow up to 8 * sizeof (uint32_t) character classes. */ | |
104 | #define MAX_NR_CHARCLASS (8 * sizeof (uint32_t)) | |
ba1ffaa1 | 105 | size_t nr_charclass; |
19bc17a9 | 106 | const char *classnames[MAX_NR_CHARCLASS]; |
4b10dd6c UD |
107 | uint32_t last_class_char; |
108 | uint32_t class256_collection[256]; | |
109 | uint32_t *class_collection; | |
19bc17a9 RM |
110 | size_t class_collection_max; |
111 | size_t class_collection_act; | |
4b10dd6c UD |
112 | uint32_t class_done; |
113 | ||
114 | struct charseq **mbdigits; | |
115 | size_t mbdigits_act; | |
116 | size_t mbdigits_max; | |
117 | uint32_t *wcdigits; | |
118 | size_t wcdigits_act; | |
119 | size_t wcdigits_max; | |
120 | ||
121 | struct charseq *mboutdigits[10]; | |
122 | uint32_t wcoutdigits[10]; | |
123 | size_t outdigits_act; | |
19bc17a9 RM |
124 | |
125 | /* If the following number ever turns out to be too small simply | |
126 | increase it. But I doubt it will. --drepper@gnu */ | |
127 | #define MAX_NR_CHARMAP 16 | |
128 | const char *mapnames[MAX_NR_CHARMAP]; | |
4b10dd6c UD |
129 | uint32_t *map_collection[MAX_NR_CHARMAP]; |
130 | uint32_t map256_collection[2][256]; | |
9a0a462c UD |
131 | size_t map_collection_max[MAX_NR_CHARMAP]; |
132 | size_t map_collection_act[MAX_NR_CHARMAP]; | |
19bc17a9 RM |
133 | size_t map_collection_nr; |
134 | size_t last_map_idx; | |
4b10dd6c UD |
135 | int tomap_done[MAX_NR_CHARMAP]; |
136 | ||
137 | /* Transliteration information. */ | |
138 | const char *translit_copy_locale; | |
139 | const char *translit_copy_repertoire; | |
140 | struct translit_t *translit; | |
19bc17a9 RM |
141 | |
142 | /* The arrays for the binary representation. */ | |
4b10dd6c UD |
143 | uint32_t plane_size; |
144 | uint32_t plane_cnt; | |
19bc17a9 RM |
145 | char_class_t *ctype_b; |
146 | char_class32_t *ctype32_b; | |
4a33c2f5 UD |
147 | uint32_t *names; |
148 | uint32_t **map; | |
49f2be5b | 149 | uint32_t **map32; |
4b10dd6c UD |
150 | uint32_t *class_name_ptr; |
151 | uint32_t *map_name_ptr; | |
75cd5204 | 152 | unsigned char *width; |
4b10dd6c | 153 | uint32_t mb_cur_max; |
6990326c | 154 | const char *codeset_name; |
4a33c2f5 UD |
155 | uint32_t translit_hash_size; |
156 | uint32_t translit_hash_layers; | |
157 | uint32_t *translit_from_idx; | |
158 | uint32_t *translit_from_tbl; | |
159 | uint32_t *translit_to_idx; | |
160 | uint32_t *translit_to_tbl; | |
4b10dd6c UD |
161 | size_t translit_idx_size; |
162 | size_t translit_from_tbl_size; | |
163 | size_t translit_to_tbl_size; | |
164 | ||
165 | struct obstack mem_pool; | |
19bc17a9 RM |
166 | }; |
167 | ||
168 | ||
4b10dd6c UD |
169 | #define obstack_chunk_alloc xmalloc |
170 | #define obstack_chunk_free free | |
171 | ||
172 | ||
19bc17a9 | 173 | /* Prototypes for local functions. */ |
4b10dd6c UD |
174 | static void ctype_startup (struct linereader *lr, struct localedef_t *locale, |
175 | struct charmap_t *charmap, int ignore_content); | |
176 | static void ctype_class_new (struct linereader *lr, | |
177 | struct locale_ctype_t *ctype, const char *name); | |
178 | static void ctype_map_new (struct linereader *lr, | |
179 | struct locale_ctype_t *ctype, | |
180 | const char *name, struct charmap_t *charmap); | |
181 | static uint32_t *find_idx (struct locale_ctype_t *ctype, uint32_t **table, | |
182 | size_t *max, size_t *act, unsigned int idx); | |
19bc17a9 | 183 | static void set_class_defaults (struct locale_ctype_t *ctype, |
4b10dd6c UD |
184 | struct charmap_t *charmap, |
185 | struct repertoire_t *repertoire); | |
75cd5204 | 186 | static void allocate_arrays (struct locale_ctype_t *ctype, |
4b10dd6c UD |
187 | struct charmap_t *charmap, |
188 | struct repertoire_t *repertoire); | |
19bc17a9 RM |
189 | |
190 | ||
4b10dd6c UD |
191 | static const char *longnames[] = |
192 | { | |
193 | "zero", "one", "two", "three", "four", | |
194 | "five", "six", "seven", "eight", "nine" | |
195 | }; | |
196 | static const unsigned char digits[] = "0123456789"; | |
197 | ||
198 | ||
199 | static void | |
19bc17a9 | 200 | ctype_startup (struct linereader *lr, struct localedef_t *locale, |
4b10dd6c | 201 | struct charmap_t *charmap, int ignore_content) |
19bc17a9 RM |
202 | { |
203 | unsigned int cnt; | |
204 | struct locale_ctype_t *ctype; | |
205 | ||
4b10dd6c | 206 | if (!ignore_content) |
19bc17a9 | 207 | { |
4b10dd6c UD |
208 | /* Allocate the needed room. */ |
209 | locale->categories[LC_CTYPE].ctype = ctype = | |
210 | (struct locale_ctype_t *) xcalloc (1, sizeof (struct locale_ctype_t)); | |
211 | ||
212 | /* We have seen no names yet. */ | |
213 | ctype->charnames_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
214 | ctype->charnames = | |
215 | (unsigned int *) xmalloc (ctype->charnames_max | |
216 | * sizeof (unsigned int)); | |
217 | for (cnt = 0; cnt < 256; ++cnt) | |
218 | ctype->charnames[cnt] = cnt; | |
219 | ctype->charnames_act = 256; | |
220 | ||
221 | /* Fill character class information. */ | |
222 | ctype->last_class_char = ILLEGAL_CHAR_VALUE; | |
223 | /* The order of the following instructions determines the bit | |
224 | positions! */ | |
225 | ctype_class_new (lr, ctype, "upper"); | |
226 | ctype_class_new (lr, ctype, "lower"); | |
227 | ctype_class_new (lr, ctype, "alpha"); | |
228 | ctype_class_new (lr, ctype, "digit"); | |
229 | ctype_class_new (lr, ctype, "xdigit"); | |
230 | ctype_class_new (lr, ctype, "space"); | |
231 | ctype_class_new (lr, ctype, "print"); | |
232 | ctype_class_new (lr, ctype, "graph"); | |
233 | ctype_class_new (lr, ctype, "blank"); | |
234 | ctype_class_new (lr, ctype, "cntrl"); | |
235 | ctype_class_new (lr, ctype, "punct"); | |
236 | ctype_class_new (lr, ctype, "alnum"); | |
011ebfab | 237 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
238 | /* The following are extensions from ISO 14652. */ |
239 | ctype_class_new (lr, ctype, "left_to_right"); | |
240 | ctype_class_new (lr, ctype, "right_to_left"); | |
241 | ctype_class_new (lr, ctype, "num_terminator"); | |
242 | ctype_class_new (lr, ctype, "num_separator"); | |
243 | ctype_class_new (lr, ctype, "segment_separator"); | |
244 | ctype_class_new (lr, ctype, "block_separator"); | |
245 | ctype_class_new (lr, ctype, "direction_control"); | |
246 | ctype_class_new (lr, ctype, "sym_swap_layout"); | |
247 | ctype_class_new (lr, ctype, "char_shape_selector"); | |
248 | ctype_class_new (lr, ctype, "num_shape_selector"); | |
249 | ctype_class_new (lr, ctype, "non_spacing"); | |
250 | ctype_class_new (lr, ctype, "non_spacing_level3"); | |
251 | ctype_class_new (lr, ctype, "normal_connect"); | |
252 | ctype_class_new (lr, ctype, "r_connect"); | |
253 | ctype_class_new (lr, ctype, "no_connect"); | |
254 | ctype_class_new (lr, ctype, "no_connect-space"); | |
255 | ctype_class_new (lr, ctype, "vowel_connect"); | |
011ebfab | 256 | #endif |
4b10dd6c UD |
257 | |
258 | ctype->class_collection_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
259 | ctype->class_collection | |
260 | = (uint32_t *) xcalloc (sizeof (unsigned long int), | |
261 | ctype->class_collection_max); | |
262 | ctype->class_collection_act = 256; | |
263 | ||
264 | /* Fill character map information. */ | |
265 | ctype->map_collection_nr = 0; | |
266 | ctype->last_map_idx = MAX_NR_CHARMAP; | |
267 | ctype_map_new (lr, ctype, "toupper", charmap); | |
268 | ctype_map_new (lr, ctype, "tolower", charmap); | |
011ebfab | 269 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 270 | ctype_map_new (lr, ctype, "tosymmetric", charmap); |
011ebfab | 271 | #endif |
4b10dd6c UD |
272 | |
273 | /* Fill first 256 entries in `toXXX' arrays. */ | |
274 | for (cnt = 0; cnt < 256; ++cnt) | |
275 | { | |
276 | ctype->map_collection[0][cnt] = cnt; | |
277 | ctype->map_collection[1][cnt] = cnt; | |
9e2b7438 | 278 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 279 | ctype->map_collection[2][cnt] = cnt; |
9e2b7438 | 280 | #endif |
4b10dd6c UD |
281 | ctype->map256_collection[0][cnt] = cnt; |
282 | ctype->map256_collection[1][cnt] = cnt; | |
283 | } | |
284 | ||
285 | obstack_init (&ctype->mem_pool); | |
19bc17a9 RM |
286 | } |
287 | } | |
288 | ||
289 | ||
290 | void | |
4b10dd6c | 291 | ctype_finish (struct localedef_t *locale, struct charmap_t *charmap) |
19bc17a9 RM |
292 | { |
293 | /* See POSIX.2, table 2-6 for the meaning of the following table. */ | |
294 | #define NCLASS 12 | |
295 | static const struct | |
296 | { | |
297 | const char *name; | |
298 | const char allow[NCLASS]; | |
299 | } | |
300 | valid_table[NCLASS] = | |
301 | { | |
302 | /* The order is important. See token.h for more information. | |
303 | M = Always, D = Default, - = Permitted, X = Mutually exclusive */ | |
304 | { "upper", "--MX-XDDXXX-" }, | |
305 | { "lower", "--MX-XDDXXX-" }, | |
306 | { "alpha", "---X-XDDXXX-" }, | |
307 | { "digit", "XXX--XDDXXX-" }, | |
308 | { "xdigit", "-----XDDXXX-" }, | |
309 | { "space", "XXXXX------X" }, | |
310 | { "print", "---------X--" }, | |
311 | { "graph", "---------X--" }, | |
312 | { "blank", "XXXXXM-----X" }, | |
313 | { "cntrl", "XXXXX-XX--XX" }, | |
314 | { "punct", "XXXXX-DD-X-X" }, | |
315 | { "alnum", "-----XDDXXX-" } | |
316 | }; | |
317 | size_t cnt; | |
318 | int cls1, cls2; | |
4b10dd6c UD |
319 | uint32_t space_value; |
320 | struct charseq *space_seq; | |
19bc17a9 | 321 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
4b10dd6c | 322 | int warned; |
19bc17a9 | 323 | |
b9eb05d6 UD |
324 | /* Now resolve copying and also handle completely missing definitions. */ |
325 | if (ctype == NULL) | |
326 | { | |
70e51ab9 UD |
327 | const char *repertoire_name; |
328 | ||
b9eb05d6 UD |
329 | /* First see whether we were supposed to copy. If yes, find the |
330 | actual definition. */ | |
331 | if (locale->copy_name[LC_CTYPE] != NULL) | |
332 | { | |
333 | /* Find the copying locale. This has to happen transitively since | |
334 | the locale we are copying from might also copying another one. */ | |
335 | struct localedef_t *from = locale; | |
336 | ||
337 | do | |
338 | from = find_locale (LC_CTYPE, from->copy_name[LC_CTYPE], | |
339 | from->repertoire_name, charmap); | |
340 | while (from->categories[LC_CTYPE].ctype == NULL | |
341 | && from->copy_name[LC_CTYPE] != NULL); | |
342 | ||
343 | ctype = locale->categories[LC_CTYPE].ctype | |
344 | = from->categories[LC_CTYPE].ctype; | |
345 | } | |
346 | ||
347 | /* If there is still no definition issue an warning and create an | |
348 | empty one. */ | |
349 | if (ctype == NULL) | |
350 | { | |
f6ada7ad UD |
351 | if (! be_quiet) |
352 | error (0, 0, _("No definition for %s category found"), "LC_CTYPE"); | |
b9eb05d6 UD |
353 | ctype_startup (NULL, locale, charmap, 0); |
354 | ctype = locale->categories[LC_CTYPE].ctype; | |
355 | } | |
70e51ab9 UD |
356 | |
357 | /* Get the repertoire we have to use. */ | |
358 | repertoire_name = locale->repertoire_name ?: repertoire_global; | |
359 | if (repertoire_name != NULL) | |
360 | ctype->repertoire = repertoire_read (repertoire_name); | |
b9eb05d6 UD |
361 | } |
362 | ||
19bc17a9 | 363 | /* Set default value for classes not specified. */ |
4b10dd6c | 364 | set_class_defaults (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
365 | |
366 | /* Check according to table. */ | |
42d7c593 | 367 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) |
19bc17a9 | 368 | { |
4b10dd6c | 369 | uint32_t tmp = ctype->class_collection[cnt]; |
19bc17a9 | 370 | |
4b10dd6c UD |
371 | if (tmp != 0) |
372 | { | |
373 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
374 | if ((tmp & _ISwbit (cls1)) != 0) | |
375 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
376 | if (valid_table[cls1].allow[cls2] != '-') | |
19bc17a9 | 377 | { |
4b10dd6c UD |
378 | int eq = (tmp & _ISwbit (cls2)) != 0; |
379 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 380 | { |
4b10dd6c UD |
381 | case 'M': |
382 | if (!eq) | |
383 | { | |
384 | uint32_t value = ctype->charnames[cnt]; | |
385 | ||
386 | if (!be_quiet) | |
387 | error (0, 0, _("\ | |
388 | character L'\\u%0*x' in class `%s' must be in class `%s'"), | |
389 | value > 0xffff ? 8 : 4, value, | |
390 | valid_table[cls1].name, | |
391 | valid_table[cls2].name); | |
392 | } | |
393 | break; | |
394 | ||
395 | case 'X': | |
396 | if (eq) | |
397 | { | |
398 | uint32_t value = ctype->charnames[cnt]; | |
399 | ||
400 | if (!be_quiet) | |
401 | error (0, 0, _("\ | |
402 | character L'\\u%0*x' in class `%s' must not be in class `%s'"), | |
403 | value > 0xffff ? 8 : 4, value, | |
404 | valid_table[cls1].name, | |
405 | valid_table[cls2].name); | |
406 | } | |
407 | break; | |
408 | ||
409 | case 'D': | |
410 | ctype->class_collection[cnt] |= _ISwbit (cls2); | |
411 | break; | |
412 | ||
413 | default: | |
414 | error (5, 0, _("internal error in %s, line %u"), | |
415 | __FUNCTION__, __LINE__); | |
19bc17a9 | 416 | } |
4b10dd6c UD |
417 | } |
418 | } | |
419 | } | |
420 | ||
421 | for (cnt = 0; cnt < 256; ++cnt) | |
422 | { | |
423 | uint32_t tmp = ctype->class256_collection[cnt]; | |
19bc17a9 | 424 | |
4b10dd6c UD |
425 | if (tmp != 0) |
426 | { | |
427 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
428 | if ((tmp & _ISbit (cls1)) != 0) | |
429 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
430 | if (valid_table[cls1].allow[cls2] != '-') | |
431 | { | |
432 | int eq = (tmp & _ISbit (cls2)) != 0; | |
433 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 434 | { |
4b10dd6c UD |
435 | case 'M': |
436 | if (!eq) | |
437 | { | |
438 | char buf[17]; | |
439 | ||
dbbbaf53 | 440 | sprintf (buf, "\\%zo", cnt); |
4b10dd6c UD |
441 | |
442 | if (!be_quiet) | |
443 | error (0, 0, _("\ | |
444 | character '%s' in class `%s' must be in class `%s'"), | |
445 | buf, valid_table[cls1].name, | |
446 | valid_table[cls2].name); | |
447 | } | |
448 | break; | |
449 | ||
450 | case 'X': | |
451 | if (eq) | |
452 | { | |
453 | char buf[17]; | |
454 | ||
dbbbaf53 | 455 | sprintf (buf, "\\%zo", cnt); |
4b10dd6c UD |
456 | |
457 | if (!be_quiet) | |
458 | error (0, 0, _("\ | |
459 | character '%s' in class `%s' must not be in class `%s'"), | |
460 | buf, valid_table[cls1].name, | |
461 | valid_table[cls2].name); | |
462 | } | |
463 | break; | |
464 | ||
465 | case 'D': | |
466 | ctype->class256_collection[cnt] |= _ISbit (cls2); | |
467 | break; | |
468 | ||
469 | default: | |
470 | error (5, 0, _("internal error in %s, line %u"), | |
471 | __FUNCTION__, __LINE__); | |
19bc17a9 | 472 | } |
4b10dd6c UD |
473 | } |
474 | } | |
19bc17a9 RM |
475 | } |
476 | ||
477 | /* ... and now test <SP> as a special case. */ | |
4b10dd6c UD |
478 | space_value = repertoire_find_value (ctype->repertoire, "SP", 2); |
479 | if (space_value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
480 | { |
481 | if (!be_quiet) | |
482 | error (0, 0, _("character <SP> not defined in character map")); | |
483 | } | |
c84142e8 UD |
484 | else if (((cnt = BITPOS (tok_space), |
485 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 486 | & BITw (tok_space)) == 0) |
c84142e8 UD |
487 | || (cnt = BITPOS (tok_blank), |
488 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 489 | & BITw (tok_blank)) == 0))) |
880f421f UD |
490 | { |
491 | if (!be_quiet) | |
492 | error (0, 0, _("<SP> character not in class `%s'"), | |
493 | valid_table[cnt].name); | |
494 | } | |
c84142e8 UD |
495 | else if (((cnt = BITPOS (tok_punct), |
496 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 497 | & BITw (tok_punct)) != 0) |
c84142e8 UD |
498 | || (cnt = BITPOS (tok_graph), |
499 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 500 | & BITw (tok_graph)) |
880f421f UD |
501 | != 0))) |
502 | { | |
503 | if (!be_quiet) | |
504 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
505 | valid_table[cnt].name); | |
506 | } | |
19bc17a9 | 507 | else |
4b10dd6c UD |
508 | ELEM (ctype, class_collection, , space_value) |= BITw (tok_print); |
509 | ||
510 | space_seq = charmap_find_value (charmap, "SP", 2); | |
511 | if (space_seq == NULL || space_seq->nbytes != 1) | |
512 | { | |
513 | if (!be_quiet) | |
514 | error (0, 0, _("character <SP> not defined in character map")); | |
515 | } | |
516 | else if (((cnt = BITPOS (tok_space), | |
517 | (ctype->class256_collection[space_seq->bytes[0]] | |
518 | & BIT (tok_space)) == 0) | |
519 | || (cnt = BITPOS (tok_blank), | |
520 | (ctype->class256_collection[space_seq->bytes[0]] | |
521 | & BIT (tok_blank)) == 0))) | |
522 | { | |
523 | if (!be_quiet) | |
524 | error (0, 0, _("<SP> character not in class `%s'"), | |
525 | valid_table[cnt].name); | |
526 | } | |
527 | else if (((cnt = BITPOS (tok_punct), | |
528 | (ctype->class256_collection[space_seq->bytes[0]] | |
529 | & BIT (tok_punct)) != 0) | |
530 | || (cnt = BITPOS (tok_graph), | |
531 | (ctype->class256_collection[space_seq->bytes[0]] | |
532 | & BIT (tok_graph)) != 0))) | |
533 | { | |
534 | if (!be_quiet) | |
535 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
536 | valid_table[cnt].name); | |
537 | } | |
538 | else | |
539 | ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print); | |
75cd5204 RM |
540 | |
541 | /* Now that the tests are done make sure the name array contains all | |
542 | characters which are handled in the WIDTH section of the | |
543 | character set definition file. */ | |
4b10dd6c UD |
544 | if (charmap->width_rules != NULL) |
545 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) | |
75cd5204 | 546 | { |
827ff758 UD |
547 | unsigned char bytes[charmap->mb_cur_max]; |
548 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
549 | ||
550 | /* We have the range of character for which the width is | |
551 | specified described using byte sequences of the multibyte | |
552 | charset. We have to convert this to UCS4 now. And we | |
553 | cannot simply convert the beginning and the end of the | |
554 | sequence, we have to iterate over the byte sequence and | |
555 | convert it for every single character. */ | |
556 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
557 | ||
558 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
559 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
560 | nbytes) <= 0) | |
561 | { | |
562 | /* Find the UCS value for `bytes'. */ | |
827ff758 | 563 | int inner; |
76e680a8 UD |
564 | uint32_t wch; |
565 | struct charseq *seq = charmap_find_symbol (charmap, bytes, nbytes); | |
566 | ||
567 | if (seq == NULL) | |
568 | wch = ILLEGAL_CHAR_VALUE; | |
569 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
570 | wch = seq->ucs4; | |
571 | else | |
572 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
573 | strlen (seq->name)); | |
827ff758 UD |
574 | |
575 | if (wch != ILLEGAL_CHAR_VALUE) | |
576 | /* We are only interested in the side-effects of the | |
577 | `find_idx' call. It will add appropriate entries in | |
578 | the name array if this is necessary. */ | |
579 | (void) find_idx (ctype, NULL, NULL, NULL, wch); | |
580 | ||
581 | /* "Increment" the bytes sequence. */ | |
582 | inner = nbytes - 1; | |
583 | while (inner >= 0 && bytes[inner] == 0xff) | |
584 | --inner; | |
585 | ||
586 | if (inner < 0) | |
587 | { | |
588 | /* We have to extend the byte sequence. */ | |
589 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
590 | break; | |
591 | ||
592 | bytes[0] = 1; | |
593 | memset (&bytes[1], 0, nbytes); | |
594 | ++nbytes; | |
595 | } | |
596 | else | |
597 | { | |
598 | ++bytes[inner]; | |
599 | while (++inner < nbytes) | |
600 | bytes[inner] = 0; | |
601 | } | |
602 | } | |
4b10dd6c UD |
603 | } |
604 | ||
605 | /* There must be a multiple of 10 digits. */ | |
606 | if (ctype->mbdigits_act % 10 != 0) | |
607 | { | |
608 | assert (ctype->mbdigits_act == ctype->wcdigits_act); | |
609 | ctype->wcdigits_act -= ctype->mbdigits_act % 10; | |
610 | ctype->mbdigits_act -= ctype->mbdigits_act % 10; | |
611 | error (0, 0, _("`digit' category has not entries in groups of ten")); | |
612 | } | |
613 | ||
614 | /* Check the input digits. There must be a multiple of ten available. | |
42d7c593 | 615 | In each group it could be that one or the other character is missing. |
4b10dd6c UD |
616 | In this case the whole group must be removed. */ |
617 | cnt = 0; | |
618 | while (cnt < ctype->mbdigits_act) | |
619 | { | |
620 | size_t inner; | |
621 | for (inner = 0; inner < 10; ++inner) | |
622 | if (ctype->mbdigits[cnt + inner] == NULL) | |
623 | break; | |
624 | ||
625 | if (inner == 10) | |
626 | cnt += 10; | |
627 | else | |
628 | { | |
629 | /* Remove the group. */ | |
630 | memmove (&ctype->mbdigits[cnt], &ctype->mbdigits[cnt + 10], | |
631 | ((ctype->wcdigits_act - cnt - 10) | |
632 | * sizeof (ctype->mbdigits[0]))); | |
633 | ctype->mbdigits_act -= 10; | |
634 | } | |
635 | } | |
636 | ||
637 | /* If no input digits are given use the default. */ | |
638 | if (ctype->mbdigits_act == 0) | |
639 | { | |
640 | if (ctype->mbdigits_max == 0) | |
641 | { | |
642 | ctype->mbdigits = obstack_alloc (&charmap->mem_pool, | |
643 | 10 * sizeof (struct charseq *)); | |
644 | ctype->mbdigits_max = 10; | |
645 | } | |
646 | ||
647 | for (cnt = 0; cnt < 10; ++cnt) | |
648 | { | |
649 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
650 | digits + cnt, 1); | |
651 | if (ctype->mbdigits[cnt] == NULL) | |
652 | { | |
653 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
654 | longnames[cnt], | |
655 | strlen (longnames[cnt])); | |
656 | if (ctype->mbdigits[cnt] == NULL) | |
657 | { | |
658 | /* Hum, this ain't good. */ | |
659 | error (0, 0, _("\ | |
660 | no input digits defined and none of the standard names in the charmap")); | |
661 | ||
662 | ctype->mbdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
663 | sizeof (struct charseq) + 1); | |
664 | ||
665 | /* This is better than nothing. */ | |
666 | ctype->mbdigits[cnt]->bytes[0] = digits[cnt]; | |
667 | ctype->mbdigits[cnt]->nbytes = 1; | |
668 | } | |
669 | } | |
670 | } | |
671 | ||
672 | ctype->mbdigits_act = 10; | |
673 | } | |
674 | ||
675 | /* Check the wide character input digits. There must be a multiple | |
42d7c593 | 676 | of ten available. In each group it could be that one or the other |
4b10dd6c UD |
677 | character is missing. In this case the whole group must be |
678 | removed. */ | |
679 | cnt = 0; | |
680 | while (cnt < ctype->wcdigits_act) | |
681 | { | |
682 | size_t inner; | |
683 | for (inner = 0; inner < 10; ++inner) | |
684 | if (ctype->wcdigits[cnt + inner] == ILLEGAL_CHAR_VALUE) | |
685 | break; | |
686 | ||
687 | if (inner == 10) | |
688 | cnt += 10; | |
689 | else | |
690 | { | |
691 | /* Remove the group. */ | |
692 | memmove (&ctype->wcdigits[cnt], &ctype->wcdigits[cnt + 10], | |
693 | ((ctype->wcdigits_act - cnt - 10) | |
694 | * sizeof (ctype->wcdigits[0]))); | |
695 | ctype->wcdigits_act -= 10; | |
696 | } | |
697 | } | |
698 | ||
699 | /* If no input digits are given use the default. */ | |
700 | if (ctype->wcdigits_act == 0) | |
701 | { | |
702 | if (ctype->wcdigits_max == 0) | |
703 | { | |
704 | ctype->wcdigits = obstack_alloc (&charmap->mem_pool, | |
705 | 10 * sizeof (uint32_t)); | |
706 | ctype->wcdigits_max = 10; | |
707 | } | |
708 | ||
709 | for (cnt = 0; cnt < 10; ++cnt) | |
710 | ctype->wcdigits[cnt] = L'0' + cnt; | |
711 | ||
712 | ctype->mbdigits_act = 10; | |
713 | } | |
714 | ||
715 | /* Check the outdigits. */ | |
716 | warned = 0; | |
717 | for (cnt = 0; cnt < 10; ++cnt) | |
718 | if (ctype->mboutdigits[cnt] == NULL) | |
719 | { | |
720 | static struct charseq replace[2]; | |
721 | ||
722 | if (!warned) | |
723 | { | |
724 | error (0, 0, _("\ | |
725 | not all characters used in `outdigit' are available in the charmap")); | |
726 | warned = 1; | |
727 | } | |
728 | ||
729 | replace[0].nbytes = 1; | |
730 | replace[0].bytes[0] = '?'; | |
731 | replace[0].bytes[1] = '\0'; | |
732 | ctype->mboutdigits[cnt] = &replace[0]; | |
733 | } | |
734 | ||
735 | warned = 0; | |
736 | for (cnt = 0; cnt < 10; ++cnt) | |
737 | if (ctype->wcoutdigits[cnt] == 0) | |
738 | { | |
739 | if (!warned) | |
740 | { | |
741 | error (0, 0, _("\ | |
742 | not all characters used in `outdigit' are available in the repertoire")); | |
743 | warned = 1; | |
744 | } | |
745 | ||
746 | ctype->wcoutdigits[cnt] = L'?'; | |
75cd5204 | 747 | } |
19bc17a9 RM |
748 | } |
749 | ||
750 | ||
751 | void | |
4b10dd6c | 752 | ctype_output (struct localedef_t *locale, struct charmap_t *charmap, |
75cd5204 | 753 | const char *output_path) |
19bc17a9 RM |
754 | { |
755 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; | |
756 | const size_t nelems = (_NL_ITEM_INDEX (_NL_NUM_LC_CTYPE) | |
5491da0d | 757 | + (ctype->map_collection_nr - 2)); |
75cd5204 RM |
758 | struct iovec iov[2 + nelems + ctype->nr_charclass |
759 | + ctype->map_collection_nr]; | |
19bc17a9 | 760 | struct locale_file data; |
4b10dd6c | 761 | uint32_t idx[nelems + 1]; |
75cd5204 | 762 | size_t elem, cnt, offset, total; |
4b10dd6c | 763 | char *cp; |
19bc17a9 RM |
764 | |
765 | /* Now prepare the output: Find the sizes of the table we can use. */ | |
4b10dd6c | 766 | allocate_arrays (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
767 | |
768 | data.magic = LIMAGIC (LC_CTYPE); | |
769 | data.n = nelems; | |
770 | iov[0].iov_base = (void *) &data; | |
771 | iov[0].iov_len = sizeof (data); | |
772 | ||
773 | iov[1].iov_base = (void *) idx; | |
774 | iov[1].iov_len = sizeof (idx); | |
775 | ||
776 | idx[0] = iov[0].iov_len + iov[1].iov_len; | |
777 | offset = 0; | |
778 | ||
779 | for (elem = 0; elem < nelems; ++elem) | |
780 | { | |
781 | if (elem < _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) | |
782 | switch (elem) | |
783 | { | |
784 | #define CTYPE_DATA(name, base, len) \ | |
785 | case _NL_ITEM_INDEX (name): \ | |
ce7a5ef4 RM |
786 | iov[2 + elem + offset].iov_base = (base); \ |
787 | iov[2 + elem + offset].iov_len = (len); \ | |
75cd5204 RM |
788 | if (elem + 1 < nelems) \ |
789 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; \ | |
19bc17a9 RM |
790 | break |
791 | ||
792 | CTYPE_DATA (_NL_CTYPE_CLASS, | |
793 | ctype->ctype_b, | |
794 | (256 + 128) * sizeof (char_class_t)); | |
795 | ||
4a33c2f5 UD |
796 | CTYPE_DATA (_NL_CTYPE_TOUPPER, |
797 | ctype->map[0], | |
f1d8b804 | 798 | (256 + 128) * sizeof (uint32_t)); |
4a33c2f5 UD |
799 | CTYPE_DATA (_NL_CTYPE_TOLOWER, |
800 | ctype->map[1], | |
f1d8b804 | 801 | (256 + 128) * sizeof (uint32_t)); |
19bc17a9 | 802 | |
49f2be5b UD |
803 | CTYPE_DATA (_NL_CTYPE_TOUPPER32, |
804 | ctype->map32[0], | |
f1d8b804 | 805 | (ctype->plane_size * ctype->plane_cnt) |
49f2be5b UD |
806 | * sizeof (uint32_t)); |
807 | CTYPE_DATA (_NL_CTYPE_TOLOWER32, | |
808 | ctype->map32[1], | |
f1d8b804 | 809 | (ctype->plane_size * ctype->plane_cnt) |
49f2be5b UD |
810 | * sizeof (uint32_t)); |
811 | ||
19bc17a9 RM |
812 | CTYPE_DATA (_NL_CTYPE_CLASS32, |
813 | ctype->ctype32_b, | |
814 | (ctype->plane_size * ctype->plane_cnt | |
815 | * sizeof (char_class32_t))); | |
816 | ||
4a33c2f5 UD |
817 | CTYPE_DATA (_NL_CTYPE_NAMES, |
818 | ctype->names, (ctype->plane_size * ctype->plane_cnt | |
819 | * sizeof (uint32_t))); | |
820 | ||
821 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_SIZE, | |
822 | &ctype->translit_hash_size, sizeof (uint32_t)); | |
823 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_LAYERS, | |
824 | &ctype->translit_hash_layers, sizeof (uint32_t)); | |
825 | ||
826 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_IDX, | |
827 | ctype->translit_from_idx, | |
4b10dd6c UD |
828 | ctype->translit_idx_size); |
829 | ||
4a33c2f5 UD |
830 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_TBL, |
831 | ctype->translit_from_tbl, | |
4b10dd6c UD |
832 | ctype->translit_from_tbl_size); |
833 | ||
4a33c2f5 UD |
834 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_IDX, |
835 | ctype->translit_to_idx, | |
4b10dd6c UD |
836 | ctype->translit_idx_size); |
837 | ||
4a33c2f5 UD |
838 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_TBL, |
839 | ctype->translit_to_tbl, ctype->translit_to_tbl_size); | |
4b10dd6c | 840 | |
4a33c2f5 | 841 | CTYPE_DATA (_NL_CTYPE_HASH_SIZE, |
4b10dd6c | 842 | &ctype->plane_size, sizeof (uint32_t)); |
4a33c2f5 | 843 | CTYPE_DATA (_NL_CTYPE_HASH_LAYERS, |
4b10dd6c | 844 | &ctype->plane_cnt, sizeof (uint32_t)); |
19bc17a9 | 845 | |
75cd5204 RM |
846 | case _NL_ITEM_INDEX (_NL_CTYPE_CLASS_NAMES): |
847 | /* The class name array. */ | |
848 | total = 0; | |
849 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt, ++offset) | |
850 | { | |
851 | iov[2 + elem + offset].iov_base | |
852 | = (void *) ctype->classnames[cnt]; | |
853 | iov[2 + elem + offset].iov_len | |
854 | = strlen (ctype->classnames[cnt]) + 1; | |
855 | total += iov[2 + elem + offset].iov_len; | |
856 | } | |
ce7a5ef4 RM |
857 | iov[2 + elem + offset].iov_base = (void *) "\0\0\0"; |
858 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); | |
859 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 860 | |
4b10dd6c | 861 | idx[elem + 1] = idx[elem] + total; |
75cd5204 RM |
862 | break; |
863 | ||
864 | case _NL_ITEM_INDEX (_NL_CTYPE_MAP_NAMES): | |
865 | /* The class name array. */ | |
866 | total = 0; | |
867 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt, ++offset) | |
868 | { | |
869 | iov[2 + elem + offset].iov_base | |
870 | = (void *) ctype->mapnames[cnt]; | |
871 | iov[2 + elem + offset].iov_len | |
872 | = strlen (ctype->mapnames[cnt]) + 1; | |
873 | total += iov[2 + elem + offset].iov_len; | |
874 | } | |
ce7a5ef4 RM |
875 | iov[2 + elem + offset].iov_base = (void *) "\0\0\0"; |
876 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); | |
877 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 878 | |
4b10dd6c | 879 | idx[elem + 1] = idx[elem] + total; |
75cd5204 | 880 | break; |
19bc17a9 RM |
881 | |
882 | CTYPE_DATA (_NL_CTYPE_WIDTH, | |
75cd5204 | 883 | ctype->width, ctype->plane_size * ctype->plane_cnt); |
19bc17a9 | 884 | |
0200214b | 885 | CTYPE_DATA (_NL_CTYPE_MB_CUR_MAX, |
4b10dd6c | 886 | &ctype->mb_cur_max, sizeof (uint32_t)); |
0200214b | 887 | |
ce7a5ef4 RM |
888 | case _NL_ITEM_INDEX (_NL_CTYPE_CODESET_NAME): |
889 | total = strlen (ctype->codeset_name) + 1; | |
890 | if (total % 4 == 0) | |
891 | iov[2 + elem + offset].iov_base = (char *) ctype->codeset_name; | |
892 | else | |
893 | { | |
894 | iov[2 + elem + offset].iov_base = alloca ((total + 3) & ~3); | |
9756dfe1 UD |
895 | memset (mempcpy (iov[2 + elem + offset].iov_base, |
896 | ctype->codeset_name, total), | |
897 | '\0', 4 - (total & 3)); | |
ce7a5ef4 RM |
898 | total = (total + 3) & ~3; |
899 | } | |
900 | iov[2 + elem + offset].iov_len = total; | |
4b10dd6c UD |
901 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
902 | break; | |
903 | ||
4a33c2f5 | 904 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN): |
4b10dd6c UD |
905 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
906 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
907 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
908 | ctype->mbdigits_act / 10; | |
a9c27b3e | 909 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
910 | break; |
911 | ||
4a33c2f5 | 912 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN): |
4b10dd6c UD |
913 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
914 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
915 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
916 | ctype->wcdigits_act / 10; | |
a9c27b3e | 917 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
918 | break; |
919 | ||
920 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_MB): | |
921 | /* Compute the length of all possible characters. For INDIGITS | |
922 | there might be more than one. We simply concatenate all of | |
923 | them with a NUL byte following. The NUL byte wouldn't be | |
924 | necessary but it makes it easier for the user. */ | |
925 | total = 0; | |
926 | for (cnt = elem - _NL_CTYPE_INDIGITS0_MB; | |
927 | cnt < ctype->mbdigits_act; cnt += 10) | |
928 | total += ctype->mbdigits[cnt]->nbytes + 1; | |
929 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
930 | iov[2 + elem + offset].iov_len = total; | |
931 | ||
932 | cp = iov[2 + elem + offset].iov_base; | |
933 | for (cnt = elem - _NL_CTYPE_INDIGITS0_MB; | |
934 | cnt < ctype->mbdigits_act; cnt += 10) | |
935 | { | |
936 | cp = mempcpy (cp, ctype->mbdigits[cnt]->bytes, | |
937 | ctype->mbdigits[cnt]->nbytes); | |
938 | *cp++ = '\0'; | |
939 | } | |
a9c27b3e | 940 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
941 | break; |
942 | ||
943 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_MB): | |
944 | /* Compute the length of all possible characters. For INDIGITS | |
945 | there might be more than one. We simply concatenate all of | |
946 | them with a NUL byte following. The NUL byte wouldn't be | |
947 | necessary but it makes it easier for the user. */ | |
948 | cnt = elem - _NL_CTYPE_OUTDIGIT0_MB; | |
949 | total = ctype->mboutdigits[cnt]->nbytes + 1; | |
950 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
951 | iov[2 + elem + offset].iov_len = total; | |
952 | ||
953 | *(char *) mempcpy (iov[2 + elem + offset].iov_base, | |
954 | ctype->mbdigits[cnt]->bytes, | |
955 | ctype->mbdigits[cnt]->nbytes) = '\0'; | |
a9c27b3e | 956 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
957 | break; |
958 | ||
4a33c2f5 | 959 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC): |
4b10dd6c UD |
960 | total = ctype->wcdigits_act / 10; |
961 | ||
962 | iov[2 + elem + offset].iov_base = | |
963 | (uint32_t *) alloca (total * sizeof (uint32_t)); | |
964 | iov[2 + elem + offset].iov_len = total * sizeof (uint32_t); | |
965 | ||
4a33c2f5 | 966 | for (cnt = elem - _NL_CTYPE_INDIGITS0_WC; |
4b10dd6c UD |
967 | cnt < ctype->wcdigits_act; cnt += 10) |
968 | ((uint32_t *) iov[2 + elem + offset].iov_base)[cnt / 10] | |
4a33c2f5 | 969 | = ctype->wcdigits[cnt]; |
a9c27b3e | 970 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
971 | break; |
972 | ||
4a33c2f5 UD |
973 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC): |
974 | cnt = elem - _NL_CTYPE_OUTDIGIT0_WC; | |
4b10dd6c UD |
975 | iov[2 + elem + offset].iov_base = &ctype->wcoutdigits[cnt]; |
976 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
a9c27b3e | 977 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
978 | break; |
979 | ||
19bc17a9 RM |
980 | default: |
981 | assert (! "unknown CTYPE element"); | |
982 | } | |
983 | else | |
984 | { | |
985 | /* Handle extra maps. */ | |
5491da0d | 986 | size_t nr = (elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) + 2; |
19bc17a9 | 987 | |
49f2be5b | 988 | iov[2 + elem + offset].iov_base = ctype->map32[nr]; |
75cd5204 | 989 | iov[2 + elem + offset].iov_len = ((ctype->plane_size |
f1d8b804 | 990 | * ctype->plane_cnt) |
4b10dd6c | 991 | * sizeof (uint32_t)); |
19bc17a9 | 992 | |
4b10dd6c | 993 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
75cd5204 | 994 | } |
19bc17a9 | 995 | } |
19bc17a9 | 996 | |
75cd5204 RM |
997 | assert (2 + elem + offset == (nelems + ctype->nr_charclass |
998 | + ctype->map_collection_nr + 2)); | |
19bc17a9 | 999 | |
75cd5204 | 1000 | write_locale_data (output_path, "LC_CTYPE", 2 + elem + offset, iov); |
19bc17a9 RM |
1001 | } |
1002 | ||
1003 | ||
4b10dd6c UD |
1004 | /* Local functions. */ |
1005 | static void | |
1006 | ctype_class_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1007 | const char *name) | |
19bc17a9 | 1008 | { |
4b10dd6c | 1009 | size_t cnt; |
19bc17a9 | 1010 | |
4b10dd6c UD |
1011 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1012 | if (strcmp (ctype->classnames[cnt], name) == 0) | |
1013 | break; | |
19bc17a9 | 1014 | |
4b10dd6c UD |
1015 | if (cnt < ctype->nr_charclass) |
1016 | { | |
1017 | lr_error (lr, _("character class `%s' already defined"), name); | |
1018 | return; | |
1019 | } | |
19bc17a9 | 1020 | |
4b10dd6c UD |
1021 | if (ctype->nr_charclass == MAX_NR_CHARCLASS) |
1022 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1023 | error (2, 0, _("\ | |
dbbbaf53 | 1024 | implementation limit: no more than %zd character classes allowed"), |
4b10dd6c | 1025 | MAX_NR_CHARCLASS); |
19bc17a9 | 1026 | |
4b10dd6c | 1027 | ctype->classnames[ctype->nr_charclass++] = name; |
19bc17a9 RM |
1028 | } |
1029 | ||
1030 | ||
4b10dd6c UD |
1031 | static void |
1032 | ctype_map_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1033 | const char *name, struct charmap_t *charmap) | |
19bc17a9 | 1034 | { |
4b10dd6c | 1035 | size_t max_chars = 0; |
ba1ffaa1 | 1036 | size_t cnt; |
19bc17a9 | 1037 | |
4b10dd6c | 1038 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
19bc17a9 | 1039 | { |
4b10dd6c UD |
1040 | if (strcmp (ctype->mapnames[cnt], name) == 0) |
1041 | break; | |
1042 | ||
1043 | if (max_chars < ctype->map_collection_max[cnt]) | |
1044 | max_chars = ctype->map_collection_max[cnt]; | |
19bc17a9 RM |
1045 | } |
1046 | ||
4b10dd6c UD |
1047 | if (cnt < ctype->map_collection_nr) |
1048 | { | |
1049 | lr_error (lr, _("character map `%s' already defined"), name); | |
1050 | return; | |
1051 | } | |
19bc17a9 | 1052 | |
4b10dd6c UD |
1053 | if (ctype->map_collection_nr == MAX_NR_CHARMAP) |
1054 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1055 | error (2, 0, _("\ | |
1056 | implementation limit: no more than %d character maps allowed"), | |
1057 | MAX_NR_CHARMAP); | |
19bc17a9 | 1058 | |
4b10dd6c UD |
1059 | ctype->mapnames[cnt] = name; |
1060 | ||
1061 | if (max_chars == 0) | |
1062 | ctype->map_collection_max[cnt] = charmap->mb_cur_max == 1 ? 256 : 512; | |
1063 | else | |
1064 | ctype->map_collection_max[cnt] = max_chars; | |
1065 | ||
1066 | ctype->map_collection[cnt] = (uint32_t *) | |
1067 | xmalloc (sizeof (uint32_t) * ctype->map_collection_max[cnt]); | |
1068 | memset (ctype->map_collection[cnt], '\0', | |
1069 | sizeof (uint32_t) * ctype->map_collection_max[cnt]); | |
1070 | ctype->map_collection_act[cnt] = 256; | |
19bc17a9 | 1071 | |
4b10dd6c | 1072 | ++ctype->map_collection_nr; |
19bc17a9 RM |
1073 | } |
1074 | ||
1075 | ||
4b10dd6c | 1076 | /* We have to be prepared that TABLE, MAX, and ACT can be NULL. This |
42d7c593 | 1077 | is possible if we only want to extend the name array. */ |
4b10dd6c UD |
1078 | static uint32_t * |
1079 | find_idx (struct locale_ctype_t *ctype, uint32_t **table, size_t *max, | |
1080 | size_t *act, uint32_t idx) | |
19bc17a9 | 1081 | { |
4b10dd6c | 1082 | size_t cnt; |
19bc17a9 | 1083 | |
4b10dd6c UD |
1084 | if (idx < 256) |
1085 | return table == NULL ? NULL : &(*table)[idx]; | |
19bc17a9 | 1086 | |
4b10dd6c UD |
1087 | for (cnt = 256; cnt < ctype->charnames_act; ++cnt) |
1088 | if (ctype->charnames[cnt] == idx) | |
1089 | break; | |
19bc17a9 | 1090 | |
4b10dd6c UD |
1091 | /* We have to distinguish two cases: the name is found or not. */ |
1092 | if (cnt == ctype->charnames_act) | |
1093 | { | |
1094 | /* Extend the name array. */ | |
1095 | if (ctype->charnames_act == ctype->charnames_max) | |
1096 | { | |
1097 | ctype->charnames_max *= 2; | |
1098 | ctype->charnames = (unsigned int *) | |
1099 | xrealloc (ctype->charnames, | |
1100 | sizeof (unsigned int) * ctype->charnames_max); | |
1101 | } | |
1102 | ctype->charnames[ctype->charnames_act++] = idx; | |
1103 | } | |
19bc17a9 | 1104 | |
4b10dd6c UD |
1105 | if (table == NULL) |
1106 | /* We have done everything we are asked to do. */ | |
1107 | return NULL; | |
19bc17a9 | 1108 | |
4b10dd6c UD |
1109 | if (cnt >= *act) |
1110 | { | |
1111 | if (cnt >= *max) | |
1112 | { | |
1113 | size_t old_max = *max; | |
1114 | do | |
1115 | *max *= 2; | |
1116 | while (*max <= cnt); | |
19bc17a9 | 1117 | |
4b10dd6c UD |
1118 | *table = |
1119 | (uint32_t *) xrealloc (*table, *max * sizeof (unsigned long int)); | |
1120 | memset (&(*table)[old_max], '\0', | |
1121 | (*max - old_max) * sizeof (uint32_t)); | |
1122 | } | |
19bc17a9 | 1123 | |
76e680a8 | 1124 | *act = cnt + 1; |
4b10dd6c | 1125 | } |
19bc17a9 | 1126 | |
4b10dd6c | 1127 | return &(*table)[cnt]; |
19bc17a9 RM |
1128 | } |
1129 | ||
1130 | ||
4b10dd6c UD |
1131 | static int |
1132 | get_character (struct token *now, struct charmap_t *charmap, | |
1133 | struct repertoire_t *repertoire, | |
1134 | struct charseq **seqp, uint32_t *wchp) | |
19bc17a9 | 1135 | { |
4b10dd6c UD |
1136 | if (now->tok == tok_bsymbol) |
1137 | { | |
1138 | /* This will hopefully be the normal case. */ | |
1139 | *wchp = repertoire_find_value (repertoire, now->val.str.startmb, | |
1140 | now->val.str.lenmb); | |
1141 | *seqp = charmap_find_value (charmap, now->val.str.startmb, | |
1142 | now->val.str.lenmb); | |
1143 | } | |
1144 | else if (now->tok == tok_ucs4) | |
1145 | { | |
1146 | *seqp = repertoire_find_seq (repertoire, now->val.ucs4); | |
19bc17a9 | 1147 | |
4b10dd6c UD |
1148 | if (*seqp == NULL) |
1149 | { | |
1150 | /* Compute the value in the charmap from the UCS value. */ | |
1151 | const char *symbol = repertoire_find_symbol (repertoire, | |
1152 | now->val.ucs4); | |
19bc17a9 | 1153 | |
4b10dd6c UD |
1154 | if (symbol == NULL) |
1155 | *seqp = NULL; | |
1156 | else | |
1157 | *seqp = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1158 | |
4b10dd6c UD |
1159 | if (*seqp == NULL) |
1160 | { | |
1161 | /* Insert a negative entry. */ | |
1162 | static const struct charseq negative | |
1163 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1164 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, 4); | |
1165 | *newp = now->val.ucs4; | |
1166 | ||
1167 | insert_entry (&repertoire->seq_table, newp, 4, | |
1168 | (void *) &negative); | |
1169 | } | |
1170 | else | |
1171 | (*seqp)->ucs4 = now->val.ucs4; | |
1172 | } | |
1173 | else if ((*seqp)->ucs4 != now->val.ucs4) | |
1174 | *seqp = NULL; | |
19bc17a9 | 1175 | |
4b10dd6c UD |
1176 | *wchp = now->val.ucs4; |
1177 | } | |
1178 | else if (now->tok == tok_charcode) | |
1179 | { | |
1180 | /* We must map from the byte code to UCS4. */ | |
1181 | *seqp = charmap_find_symbol (charmap, now->val.str.startmb, | |
1182 | now->val.str.lenmb); | |
19bc17a9 | 1183 | |
4b10dd6c UD |
1184 | if (*seqp == NULL) |
1185 | *wchp = ILLEGAL_CHAR_VALUE; | |
1186 | else | |
1187 | { | |
1188 | if ((*seqp)->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1189 | (*seqp)->ucs4 = repertoire_find_value (repertoire, (*seqp)->name, | |
1190 | strlen ((*seqp)->name)); | |
1191 | *wchp = (*seqp)->ucs4; | |
1192 | } | |
1193 | } | |
1194 | else | |
1195 | return 1; | |
19bc17a9 RM |
1196 | |
1197 | return 0; | |
1198 | } | |
1199 | ||
1200 | ||
4b10dd6c UD |
1201 | /* Ellipsis like in `<foo123>..<foo12a>' or `<j1234>....<j1245>'. */ |
1202 | static void | |
1203 | charclass_symbolic_ellipsis (struct linereader *ldfile, | |
1204 | struct locale_ctype_t *ctype, | |
1205 | struct charmap_t *charmap, | |
1206 | struct repertoire_t *repertoire, | |
1207 | struct token *now, | |
1208 | const char *last_str, | |
1209 | unsigned long int class256_bit, | |
1210 | unsigned long int class_bit, int base, | |
1211 | int ignore_content, int handle_digits) | |
19bc17a9 | 1212 | { |
4b10dd6c UD |
1213 | const char *nowstr = now->val.str.startmb; |
1214 | char tmp[now->val.str.lenmb + 1]; | |
1215 | const char *cp; | |
1216 | char *endp; | |
1217 | unsigned long int from; | |
1218 | unsigned long int to; | |
19bc17a9 | 1219 | |
4b10dd6c UD |
1220 | /* We have to compute the ellipsis values using the symbolic names. */ |
1221 | assert (last_str != NULL); | |
1222 | ||
1223 | if (strlen (last_str) != now->val.str.lenmb) | |
19bc17a9 | 1224 | { |
4b10dd6c UD |
1225 | invalid_range: |
1226 | lr_error (ldfile, | |
549b3c3a | 1227 | _("`%s' and `%.*s' are no valid names for symbolic range"), |
f6ada7ad | 1228 | last_str, (int) now->val.str.lenmb, nowstr); |
4b10dd6c | 1229 | return; |
19bc17a9 RM |
1230 | } |
1231 | ||
4b10dd6c UD |
1232 | if (memcmp (last_str, nowstr, now->val.str.lenmb) == 0) |
1233 | /* Nothing to do, the names are the same. */ | |
1234 | return; | |
19bc17a9 | 1235 | |
4b10dd6c UD |
1236 | for (cp = last_str; *cp == *(nowstr + (cp - last_str)); ++cp) |
1237 | ; | |
19bc17a9 | 1238 | |
4b10dd6c UD |
1239 | errno = 0; |
1240 | from = strtoul (cp, &endp, base); | |
1241 | if ((from == UINT_MAX && errno == ERANGE) || *endp != '\0') | |
1242 | goto invalid_range; | |
19bc17a9 | 1243 | |
4b10dd6c | 1244 | to = strtoul (nowstr + (cp - last_str), &endp, base); |
549b3c3a UD |
1245 | if ((to == UINT_MAX && errno == ERANGE) |
1246 | || (endp - nowstr) != now->val.str.lenmb || from >= to) | |
4b10dd6c | 1247 | goto invalid_range; |
19bc17a9 | 1248 | |
4b10dd6c UD |
1249 | /* OK, we have a range FROM - TO. Now we can create the symbolic names. */ |
1250 | if (!ignore_content) | |
1251 | { | |
1252 | now->val.str.startmb = tmp; | |
1253 | while (++from <= to) | |
1254 | { | |
1255 | struct charseq *seq; | |
1256 | uint32_t wch; | |
19bc17a9 | 1257 | |
4b10dd6c UD |
1258 | sprintf (tmp, (base == 10 ? "%.*s%0*d" : "%.*s%0*X"), cp - last_str, |
1259 | last_str, now->val.str.lenmb - (cp - last_str), from); | |
19bc17a9 | 1260 | |
4b10dd6c UD |
1261 | get_character (now, charmap, repertoire, &seq, &wch); |
1262 | ||
1263 | if (seq != NULL && seq->nbytes == 1) | |
1264 | /* Yep, we can store information about this byte sequence. */ | |
1265 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
19bc17a9 | 1266 | |
4b10dd6c UD |
1267 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) |
1268 | /* We have the UCS4 position. */ | |
1269 | *find_idx (ctype, &ctype->class_collection, | |
1270 | &ctype->class_collection_max, | |
1271 | &ctype->class_collection_act, wch) |= class_bit; | |
19bc17a9 | 1272 | |
4b10dd6c UD |
1273 | if (handle_digits == 1) |
1274 | { | |
1275 | /* We must store the digit values. */ | |
1276 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1277 | { | |
1278 | ctype->mbdigits_max *= 2; | |
1279 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1280 | (ctype->mbdigits_max | |
1281 | * sizeof (char *))); | |
1282 | ctype->wcdigits_max *= 2; | |
1283 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1284 | (ctype->wcdigits_max | |
1285 | * sizeof (uint32_t))); | |
1286 | } | |
1287 | ||
1288 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1289 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1290 | } | |
1291 | else if (handle_digits == 2) | |
1292 | { | |
1293 | /* We must store the digit values. */ | |
1294 | if (ctype->outdigits_act >= 10) | |
1295 | { | |
1296 | lr_error (ldfile, _("\ | |
1297 | %s: field `%s' does not contain exactly ten entries"), | |
1298 | "LC_CTYPE", "outdigit"); | |
1299 | return; | |
1300 | } | |
1301 | ||
1302 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1303 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1304 | ++ctype->outdigits_act; | |
1305 | } | |
1306 | } | |
1307 | } | |
19bc17a9 RM |
1308 | } |
1309 | ||
1310 | ||
4b10dd6c UD |
1311 | /* Ellipsis like in `<U1234>..<U2345>'. */ |
1312 | static void | |
1313 | charclass_ucs4_ellipsis (struct linereader *ldfile, | |
1314 | struct locale_ctype_t *ctype, | |
1315 | struct charmap_t *charmap, | |
1316 | struct repertoire_t *repertoire, | |
1317 | struct token *now, uint32_t last_wch, | |
1318 | unsigned long int class256_bit, | |
1319 | unsigned long int class_bit, int ignore_content, | |
1320 | int handle_digits) | |
19bc17a9 | 1321 | { |
4b10dd6c | 1322 | if (last_wch > now->val.ucs4) |
19bc17a9 | 1323 | { |
4b10dd6c UD |
1324 | lr_error (ldfile, _("\ |
1325 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1326 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, now->val.ucs4, | |
1327 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, last_wch); | |
19bc17a9 RM |
1328 | return; |
1329 | } | |
1330 | ||
4b10dd6c UD |
1331 | if (!ignore_content) |
1332 | while (++last_wch <= now->val.ucs4) | |
1333 | { | |
1334 | /* We have to find out whether there is a byte sequence corresponding | |
1335 | to this UCS4 value. */ | |
1336 | struct charseq *seq = repertoire_find_seq (repertoire, last_wch); | |
19bc17a9 | 1337 | |
4b10dd6c UD |
1338 | /* If this is the first time we look for this sequence create a new |
1339 | entry. */ | |
1340 | if (seq == NULL) | |
1341 | { | |
1342 | /* Find the symbolic name for this UCS4 value. */ | |
1343 | const char *symbol = repertoire_find_symbol (repertoire, last_wch); | |
1344 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, 4); | |
1345 | *newp = last_wch; | |
19bc17a9 | 1346 | |
4b10dd6c UD |
1347 | if (symbol != NULL) |
1348 | /* We have a name, now search the multibyte value. */ | |
1349 | seq = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1350 | |
4b10dd6c UD |
1351 | if (seq == NULL) |
1352 | { | |
1353 | /* We have to create a fake entry. */ | |
1354 | static const struct charseq negative | |
1355 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1356 | seq = (struct charseq *) &negative; | |
1357 | } | |
1358 | else | |
1359 | seq->ucs4 = last_wch; | |
1360 | ||
1361 | insert_entry (&repertoire->seq_table, newp, 4, seq); | |
1362 | } | |
1363 | ||
1364 | /* We have a name, now search the multibyte value. */ | |
1365 | if (seq->ucs4 == last_wch && seq->nbytes == 1) | |
1366 | /* Yep, we can store information about this byte sequence. */ | |
1367 | ctype->class256_collection[(size_t) seq->bytes[0]] | |
1368 | |= class256_bit; | |
1369 | ||
1370 | /* And of course we have the UCS4 position. */ | |
1371 | if (class_bit != 0 && class_bit != 0) | |
1372 | *find_idx (ctype, &ctype->class_collection, | |
1373 | &ctype->class_collection_max, | |
1374 | &ctype->class_collection_act, last_wch) |= class_bit; | |
1375 | ||
1376 | if (handle_digits == 1) | |
1377 | { | |
1378 | /* We must store the digit values. */ | |
1379 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1380 | { | |
1381 | ctype->mbdigits_max *= 2; | |
1382 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1383 | (ctype->mbdigits_max | |
1384 | * sizeof (char *))); | |
1385 | ctype->wcdigits_max *= 2; | |
1386 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1387 | (ctype->wcdigits_max | |
1388 | * sizeof (uint32_t))); | |
1389 | } | |
1390 | ||
1391 | ctype->mbdigits[ctype->mbdigits_act++] = (seq->ucs4 == last_wch | |
1392 | ? seq : NULL); | |
1393 | ctype->wcdigits[ctype->wcdigits_act++] = last_wch; | |
1394 | } | |
1395 | else if (handle_digits == 2) | |
1396 | { | |
1397 | /* We must store the digit values. */ | |
1398 | if (ctype->outdigits_act >= 10) | |
1399 | { | |
1400 | lr_error (ldfile, _("\ | |
1401 | %s: field `%s' does not contain exactly ten entries"), | |
1402 | "LC_CTYPE", "outdigit"); | |
1403 | return; | |
1404 | } | |
19bc17a9 | 1405 | |
4b10dd6c UD |
1406 | ctype->mboutdigits[ctype->outdigits_act] = (seq->ucs4 == last_wch |
1407 | ? seq : NULL); | |
1408 | ctype->wcoutdigits[ctype->outdigits_act] = last_wch; | |
1409 | ++ctype->outdigits_act; | |
1410 | } | |
1411 | } | |
19bc17a9 RM |
1412 | } |
1413 | ||
1414 | ||
4b10dd6c | 1415 | /* Ellipsis as in `/xea/x12.../xea/x34'. */ |
19bc17a9 | 1416 | static void |
4b10dd6c UD |
1417 | charclass_charcode_ellipsis (struct linereader *ldfile, |
1418 | struct locale_ctype_t *ctype, | |
1419 | struct charmap_t *charmap, | |
1420 | struct repertoire_t *repertoire, | |
1421 | struct token *now, char *last_charcode, | |
1422 | uint32_t last_charcode_len, | |
1423 | unsigned long int class256_bit, | |
1424 | unsigned long int class_bit, int ignore_content, | |
1425 | int handle_digits) | |
19bc17a9 | 1426 | { |
4b10dd6c UD |
1427 | /* First check whether the to-value is larger. */ |
1428 | if (now->val.charcode.nbytes != last_charcode_len) | |
1429 | { | |
1430 | lr_error (ldfile, _("\ | |
1431 | start end end character sequence of range must have the same length")); | |
1432 | return; | |
1433 | } | |
19bc17a9 | 1434 | |
4b10dd6c | 1435 | if (memcmp (last_charcode, now->val.charcode.bytes, last_charcode_len) > 0) |
19bc17a9 | 1436 | { |
4b10dd6c UD |
1437 | lr_error (ldfile, _("\ |
1438 | to-value character sequence is smaller than from-value sequence")); | |
19bc17a9 RM |
1439 | return; |
1440 | } | |
1441 | ||
4b10dd6c UD |
1442 | if (!ignore_content) |
1443 | { | |
1444 | do | |
1445 | { | |
1446 | /* Increment the byte sequence value. */ | |
1447 | struct charseq *seq; | |
1448 | uint32_t wch; | |
1449 | int i; | |
1450 | ||
1451 | for (i = last_charcode_len - 1; i >= 0; --i) | |
1452 | if (++last_charcode[i] != 0) | |
1453 | break; | |
1454 | ||
1455 | if (last_charcode_len == 1) | |
1456 | /* Of course we have the charcode value. */ | |
1457 | ctype->class256_collection[(size_t) last_charcode[0]] | |
1458 | |= class256_bit; | |
1459 | ||
1460 | /* Find the symbolic name. */ | |
1461 | seq = charmap_find_symbol (charmap, last_charcode, | |
1462 | last_charcode_len); | |
1463 | if (seq != NULL) | |
1464 | { | |
1465 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1466 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1467 | strlen (seq->name)); | |
1468 | wch = seq->ucs4; | |
1469 | ||
1470 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) | |
1471 | *find_idx (ctype, &ctype->class_collection, | |
1472 | &ctype->class_collection_max, | |
1473 | &ctype->class_collection_act, wch) |= class_bit; | |
1474 | } | |
1475 | else | |
1476 | wch = ILLEGAL_CHAR_VALUE; | |
19bc17a9 | 1477 | |
4b10dd6c UD |
1478 | if (handle_digits == 1) |
1479 | { | |
1480 | /* We must store the digit values. */ | |
1481 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1482 | { | |
1483 | ctype->mbdigits_max *= 2; | |
1484 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1485 | (ctype->mbdigits_max | |
1486 | * sizeof (char *))); | |
1487 | ctype->wcdigits_max *= 2; | |
1488 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1489 | (ctype->wcdigits_max | |
1490 | * sizeof (uint32_t))); | |
1491 | } | |
1492 | ||
1493 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1494 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1495 | seq->nbytes = last_charcode_len; | |
1496 | ||
1497 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1498 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1499 | } | |
1500 | else if (handle_digits == 2) | |
1501 | { | |
1502 | struct charseq *seq; | |
1503 | /* We must store the digit values. */ | |
1504 | if (ctype->outdigits_act >= 10) | |
1505 | { | |
1506 | lr_error (ldfile, _("\ | |
1507 | %s: field `%s' does not contain exactly ten entries"), | |
1508 | "LC_CTYPE", "outdigit"); | |
1509 | return; | |
1510 | } | |
1511 | ||
1512 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1513 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1514 | seq->nbytes = last_charcode_len; | |
1515 | ||
1516 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1517 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1518 | ++ctype->outdigits_act; | |
1519 | } | |
1520 | } | |
1521 | while (memcmp (last_charcode, now->val.charcode.bytes, | |
1522 | last_charcode_len) != 0); | |
1523 | } | |
19bc17a9 RM |
1524 | } |
1525 | ||
1526 | ||
4b10dd6c UD |
1527 | /* Read one transliteration entry. */ |
1528 | static uint32_t * | |
1529 | read_widestring (struct linereader *ldfile, struct token *now, | |
1530 | struct charmap_t *charmap, struct repertoire_t *repertoire) | |
19bc17a9 | 1531 | { |
4b10dd6c | 1532 | uint32_t *wstr; |
19bc17a9 | 1533 | |
4b10dd6c UD |
1534 | if (now->tok == tok_default_missing) |
1535 | /* The special name "" will denote this case. */ | |
1536 | wstr = (uint32_t *) L""; | |
1537 | else if (now->tok == tok_bsymbol) | |
19bc17a9 | 1538 | { |
4b10dd6c UD |
1539 | /* Get the value from the repertoire. */ |
1540 | wstr = xmalloc (2 * sizeof (uint32_t)); | |
1541 | wstr[0] = repertoire_find_value (repertoire, now->val.str.startmb, | |
1542 | now->val.str.lenmb); | |
1543 | if (wstr[0] == ILLEGAL_CHAR_VALUE) | |
1544 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1545 | return NULL; | |
1546 | ||
1547 | wstr[1] = 0; | |
19bc17a9 | 1548 | } |
4b10dd6c | 1549 | else if (now->tok == tok_ucs4) |
19bc17a9 | 1550 | { |
4b10dd6c UD |
1551 | wstr = xmalloc (2 * sizeof (uint32_t)); |
1552 | wstr[0] = now->val.ucs4; | |
1553 | wstr[1] = 0; | |
1554 | } | |
1555 | else if (now->tok == tok_charcode) | |
1556 | { | |
1557 | /* Argh, we have to convert to the symbol name first and then to the | |
1558 | UCS4 value. */ | |
1559 | struct charseq *seq = charmap_find_symbol (charmap, | |
1560 | now->val.str.startmb, | |
1561 | now->val.str.lenmb); | |
1562 | if (seq == NULL) | |
1563 | /* Cannot find the UCS4 value. */ | |
1564 | return NULL; | |
1565 | ||
1566 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1567 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1568 | strlen (seq->name)); | |
1569 | if (seq->ucs4 == ILLEGAL_CHAR_VALUE) | |
1570 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1571 | return NULL; | |
1572 | ||
1573 | wstr = xmalloc (2 * sizeof (uint32_t)); | |
1574 | wstr[0] = seq->ucs4; | |
1575 | wstr[1] = 0; | |
1576 | } | |
1577 | else if (now->tok == tok_string) | |
1578 | { | |
1579 | wstr = now->val.str.startwc; | |
1580 | if (wstr[0] == 0) | |
1581 | return NULL; | |
1582 | } | |
1583 | else | |
1584 | { | |
1585 | if (now->tok != tok_eol && now->tok != tok_eof) | |
1586 | lr_ignore_rest (ldfile, 0); | |
1587 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
1588 | return (uint32_t *) -1l; | |
19bc17a9 RM |
1589 | } |
1590 | ||
4b10dd6c UD |
1591 | return wstr; |
1592 | } | |
19bc17a9 | 1593 | |
19bc17a9 | 1594 | |
4b10dd6c UD |
1595 | static void |
1596 | read_translit_entry (struct linereader *ldfile, struct locale_ctype_t *ctype, | |
1597 | struct token *now, struct charmap_t *charmap, | |
1598 | struct repertoire_t *repertoire) | |
1599 | { | |
1600 | uint32_t *from_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1601 | struct translit_t *result; | |
1602 | struct translit_to_t **top; | |
1603 | struct obstack *ob = &ctype->mem_pool; | |
1604 | int first; | |
1605 | int ignore; | |
1606 | ||
1607 | if (from_wstr == NULL) | |
1608 | /* There is no valid from string. */ | |
1609 | return; | |
19bc17a9 | 1610 | |
4b10dd6c UD |
1611 | result = (struct translit_t *) obstack_alloc (ob, |
1612 | sizeof (struct translit_t)); | |
1613 | result->from = from_wstr; | |
1614 | result->next = NULL; | |
1615 | result->to = NULL; | |
1616 | top = &result->to; | |
1617 | first = 1; | |
1618 | ignore = 0; | |
1619 | ||
1620 | while (1) | |
1621 | { | |
1622 | uint32_t *to_wstr; | |
1623 | ||
1624 | /* Next we have one or more transliterations. They are | |
1625 | separated by semicolons. */ | |
1626 | now = lr_token (ldfile, charmap, repertoire); | |
1627 | ||
1628 | if (!first && (now->tok == tok_semicolon || now->tok == tok_eol)) | |
1629 | { | |
1630 | /* One string read. */ | |
1631 | const uint32_t zero = 0; | |
1632 | ||
1633 | if (!ignore) | |
1634 | { | |
1635 | obstack_grow (ob, &zero, 4); | |
1636 | to_wstr = obstack_finish (ob); | |
1637 | ||
1638 | *top = obstack_alloc (ob, sizeof (struct translit_to_t)); | |
1639 | (*top)->str = to_wstr; | |
1640 | (*top)->next = NULL; | |
1641 | } | |
1642 | ||
1643 | if (now->tok == tok_eol) | |
1644 | { | |
1645 | result->next = ctype->translit; | |
1646 | ctype->translit = result; | |
1647 | return; | |
1648 | } | |
1649 | ||
1650 | if (!ignore) | |
1651 | top = &(*top)->next; | |
1652 | ignore = 0; | |
1653 | } | |
1654 | else | |
1655 | { | |
1656 | to_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1657 | if (to_wstr == (uint32_t *) -1l) | |
1658 | { | |
1659 | /* An error occurred. */ | |
1660 | obstack_free (ob, result); | |
1661 | return; | |
1662 | } | |
1663 | ||
1664 | if (to_wstr == NULL) | |
1665 | ignore = 1; | |
1666 | else | |
1667 | /* This value is usable. */ | |
1668 | obstack_grow (ob, to_wstr, wcslen ((wchar_t *) to_wstr) * 4); | |
19bc17a9 | 1669 | |
4b10dd6c UD |
1670 | first = 0; |
1671 | } | |
1672 | } | |
19bc17a9 RM |
1673 | } |
1674 | ||
1675 | ||
4b10dd6c UD |
1676 | /* The parser for the LC_CTYPE section of the locale definition. */ |
1677 | void | |
1678 | ctype_read (struct linereader *ldfile, struct localedef_t *result, | |
1679 | struct charmap_t *charmap, const char *repertoire_name, | |
1680 | int ignore_content) | |
19bc17a9 | 1681 | { |
4b10dd6c UD |
1682 | struct repertoire_t *repertoire = NULL; |
1683 | struct locale_ctype_t *ctype; | |
1684 | struct token *now; | |
1685 | enum token_t nowtok; | |
19bc17a9 | 1686 | size_t cnt; |
4b10dd6c UD |
1687 | struct charseq *last_seq; |
1688 | uint32_t last_wch = 0; | |
1689 | enum token_t last_token; | |
1690 | enum token_t ellipsis_token; | |
1691 | char last_charcode[16]; | |
1692 | size_t last_charcode_len = 0; | |
1693 | const char *last_str = NULL; | |
1694 | int mapidx; | |
19bc17a9 | 1695 | |
4b10dd6c UD |
1696 | /* Get the repertoire we have to use. */ |
1697 | if (repertoire_name != NULL) | |
1698 | repertoire = repertoire_read (repertoire_name); | |
19bc17a9 | 1699 | |
4b10dd6c UD |
1700 | /* The rest of the line containing `LC_CTYPE' must be free. */ |
1701 | lr_ignore_rest (ldfile, 1); | |
19bc17a9 | 1702 | |
4b10dd6c UD |
1703 | |
1704 | do | |
19bc17a9 | 1705 | { |
4b10dd6c UD |
1706 | now = lr_token (ldfile, charmap, NULL); |
1707 | nowtok = now->tok; | |
19bc17a9 | 1708 | } |
4b10dd6c | 1709 | while (nowtok == tok_eol); |
19bc17a9 | 1710 | |
4b10dd6c UD |
1711 | /* If we see `copy' now we are almost done. */ |
1712 | if (nowtok == tok_copy) | |
1713 | { | |
b9eb05d6 | 1714 | handle_copy (ldfile, charmap, repertoire, result, tok_lc_ctype, LC_CTYPE, |
4b10dd6c UD |
1715 | "LC_CTYPE", ignore_content); |
1716 | return; | |
1717 | } | |
75cd5204 | 1718 | |
4b10dd6c UD |
1719 | /* Prepare the data structures. */ |
1720 | ctype_startup (ldfile, result, charmap, ignore_content); | |
1721 | ctype = result->categories[LC_CTYPE].ctype; | |
1722 | ||
1723 | /* Remember the repertoire we use. */ | |
1724 | if (!ignore_content) | |
1725 | ctype->repertoire = repertoire; | |
1726 | ||
1727 | while (1) | |
19bc17a9 | 1728 | { |
4b10dd6c UD |
1729 | unsigned long int class_bit = 0; |
1730 | unsigned long int class256_bit = 0; | |
1731 | int handle_digits = 0; | |
1732 | ||
1733 | /* Of course we don't proceed beyond the end of file. */ | |
1734 | if (nowtok == tok_eof) | |
1735 | break; | |
1736 | ||
1737 | /* Ingore empty lines. */ | |
1738 | if (nowtok == tok_eol) | |
19bc17a9 | 1739 | { |
4b10dd6c UD |
1740 | now = lr_token (ldfile, charmap, NULL); |
1741 | nowtok = now->tok; | |
1742 | continue; | |
1743 | } | |
19bc17a9 | 1744 | |
4b10dd6c UD |
1745 | switch (nowtok) |
1746 | { | |
5491da0d UD |
1747 | case tok_charclass: |
1748 | now = lr_token (ldfile, charmap, NULL); | |
1749 | while (now->tok == tok_ident || now->tok == tok_string) | |
1750 | { | |
1751 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
1752 | now = lr_token (ldfile, charmap, NULL); | |
1753 | if (now->tok != tok_semicolon) | |
1754 | break; | |
1755 | now = lr_token (ldfile, charmap, NULL); | |
1756 | } | |
1757 | if (now->tok != tok_eol) | |
1758 | SYNTAX_ERROR (_("\ | |
1759 | %s: syntax error in definition of new character class"), "LC_CTYPE"); | |
1760 | break; | |
1761 | ||
1762 | case tok_charconv: | |
1763 | now = lr_token (ldfile, charmap, NULL); | |
1764 | while (now->tok == tok_ident || now->tok == tok_string) | |
1765 | { | |
1766 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
1767 | now = lr_token (ldfile, charmap, NULL); | |
1768 | if (now->tok != tok_semicolon) | |
1769 | break; | |
1770 | now = lr_token (ldfile, charmap, NULL); | |
1771 | } | |
1772 | if (now->tok != tok_eol) | |
1773 | SYNTAX_ERROR (_("\ | |
1774 | %s: syntax error in definition of new character map"), "LC_CTYPE"); | |
1775 | break; | |
1776 | ||
4b10dd6c | 1777 | case tok_class: |
b9eb05d6 UD |
1778 | /* Ignore the rest of the line if we don't need the input of |
1779 | this line. */ | |
1780 | if (ignore_content) | |
1781 | { | |
1782 | lr_ignore_rest (ldfile, 0); | |
1783 | break; | |
1784 | } | |
1785 | ||
4b10dd6c UD |
1786 | /* We simply forget the `class' keyword and use the following |
1787 | operand to determine the bit. */ | |
1788 | now = lr_token (ldfile, charmap, NULL); | |
1789 | if (now->tok == tok_ident || now->tok == tok_string) | |
1790 | { | |
87372aa9 | 1791 | /* Must can be one of the predefined class names. */ |
4b10dd6c UD |
1792 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1793 | if (strcmp (ctype->classnames[cnt], now->val.str.startmb) == 0) | |
1794 | break; | |
1795 | if (cnt >= ctype->nr_charclass) | |
1796 | { | |
011ebfab | 1797 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
1798 | if (now->val.str.lenmb == 8 |
1799 | && memcmp ("special1", now->val.str.startmb, 8) == 0) | |
1800 | class_bit = _ISwspecial1; | |
1801 | else if (now->val.str.lenmb == 8 | |
1802 | && memcmp ("special2", now->val.str.startmb, 8) == 0) | |
1803 | class_bit = _ISwspecial2; | |
1804 | else if (now->val.str.lenmb == 8 | |
1805 | && memcmp ("special3", now->val.str.startmb, 8) == 0) | |
1806 | class_bit = _ISwspecial3; | |
1807 | else | |
011ebfab | 1808 | #endif |
4b10dd6c | 1809 | { |
87372aa9 UD |
1810 | /* OK, it's a new class. */ |
1811 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
4b10dd6c | 1812 | |
87372aa9 | 1813 | class_bit = _ISwbit (ctype->nr_charclass - 1); |
4b10dd6c UD |
1814 | } |
1815 | } | |
1816 | else | |
7f653277 UD |
1817 | { |
1818 | class_bit = _ISwbit (cnt); | |
4b10dd6c | 1819 | |
7f653277 UD |
1820 | free (now->val.str.startmb); |
1821 | } | |
4b10dd6c UD |
1822 | } |
1823 | else if (now->tok == tok_digit) | |
1824 | goto handle_tok_digit; | |
1825 | else if (now->tok < tok_upper || now->tok > tok_blank) | |
1826 | goto err_label; | |
1827 | else | |
1828 | { | |
1829 | class_bit = BITw (now->tok); | |
1830 | class256_bit = BIT (now->tok); | |
1831 | } | |
1832 | ||
1833 | /* The next character must be a semicolon. */ | |
1834 | now = lr_token (ldfile, charmap, NULL); | |
1835 | if (now->tok != tok_semicolon) | |
1836 | goto err_label; | |
1837 | goto read_charclass; | |
1838 | ||
1839 | case tok_upper: | |
1840 | case tok_lower: | |
1841 | case tok_alpha: | |
1842 | case tok_alnum: | |
1843 | case tok_space: | |
1844 | case tok_cntrl: | |
1845 | case tok_punct: | |
1846 | case tok_graph: | |
1847 | case tok_print: | |
1848 | case tok_xdigit: | |
1849 | case tok_blank: | |
b9eb05d6 UD |
1850 | /* Ignore the rest of the line if we don't need the input of |
1851 | this line. */ | |
1852 | if (ignore_content) | |
1853 | { | |
1854 | lr_ignore_rest (ldfile, 0); | |
1855 | break; | |
1856 | } | |
1857 | ||
4b10dd6c UD |
1858 | class_bit = BITw (now->tok); |
1859 | class256_bit = BIT (now->tok); | |
1860 | handle_digits = 0; | |
1861 | read_charclass: | |
1862 | ctype->class_done |= class_bit; | |
1863 | last_token = tok_none; | |
1864 | ellipsis_token = tok_none; | |
1865 | now = lr_token (ldfile, charmap, NULL); | |
1866 | while (now->tok != tok_eol && now->tok != tok_eof) | |
1867 | { | |
1868 | uint32_t wch; | |
1869 | struct charseq *seq; | |
1870 | ||
1871 | if (ellipsis_token == tok_none) | |
1872 | { | |
1873 | if (get_character (now, charmap, repertoire, &seq, &wch)) | |
1874 | goto err_label; | |
1875 | ||
1876 | if (!ignore_content && seq != NULL && seq->nbytes == 1) | |
1877 | /* Yep, we can store information about this byte | |
1878 | sequence. */ | |
1879 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
1880 | ||
1881 | if (!ignore_content && wch != ILLEGAL_CHAR_VALUE | |
1882 | && class_bit != 0) | |
1883 | /* We have the UCS4 position. */ | |
1884 | *find_idx (ctype, &ctype->class_collection, | |
1885 | &ctype->class_collection_max, | |
1886 | &ctype->class_collection_act, wch) |= class_bit; | |
1887 | ||
1888 | last_token = now->tok; | |
549b3c3a | 1889 | /* Terminate the string. */ |
9e2b7438 UD |
1890 | if (last_token == tok_bsymbol) |
1891 | { | |
1892 | now->val.str.startmb[now->val.str.lenmb] = '\0'; | |
1893 | last_str = now->val.str.startmb; | |
1894 | } | |
1895 | else | |
1896 | last_str = NULL; | |
4b10dd6c UD |
1897 | last_seq = seq; |
1898 | last_wch = wch; | |
1899 | memcpy (last_charcode, now->val.charcode.bytes, 16); | |
1900 | last_charcode_len = now->val.charcode.nbytes; | |
1901 | ||
1902 | if (!ignore_content && handle_digits == 1) | |
1903 | { | |
1904 | /* We must store the digit values. */ | |
1905 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1906 | { | |
b9eb05d6 | 1907 | ctype->mbdigits_max += 10; |
4b10dd6c UD |
1908 | ctype->mbdigits = xrealloc (ctype->mbdigits, |
1909 | (ctype->mbdigits_max | |
1910 | * sizeof (char *))); | |
b9eb05d6 | 1911 | ctype->wcdigits_max += 10; |
4b10dd6c UD |
1912 | ctype->wcdigits = xrealloc (ctype->wcdigits, |
1913 | (ctype->wcdigits_max | |
1914 | * sizeof (uint32_t))); | |
1915 | } | |
1916 | ||
1917 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1918 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1919 | } | |
1920 | else if (!ignore_content && handle_digits == 2) | |
1921 | { | |
1922 | /* We must store the digit values. */ | |
1923 | if (ctype->outdigits_act >= 10) | |
1924 | { | |
1925 | lr_error (ldfile, _("\ | |
1926 | %s: field `%s' does not contain exactly ten entries"), | |
1927 | "LC_CTYPE", "outdigit"); | |
1928 | goto err_label; | |
1929 | } | |
1930 | ||
1931 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1932 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1933 | ++ctype->outdigits_act; | |
1934 | } | |
1935 | } | |
1936 | else | |
1937 | { | |
1938 | /* Now it gets complicated. We have to resolve the | |
1939 | ellipsis problem. First we must distinguish between | |
1940 | the different kind of ellipsis and this must match the | |
1941 | tokens we have seen. */ | |
1942 | assert (last_token != tok_none); | |
1943 | ||
1944 | if (last_token != now->tok) | |
1945 | { | |
1946 | lr_error (ldfile, _("\ | |
1947 | ellipsis range must be marked by two operands of same type")); | |
1948 | lr_ignore_rest (ldfile, 0); | |
1949 | break; | |
1950 | } | |
1951 | ||
1952 | if (last_token == tok_bsymbol) | |
1953 | { | |
1954 | if (ellipsis_token == tok_ellipsis3) | |
1955 | lr_error (ldfile, _("with symbolic name range values \ | |
1956 | the absolute ellipsis `...' must not be used")); | |
1957 | ||
1958 | charclass_symbolic_ellipsis (ldfile, ctype, charmap, | |
1959 | repertoire, now, last_str, | |
1960 | class256_bit, class_bit, | |
1961 | (ellipsis_token | |
1962 | == tok_ellipsis4 | |
1963 | ? 10 : 16), | |
1964 | ignore_content, | |
1965 | handle_digits); | |
1966 | } | |
1967 | else if (last_token == tok_ucs4) | |
1968 | { | |
1969 | if (ellipsis_token != tok_ellipsis2) | |
1970 | lr_error (ldfile, _("\ | |
1971 | with UCS range values one must use the hexadecimal symbolic ellipsis `..'")); | |
1972 | ||
1973 | charclass_ucs4_ellipsis (ldfile, ctype, charmap, | |
1974 | repertoire, now, last_wch, | |
1975 | class256_bit, class_bit, | |
1976 | ignore_content, handle_digits); | |
1977 | } | |
1978 | else | |
1979 | { | |
1980 | assert (last_token == tok_charcode); | |
1981 | ||
1982 | if (ellipsis_token != tok_ellipsis3) | |
1983 | lr_error (ldfile, _("\ | |
1984 | with character code range values one must use the absolute ellipsis `...'")); | |
1985 | ||
1986 | charclass_charcode_ellipsis (ldfile, ctype, charmap, | |
1987 | repertoire, now, | |
1988 | last_charcode, | |
1989 | last_charcode_len, | |
1990 | class256_bit, class_bit, | |
1991 | ignore_content, | |
1992 | handle_digits); | |
1993 | } | |
1994 | ||
1995 | /* Now we have used the last value. */ | |
1996 | last_token = tok_none; | |
1997 | } | |
1998 | ||
1999 | /* Next we expect a semicolon or the end of the line. */ | |
2000 | now = lr_token (ldfile, charmap, NULL); | |
2001 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2002 | break; | |
2003 | ||
2004 | if (last_token != tok_none | |
2005 | && now->tok >= tok_ellipsis2 && now->tok <= tok_ellipsis4) | |
2006 | { | |
2007 | ellipsis_token = now->tok; | |
2008 | now = lr_token (ldfile, charmap, NULL); | |
2009 | continue; | |
2010 | } | |
2011 | ||
2012 | if (now->tok != tok_semicolon) | |
2013 | goto err_label; | |
2014 | ||
2015 | /* And get the next character. */ | |
2016 | now = lr_token (ldfile, charmap, NULL); | |
2017 | ||
2018 | ellipsis_token = tok_none; | |
2019 | } | |
2020 | break; | |
2021 | ||
2022 | case tok_digit: | |
b9eb05d6 UD |
2023 | /* Ignore the rest of the line if we don't need the input of |
2024 | this line. */ | |
2025 | if (ignore_content) | |
42d7c593 UD |
2026 | { |
2027 | lr_ignore_rest (ldfile, 0); | |
2028 | break; | |
2029 | } | |
b9eb05d6 | 2030 | |
4b10dd6c UD |
2031 | handle_tok_digit: |
2032 | class_bit = _ISwdigit; | |
2033 | class256_bit = _ISdigit; | |
2034 | handle_digits = 1; | |
2035 | goto read_charclass; | |
2036 | ||
2037 | case tok_outdigit: | |
b9eb05d6 UD |
2038 | /* Ignore the rest of the line if we don't need the input of |
2039 | this line. */ | |
2040 | if (ignore_content) | |
2041 | { | |
2042 | lr_ignore_rest (ldfile, 0); | |
2043 | break; | |
2044 | } | |
2045 | ||
4b10dd6c UD |
2046 | if (ctype->outdigits_act != 0) |
2047 | lr_error (ldfile, _("\ | |
2048 | %s: field `%s' declared more than once"), | |
2049 | "LC_CTYPE", "outdigit"); | |
2050 | class_bit = 0; | |
2051 | class256_bit = 0; | |
2052 | handle_digits = 2; | |
2053 | goto read_charclass; | |
2054 | ||
2055 | case tok_toupper: | |
b9eb05d6 UD |
2056 | /* Ignore the rest of the line if we don't need the input of |
2057 | this line. */ | |
2058 | if (ignore_content) | |
2059 | { | |
2060 | lr_ignore_rest (ldfile, 0); | |
2061 | break; | |
2062 | } | |
2063 | ||
4b10dd6c UD |
2064 | mapidx = 0; |
2065 | goto read_mapping; | |
2066 | ||
2067 | case tok_tolower: | |
b9eb05d6 UD |
2068 | /* Ignore the rest of the line if we don't need the input of |
2069 | this line. */ | |
2070 | if (ignore_content) | |
2071 | { | |
2072 | lr_ignore_rest (ldfile, 0); | |
2073 | break; | |
2074 | } | |
2075 | ||
4b10dd6c UD |
2076 | mapidx = 1; |
2077 | goto read_mapping; | |
2078 | ||
2079 | case tok_map: | |
b9eb05d6 UD |
2080 | /* Ignore the rest of the line if we don't need the input of |
2081 | this line. */ | |
2082 | if (ignore_content) | |
2083 | { | |
2084 | lr_ignore_rest (ldfile, 0); | |
2085 | break; | |
2086 | } | |
2087 | ||
4b10dd6c UD |
2088 | /* We simply forget the `map' keyword and use the following |
2089 | operand to determine the mapping. */ | |
2090 | now = lr_token (ldfile, charmap, NULL); | |
2091 | if (now->tok == tok_ident || now->tok == tok_string) | |
2092 | { | |
2093 | size_t cnt; | |
2094 | ||
2095 | for (cnt = 2; cnt < ctype->map_collection_nr; ++cnt) | |
2096 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2097 | break; | |
2098 | ||
7f653277 UD |
2099 | if (cnt < ctype->map_collection_nr) |
2100 | free (now->val.str.startmb); | |
2101 | else | |
87372aa9 UD |
2102 | /* OK, it's a new map. */ |
2103 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2104 | ||
2105 | mapidx = cnt; | |
4b10dd6c UD |
2106 | } |
2107 | else if (now->tok < tok_toupper || now->tok > tok_tolower) | |
2108 | goto err_label; | |
2109 | else | |
2110 | mapidx = now->tok - tok_toupper; | |
2111 | ||
2112 | now = lr_token (ldfile, charmap, NULL); | |
2113 | /* This better should be a semicolon. */ | |
2114 | if (now->tok != tok_semicolon) | |
2115 | goto err_label; | |
2116 | ||
2117 | read_mapping: | |
2118 | /* Test whether this mapping was already defined. */ | |
2119 | if (ctype->tomap_done[mapidx]) | |
2120 | { | |
2121 | lr_error (ldfile, _("duplicated definition for mapping `%s'"), | |
2122 | ctype->mapnames[mapidx]); | |
2123 | lr_ignore_rest (ldfile, 0); | |
2124 | break; | |
2125 | } | |
2126 | ctype->tomap_done[mapidx] = 1; | |
2127 | ||
2128 | now = lr_token (ldfile, charmap, NULL); | |
2129 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2130 | { | |
2131 | struct charseq *from_seq; | |
2132 | uint32_t from_wch; | |
2133 | struct charseq *to_seq; | |
2134 | uint32_t to_wch; | |
2135 | ||
2136 | /* Every pair starts with an opening brace. */ | |
2137 | if (now->tok != tok_open_brace) | |
2138 | goto err_label; | |
2139 | ||
2140 | /* Next comes the from-value. */ | |
2141 | now = lr_token (ldfile, charmap, NULL); | |
2142 | if (get_character (now, charmap, repertoire, &from_seq, | |
2143 | &from_wch) != 0) | |
2144 | goto err_label; | |
2145 | ||
2146 | /* The next is a comma. */ | |
2147 | now = lr_token (ldfile, charmap, NULL); | |
2148 | if (now->tok != tok_comma) | |
2149 | goto err_label; | |
2150 | ||
2151 | /* And the other value. */ | |
2152 | now = lr_token (ldfile, charmap, NULL); | |
2153 | if (get_character (now, charmap, repertoire, &to_seq, | |
2154 | &to_wch) != 0) | |
2155 | goto err_label; | |
2156 | ||
2157 | /* And the last thing is the closing brace. */ | |
2158 | now = lr_token (ldfile, charmap, NULL); | |
2159 | if (now->tok != tok_close_brace) | |
2160 | goto err_label; | |
2161 | ||
2162 | if (!ignore_content) | |
2163 | { | |
2164 | if (mapidx < 2 && from_seq != NULL && to_seq != NULL | |
2165 | && from_seq->nbytes == 1 && to_seq->nbytes == 1) | |
2166 | /* We can use this value. */ | |
2167 | ctype->map256_collection[mapidx][from_seq->bytes[0]] | |
2168 | = to_seq->bytes[0]; | |
2169 | ||
2170 | if (from_wch != ILLEGAL_CHAR_VALUE | |
2171 | && to_wch != ILLEGAL_CHAR_VALUE) | |
2172 | /* Both correct values. */ | |
2173 | *find_idx (ctype, &ctype->map_collection[mapidx], | |
2174 | &ctype->map_collection_max[mapidx], | |
2175 | &ctype->map_collection_act[mapidx], | |
2176 | from_wch) = to_wch; | |
2177 | } | |
2178 | ||
2179 | /* Now comes a semicolon or the end of the line/file. */ | |
2180 | now = lr_token (ldfile, charmap, NULL); | |
2181 | if (now->tok == tok_semicolon) | |
2182 | now = lr_token (ldfile, charmap, NULL); | |
2183 | } | |
2184 | break; | |
2185 | ||
2186 | case tok_translit_start: | |
b9eb05d6 UD |
2187 | /* Ignore the rest of the line if we don't need the input of |
2188 | this line. */ | |
2189 | if (ignore_content) | |
2190 | { | |
2191 | lr_ignore_rest (ldfile, 0); | |
2192 | break; | |
2193 | } | |
2194 | ||
4b10dd6c UD |
2195 | /* The rest of the line better should be empty. */ |
2196 | lr_ignore_rest (ldfile, 1); | |
2197 | ||
2198 | /* We count here the number of allocated entries in the `translit' | |
2199 | array. */ | |
2200 | cnt = 0; | |
2201 | ||
2202 | /* We proceed until we see the `translit_end' token. */ | |
2203 | while (now = lr_token (ldfile, charmap, repertoire), | |
2204 | now->tok != tok_translit_end && now->tok != tok_eof) | |
2205 | { | |
2206 | if (now->tok == tok_eol) | |
2207 | /* Ignore empty lines. */ | |
2208 | continue; | |
2209 | ||
2210 | if (now->tok == tok_translit_end) | |
2211 | { | |
2212 | lr_ignore_rest (ldfile, 0); | |
2213 | break; | |
2214 | } | |
2215 | ||
2216 | if (now->tok == tok_include) | |
2217 | { | |
2218 | /* We have to include locale. */ | |
2219 | const char *locale_name; | |
2220 | const char *repertoire_name; | |
2221 | ||
2222 | now = lr_token (ldfile, charmap, NULL); | |
2223 | /* This should be a string or an identifier. In any | |
2224 | case something to name a locale. */ | |
2225 | if (now->tok != tok_string && now->tok != tok_ident) | |
2226 | { | |
2227 | translit_syntax: | |
2228 | lr_error (ldfile, _("%s: syntax error"), "LC_CTYPE"); | |
2229 | lr_ignore_rest (ldfile, 0); | |
2230 | continue; | |
2231 | } | |
2232 | locale_name = now->val.str.startmb; | |
2233 | ||
2234 | /* Next should be a semicolon. */ | |
2235 | now = lr_token (ldfile, charmap, NULL); | |
2236 | if (now->tok != tok_semicolon) | |
2237 | goto translit_syntax; | |
2238 | ||
2239 | /* Now the repertoire name. */ | |
2240 | now = lr_token (ldfile, charmap, NULL); | |
2241 | if ((now->tok != tok_string && now->tok != tok_ident) | |
2242 | || now->val.str.startmb == NULL) | |
2243 | goto translit_syntax; | |
2244 | repertoire_name = now->val.str.startmb; | |
2245 | ||
2246 | /* We must not have more than one `include'. */ | |
2247 | if (ctype->translit_copy_locale != NULL) | |
2248 | { | |
2249 | lr_error (ldfile, _("\ | |
2250 | %s: only one `include' instruction allowed"), "LC_CTYPE"); | |
2251 | lr_ignore_rest (ldfile, 0); | |
2252 | continue; | |
2253 | } | |
2254 | ||
2255 | ctype->translit_copy_locale = locale_name; | |
2256 | ctype->translit_copy_repertoire = repertoire_name; | |
2257 | ||
2258 | /* The rest of the line must be empty. */ | |
2259 | lr_ignore_rest (ldfile, 1); | |
2260 | continue; | |
2261 | } | |
2262 | ||
2263 | read_translit_entry (ldfile, ctype, now, charmap, repertoire); | |
2264 | } | |
2265 | break; | |
2266 | ||
2267 | case tok_ident: | |
b9eb05d6 UD |
2268 | /* Ignore the rest of the line if we don't need the input of |
2269 | this line. */ | |
2270 | if (ignore_content) | |
2271 | { | |
2272 | lr_ignore_rest (ldfile, 0); | |
2273 | break; | |
2274 | } | |
2275 | ||
4b10dd6c UD |
2276 | /* This could mean one of several things. First test whether |
2277 | it's a character class name. */ | |
2278 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
2279 | if (strcmp (now->val.str.startmb, ctype->classnames[cnt]) == 0) | |
2280 | break; | |
2281 | if (cnt < ctype->nr_charclass) | |
2282 | { | |
2283 | class_bit = _ISwbit (cnt); | |
2284 | class256_bit = cnt <= 11 ? _ISbit (cnt) : 0; | |
2285 | free (now->val.str.startmb); | |
2286 | goto read_charclass; | |
2287 | } | |
5491da0d UD |
2288 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
2289 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2290 | break; | |
2291 | if (cnt < ctype->map_collection_nr) | |
2292 | { | |
2293 | mapidx = cnt; | |
2294 | free (now->val.str.startmb); | |
2295 | goto read_mapping; | |
2296 | } | |
011ebfab | 2297 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2298 | if (strcmp (now->val.str.startmb, "special1") == 0) |
2299 | { | |
2300 | class_bit = _ISwspecial1; | |
2301 | free (now->val.str.startmb); | |
2302 | goto read_charclass; | |
2303 | } | |
2304 | if (strcmp (now->val.str.startmb, "special2") == 0) | |
2305 | { | |
2306 | class_bit = _ISwspecial2; | |
2307 | free (now->val.str.startmb); | |
2308 | goto read_charclass; | |
2309 | } | |
2310 | if (strcmp (now->val.str.startmb, "special3") == 0) | |
2311 | { | |
2312 | class_bit = _ISwspecial3; | |
2313 | free (now->val.str.startmb); | |
2314 | goto read_charclass; | |
2315 | } | |
2316 | if (strcmp (now->val.str.startmb, "tosymmetric") == 0) | |
2317 | { | |
2318 | mapidx = 2; | |
2319 | goto read_mapping; | |
2320 | } | |
011ebfab | 2321 | #endif |
4b10dd6c UD |
2322 | break; |
2323 | ||
2324 | case tok_end: | |
2325 | /* Next we assume `LC_CTYPE'. */ | |
2326 | now = lr_token (ldfile, charmap, NULL); | |
2327 | if (now->tok == tok_eof) | |
2328 | break; | |
2329 | if (now->tok == tok_eol) | |
2330 | lr_error (ldfile, _("%s: incomplete `END' line"), | |
2331 | "LC_CTYPE"); | |
2332 | else if (now->tok != tok_lc_ctype) | |
2333 | lr_error (ldfile, _("\ | |
2334 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2335 | lr_ignore_rest (ldfile, now->tok == tok_lc_ctype); | |
2336 | return; | |
2337 | ||
2338 | default: | |
2339 | err_label: | |
2340 | if (now->tok != tok_eof) | |
2341 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
19bc17a9 RM |
2342 | } |
2343 | ||
4b10dd6c UD |
2344 | /* Prepare for the next round. */ |
2345 | now = lr_token (ldfile, charmap, NULL); | |
2346 | nowtok = now->tok; | |
19bc17a9 RM |
2347 | } |
2348 | ||
4b10dd6c UD |
2349 | /* When we come here we reached the end of the file. */ |
2350 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); | |
19bc17a9 RM |
2351 | } |
2352 | ||
2353 | ||
2354 | static void | |
4b10dd6c UD |
2355 | set_class_defaults (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2356 | struct repertoire_t *repertoire) | |
19bc17a9 | 2357 | { |
4b10dd6c UD |
2358 | size_t cnt; |
2359 | ||
19bc17a9 RM |
2360 | /* These function defines the default values for the classes and conversions |
2361 | according to POSIX.2 2.5.2.1. | |
2362 | It may seem that the order of these if-blocks is arbitrary but it is NOT. | |
2363 | Don't move them unless you know what you do! */ | |
2364 | ||
4b10dd6c | 2365 | void set_default (int bitpos, int from, int to) |
19bc17a9 RM |
2366 | { |
2367 | char tmp[2]; | |
2368 | int ch; | |
4b10dd6c UD |
2369 | int bit = _ISbit (bitpos); |
2370 | int bitw = _ISwbit (bitpos); | |
19bc17a9 RM |
2371 | /* Define string. */ |
2372 | strcpy (tmp, "?"); | |
2373 | ||
2374 | for (ch = from; ch <= to; ++ch) | |
2375 | { | |
4b10dd6c UD |
2376 | uint32_t value; |
2377 | struct charseq *seq; | |
19bc17a9 RM |
2378 | tmp[0] = ch; |
2379 | ||
4b10dd6c UD |
2380 | value = repertoire_find_value (repertoire, tmp, 1); |
2381 | if (value == ILLEGAL_CHAR_VALUE) | |
19bc17a9 | 2382 | { |
880f421f UD |
2383 | if (!be_quiet) |
2384 | error (0, 0, _("\ | |
4b10dd6c UD |
2385 | %s: character `%s' not defined in repertoire while needed as default value"), |
2386 | "LC_CTYPE", tmp); | |
2387 | } | |
2388 | else | |
2389 | ELEM (ctype, class_collection, , value) |= bitw; | |
2390 | ||
2391 | seq = charmap_find_value (charmap, tmp, 1); | |
2392 | if (seq == NULL) | |
2393 | { | |
2394 | if (!be_quiet) | |
2395 | error (0, 0, _("\ | |
2396 | %s: character `%s' not defined in charmap while needed as default value"), | |
2397 | "LC_CTYPE", tmp); | |
19bc17a9 | 2398 | } |
4b10dd6c UD |
2399 | else if (seq->nbytes != 1) |
2400 | error (0, 0, _("\ | |
2401 | %s: character `%s' in charmap not representable with one byte"), | |
2402 | "LC_CTYPE", tmp); | |
19bc17a9 | 2403 | else |
4b10dd6c | 2404 | ctype->class256_collection[seq->bytes[0]] |= bit; |
19bc17a9 RM |
2405 | } |
2406 | } | |
2407 | ||
2408 | /* Set default values if keyword was not present. */ | |
4b10dd6c | 2409 | if ((ctype->class_done & BITw (tok_upper)) == 0) |
19bc17a9 RM |
2410 | /* "If this keyword [lower] is not specified, the lowercase letters |
2411 | `A' through `Z', ..., shall automatically belong to this class, | |
2412 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2413 | set_default (BITPOS (tok_upper), 'A', 'Z'); |
19bc17a9 | 2414 | |
4b10dd6c | 2415 | if ((ctype->class_done & BITw (tok_lower)) == 0) |
19bc17a9 RM |
2416 | /* "If this keyword [lower] is not specified, the lowercase letters |
2417 | `a' through `z', ..., shall automatically belong to this class, | |
2418 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2419 | set_default (BITPOS (tok_lower), 'a', 'z'); |
19bc17a9 | 2420 | |
4b10dd6c | 2421 | if ((ctype->class_done & BITw (tok_alpha)) == 0) |
19bc17a9 RM |
2422 | { |
2423 | /* Table 2-6 in P1003.2 says that characters in class `upper' or | |
2424 | class `lower' *must* be in class `alpha'. */ | |
2425 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower); | |
96f0d1f5 UD |
2426 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower); |
2427 | ||
2428 | for (cnt = 0; cnt < 256; ++cnt) | |
2429 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2430 | ctype->class256_collection[cnt] |= BIT (tok_alpha); | |
19bc17a9 RM |
2431 | |
2432 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2433 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2434 | ctype->class_collection[cnt] |= BITw (tok_alpha); | |
19bc17a9 RM |
2435 | } |
2436 | ||
4b10dd6c | 2437 | if ((ctype->class_done & BITw (tok_digit)) == 0) |
19bc17a9 RM |
2438 | /* "If this keyword [digit] is not specified, the digits `0' through |
2439 | `9', ..., shall automatically belong to this class, with | |
2440 | implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2441 | set_default (BITPOS (tok_digit), '0', '9'); |
19bc17a9 RM |
2442 | |
2443 | /* "Only characters specified for the `alpha' and `digit' keyword | |
2444 | shall be specified. Characters specified for the keyword `alpha' | |
2445 | and `digit' are automatically included in this class. */ | |
2446 | { | |
2447 | unsigned long int mask = BIT (tok_alpha) | BIT (tok_digit); | |
96f0d1f5 UD |
2448 | unsigned long int maskw = BITw (tok_alpha) | BITw (tok_digit); |
2449 | ||
2450 | for (cnt = 0; cnt < 256; ++cnt) | |
2451 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2452 | ctype->class256_collection[cnt] |= BIT (tok_alnum); | |
19bc17a9 RM |
2453 | |
2454 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2455 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2456 | ctype->class_collection[cnt] |= BITw (tok_alnum); | |
19bc17a9 RM |
2457 | } |
2458 | ||
4b10dd6c | 2459 | if ((ctype->class_done & BITw (tok_space)) == 0) |
19bc17a9 RM |
2460 | /* "If this keyword [space] is not specified, the characters <space>, |
2461 | <form-feed>, <newline>, <carriage-return>, <tab>, and | |
2462 | <vertical-tab>, ..., shall automatically belong to this class, | |
2463 | with implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
2464 | { | |
4b10dd6c UD |
2465 | uint32_t value; |
2466 | struct charseq *seq; | |
19bc17a9 | 2467 | |
4b10dd6c UD |
2468 | value = repertoire_find_value (repertoire, "space", 5); |
2469 | if (value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
2470 | { |
2471 | if (!be_quiet) | |
2472 | error (0, 0, _("\ | |
4b10dd6c UD |
2473 | %s: character `%s' not defined while needed as default value"), |
2474 | "LC_CTYPE", "<space>"); | |
880f421f | 2475 | } |
19bc17a9 RM |
2476 | else |
2477 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2478 | ||
4b10dd6c UD |
2479 | seq = charmap_find_value (charmap, "space", 5); |
2480 | if (seq == NULL) | |
880f421f UD |
2481 | { |
2482 | if (!be_quiet) | |
2483 | error (0, 0, _("\ | |
4b10dd6c UD |
2484 | %s: character `%s' not defined while needed as default value"), |
2485 | "LC_CTYPE", "<space>"); | |
2486 | } | |
2487 | else if (seq->nbytes != 1) | |
2488 | error (0, 0, _("\ | |
2489 | %s: character `%s' in charmap not representable with one byte"), | |
2490 | "LC_CTYPE", "<space>"); | |
2491 | else | |
2492 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2493 | ||
2494 | ||
2495 | value = repertoire_find_value (repertoire, "form-feed", 9); | |
2496 | if (value == ILLEGAL_CHAR_VALUE) | |
2497 | { | |
2498 | if (!be_quiet) | |
2499 | error (0, 0, _("\ | |
2500 | %s: character `%s' not defined while needed as default value"), | |
2501 | "LC_CTYPE", "<form-feed>"); | |
880f421f | 2502 | } |
19bc17a9 RM |
2503 | else |
2504 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2505 | ||
4b10dd6c UD |
2506 | seq = charmap_find_value (charmap, "form-feed", 9); |
2507 | if (seq == NULL) | |
880f421f UD |
2508 | { |
2509 | if (!be_quiet) | |
2510 | error (0, 0, _("\ | |
4b10dd6c UD |
2511 | %s: character `%s' not defined while needed as default value"), |
2512 | "LC_CTYPE", "<form-feed>"); | |
2513 | } | |
2514 | else if (seq->nbytes != 1) | |
2515 | error (0, 0, _("\ | |
2516 | %s: character `%s' in charmap not representable with one byte"), | |
2517 | "LC_CTYPE", "<form-feed>"); | |
2518 | else | |
2519 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2520 | ||
2521 | ||
2522 | value = repertoire_find_value (repertoire, "newline", 7); | |
2523 | if (value == ILLEGAL_CHAR_VALUE) | |
2524 | { | |
2525 | if (!be_quiet) | |
2526 | error (0, 0, _("\ | |
2527 | %s: character `%s' not defined while needed as default value"), | |
2528 | "LC_CTYPE", "<newline>"); | |
880f421f | 2529 | } |
19bc17a9 RM |
2530 | else |
2531 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2532 | ||
4b10dd6c UD |
2533 | seq = charmap_find_value (charmap, "newline", 7); |
2534 | if (seq == NULL) | |
880f421f UD |
2535 | { |
2536 | if (!be_quiet) | |
2537 | error (0, 0, _("\ | |
19bc17a9 | 2538 | character `%s' not defined while needed as default value"), |
4b10dd6c UD |
2539 | "<newline>"); |
2540 | } | |
2541 | else if (seq->nbytes != 1) | |
2542 | error (0, 0, _("\ | |
2543 | %s: character `%s' in charmap not representable with one byte"), | |
2544 | "LC_CTYPE", "<newline>"); | |
2545 | else | |
2546 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2547 | ||
2548 | ||
2549 | value = repertoire_find_value (repertoire, "carriage-return", 15); | |
2550 | if (value == ILLEGAL_CHAR_VALUE) | |
2551 | { | |
2552 | if (!be_quiet) | |
2553 | error (0, 0, _("\ | |
2554 | %s: character `%s' not defined while needed as default value"), | |
2555 | "LC_CTYPE", "<carriage-return>"); | |
880f421f | 2556 | } |
19bc17a9 RM |
2557 | else |
2558 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2559 | ||
4b10dd6c UD |
2560 | seq = charmap_find_value (charmap, "carriage-return", 15); |
2561 | if (seq == NULL) | |
880f421f UD |
2562 | { |
2563 | if (!be_quiet) | |
2564 | error (0, 0, _("\ | |
4b10dd6c UD |
2565 | %s: character `%s' not defined while needed as default value"), |
2566 | "LC_CTYPE", "<carriage-return>"); | |
2567 | } | |
2568 | else if (seq->nbytes != 1) | |
2569 | error (0, 0, _("\ | |
2570 | %s: character `%s' in charmap not representable with one byte"), | |
2571 | "LC_CTYPE", "<carriage-return>"); | |
2572 | else | |
2573 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2574 | ||
2575 | ||
2576 | value = repertoire_find_value (repertoire, "tab", 3); | |
2577 | if (value == ILLEGAL_CHAR_VALUE) | |
2578 | { | |
2579 | if (!be_quiet) | |
2580 | error (0, 0, _("\ | |
2581 | %s: character `%s' not defined while needed as default value"), | |
2582 | "LC_CTYPE", "<tab>"); | |
880f421f | 2583 | } |
19bc17a9 RM |
2584 | else |
2585 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2586 | ||
4b10dd6c UD |
2587 | seq = charmap_find_value (charmap, "tab", 3); |
2588 | if (seq == NULL) | |
880f421f UD |
2589 | { |
2590 | if (!be_quiet) | |
2591 | error (0, 0, _("\ | |
4b10dd6c UD |
2592 | %s: character `%s' not defined while needed as default value"), |
2593 | "LC_CTYPE", "<tab>"); | |
2594 | } | |
2595 | else if (seq->nbytes != 1) | |
2596 | error (0, 0, _("\ | |
2597 | %s: character `%s' in charmap not representable with one byte"), | |
2598 | "LC_CTYPE", "<tab>"); | |
2599 | else | |
2600 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2601 | ||
2602 | ||
2603 | value = repertoire_find_value (repertoire, "vertical-tab", 12); | |
2604 | if (value == ILLEGAL_CHAR_VALUE) | |
2605 | { | |
2606 | if (!be_quiet) | |
2607 | error (0, 0, _("\ | |
2608 | %s: character `%s' not defined while needed as default value"), | |
2609 | "LC_CTYPE", "<vertical-tab>"); | |
880f421f | 2610 | } |
19bc17a9 RM |
2611 | else |
2612 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
4b10dd6c UD |
2613 | |
2614 | seq = charmap_find_value (charmap, "vertical-tab", 12); | |
2615 | if (seq == NULL) | |
2616 | { | |
2617 | if (!be_quiet) | |
2618 | error (0, 0, _("\ | |
2619 | %s: character `%s' not defined while needed as default value"), | |
2620 | "LC_CTYPE", "<vertical-tab>"); | |
2621 | } | |
2622 | else if (seq->nbytes != 1) | |
2623 | error (0, 0, _("\ | |
2624 | %s: character `%s' in charmap not representable with one byte"), | |
2625 | "LC_CTYPE", "<vertical-tab>"); | |
2626 | else | |
2627 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
19bc17a9 RM |
2628 | } |
2629 | ||
4b10dd6c | 2630 | if ((ctype->class_done & BITw (tok_xdigit)) == 0) |
19bc17a9 RM |
2631 | /* "If this keyword is not specified, the digits `0' to `9', the |
2632 | uppercase letters `A' through `F', and the lowercase letters `a' | |
2633 | through `f', ..., shell automatically belong to this class, with | |
2634 | implementation defined character values." [P1003.2, 2.5.2.1] */ | |
2635 | { | |
4b10dd6c UD |
2636 | set_default (BITPOS (tok_xdigit), '0', '9'); |
2637 | set_default (BITPOS (tok_xdigit), 'A', 'F'); | |
2638 | set_default (BITPOS (tok_xdigit), 'a', 'f'); | |
19bc17a9 RM |
2639 | } |
2640 | ||
4b10dd6c | 2641 | if ((ctype->class_done & BITw (tok_blank)) == 0) |
19bc17a9 RM |
2642 | /* "If this keyword [blank] is unspecified, the characters <space> and |
2643 | <tab> shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
2644 | { | |
4b10dd6c UD |
2645 | uint32_t value; |
2646 | struct charseq *seq; | |
19bc17a9 | 2647 | |
4b10dd6c UD |
2648 | value = repertoire_find_value (repertoire, "space", 5); |
2649 | if (value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
2650 | { |
2651 | if (!be_quiet) | |
2652 | error (0, 0, _("\ | |
4b10dd6c UD |
2653 | %s: character `%s' not defined while needed as default value"), |
2654 | "LC_CTYPE", "<space>"); | |
880f421f | 2655 | } |
19bc17a9 RM |
2656 | else |
2657 | ELEM (ctype, class_collection, , value) |= BIT (tok_blank); | |
2658 | ||
4b10dd6c UD |
2659 | seq = charmap_find_value (charmap, "space", 5); |
2660 | if (seq == NULL) | |
880f421f UD |
2661 | { |
2662 | if (!be_quiet) | |
2663 | error (0, 0, _("\ | |
4b10dd6c UD |
2664 | %s: character `%s' not defined while needed as default value"), |
2665 | "LC_CTYPE", "<space>"); | |
2666 | } | |
2667 | else if (seq->nbytes != 1) | |
2668 | error (0, 0, _("\ | |
2669 | %s: character `%s' in charmap not representable with one byte"), | |
2670 | "LC_CTYPE", "<space>"); | |
2671 | else | |
2672 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
2673 | ||
2674 | ||
2675 | value = repertoire_find_value (repertoire, "tab", 3); | |
2676 | if (value == ILLEGAL_CHAR_VALUE) | |
2677 | { | |
2678 | if (!be_quiet) | |
2679 | error (0, 0, _("\ | |
2680 | %s: character `%s' not defined while needed as default value"), | |
2681 | "LC_CTYPE", "<tab>"); | |
880f421f | 2682 | } |
19bc17a9 RM |
2683 | else |
2684 | ELEM (ctype, class_collection, , value) |= BIT (tok_blank); | |
4b10dd6c UD |
2685 | |
2686 | seq = charmap_find_value (charmap, "tab", 3); | |
2687 | if (seq == NULL) | |
2688 | { | |
2689 | if (!be_quiet) | |
2690 | error (0, 0, _("\ | |
2691 | %s: character `%s' not defined while needed as default value"), | |
2692 | "LC_CTYPE", "<tab>"); | |
2693 | } | |
2694 | else if (seq->nbytes != 1) | |
2695 | error (0, 0, _("\ | |
2696 | %s: character `%s' in charmap not representable with one byte"), | |
2697 | "LC_CTYPE", "<tab>"); | |
2698 | else | |
2699 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
19bc17a9 RM |
2700 | } |
2701 | ||
4b10dd6c | 2702 | if ((ctype->class_done & BITw (tok_graph)) == 0) |
19bc17a9 RM |
2703 | /* "If this keyword [graph] is not specified, characters specified for |
2704 | the keywords `upper', `lower', `alpha', `digit', `xdigit' and `punct', | |
2705 | shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
2706 | { | |
2707 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
2708 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
2709 | size_t cnt; | |
2710 | ||
2711 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2712 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2713 | ctype->class_collection[cnt] |= BIT (tok_graph); | |
4b10dd6c UD |
2714 | |
2715 | for (cnt = 0; cnt < 256; ++cnt) | |
2716 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2717 | ctype->class256_collection[cnt] |= BIT (tok_graph); | |
19bc17a9 RM |
2718 | } |
2719 | ||
4b10dd6c | 2720 | if ((ctype->class_done & BITw (tok_print)) == 0) |
19bc17a9 RM |
2721 | /* "If this keyword [print] is not provided, characters specified for |
2722 | the keywords `upper', `lower', `alpha', `digit', `xdigit', `punct', | |
2723 | and the <space> character shall belong to this character class." | |
2724 | [P1003.2, 2.5.2.1] */ | |
2725 | { | |
2726 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
2727 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
2728 | size_t cnt; | |
4b10dd6c UD |
2729 | uint32_t space; |
2730 | struct charseq *seq; | |
19bc17a9 RM |
2731 | |
2732 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2733 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2734 | ctype->class_collection[cnt] |= BIT (tok_print); | |
2735 | ||
4b10dd6c UD |
2736 | for (cnt = 0; cnt < 256; ++cnt) |
2737 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2738 | ctype->class256_collection[cnt] |= BIT (tok_print); | |
2739 | ||
2740 | ||
2741 | space = repertoire_find_value (repertoire, "space", 5); | |
880f421f UD |
2742 | if (space == ILLEGAL_CHAR_VALUE) |
2743 | { | |
2744 | if (!be_quiet) | |
2745 | error (0, 0, _("\ | |
4b10dd6c UD |
2746 | %s: character `%s' not defined while needed as default value"), |
2747 | "LC_CTYPE", "<space>"); | |
880f421f | 2748 | } |
19bc17a9 RM |
2749 | else |
2750 | ELEM (ctype, class_collection, , space) |= BIT (tok_print); | |
4b10dd6c UD |
2751 | |
2752 | seq = charmap_find_value (charmap, "space", 5); | |
2753 | if (seq == NULL) | |
2754 | { | |
2755 | if (!be_quiet) | |
2756 | error (0, 0, _("\ | |
2757 | %s: character `%s' not defined while needed as default value"), | |
2758 | "LC_CTYPE", "<space>"); | |
2759 | } | |
2760 | else if (seq->nbytes != 1) | |
2761 | error (0, 0, _("\ | |
2762 | %s: character `%s' in charmap not representable with one byte"), | |
2763 | "LC_CTYPE", "<space>"); | |
2764 | else | |
2765 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_print); | |
19bc17a9 RM |
2766 | } |
2767 | ||
4b10dd6c | 2768 | if (ctype->tomap_done[0] == 0) |
6d52618b | 2769 | /* "If this keyword [toupper] is not specified, the lowercase letters |
19bc17a9 RM |
2770 | `a' through `z', and their corresponding uppercase letters `A' to |
2771 | `Z', ..., shall automatically be included, with implementation- | |
2772 | defined character values." [P1003.2, 2.5.2.1] */ | |
2773 | { | |
2774 | char tmp[4]; | |
2775 | int ch; | |
2776 | ||
2777 | strcpy (tmp, "<?>"); | |
2778 | ||
2779 | for (ch = 'a'; ch <= 'z'; ++ch) | |
2780 | { | |
4b10dd6c UD |
2781 | uint32_t value_from, value_to; |
2782 | struct charseq *seq_from, *seq_to; | |
19bc17a9 RM |
2783 | |
2784 | tmp[1] = (char) ch; | |
2785 | ||
4b10dd6c UD |
2786 | value_from = repertoire_find_value (repertoire, &tmp[1], 1); |
2787 | if (value_from == ILLEGAL_CHAR_VALUE) | |
19bc17a9 | 2788 | { |
880f421f UD |
2789 | if (!be_quiet) |
2790 | error (0, 0, _("\ | |
4b10dd6c UD |
2791 | %s: character `%s' not defined while needed as default value"), |
2792 | "LC_CTYPE", tmp); | |
2793 | } | |
2794 | else | |
2795 | { | |
2796 | /* This conversion is implementation defined. */ | |
2797 | tmp[1] = (char) (ch + ('A' - 'a')); | |
2798 | value_to = repertoire_find_value (repertoire, &tmp[1], 1); | |
2799 | if (value_to == ILLEGAL_CHAR_VALUE) | |
2800 | { | |
2801 | if (!be_quiet) | |
2802 | error (0, 0, _("\ | |
2803 | %s: character `%s' not defined while needed as default value"), | |
2804 | "LC_CTYPE", tmp); | |
2805 | } | |
2806 | else | |
2807 | /* The index [0] is determined by the order of the | |
2808 | `ctype_map_newP' calls in `ctype_startup'. */ | |
2809 | ELEM (ctype, map_collection, [0], value_from) = value_to; | |
19bc17a9 RM |
2810 | } |
2811 | ||
4b10dd6c UD |
2812 | seq_from = charmap_find_value (charmap, &tmp[1], 1); |
2813 | if (seq_from == NULL) | |
19bc17a9 | 2814 | { |
880f421f UD |
2815 | if (!be_quiet) |
2816 | error (0, 0, _("\ | |
4b10dd6c UD |
2817 | %s: character `%s' not defined while needed as default value"), |
2818 | "LC_CTYPE", tmp); | |
2819 | } | |
2820 | else if (seq_from->nbytes != 1) | |
2821 | { | |
2822 | if (!be_quiet) | |
2823 | error (0, 0, _("\ | |
2824 | %s: character `%s' needed as default value not representable with one byte"), | |
2825 | "LC_CTYPE", tmp); | |
2826 | } | |
2827 | else | |
2828 | { | |
2829 | /* This conversion is implementation defined. */ | |
2830 | tmp[1] = (char) (ch + ('A' - 'a')); | |
2831 | seq_to = charmap_find_value (charmap, &tmp[1], 1); | |
2832 | if (seq_to == NULL) | |
2833 | { | |
2834 | if (!be_quiet) | |
2835 | error (0, 0, _("\ | |
2836 | %s: character `%s' not defined while needed as default value"), | |
2837 | "LC_CTYPE", tmp); | |
2838 | } | |
2839 | else if (seq_to->nbytes != 1) | |
2840 | { | |
2841 | if (!be_quiet) | |
2842 | error (0, 0, _("\ | |
2843 | %s: character `%s' needed as default value not representable with one byte"), | |
2844 | "LC_CTYPE", tmp); | |
2845 | } | |
2846 | else | |
2847 | /* The index [0] is determined by the order of the | |
2848 | `ctype_map_newP' calls in `ctype_startup'. */ | |
2849 | ctype->map256_collection[0][seq_from->bytes[0]] | |
2850 | = seq_to->bytes[0]; | |
19bc17a9 | 2851 | } |
19bc17a9 RM |
2852 | } |
2853 | } | |
2854 | ||
4b10dd6c | 2855 | if (ctype->tomap_done[1] == 0) |
19bc17a9 RM |
2856 | /* "If this keyword [tolower] is not specified, the mapping shall be |
2857 | the reverse mapping of the one specified to `toupper'." [P1003.2] */ | |
2858 | { | |
19bc17a9 RM |
2859 | for (cnt = 0; cnt < ctype->map_collection_act[0]; ++cnt) |
2860 | if (ctype->map_collection[0][cnt] != 0) | |
2861 | ELEM (ctype, map_collection, [1], | |
2862 | ctype->map_collection[0][cnt]) | |
2863 | = ctype->charnames[cnt]; | |
4b10dd6c UD |
2864 | |
2865 | for (cnt = 0; cnt < 256; ++cnt) | |
2866 | if (ctype->map256_collection[0][cnt] != 0) | |
85cb60ff | 2867 | ctype->map256_collection[1][ctype->map256_collection[0][cnt]] = cnt; |
4b10dd6c UD |
2868 | } |
2869 | ||
2870 | if (ctype->outdigits_act == 0) | |
2871 | { | |
2872 | for (cnt = 0; cnt < 10; ++cnt) | |
2873 | { | |
2874 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
2875 | digits + cnt, 1); | |
2876 | ||
2877 | if (ctype->mboutdigits[cnt] == NULL) | |
2878 | { | |
2879 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
2880 | longnames[cnt], | |
2881 | strlen (longnames[cnt])); | |
2882 | ||
2883 | if (ctype->mboutdigits[cnt] == NULL) | |
2884 | { | |
2885 | /* Provide a replacement. */ | |
2886 | error (0, 0, _("\ | |
2887 | no output digits defined and none of the standard names in the charmap")); | |
2888 | ||
2889 | ctype->mboutdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
2890 | sizeof (struct charseq) + 1); | |
2891 | ||
2892 | /* This is better than nothing. */ | |
2893 | ctype->mboutdigits[cnt]->bytes[0] = digits[cnt]; | |
2894 | ctype->mboutdigits[cnt]->nbytes = 1; | |
2895 | } | |
2896 | } | |
b9eb05d6 UD |
2897 | |
2898 | ctype->wcoutdigits[cnt] = repertoire_find_value (repertoire, | |
2899 | digits + cnt, 1); | |
2900 | ||
2901 | if (ctype->wcoutdigits[cnt] == ILLEGAL_CHAR_VALUE) | |
2902 | { | |
2903 | ctype->wcoutdigits[cnt] = repertoire_find_value (repertoire, | |
2904 | longnames[cnt], | |
2905 | strlen (longnames[cnt])); | |
2906 | ||
2907 | if (ctype->wcoutdigits[cnt] == ILLEGAL_CHAR_VALUE) | |
2908 | { | |
2909 | /* Provide a replacement. */ | |
2910 | error (0, 0, _("\ | |
2911 | no output digits defined and none of the standard names in the repertoire")); | |
2912 | ||
2913 | /* This is better than nothing. */ | |
2914 | ctype->wcoutdigits[cnt] = (uint32_t) digits[cnt]; | |
2915 | } | |
2916 | } | |
4b10dd6c UD |
2917 | } |
2918 | ||
2919 | ctype->outdigits_act = 10; | |
19bc17a9 RM |
2920 | } |
2921 | } | |
2922 | ||
2923 | ||
2924 | static void | |
4b10dd6c UD |
2925 | allocate_arrays (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2926 | struct repertoire_t *repertoire) | |
19bc17a9 RM |
2927 | { |
2928 | size_t idx; | |
2929 | ||
6d52618b UD |
2930 | /* First we have to decide how we organize the arrays. It is easy |
2931 | for a one-byte character set. But multi-byte character set | |
2932 | cannot be stored flat because the chars might be sparsely used. | |
2933 | So we determine an optimal hashing function for the used | |
2934 | characters. | |
2935 | ||
2936 | We use a very trivial hashing function to store the sparse | |
2937 | table. CH % TABSIZE is used as an index. To solve multiple hits | |
2938 | we have N planes. This guarantees a fixed search time for a | |
42d7c593 | 2939 | character [N / 2]. In the following code we determine the minimum |
66ac0abe UD |
2940 | value for TABSIZE * N, where TABSIZE >= 256. |
2941 | ||
2942 | Some people complained that this algorithm takes too long. Well, | |
2943 | go on, improve it. But changing the step size is *not* an | |
2944 | option. Some people changed this to use only sizes of prime | |
2945 | numbers. Think again, do some math. We are looking for the | |
2946 | optimal solution, not something which works in general. Unless | |
2947 | somebody can provide a dynamic programming solution I think this | |
2948 | implementation is as good as it can get. */ | |
19bc17a9 RM |
2949 | size_t min_total = UINT_MAX; |
2950 | size_t act_size = 256; | |
2951 | ||
66ac0abe | 2952 | if (!be_quiet && ctype->charnames_act > 512) |
c84142e8 | 2953 | fputs (_("\ |
19bc17a9 | 2954 | Computing table size for character classes might take a while..."), |
c84142e8 | 2955 | stderr); |
19bc17a9 | 2956 | |
66ac0abe UD |
2957 | /* While we want to have a small total size we are willing to use a |
2958 | little bit larger table if this reduces the number of layers. | |
2959 | Therefore we add a little penalty to the number of planes. | |
2960 | Maybe this constant has to be adjusted a bit. */ | |
2961 | #define PENALTY 128 | |
2962 | do | |
19bc17a9 RM |
2963 | { |
2964 | size_t cnt[act_size]; | |
2965 | size_t act_planes = 1; | |
2966 | ||
2967 | memset (cnt, '\0', sizeof cnt); | |
2968 | ||
2969 | for (idx = 0; idx < 256; ++idx) | |
2970 | cnt[idx] = 1; | |
2971 | ||
2972 | for (idx = 0; idx < ctype->charnames_act; ++idx) | |
2973 | if (ctype->charnames[idx] >= 256) | |
2974 | { | |
2975 | size_t nr = ctype->charnames[idx] % act_size; | |
2976 | ||
2977 | if (++cnt[nr] > act_planes) | |
2978 | { | |
2979 | act_planes = cnt[nr]; | |
66ac0abe | 2980 | if ((act_size + PENALTY) * act_planes >= min_total) |
19bc17a9 RM |
2981 | break; |
2982 | } | |
2983 | } | |
2984 | ||
66ac0abe | 2985 | if ((act_size + PENALTY) * act_planes < min_total) |
19bc17a9 | 2986 | { |
66ac0abe | 2987 | min_total = (act_size + PENALTY) * act_planes; |
19bc17a9 RM |
2988 | ctype->plane_size = act_size; |
2989 | ctype->plane_cnt = act_planes; | |
2990 | } | |
2991 | ||
2992 | ++act_size; | |
2993 | } | |
66ac0abe | 2994 | while (act_size < min_total); |
19bc17a9 | 2995 | |
66ac0abe | 2996 | if (!be_quiet && ctype->charnames_act > 512) |
c84142e8 | 2997 | fputs (_(" done\n"), stderr); |
19bc17a9 | 2998 | |
75cd5204 | 2999 | |
4a33c2f5 UD |
3000 | ctype->names = (uint32_t *) xcalloc (ctype->plane_size |
3001 | * ctype->plane_cnt, | |
3002 | sizeof (uint32_t)); | |
19bc17a9 RM |
3003 | |
3004 | for (idx = 1; idx < 256; ++idx) | |
4a33c2f5 | 3005 | ctype->names[idx] = idx; |
19bc17a9 RM |
3006 | |
3007 | /* Trick: change the 0th entry's name to 1 to mark the cell occupied. */ | |
4a33c2f5 | 3008 | ctype->names[0] = 1; |
19bc17a9 RM |
3009 | |
3010 | for (idx = 256; idx < ctype->charnames_act; ++idx) | |
3011 | { | |
3012 | size_t nr = (ctype->charnames[idx] % ctype->plane_size); | |
3013 | size_t depth = 0; | |
3014 | ||
4a33c2f5 | 3015 | while (ctype->names[nr + depth * ctype->plane_size]) |
19bc17a9 RM |
3016 | ++depth; |
3017 | assert (depth < ctype->plane_cnt); | |
3018 | ||
4a33c2f5 | 3019 | ctype->names[nr + depth * ctype->plane_size] = ctype->charnames[idx]; |
19bc17a9 RM |
3020 | |
3021 | /* Now for faster access remember the index in the NAMES_B array. */ | |
3022 | ctype->charnames[idx] = nr + depth * ctype->plane_size; | |
3023 | } | |
4a33c2f5 | 3024 | ctype->names[0] = 0; |
19bc17a9 RM |
3025 | |
3026 | ||
3027 | /* You wonder about this amount of memory? This is only because some | |
3028 | users do not manage to address the array with unsigned values or | |
3029 | data types with range >= 256. '\200' would result in the array | |
3030 | index -128. To help these poor people we duplicate the entries for | |
3031 | 128 up to 255 below the entry for \0. */ | |
3032 | ctype->ctype_b = (char_class_t *) xcalloc (256 + 128, | |
3033 | sizeof (char_class_t)); | |
3034 | ctype->ctype32_b = (char_class32_t *) xcalloc (ctype->plane_size | |
3035 | * ctype->plane_cnt, | |
3036 | sizeof (char_class32_t)); | |
3037 | ||
4a33c2f5 | 3038 | /* This is the array accessed using the multibyte string elements. */ |
4b10dd6c | 3039 | for (idx = 0; idx < 256; ++idx) |
4a33c2f5 | 3040 | ctype->ctype_b[128 + idx] = ctype->class256_collection[idx]; |
19bc17a9 | 3041 | |
75cd5204 RM |
3042 | /* Mirror first 127 entries. We must take care that entry -1 is not |
3043 | mirrored because EOF == -1. */ | |
3044 | for (idx = 0; idx < 127; ++idx) | |
19bc17a9 RM |
3045 | ctype->ctype_b[idx] = ctype->ctype_b[256 + idx]; |
3046 | ||
3047 | /* The 32 bit array contains all characters. */ | |
3048 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
4a33c2f5 | 3049 | ctype->ctype32_b[ctype->charnames[idx]] = ctype->class_collection[idx]; |
19bc17a9 RM |
3050 | |
3051 | /* Room for table of mappings. */ | |
49f2be5b UD |
3052 | ctype->map = (uint32_t **) xmalloc (2 * sizeof (uint32_t *)); |
3053 | ctype->map32 = (uint32_t **) xmalloc (ctype->map_collection_nr | |
4a33c2f5 | 3054 | * sizeof (uint32_t *)); |
19bc17a9 RM |
3055 | |
3056 | /* Fill in all mappings. */ | |
49f2be5b | 3057 | for (idx = 0; idx < 2; ++idx) |
19bc17a9 RM |
3058 | { |
3059 | unsigned int idx2; | |
3060 | ||
3061 | /* Allocate table. */ | |
49f2be5b | 3062 | ctype->map[idx] = (uint32_t *) xmalloc ((256 + 128) * sizeof (uint32_t)); |
19bc17a9 RM |
3063 | |
3064 | /* Copy values from collection. */ | |
4b10dd6c | 3065 | for (idx2 = 0; idx2 < 256; ++idx2) |
4a33c2f5 | 3066 | ctype->map[idx][128 + idx2] = ctype->map256_collection[idx][idx2]; |
19bc17a9 | 3067 | |
75cd5204 RM |
3068 | /* Mirror first 127 entries. We must take care not to map entry |
3069 | -1 because EOF == -1. */ | |
3070 | for (idx2 = 0; idx2 < 127; ++idx2) | |
4a33c2f5 | 3071 | ctype->map[idx][idx2] = ctype->map[idx][256 + idx2]; |
19bc17a9 | 3072 | |
75cd5204 | 3073 | /* EOF must map to EOF. */ |
4a33c2f5 | 3074 | ctype->map[idx][127] = EOF; |
49f2be5b | 3075 | } |
a9c27b3e | 3076 | |
49f2be5b UD |
3077 | for (idx = 0; idx < ctype->map_collection_nr; ++idx) |
3078 | { | |
3079 | unsigned int idx2; | |
3080 | ||
3081 | /* Allocate table. */ | |
f1d8b804 UD |
3082 | ctype->map32[idx] = (uint32_t *) xmalloc (ctype->plane_size |
3083 | * ctype->plane_cnt | |
3084 | * sizeof (uint32_t)); | |
49f2be5b UD |
3085 | |
3086 | /* Copy default value (identity mapping). */ | |
f1d8b804 | 3087 | memcpy (ctype->map32[idx], ctype->names, |
49f2be5b UD |
3088 | ctype->plane_size * ctype->plane_cnt * sizeof (uint32_t)); |
3089 | ||
3090 | /* Copy values from collection. */ | |
3091 | for (idx2 = 0; idx2 < 256; ++idx2) | |
a9c27b3e | 3092 | if (ctype->map_collection[idx][idx2] != 0) |
f1d8b804 UD |
3093 | ctype->map32[idx][idx2] = ctype->map_collection[idx][idx2]; |
3094 | ||
3095 | while (idx2 < ctype->map_collection_act[idx]) | |
b06c53e7 UD |
3096 | { |
3097 | if (ctype->map_collection[idx][idx2] != 0) | |
450bf66e UD |
3098 | ctype->map32[idx][ctype->charnames[idx2]] = |
3099 | ctype->map_collection[idx][idx2]; | |
b06c53e7 UD |
3100 | ++idx2; |
3101 | } | |
19bc17a9 RM |
3102 | } |
3103 | ||
3104 | /* Extra array for class and map names. */ | |
4b10dd6c UD |
3105 | ctype->class_name_ptr = (uint32_t *) xmalloc (ctype->nr_charclass |
3106 | * sizeof (uint32_t)); | |
3107 | ctype->map_name_ptr = (uint32_t *) xmalloc (ctype->map_collection_nr | |
3108 | * sizeof (uint32_t)); | |
75cd5204 RM |
3109 | |
3110 | /* Array for width information. Because the expected width are very | |
3111 | small we use only one single byte. This save space and we need | |
3112 | not provide the information twice with both endianesses. */ | |
3113 | ctype->width = (unsigned char *) xmalloc (ctype->plane_size | |
3114 | * ctype->plane_cnt); | |
3115 | /* Initialize with default width value. */ | |
4b10dd6c | 3116 | memset (ctype->width, charmap->width_default, |
75cd5204 | 3117 | ctype->plane_size * ctype->plane_cnt); |
4b10dd6c | 3118 | if (charmap->width_rules != NULL) |
75cd5204 RM |
3119 | { |
3120 | size_t cnt; | |
3121 | ||
4b10dd6c | 3122 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
827ff758 UD |
3123 | { |
3124 | unsigned char bytes[charmap->mb_cur_max]; | |
3125 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
3126 | ||
3127 | /* We have the range of character for which the width is | |
3128 | specified described using byte sequences of the multibyte | |
3129 | charset. We have to convert this to UCS4 now. And we | |
3130 | cannot simply convert the beginning and the end of the | |
3131 | sequence, we have to iterate over the byte sequence and | |
3132 | convert it for every single character. */ | |
3133 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
3134 | ||
3135 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
3136 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
3137 | nbytes) <= 0) | |
75cd5204 | 3138 | { |
827ff758 | 3139 | /* Find the UCS value for `bytes'. */ |
827ff758 | 3140 | int inner; |
76e680a8 UD |
3141 | uint32_t wch; |
3142 | struct charseq *seq = | |
3143 | charmap_find_symbol (charmap, bytes, nbytes); | |
3144 | ||
3145 | if (seq == NULL) | |
3146 | wch = ILLEGAL_CHAR_VALUE; | |
3147 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
3148 | wch = seq->ucs4; | |
3149 | else | |
3150 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
3151 | strlen (seq->name)); | |
827ff758 UD |
3152 | |
3153 | if (wch != ILLEGAL_CHAR_VALUE) | |
3154 | { | |
3155 | /* Store the value. */ | |
b1c9ad82 | 3156 | size_t nr = wch % ctype->plane_size; |
827ff758 UD |
3157 | size_t depth = 0; |
3158 | ||
b1c9ad82 | 3159 | while (ctype->names[nr + depth * ctype->plane_size] != wch) |
827ff758 UD |
3160 | ++depth; |
3161 | assert (depth < ctype->plane_cnt); | |
3162 | ||
3163 | ctype->width[nr + depth * ctype->plane_size] | |
3164 | = charmap->width_rules[cnt].width; | |
3165 | } | |
3166 | ||
3167 | /* "Increment" the bytes sequence. */ | |
3168 | inner = nbytes - 1; | |
3169 | while (inner >= 0 && bytes[inner] == 0xff) | |
3170 | --inner; | |
75cd5204 | 3171 | |
827ff758 UD |
3172 | if (inner < 0) |
3173 | { | |
3174 | /* We have to extend the byte sequence. */ | |
3175 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
3176 | break; | |
75cd5204 | 3177 | |
827ff758 UD |
3178 | bytes[0] = 1; |
3179 | memset (&bytes[1], 0, nbytes); | |
3180 | ++nbytes; | |
3181 | } | |
3182 | else | |
3183 | { | |
3184 | ++bytes[inner]; | |
3185 | while (++inner < nbytes) | |
3186 | bytes[inner] = 0; | |
3187 | } | |
75cd5204 | 3188 | } |
827ff758 | 3189 | } |
75cd5204 | 3190 | } |
0200214b | 3191 | |
4b10dd6c UD |
3192 | /* Set MB_CUR_MAX. */ |
3193 | ctype->mb_cur_max = charmap->mb_cur_max; | |
6990326c RM |
3194 | |
3195 | /* We need the name of the currently used 8-bit character set to | |
3196 | make correct conversion between this 8-bit representation and the | |
3197 | ISO 10646 character set used internally for wide characters. */ | |
4b10dd6c UD |
3198 | ctype->codeset_name = charmap->code_set_name; |
3199 | ||
3200 | /* Now determine the table for the transliteration information. | |
3201 | ||
3202 | XXX It is not yet clear to me whether it is worth implementing a | |
3203 | complicated algorithm which uses a hash table to locate the entries. | |
3204 | For now I'll use a simple array which can be searching using binary | |
3205 | search. */ | |
3206 | if (ctype->translit_copy_locale != NULL) | |
3207 | { | |
3208 | /* Fold in the transliteration information from the locale mentioned | |
3209 | in the `include' statement. */ | |
3210 | struct locale_ctype_t *here = ctype; | |
3211 | ||
3212 | do | |
3213 | { | |
3214 | struct localedef_t *other = find_locale (LC_CTYPE, | |
3215 | here->translit_copy_locale, | |
3216 | repertoire->name, charmap); | |
3217 | ||
3218 | if (other == NULL) | |
3219 | { | |
3220 | error (0, 0, _("\ | |
3221 | %s: transliteration data from locale `%s' not available"), | |
3222 | "LC_CTYPE", here->translit_copy_locale); | |
3223 | break; | |
3224 | } | |
3225 | ||
3226 | here = other->categories[LC_CTYPE].ctype; | |
3227 | ||
3228 | /* Enqueue the information if necessary. */ | |
3229 | if (here->translit != NULL) | |
3230 | { | |
3231 | struct translit_t *endp = here->translit; | |
3232 | while (endp->next != NULL) | |
3233 | endp = endp->next; | |
3234 | ||
3235 | endp->next = ctype->translit; | |
3236 | ctype->translit = here->translit; | |
3237 | } | |
3238 | } | |
3239 | while (here->translit_copy_locale != NULL); | |
3240 | } | |
3241 | ||
3242 | if (ctype->translit != NULL) | |
3243 | { | |
3244 | /* First count how many entries we have. This is the upper limit | |
3245 | since some entries from the included files might be overwritten. */ | |
3246 | size_t number = 0; | |
3247 | size_t cnt; | |
3248 | struct translit_t *runp = ctype->translit; | |
3249 | struct translit_t **sorted; | |
3250 | size_t from_len, to_len; | |
3251 | ||
3252 | while (runp != NULL) | |
3253 | { | |
3254 | ++number; | |
3255 | runp = runp->next; | |
3256 | } | |
3257 | ||
3258 | /* Next we allocate an array large enough and fill in the values. */ | |
a9c27b3e UD |
3259 | sorted = (struct translit_t **) alloca (number |
3260 | * sizeof (struct translit_t **)); | |
4b10dd6c UD |
3261 | runp = ctype->translit; |
3262 | number = 0; | |
3263 | do | |
3264 | { | |
3265 | /* Search for the place where to insert this string. | |
3266 | XXX Better use a real sorting algorithm later. */ | |
3267 | size_t idx = 0; | |
3268 | int replace = 0; | |
3269 | ||
3270 | while (idx < number) | |
3271 | { | |
3272 | int res = wcscmp ((const wchar_t *) sorted[idx]->from, | |
3273 | (const wchar_t *) runp->from); | |
3274 | if (res == 0) | |
3275 | { | |
3276 | replace = 1; | |
3277 | break; | |
3278 | } | |
3279 | if (res > 0) | |
3280 | break; | |
3281 | ++idx; | |
3282 | } | |
3283 | ||
3284 | if (replace) | |
3285 | sorted[idx] = runp; | |
3286 | else | |
3287 | { | |
3288 | memmove (&sorted[idx + 1], &sorted[idx], | |
3289 | (number - idx) * sizeof (struct translit_t *)); | |
3290 | sorted[idx] = runp; | |
3291 | ++number; | |
3292 | } | |
3293 | ||
3294 | runp = runp->next; | |
3295 | } | |
3296 | while (runp != NULL); | |
3297 | ||
3298 | /* The next step is putting all the possible transliteration | |
3299 | strings in one memory block so that we can write it out. | |
3300 | We need several different blocks: | |
3301 | - index to the tfromstring array | |
3302 | - from-string array | |
3303 | - index to the to-string array | |
3304 | - to-string array. | |
3305 | And this all must be available for both endianes variants. | |
3306 | */ | |
3307 | from_len = to_len = 0; | |
3308 | for (cnt = 0; cnt < number; ++cnt) | |
3309 | { | |
3310 | struct translit_to_t *srunp; | |
3311 | from_len += wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
3312 | srunp = sorted[cnt]->to; | |
3313 | while (srunp != NULL) | |
3314 | { | |
3315 | to_len += wcslen ((const wchar_t *) srunp->str) + 1; | |
3316 | srunp = srunp->next; | |
3317 | } | |
3318 | /* Plus one for the extra NUL character marking the end of | |
3319 | the list for the current entry. */ | |
3320 | ++to_len; | |
3321 | } | |
3322 | ||
3323 | /* We can allocate the arrays for the results. */ | |
4a33c2f5 UD |
3324 | ctype->translit_from_idx = xmalloc (number * sizeof (uint32_t)); |
3325 | ctype->translit_from_tbl = xmalloc (from_len * sizeof (uint32_t)); | |
3326 | ctype->translit_to_idx = xmalloc (number * sizeof (uint32_t)); | |
3327 | ctype->translit_to_tbl = xmalloc (to_len * sizeof (uint32_t)); | |
4b10dd6c UD |
3328 | |
3329 | from_len = 0; | |
3330 | to_len = 0; | |
3331 | for (cnt = 0; cnt < number; ++cnt) | |
3332 | { | |
3333 | size_t len; | |
3334 | struct translit_to_t *srunp; | |
3335 | ||
4a33c2f5 UD |
3336 | ctype->translit_from_idx[cnt] = from_len; |
3337 | ctype->translit_to_idx[cnt] = to_len; | |
4b10dd6c UD |
3338 | |
3339 | len = wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4a33c2f5 | 3340 | wmemcpy ((wchar_t *) &ctype->translit_from_tbl[from_len], |
4b10dd6c UD |
3341 | (const wchar_t *) sorted[cnt]->from, len); |
3342 | from_len += len; | |
3343 | ||
4a33c2f5 | 3344 | ctype->translit_to_idx[cnt] = to_len; |
4b10dd6c UD |
3345 | srunp = sorted[cnt]->to; |
3346 | while (srunp != NULL) | |
3347 | { | |
3348 | len = wcslen ((const wchar_t *) srunp->str) + 1; | |
4a33c2f5 | 3349 | wmemcpy ((wchar_t *) &ctype->translit_to_tbl[to_len], |
4b10dd6c UD |
3350 | (const wchar_t *) srunp->str, len); |
3351 | to_len += len; | |
3352 | srunp = srunp->next; | |
3353 | } | |
4a33c2f5 | 3354 | ctype->translit_to_tbl[to_len++] = L'\0'; |
4b10dd6c | 3355 | } |
4b10dd6c UD |
3356 | |
3357 | /* Store the information about the length. */ | |
3358 | ctype->translit_idx_size = number * sizeof (uint32_t); | |
3359 | ctype->translit_from_tbl_size = from_len * sizeof (uint32_t); | |
3360 | ctype->translit_to_tbl_size = to_len * sizeof (uint32_t); | |
3361 | } | |
3362 | else | |
3363 | { | |
3364 | /* Provide some dummy pointers since we have nothing to write out. */ | |
3365 | static uint32_t no_str = { 0 }; | |
3366 | ||
4a33c2f5 UD |
3367 | ctype->translit_from_idx = &no_str; |
3368 | ctype->translit_from_tbl = &no_str; | |
3369 | ctype->translit_to_tbl = &no_str; | |
4b10dd6c UD |
3370 | ctype->translit_idx_size = 0; |
3371 | ctype->translit_from_tbl_size = 0; | |
3372 | ctype->translit_to_tbl_size = 0; | |
3373 | } | |
19bc17a9 | 3374 | } |