]>
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 | ||
4b10dd6c UD |
47 | /* These are the extra bits not in wctype.h since these are not preallocated |
48 | classes. */ | |
49 | #define _ISwspecial1 (1 << 29) | |
50 | #define _ISwspecial2 (1 << 30) | |
51 | #define _ISwspecial3 (1 << 31) | |
19bc17a9 RM |
52 | |
53 | ||
54 | /* The bit used for representing a special class. */ | |
55 | #define BITPOS(class) ((class) - tok_upper) | |
4b10dd6c UD |
56 | #define BIT(class) (_ISbit (BITPOS (class))) |
57 | #define BITw(class) (_ISwbit (BITPOS (class))) | |
19bc17a9 RM |
58 | |
59 | #define ELEM(ctype, collection, idx, value) \ | |
60 | *find_idx (ctype, &ctype->collection idx, &ctype->collection##_max idx, \ | |
61 | &ctype->collection##_act idx, value) | |
62 | ||
19bc17a9 RM |
63 | |
64 | /* To be compatible with former implementations we for now restrict | |
65 | the number of bits for character classes to 16. When compatibility | |
66 | is not necessary anymore increase the number to 32. */ | |
4b10dd6c UD |
67 | #define char_class_t uint16_t |
68 | #define CHAR_CLASS_TRANS bswap_16 | |
69 | #define char_class32_t uint32_t | |
70 | #define CHAR_CLASS32_TRANS bswap_32 | |
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; | |
4b10dd6c UD |
147 | uint32_t *names_el; |
148 | uint32_t *names_eb; | |
149 | uint32_t **map_eb; | |
150 | uint32_t **map_el; | |
151 | uint32_t *class_name_ptr; | |
152 | uint32_t *map_name_ptr; | |
75cd5204 | 153 | unsigned char *width; |
4b10dd6c | 154 | uint32_t mb_cur_max; |
6990326c | 155 | const char *codeset_name; |
4b10dd6c UD |
156 | uint32_t translit_hash_size_eb; |
157 | uint32_t translit_hash_size_el; | |
158 | uint32_t translit_hash_layers_eb; | |
159 | uint32_t translit_hash_layers_el; | |
160 | uint32_t *translit_from_idx_eb; | |
161 | uint32_t *translit_from_idx_el; | |
162 | uint32_t *translit_from_tbl_eb; | |
163 | uint32_t *translit_from_tbl_el; | |
164 | uint32_t *translit_to_idx_eb; | |
165 | uint32_t *translit_to_idx_el; | |
166 | uint32_t *translit_to_tbl_eb; | |
167 | uint32_t *translit_to_tbl_el; | |
168 | size_t translit_idx_size; | |
169 | size_t translit_from_tbl_size; | |
170 | size_t translit_to_tbl_size; | |
171 | ||
172 | struct obstack mem_pool; | |
19bc17a9 RM |
173 | }; |
174 | ||
175 | ||
4b10dd6c UD |
176 | #define obstack_chunk_alloc xmalloc |
177 | #define obstack_chunk_free free | |
178 | ||
179 | ||
19bc17a9 | 180 | /* Prototypes for local functions. */ |
4b10dd6c UD |
181 | static void ctype_startup (struct linereader *lr, struct localedef_t *locale, |
182 | struct charmap_t *charmap, int ignore_content); | |
183 | static void ctype_class_new (struct linereader *lr, | |
184 | struct locale_ctype_t *ctype, const char *name); | |
185 | static void ctype_map_new (struct linereader *lr, | |
186 | struct locale_ctype_t *ctype, | |
187 | const char *name, struct charmap_t *charmap); | |
188 | static uint32_t *find_idx (struct locale_ctype_t *ctype, uint32_t **table, | |
189 | size_t *max, size_t *act, unsigned int idx); | |
19bc17a9 | 190 | static void set_class_defaults (struct locale_ctype_t *ctype, |
4b10dd6c UD |
191 | struct charmap_t *charmap, |
192 | struct repertoire_t *repertoire); | |
75cd5204 | 193 | static void allocate_arrays (struct locale_ctype_t *ctype, |
4b10dd6c UD |
194 | struct charmap_t *charmap, |
195 | struct repertoire_t *repertoire); | |
19bc17a9 RM |
196 | |
197 | ||
4b10dd6c UD |
198 | static const char *longnames[] = |
199 | { | |
200 | "zero", "one", "two", "three", "four", | |
201 | "five", "six", "seven", "eight", "nine" | |
202 | }; | |
203 | static const unsigned char digits[] = "0123456789"; | |
204 | ||
205 | ||
206 | static void | |
19bc17a9 | 207 | ctype_startup (struct linereader *lr, struct localedef_t *locale, |
4b10dd6c | 208 | struct charmap_t *charmap, int ignore_content) |
19bc17a9 RM |
209 | { |
210 | unsigned int cnt; | |
211 | struct locale_ctype_t *ctype; | |
212 | ||
4b10dd6c | 213 | if (!ignore_content) |
19bc17a9 | 214 | { |
4b10dd6c UD |
215 | /* Allocate the needed room. */ |
216 | locale->categories[LC_CTYPE].ctype = ctype = | |
217 | (struct locale_ctype_t *) xcalloc (1, sizeof (struct locale_ctype_t)); | |
218 | ||
219 | /* We have seen no names yet. */ | |
220 | ctype->charnames_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
221 | ctype->charnames = | |
222 | (unsigned int *) xmalloc (ctype->charnames_max | |
223 | * sizeof (unsigned int)); | |
224 | for (cnt = 0; cnt < 256; ++cnt) | |
225 | ctype->charnames[cnt] = cnt; | |
226 | ctype->charnames_act = 256; | |
227 | ||
228 | /* Fill character class information. */ | |
229 | ctype->last_class_char = ILLEGAL_CHAR_VALUE; | |
230 | /* The order of the following instructions determines the bit | |
231 | positions! */ | |
232 | ctype_class_new (lr, ctype, "upper"); | |
233 | ctype_class_new (lr, ctype, "lower"); | |
234 | ctype_class_new (lr, ctype, "alpha"); | |
235 | ctype_class_new (lr, ctype, "digit"); | |
236 | ctype_class_new (lr, ctype, "xdigit"); | |
237 | ctype_class_new (lr, ctype, "space"); | |
238 | ctype_class_new (lr, ctype, "print"); | |
239 | ctype_class_new (lr, ctype, "graph"); | |
240 | ctype_class_new (lr, ctype, "blank"); | |
241 | ctype_class_new (lr, ctype, "cntrl"); | |
242 | ctype_class_new (lr, ctype, "punct"); | |
243 | ctype_class_new (lr, ctype, "alnum"); | |
244 | /* The following are extensions from ISO 14652. */ | |
245 | ctype_class_new (lr, ctype, "left_to_right"); | |
246 | ctype_class_new (lr, ctype, "right_to_left"); | |
247 | ctype_class_new (lr, ctype, "num_terminator"); | |
248 | ctype_class_new (lr, ctype, "num_separator"); | |
249 | ctype_class_new (lr, ctype, "segment_separator"); | |
250 | ctype_class_new (lr, ctype, "block_separator"); | |
251 | ctype_class_new (lr, ctype, "direction_control"); | |
252 | ctype_class_new (lr, ctype, "sym_swap_layout"); | |
253 | ctype_class_new (lr, ctype, "char_shape_selector"); | |
254 | ctype_class_new (lr, ctype, "num_shape_selector"); | |
255 | ctype_class_new (lr, ctype, "non_spacing"); | |
256 | ctype_class_new (lr, ctype, "non_spacing_level3"); | |
257 | ctype_class_new (lr, ctype, "normal_connect"); | |
258 | ctype_class_new (lr, ctype, "r_connect"); | |
259 | ctype_class_new (lr, ctype, "no_connect"); | |
260 | ctype_class_new (lr, ctype, "no_connect-space"); | |
261 | ctype_class_new (lr, ctype, "vowel_connect"); | |
262 | ||
263 | ctype->class_collection_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
264 | ctype->class_collection | |
265 | = (uint32_t *) xcalloc (sizeof (unsigned long int), | |
266 | ctype->class_collection_max); | |
267 | ctype->class_collection_act = 256; | |
268 | ||
269 | /* Fill character map information. */ | |
270 | ctype->map_collection_nr = 0; | |
271 | ctype->last_map_idx = MAX_NR_CHARMAP; | |
272 | ctype_map_new (lr, ctype, "toupper", charmap); | |
273 | ctype_map_new (lr, ctype, "tolower", charmap); | |
274 | ctype_map_new (lr, ctype, "tosymmetric", charmap); | |
275 | ||
276 | /* Fill first 256 entries in `toXXX' arrays. */ | |
277 | for (cnt = 0; cnt < 256; ++cnt) | |
278 | { | |
279 | ctype->map_collection[0][cnt] = cnt; | |
280 | ctype->map_collection[1][cnt] = cnt; | |
281 | ctype->map_collection[2][cnt] = cnt; | |
282 | ctype->map256_collection[0][cnt] = cnt; | |
283 | ctype->map256_collection[1][cnt] = cnt; | |
284 | } | |
285 | ||
286 | obstack_init (&ctype->mem_pool); | |
19bc17a9 RM |
287 | } |
288 | } | |
289 | ||
290 | ||
291 | void | |
4b10dd6c | 292 | ctype_finish (struct localedef_t *locale, struct charmap_t *charmap) |
19bc17a9 RM |
293 | { |
294 | /* See POSIX.2, table 2-6 for the meaning of the following table. */ | |
295 | #define NCLASS 12 | |
296 | static const struct | |
297 | { | |
298 | const char *name; | |
299 | const char allow[NCLASS]; | |
300 | } | |
301 | valid_table[NCLASS] = | |
302 | { | |
303 | /* The order is important. See token.h for more information. | |
304 | M = Always, D = Default, - = Permitted, X = Mutually exclusive */ | |
305 | { "upper", "--MX-XDDXXX-" }, | |
306 | { "lower", "--MX-XDDXXX-" }, | |
307 | { "alpha", "---X-XDDXXX-" }, | |
308 | { "digit", "XXX--XDDXXX-" }, | |
309 | { "xdigit", "-----XDDXXX-" }, | |
310 | { "space", "XXXXX------X" }, | |
311 | { "print", "---------X--" }, | |
312 | { "graph", "---------X--" }, | |
313 | { "blank", "XXXXXM-----X" }, | |
314 | { "cntrl", "XXXXX-XX--XX" }, | |
315 | { "punct", "XXXXX-DD-X-X" }, | |
316 | { "alnum", "-----XDDXXX-" } | |
317 | }; | |
318 | size_t cnt; | |
319 | int cls1, cls2; | |
4b10dd6c UD |
320 | uint32_t space_value; |
321 | struct charseq *space_seq; | |
19bc17a9 | 322 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
4b10dd6c | 323 | int warned; |
19bc17a9 RM |
324 | |
325 | /* Set default value for classes not specified. */ | |
4b10dd6c | 326 | set_class_defaults (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
327 | |
328 | /* Check according to table. */ | |
329 | for (cnt = 0; cnt < ctype->class_collection_max; ++cnt) | |
330 | { | |
4b10dd6c | 331 | uint32_t tmp = ctype->class_collection[cnt]; |
19bc17a9 | 332 | |
4b10dd6c UD |
333 | if (tmp != 0) |
334 | { | |
335 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
336 | if ((tmp & _ISwbit (cls1)) != 0) | |
337 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
338 | if (valid_table[cls1].allow[cls2] != '-') | |
19bc17a9 | 339 | { |
4b10dd6c UD |
340 | int eq = (tmp & _ISwbit (cls2)) != 0; |
341 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 342 | { |
4b10dd6c UD |
343 | case 'M': |
344 | if (!eq) | |
345 | { | |
346 | uint32_t value = ctype->charnames[cnt]; | |
347 | ||
348 | if (!be_quiet) | |
349 | error (0, 0, _("\ | |
350 | character L'\\u%0*x' in class `%s' must be in class `%s'"), | |
351 | value > 0xffff ? 8 : 4, value, | |
352 | valid_table[cls1].name, | |
353 | valid_table[cls2].name); | |
354 | } | |
355 | break; | |
356 | ||
357 | case 'X': | |
358 | if (eq) | |
359 | { | |
360 | uint32_t value = ctype->charnames[cnt]; | |
361 | ||
362 | if (!be_quiet) | |
363 | error (0, 0, _("\ | |
364 | character L'\\u%0*x' in class `%s' must not be in class `%s'"), | |
365 | value > 0xffff ? 8 : 4, value, | |
366 | valid_table[cls1].name, | |
367 | valid_table[cls2].name); | |
368 | } | |
369 | break; | |
370 | ||
371 | case 'D': | |
372 | ctype->class_collection[cnt] |= _ISwbit (cls2); | |
373 | break; | |
374 | ||
375 | default: | |
376 | error (5, 0, _("internal error in %s, line %u"), | |
377 | __FUNCTION__, __LINE__); | |
19bc17a9 | 378 | } |
4b10dd6c UD |
379 | } |
380 | } | |
381 | } | |
382 | ||
383 | for (cnt = 0; cnt < 256; ++cnt) | |
384 | { | |
385 | uint32_t tmp = ctype->class256_collection[cnt]; | |
19bc17a9 | 386 | |
4b10dd6c UD |
387 | if (tmp != 0) |
388 | { | |
389 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
390 | if ((tmp & _ISbit (cls1)) != 0) | |
391 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
392 | if (valid_table[cls1].allow[cls2] != '-') | |
393 | { | |
394 | int eq = (tmp & _ISbit (cls2)) != 0; | |
395 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 396 | { |
4b10dd6c UD |
397 | case 'M': |
398 | if (!eq) | |
399 | { | |
400 | char buf[17]; | |
401 | ||
402 | sprintf (buf, "\\%o", cnt); | |
403 | ||
404 | if (!be_quiet) | |
405 | error (0, 0, _("\ | |
406 | character '%s' in class `%s' must be in class `%s'"), | |
407 | buf, valid_table[cls1].name, | |
408 | valid_table[cls2].name); | |
409 | } | |
410 | break; | |
411 | ||
412 | case 'X': | |
413 | if (eq) | |
414 | { | |
415 | char buf[17]; | |
416 | ||
417 | sprintf (buf, "\\%o", cnt); | |
418 | ||
419 | if (!be_quiet) | |
420 | error (0, 0, _("\ | |
421 | character '%s' in class `%s' must not be in class `%s'"), | |
422 | buf, valid_table[cls1].name, | |
423 | valid_table[cls2].name); | |
424 | } | |
425 | break; | |
426 | ||
427 | case 'D': | |
428 | ctype->class256_collection[cnt] |= _ISbit (cls2); | |
429 | break; | |
430 | ||
431 | default: | |
432 | error (5, 0, _("internal error in %s, line %u"), | |
433 | __FUNCTION__, __LINE__); | |
19bc17a9 | 434 | } |
4b10dd6c UD |
435 | } |
436 | } | |
19bc17a9 RM |
437 | } |
438 | ||
439 | /* ... and now test <SP> as a special case. */ | |
4b10dd6c UD |
440 | space_value = repertoire_find_value (ctype->repertoire, "SP", 2); |
441 | if (space_value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
442 | { |
443 | if (!be_quiet) | |
444 | error (0, 0, _("character <SP> not defined in character map")); | |
445 | } | |
c84142e8 UD |
446 | else if (((cnt = BITPOS (tok_space), |
447 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 448 | & BITw (tok_space)) == 0) |
c84142e8 UD |
449 | || (cnt = BITPOS (tok_blank), |
450 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 451 | & BITw (tok_blank)) == 0))) |
880f421f UD |
452 | { |
453 | if (!be_quiet) | |
454 | error (0, 0, _("<SP> character not in class `%s'"), | |
455 | valid_table[cnt].name); | |
456 | } | |
c84142e8 UD |
457 | else if (((cnt = BITPOS (tok_punct), |
458 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 459 | & BITw (tok_punct)) != 0) |
c84142e8 UD |
460 | || (cnt = BITPOS (tok_graph), |
461 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 462 | & BITw (tok_graph)) |
880f421f UD |
463 | != 0))) |
464 | { | |
465 | if (!be_quiet) | |
466 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
467 | valid_table[cnt].name); | |
468 | } | |
19bc17a9 | 469 | else |
4b10dd6c UD |
470 | ELEM (ctype, class_collection, , space_value) |= BITw (tok_print); |
471 | ||
472 | space_seq = charmap_find_value (charmap, "SP", 2); | |
473 | if (space_seq == NULL || space_seq->nbytes != 1) | |
474 | { | |
475 | if (!be_quiet) | |
476 | error (0, 0, _("character <SP> not defined in character map")); | |
477 | } | |
478 | else if (((cnt = BITPOS (tok_space), | |
479 | (ctype->class256_collection[space_seq->bytes[0]] | |
480 | & BIT (tok_space)) == 0) | |
481 | || (cnt = BITPOS (tok_blank), | |
482 | (ctype->class256_collection[space_seq->bytes[0]] | |
483 | & BIT (tok_blank)) == 0))) | |
484 | { | |
485 | if (!be_quiet) | |
486 | error (0, 0, _("<SP> character not in class `%s'"), | |
487 | valid_table[cnt].name); | |
488 | } | |
489 | else if (((cnt = BITPOS (tok_punct), | |
490 | (ctype->class256_collection[space_seq->bytes[0]] | |
491 | & BIT (tok_punct)) != 0) | |
492 | || (cnt = BITPOS (tok_graph), | |
493 | (ctype->class256_collection[space_seq->bytes[0]] | |
494 | & BIT (tok_graph)) != 0))) | |
495 | { | |
496 | if (!be_quiet) | |
497 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
498 | valid_table[cnt].name); | |
499 | } | |
500 | else | |
501 | ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print); | |
75cd5204 RM |
502 | |
503 | /* Now that the tests are done make sure the name array contains all | |
504 | characters which are handled in the WIDTH section of the | |
505 | character set definition file. */ | |
4b10dd6c UD |
506 | if (charmap->width_rules != NULL) |
507 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) | |
75cd5204 | 508 | { |
4b10dd6c | 509 | #if 0 |
75cd5204 | 510 | size_t inner; |
4b10dd6c UD |
511 | for (inner = charmap->width_rules[cnt].from; |
512 | inner <= charmap->width_rules[cnt].to; ++inner) | |
75cd5204 | 513 | (void) find_idx (ctype, NULL, NULL, NULL, inner); |
4b10dd6c UD |
514 | #else |
515 | /* XXX Handle width. We must convert from the charseq to the | |
516 | repertoire value */ | |
517 | abort (); | |
518 | #endif | |
519 | } | |
520 | ||
521 | /* There must be a multiple of 10 digits. */ | |
522 | if (ctype->mbdigits_act % 10 != 0) | |
523 | { | |
524 | assert (ctype->mbdigits_act == ctype->wcdigits_act); | |
525 | ctype->wcdigits_act -= ctype->mbdigits_act % 10; | |
526 | ctype->mbdigits_act -= ctype->mbdigits_act % 10; | |
527 | error (0, 0, _("`digit' category has not entries in groups of ten")); | |
528 | } | |
529 | ||
530 | /* Check the input digits. There must be a multiple of ten available. | |
531 | In each group I could be that one or the other character is missing. | |
532 | In this case the whole group must be removed. */ | |
533 | cnt = 0; | |
534 | while (cnt < ctype->mbdigits_act) | |
535 | { | |
536 | size_t inner; | |
537 | for (inner = 0; inner < 10; ++inner) | |
538 | if (ctype->mbdigits[cnt + inner] == NULL) | |
539 | break; | |
540 | ||
541 | if (inner == 10) | |
542 | cnt += 10; | |
543 | else | |
544 | { | |
545 | /* Remove the group. */ | |
546 | memmove (&ctype->mbdigits[cnt], &ctype->mbdigits[cnt + 10], | |
547 | ((ctype->wcdigits_act - cnt - 10) | |
548 | * sizeof (ctype->mbdigits[0]))); | |
549 | ctype->mbdigits_act -= 10; | |
550 | } | |
551 | } | |
552 | ||
553 | /* If no input digits are given use the default. */ | |
554 | if (ctype->mbdigits_act == 0) | |
555 | { | |
556 | if (ctype->mbdigits_max == 0) | |
557 | { | |
558 | ctype->mbdigits = obstack_alloc (&charmap->mem_pool, | |
559 | 10 * sizeof (struct charseq *)); | |
560 | ctype->mbdigits_max = 10; | |
561 | } | |
562 | ||
563 | for (cnt = 0; cnt < 10; ++cnt) | |
564 | { | |
565 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
566 | digits + cnt, 1); | |
567 | if (ctype->mbdigits[cnt] == NULL) | |
568 | { | |
569 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
570 | longnames[cnt], | |
571 | strlen (longnames[cnt])); | |
572 | if (ctype->mbdigits[cnt] == NULL) | |
573 | { | |
574 | /* Hum, this ain't good. */ | |
575 | error (0, 0, _("\ | |
576 | no input digits defined and none of the standard names in the charmap")); | |
577 | ||
578 | ctype->mbdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
579 | sizeof (struct charseq) + 1); | |
580 | ||
581 | /* This is better than nothing. */ | |
582 | ctype->mbdigits[cnt]->bytes[0] = digits[cnt]; | |
583 | ctype->mbdigits[cnt]->nbytes = 1; | |
584 | } | |
585 | } | |
586 | } | |
587 | ||
588 | ctype->mbdigits_act = 10; | |
589 | } | |
590 | ||
591 | /* Check the wide character input digits. There must be a multiple | |
592 | of ten available. In each group I could be that one or the other | |
593 | character is missing. In this case the whole group must be | |
594 | removed. */ | |
595 | cnt = 0; | |
596 | while (cnt < ctype->wcdigits_act) | |
597 | { | |
598 | size_t inner; | |
599 | for (inner = 0; inner < 10; ++inner) | |
600 | if (ctype->wcdigits[cnt + inner] == ILLEGAL_CHAR_VALUE) | |
601 | break; | |
602 | ||
603 | if (inner == 10) | |
604 | cnt += 10; | |
605 | else | |
606 | { | |
607 | /* Remove the group. */ | |
608 | memmove (&ctype->wcdigits[cnt], &ctype->wcdigits[cnt + 10], | |
609 | ((ctype->wcdigits_act - cnt - 10) | |
610 | * sizeof (ctype->wcdigits[0]))); | |
611 | ctype->wcdigits_act -= 10; | |
612 | } | |
613 | } | |
614 | ||
615 | /* If no input digits are given use the default. */ | |
616 | if (ctype->wcdigits_act == 0) | |
617 | { | |
618 | if (ctype->wcdigits_max == 0) | |
619 | { | |
620 | ctype->wcdigits = obstack_alloc (&charmap->mem_pool, | |
621 | 10 * sizeof (uint32_t)); | |
622 | ctype->wcdigits_max = 10; | |
623 | } | |
624 | ||
625 | for (cnt = 0; cnt < 10; ++cnt) | |
626 | ctype->wcdigits[cnt] = L'0' + cnt; | |
627 | ||
628 | ctype->mbdigits_act = 10; | |
629 | } | |
630 | ||
631 | /* Check the outdigits. */ | |
632 | warned = 0; | |
633 | for (cnt = 0; cnt < 10; ++cnt) | |
634 | if (ctype->mboutdigits[cnt] == NULL) | |
635 | { | |
636 | static struct charseq replace[2]; | |
637 | ||
638 | if (!warned) | |
639 | { | |
640 | error (0, 0, _("\ | |
641 | not all characters used in `outdigit' are available in the charmap")); | |
642 | warned = 1; | |
643 | } | |
644 | ||
645 | replace[0].nbytes = 1; | |
646 | replace[0].bytes[0] = '?'; | |
647 | replace[0].bytes[1] = '\0'; | |
648 | ctype->mboutdigits[cnt] = &replace[0]; | |
649 | } | |
650 | ||
651 | warned = 0; | |
652 | for (cnt = 0; cnt < 10; ++cnt) | |
653 | if (ctype->wcoutdigits[cnt] == 0) | |
654 | { | |
655 | if (!warned) | |
656 | { | |
657 | error (0, 0, _("\ | |
658 | not all characters used in `outdigit' are available in the repertoire")); | |
659 | warned = 1; | |
660 | } | |
661 | ||
662 | ctype->wcoutdigits[cnt] = L'?'; | |
75cd5204 | 663 | } |
19bc17a9 RM |
664 | } |
665 | ||
666 | ||
667 | void | |
4b10dd6c | 668 | ctype_output (struct localedef_t *locale, struct charmap_t *charmap, |
75cd5204 | 669 | const char *output_path) |
19bc17a9 RM |
670 | { |
671 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; | |
672 | const size_t nelems = (_NL_ITEM_INDEX (_NL_NUM_LC_CTYPE) | |
673 | + 2 * (ctype->map_collection_nr - 2)); | |
75cd5204 RM |
674 | struct iovec iov[2 + nelems + ctype->nr_charclass |
675 | + ctype->map_collection_nr]; | |
19bc17a9 | 676 | struct locale_file data; |
4b10dd6c | 677 | uint32_t idx[nelems + 1]; |
75cd5204 | 678 | size_t elem, cnt, offset, total; |
4b10dd6c | 679 | char *cp; |
19bc17a9 RM |
680 | |
681 | /* Now prepare the output: Find the sizes of the table we can use. */ | |
4b10dd6c | 682 | allocate_arrays (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
683 | |
684 | data.magic = LIMAGIC (LC_CTYPE); | |
685 | data.n = nelems; | |
686 | iov[0].iov_base = (void *) &data; | |
687 | iov[0].iov_len = sizeof (data); | |
688 | ||
689 | iov[1].iov_base = (void *) idx; | |
690 | iov[1].iov_len = sizeof (idx); | |
691 | ||
692 | idx[0] = iov[0].iov_len + iov[1].iov_len; | |
693 | offset = 0; | |
694 | ||
695 | for (elem = 0; elem < nelems; ++elem) | |
696 | { | |
697 | if (elem < _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) | |
698 | switch (elem) | |
699 | { | |
700 | #define CTYPE_DATA(name, base, len) \ | |
701 | case _NL_ITEM_INDEX (name): \ | |
ce7a5ef4 RM |
702 | iov[2 + elem + offset].iov_base = (base); \ |
703 | iov[2 + elem + offset].iov_len = (len); \ | |
75cd5204 RM |
704 | if (elem + 1 < nelems) \ |
705 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; \ | |
19bc17a9 RM |
706 | break |
707 | ||
708 | CTYPE_DATA (_NL_CTYPE_CLASS, | |
709 | ctype->ctype_b, | |
710 | (256 + 128) * sizeof (char_class_t)); | |
711 | ||
712 | CTYPE_DATA (_NL_CTYPE_TOUPPER_EB, | |
713 | ctype->map_eb[0], | |
714 | (ctype->plane_size * ctype->plane_cnt + 128) | |
4b10dd6c | 715 | * sizeof (uint32_t)); |
19bc17a9 RM |
716 | CTYPE_DATA (_NL_CTYPE_TOLOWER_EB, |
717 | ctype->map_eb[1], | |
718 | (ctype->plane_size * ctype->plane_cnt + 128) | |
4b10dd6c | 719 | * sizeof (uint32_t)); |
19bc17a9 RM |
720 | |
721 | CTYPE_DATA (_NL_CTYPE_TOUPPER_EL, | |
722 | ctype->map_el[0], | |
723 | (ctype->plane_size * ctype->plane_cnt + 128) | |
4b10dd6c | 724 | * sizeof (uint32_t)); |
19bc17a9 RM |
725 | CTYPE_DATA (_NL_CTYPE_TOLOWER_EL, |
726 | ctype->map_el[1], | |
727 | (ctype->plane_size * ctype->plane_cnt + 128) | |
4b10dd6c | 728 | * sizeof (uint32_t)); |
19bc17a9 RM |
729 | |
730 | CTYPE_DATA (_NL_CTYPE_CLASS32, | |
731 | ctype->ctype32_b, | |
732 | (ctype->plane_size * ctype->plane_cnt | |
733 | * sizeof (char_class32_t))); | |
734 | ||
735 | CTYPE_DATA (_NL_CTYPE_NAMES_EB, | |
7a12c6bb | 736 | ctype->names_eb, (ctype->plane_size * ctype->plane_cnt |
4b10dd6c | 737 | * sizeof (uint32_t))); |
19bc17a9 | 738 | CTYPE_DATA (_NL_CTYPE_NAMES_EL, |
7a12c6bb | 739 | ctype->names_el, (ctype->plane_size * ctype->plane_cnt |
4b10dd6c UD |
740 | * sizeof (uint32_t))); |
741 | ||
742 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_SIZE_EB, | |
743 | &ctype->translit_hash_size_eb, sizeof (uint32_t)); | |
744 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_SIZE_EL, | |
745 | &ctype->translit_hash_size_el, sizeof (uint32_t)); | |
746 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_LAYERS_EB, | |
747 | &ctype->translit_hash_layers_eb, sizeof (uint32_t)); | |
748 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_HASH_LAYERS_EL, | |
749 | &ctype->translit_hash_layers_el, sizeof (uint32_t)); | |
750 | ||
751 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_IDX_EB, | |
752 | ctype->translit_from_idx_eb, | |
753 | ctype->translit_idx_size); | |
754 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_IDX_EL, | |
755 | ctype->translit_from_idx_el, | |
756 | ctype->translit_idx_size); | |
757 | ||
758 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_TBL_EB, | |
759 | ctype->translit_from_tbl_eb, | |
760 | ctype->translit_from_tbl_size); | |
761 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_TBL_EL, | |
762 | ctype->translit_from_tbl_el, | |
763 | ctype->translit_from_tbl_size); | |
764 | ||
765 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_IDX_EB, | |
766 | ctype->translit_to_idx_eb, | |
767 | ctype->translit_idx_size); | |
768 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_IDX_EL, | |
769 | ctype->translit_to_idx_el, | |
770 | ctype->translit_idx_size); | |
771 | ||
772 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_TBL_EB, | |
773 | ctype->translit_to_tbl_eb, ctype->translit_to_tbl_size); | |
774 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_TBL_EL, | |
775 | ctype->translit_to_tbl_el, ctype->translit_to_tbl_size); | |
776 | ||
777 | #if __BYTE_ORDER == __BIG_ENDIAN | |
778 | CTYPE_DATA (_NL_CTYPE_HASH_SIZE_EB, | |
779 | &ctype->plane_size, sizeof (uint32_t)); | |
780 | CTYPE_DATA (_NL_CTYPE_HASH_LAYERS_EB, | |
781 | &ctype->plane_cnt, sizeof (uint32_t)); | |
782 | #else | |
783 | case _NL_ITEM_INDEX (_NL_CTYPE_HASH_SIZE_EB): | |
784 | iov[2 + elem + offset].iov_base = | |
785 | (uint32_t *) alloca (sizeof (uint32_t)); | |
786 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
787 | bswap_32 (ctype->plane_size); | |
788 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
789 | break; | |
790 | case _NL_ITEM_INDEX (_NL_CTYPE_HASH_LAYERS_EB): | |
791 | iov[2 + elem + offset].iov_base = | |
792 | (uint32_t *) alloca (sizeof (uint32_t)); | |
793 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
794 | bswap_32 (ctype->plane_cnt); | |
795 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
796 | break; | |
797 | #endif | |
798 | #if __BYTE_ORDER == __BIG_ENDIAN | |
799 | CTYPE_DATA (_NL_CTYPE_HASH_SIZE_EL, | |
800 | &ctype->plane_size, sizeof (uint32_t)); | |
801 | CTYPE_DATA (_NL_CTYPE_HASH_LAYERS_EL, | |
802 | &ctype->plane_cnt, sizeof (uint32_t)); | |
803 | #else | |
804 | case _NL_ITEM_INDEX (_NL_CTYPE_HASH_SIZE_EL): | |
805 | iov[2 + elem + offset].iov_base = | |
806 | (uint32_t *) alloca (sizeof (uint32_t)); | |
807 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
808 | bswap_32 (ctype->plane_size); | |
809 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
810 | break; | |
811 | case _NL_ITEM_INDEX (_NL_CTYPE_HASH_LAYERS_EL): | |
812 | iov[2 + elem + offset].iov_base = | |
813 | (uint32_t *) alloca (sizeof (uint32_t)); | |
814 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
815 | bswap_32 (ctype->plane_cnt); | |
816 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
817 | break; | |
818 | #endif | |
19bc17a9 | 819 | |
75cd5204 RM |
820 | case _NL_ITEM_INDEX (_NL_CTYPE_CLASS_NAMES): |
821 | /* The class name array. */ | |
822 | total = 0; | |
823 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt, ++offset) | |
824 | { | |
825 | iov[2 + elem + offset].iov_base | |
826 | = (void *) ctype->classnames[cnt]; | |
827 | iov[2 + elem + offset].iov_len | |
828 | = strlen (ctype->classnames[cnt]) + 1; | |
829 | total += iov[2 + elem + offset].iov_len; | |
830 | } | |
ce7a5ef4 RM |
831 | iov[2 + elem + offset].iov_base = (void *) "\0\0\0"; |
832 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); | |
833 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 834 | |
4b10dd6c | 835 | idx[elem + 1] = idx[elem] + total; |
75cd5204 RM |
836 | break; |
837 | ||
838 | case _NL_ITEM_INDEX (_NL_CTYPE_MAP_NAMES): | |
839 | /* The class name array. */ | |
840 | total = 0; | |
841 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt, ++offset) | |
842 | { | |
843 | iov[2 + elem + offset].iov_base | |
844 | = (void *) ctype->mapnames[cnt]; | |
845 | iov[2 + elem + offset].iov_len | |
846 | = strlen (ctype->mapnames[cnt]) + 1; | |
847 | total += iov[2 + elem + offset].iov_len; | |
848 | } | |
ce7a5ef4 RM |
849 | iov[2 + elem + offset].iov_base = (void *) "\0\0\0"; |
850 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); | |
851 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 852 | |
4b10dd6c | 853 | idx[elem + 1] = idx[elem] + total; |
75cd5204 | 854 | break; |
19bc17a9 RM |
855 | |
856 | CTYPE_DATA (_NL_CTYPE_WIDTH, | |
75cd5204 | 857 | ctype->width, ctype->plane_size * ctype->plane_cnt); |
19bc17a9 | 858 | |
0200214b | 859 | CTYPE_DATA (_NL_CTYPE_MB_CUR_MAX, |
4b10dd6c | 860 | &ctype->mb_cur_max, sizeof (uint32_t)); |
0200214b | 861 | |
ce7a5ef4 RM |
862 | case _NL_ITEM_INDEX (_NL_CTYPE_CODESET_NAME): |
863 | total = strlen (ctype->codeset_name) + 1; | |
864 | if (total % 4 == 0) | |
865 | iov[2 + elem + offset].iov_base = (char *) ctype->codeset_name; | |
866 | else | |
867 | { | |
868 | iov[2 + elem + offset].iov_base = alloca ((total + 3) & ~3); | |
9756dfe1 UD |
869 | memset (mempcpy (iov[2 + elem + offset].iov_base, |
870 | ctype->codeset_name, total), | |
871 | '\0', 4 - (total & 3)); | |
ce7a5ef4 RM |
872 | total = (total + 3) & ~3; |
873 | } | |
874 | iov[2 + elem + offset].iov_len = total; | |
4b10dd6c UD |
875 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
876 | break; | |
877 | ||
878 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN_EB): | |
879 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN_EL): | |
880 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); | |
881 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
882 | if ((elem == _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN_EB) | |
883 | && __BYTE_ORDER == __BIG_ENDIAN) | |
884 | || (elem == _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN_EL) | |
885 | && __BYTE_ORDER == __LITTLE_ENDIAN)) | |
886 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
887 | ctype->mbdigits_act / 10; | |
888 | else | |
889 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
890 | bswap_32 (ctype->mbdigits_act / 10); | |
891 | break; | |
892 | ||
893 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN_EB): | |
894 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN_EL): | |
895 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); | |
896 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
897 | if ((elem == _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN_EB) | |
898 | && __BYTE_ORDER == __BIG_ENDIAN) | |
899 | || (elem == _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN_EL) | |
900 | && __BYTE_ORDER == __LITTLE_ENDIAN)) | |
901 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
902 | ctype->wcdigits_act / 10; | |
903 | else | |
904 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
905 | bswap_32 (ctype->wcdigits_act / 10); | |
906 | break; | |
907 | ||
908 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_MB): | |
909 | /* Compute the length of all possible characters. For INDIGITS | |
910 | there might be more than one. We simply concatenate all of | |
911 | them with a NUL byte following. The NUL byte wouldn't be | |
912 | necessary but it makes it easier for the user. */ | |
913 | total = 0; | |
914 | for (cnt = elem - _NL_CTYPE_INDIGITS0_MB; | |
915 | cnt < ctype->mbdigits_act; cnt += 10) | |
916 | total += ctype->mbdigits[cnt]->nbytes + 1; | |
917 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
918 | iov[2 + elem + offset].iov_len = total; | |
919 | ||
920 | cp = iov[2 + elem + offset].iov_base; | |
921 | for (cnt = elem - _NL_CTYPE_INDIGITS0_MB; | |
922 | cnt < ctype->mbdigits_act; cnt += 10) | |
923 | { | |
924 | cp = mempcpy (cp, ctype->mbdigits[cnt]->bytes, | |
925 | ctype->mbdigits[cnt]->nbytes); | |
926 | *cp++ = '\0'; | |
927 | } | |
928 | break; | |
929 | ||
930 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_MB): | |
931 | /* Compute the length of all possible characters. For INDIGITS | |
932 | there might be more than one. We simply concatenate all of | |
933 | them with a NUL byte following. The NUL byte wouldn't be | |
934 | necessary but it makes it easier for the user. */ | |
935 | cnt = elem - _NL_CTYPE_OUTDIGIT0_MB; | |
936 | total = ctype->mboutdigits[cnt]->nbytes + 1; | |
937 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
938 | iov[2 + elem + offset].iov_len = total; | |
939 | ||
940 | *(char *) mempcpy (iov[2 + elem + offset].iov_base, | |
941 | ctype->mbdigits[cnt]->bytes, | |
942 | ctype->mbdigits[cnt]->nbytes) = '\0'; | |
943 | break; | |
944 | ||
945 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC_EB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC_EB): | |
946 | total = ctype->wcdigits_act / 10; | |
947 | ||
948 | iov[2 + elem + offset].iov_base = | |
949 | (uint32_t *) alloca (total * sizeof (uint32_t)); | |
950 | iov[2 + elem + offset].iov_len = total * sizeof (uint32_t); | |
951 | ||
952 | for (cnt = elem - _NL_CTYPE_INDIGITS0_WC_EB; | |
953 | cnt < ctype->wcdigits_act; cnt += 10) | |
954 | ((uint32_t *) iov[2 + elem + offset].iov_base)[cnt / 10] | |
955 | = (__BYTE_ORDER == __LITTLE_ENDIAN | |
956 | ? bswap_32 (ctype->wcdigits[cnt]) : ctype->wcdigits[cnt]); | |
957 | break; | |
958 | ||
959 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC_EL) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC_EL): | |
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 | ||
966 | for (cnt = elem - _NL_CTYPE_INDIGITS0_WC_EL; | |
967 | cnt < ctype->wcdigits_act; cnt += 10) | |
968 | ((uint32_t *) iov[2 + elem + offset].iov_base)[cnt / 10] | |
969 | = (__BYTE_ORDER == __BIG_ENDIAN | |
970 | ? bswap_32 (ctype->wcdigits[cnt]) : ctype->wcdigits[cnt]); | |
971 | break; | |
972 | ||
973 | #if __BYTE_ORDER == __BIG_ENDIAN | |
974 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC_EB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC_EB): | |
975 | cnt = elem - _NL_CTYPE_OUTDIGIT0_WC_EB; | |
976 | #else | |
977 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC_EL) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC_EL): | |
978 | cnt = elem - _NL_CTYPE_OUTDIGIT0_WC_EL; | |
979 | #endif | |
980 | iov[2 + elem + offset].iov_base = &ctype->wcoutdigits[cnt]; | |
981 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
982 | break; | |
983 | ||
984 | #if __BYTE_ORDER == __LITTLE_ENDIAN | |
985 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC_EB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC_EB): | |
986 | cnt = elem - _NL_CTYPE_OUTDIGIT0_WC_EB; | |
987 | #else | |
988 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC_EL) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC_EL): | |
989 | cnt = elem - _NL_CTYPE_OUTDIGIT0_WC_EL; | |
990 | #endif | |
991 | iov[2 + elem + offset].iov_base = | |
992 | (uint32_t *) alloca (sizeof (uint32_t)); | |
993 | *(uint32_t *) iov[2 + elem + offset].iov_base = | |
994 | bswap_32 (ctype->wcoutdigits[cnt]); | |
995 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
ce7a5ef4 | 996 | break; |
6990326c | 997 | |
19bc17a9 RM |
998 | default: |
999 | assert (! "unknown CTYPE element"); | |
1000 | } | |
1001 | else | |
1002 | { | |
1003 | /* Handle extra maps. */ | |
1004 | size_t nr = (elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) >> 1; | |
1005 | ||
1006 | if (((elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) & 1) == 0) | |
75cd5204 | 1007 | iov[2 + elem + offset].iov_base = ctype->map_eb[nr]; |
19bc17a9 | 1008 | else |
75cd5204 | 1009 | iov[2 + elem + offset].iov_base = ctype->map_el[nr]; |
19bc17a9 | 1010 | |
75cd5204 RM |
1011 | iov[2 + elem + offset].iov_len = ((ctype->plane_size |
1012 | * ctype->plane_cnt + 128) | |
4b10dd6c | 1013 | * sizeof (uint32_t)); |
19bc17a9 | 1014 | |
4b10dd6c | 1015 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
75cd5204 | 1016 | } |
19bc17a9 | 1017 | } |
19bc17a9 | 1018 | |
75cd5204 RM |
1019 | assert (2 + elem + offset == (nelems + ctype->nr_charclass |
1020 | + ctype->map_collection_nr + 2)); | |
19bc17a9 | 1021 | |
75cd5204 | 1022 | write_locale_data (output_path, "LC_CTYPE", 2 + elem + offset, iov); |
19bc17a9 RM |
1023 | } |
1024 | ||
1025 | ||
4b10dd6c UD |
1026 | /* Local functions. */ |
1027 | static void | |
1028 | ctype_class_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1029 | const char *name) | |
19bc17a9 | 1030 | { |
4b10dd6c | 1031 | size_t cnt; |
19bc17a9 | 1032 | |
4b10dd6c UD |
1033 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1034 | if (strcmp (ctype->classnames[cnt], name) == 0) | |
1035 | break; | |
19bc17a9 | 1036 | |
4b10dd6c UD |
1037 | if (cnt < ctype->nr_charclass) |
1038 | { | |
1039 | lr_error (lr, _("character class `%s' already defined"), name); | |
1040 | return; | |
1041 | } | |
19bc17a9 | 1042 | |
4b10dd6c UD |
1043 | if (ctype->nr_charclass == MAX_NR_CHARCLASS) |
1044 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1045 | error (2, 0, _("\ | |
1046 | implementation limit: no more than %d character classes allowed"), | |
1047 | MAX_NR_CHARCLASS); | |
19bc17a9 | 1048 | |
4b10dd6c | 1049 | ctype->classnames[ctype->nr_charclass++] = name; |
19bc17a9 RM |
1050 | } |
1051 | ||
1052 | ||
4b10dd6c UD |
1053 | static void |
1054 | ctype_map_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1055 | const char *name, struct charmap_t *charmap) | |
19bc17a9 | 1056 | { |
4b10dd6c | 1057 | size_t max_chars = 0; |
ba1ffaa1 | 1058 | size_t cnt; |
19bc17a9 | 1059 | |
4b10dd6c | 1060 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
19bc17a9 | 1061 | { |
4b10dd6c UD |
1062 | if (strcmp (ctype->mapnames[cnt], name) == 0) |
1063 | break; | |
1064 | ||
1065 | if (max_chars < ctype->map_collection_max[cnt]) | |
1066 | max_chars = ctype->map_collection_max[cnt]; | |
19bc17a9 RM |
1067 | } |
1068 | ||
4b10dd6c UD |
1069 | if (cnt < ctype->map_collection_nr) |
1070 | { | |
1071 | lr_error (lr, _("character map `%s' already defined"), name); | |
1072 | return; | |
1073 | } | |
19bc17a9 | 1074 | |
4b10dd6c UD |
1075 | if (ctype->map_collection_nr == MAX_NR_CHARMAP) |
1076 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1077 | error (2, 0, _("\ | |
1078 | implementation limit: no more than %d character maps allowed"), | |
1079 | MAX_NR_CHARMAP); | |
19bc17a9 | 1080 | |
4b10dd6c UD |
1081 | ctype->mapnames[cnt] = name; |
1082 | ||
1083 | if (max_chars == 0) | |
1084 | ctype->map_collection_max[cnt] = charmap->mb_cur_max == 1 ? 256 : 512; | |
1085 | else | |
1086 | ctype->map_collection_max[cnt] = max_chars; | |
1087 | ||
1088 | ctype->map_collection[cnt] = (uint32_t *) | |
1089 | xmalloc (sizeof (uint32_t) * ctype->map_collection_max[cnt]); | |
1090 | memset (ctype->map_collection[cnt], '\0', | |
1091 | sizeof (uint32_t) * ctype->map_collection_max[cnt]); | |
1092 | ctype->map_collection_act[cnt] = 256; | |
19bc17a9 | 1093 | |
4b10dd6c | 1094 | ++ctype->map_collection_nr; |
19bc17a9 RM |
1095 | } |
1096 | ||
1097 | ||
4b10dd6c UD |
1098 | /* We have to be prepared that TABLE, MAX, and ACT can be NULL. This |
1099 | is possible if we only want ot extend the name array. */ | |
1100 | static uint32_t * | |
1101 | find_idx (struct locale_ctype_t *ctype, uint32_t **table, size_t *max, | |
1102 | size_t *act, uint32_t idx) | |
19bc17a9 | 1103 | { |
4b10dd6c | 1104 | size_t cnt; |
19bc17a9 | 1105 | |
4b10dd6c UD |
1106 | if (idx < 256) |
1107 | return table == NULL ? NULL : &(*table)[idx]; | |
19bc17a9 | 1108 | |
4b10dd6c UD |
1109 | for (cnt = 256; cnt < ctype->charnames_act; ++cnt) |
1110 | if (ctype->charnames[cnt] == idx) | |
1111 | break; | |
19bc17a9 | 1112 | |
4b10dd6c UD |
1113 | /* We have to distinguish two cases: the name is found or not. */ |
1114 | if (cnt == ctype->charnames_act) | |
1115 | { | |
1116 | /* Extend the name array. */ | |
1117 | if (ctype->charnames_act == ctype->charnames_max) | |
1118 | { | |
1119 | ctype->charnames_max *= 2; | |
1120 | ctype->charnames = (unsigned int *) | |
1121 | xrealloc (ctype->charnames, | |
1122 | sizeof (unsigned int) * ctype->charnames_max); | |
1123 | } | |
1124 | ctype->charnames[ctype->charnames_act++] = idx; | |
1125 | } | |
19bc17a9 | 1126 | |
4b10dd6c UD |
1127 | if (table == NULL) |
1128 | /* We have done everything we are asked to do. */ | |
1129 | return NULL; | |
19bc17a9 | 1130 | |
4b10dd6c UD |
1131 | if (cnt >= *act) |
1132 | { | |
1133 | if (cnt >= *max) | |
1134 | { | |
1135 | size_t old_max = *max; | |
1136 | do | |
1137 | *max *= 2; | |
1138 | while (*max <= cnt); | |
19bc17a9 | 1139 | |
4b10dd6c UD |
1140 | *table = |
1141 | (uint32_t *) xrealloc (*table, *max * sizeof (unsigned long int)); | |
1142 | memset (&(*table)[old_max], '\0', | |
1143 | (*max - old_max) * sizeof (uint32_t)); | |
1144 | } | |
19bc17a9 | 1145 | |
4b10dd6c UD |
1146 | *act = cnt; |
1147 | } | |
19bc17a9 | 1148 | |
4b10dd6c | 1149 | return &(*table)[cnt]; |
19bc17a9 RM |
1150 | } |
1151 | ||
1152 | ||
4b10dd6c UD |
1153 | static int |
1154 | get_character (struct token *now, struct charmap_t *charmap, | |
1155 | struct repertoire_t *repertoire, | |
1156 | struct charseq **seqp, uint32_t *wchp) | |
19bc17a9 | 1157 | { |
4b10dd6c UD |
1158 | if (now->tok == tok_bsymbol) |
1159 | { | |
1160 | /* This will hopefully be the normal case. */ | |
1161 | *wchp = repertoire_find_value (repertoire, now->val.str.startmb, | |
1162 | now->val.str.lenmb); | |
1163 | *seqp = charmap_find_value (charmap, now->val.str.startmb, | |
1164 | now->val.str.lenmb); | |
1165 | } | |
1166 | else if (now->tok == tok_ucs4) | |
1167 | { | |
1168 | *seqp = repertoire_find_seq (repertoire, now->val.ucs4); | |
19bc17a9 | 1169 | |
4b10dd6c UD |
1170 | if (*seqp == NULL) |
1171 | { | |
1172 | /* Compute the value in the charmap from the UCS value. */ | |
1173 | const char *symbol = repertoire_find_symbol (repertoire, | |
1174 | now->val.ucs4); | |
19bc17a9 | 1175 | |
4b10dd6c UD |
1176 | if (symbol == NULL) |
1177 | *seqp = NULL; | |
1178 | else | |
1179 | *seqp = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1180 | |
4b10dd6c UD |
1181 | if (*seqp == NULL) |
1182 | { | |
1183 | /* Insert a negative entry. */ | |
1184 | static const struct charseq negative | |
1185 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1186 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, 4); | |
1187 | *newp = now->val.ucs4; | |
1188 | ||
1189 | insert_entry (&repertoire->seq_table, newp, 4, | |
1190 | (void *) &negative); | |
1191 | } | |
1192 | else | |
1193 | (*seqp)->ucs4 = now->val.ucs4; | |
1194 | } | |
1195 | else if ((*seqp)->ucs4 != now->val.ucs4) | |
1196 | *seqp = NULL; | |
19bc17a9 | 1197 | |
4b10dd6c UD |
1198 | *wchp = now->val.ucs4; |
1199 | } | |
1200 | else if (now->tok == tok_charcode) | |
1201 | { | |
1202 | /* We must map from the byte code to UCS4. */ | |
1203 | *seqp = charmap_find_symbol (charmap, now->val.str.startmb, | |
1204 | now->val.str.lenmb); | |
19bc17a9 | 1205 | |
4b10dd6c UD |
1206 | if (*seqp == NULL) |
1207 | *wchp = ILLEGAL_CHAR_VALUE; | |
1208 | else | |
1209 | { | |
1210 | if ((*seqp)->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1211 | (*seqp)->ucs4 = repertoire_find_value (repertoire, (*seqp)->name, | |
1212 | strlen ((*seqp)->name)); | |
1213 | *wchp = (*seqp)->ucs4; | |
1214 | } | |
1215 | } | |
1216 | else | |
1217 | return 1; | |
19bc17a9 RM |
1218 | |
1219 | return 0; | |
1220 | } | |
1221 | ||
1222 | ||
4b10dd6c UD |
1223 | /* Ellipsis like in `<foo123>..<foo12a>' or `<j1234>....<j1245>'. */ |
1224 | static void | |
1225 | charclass_symbolic_ellipsis (struct linereader *ldfile, | |
1226 | struct locale_ctype_t *ctype, | |
1227 | struct charmap_t *charmap, | |
1228 | struct repertoire_t *repertoire, | |
1229 | struct token *now, | |
1230 | const char *last_str, | |
1231 | unsigned long int class256_bit, | |
1232 | unsigned long int class_bit, int base, | |
1233 | int ignore_content, int handle_digits) | |
19bc17a9 | 1234 | { |
4b10dd6c UD |
1235 | const char *nowstr = now->val.str.startmb; |
1236 | char tmp[now->val.str.lenmb + 1]; | |
1237 | const char *cp; | |
1238 | char *endp; | |
1239 | unsigned long int from; | |
1240 | unsigned long int to; | |
19bc17a9 | 1241 | |
4b10dd6c UD |
1242 | /* We have to compute the ellipsis values using the symbolic names. */ |
1243 | assert (last_str != NULL); | |
1244 | ||
1245 | if (strlen (last_str) != now->val.str.lenmb) | |
19bc17a9 | 1246 | { |
4b10dd6c UD |
1247 | invalid_range: |
1248 | lr_error (ldfile, | |
1249 | _("`%s' and `%s' are no valid names for symbolic range"), | |
1250 | last_str, nowstr); | |
1251 | return; | |
19bc17a9 RM |
1252 | } |
1253 | ||
4b10dd6c UD |
1254 | if (memcmp (last_str, nowstr, now->val.str.lenmb) == 0) |
1255 | /* Nothing to do, the names are the same. */ | |
1256 | return; | |
19bc17a9 | 1257 | |
4b10dd6c UD |
1258 | for (cp = last_str; *cp == *(nowstr + (cp - last_str)); ++cp) |
1259 | ; | |
19bc17a9 | 1260 | |
4b10dd6c UD |
1261 | errno = 0; |
1262 | from = strtoul (cp, &endp, base); | |
1263 | if ((from == UINT_MAX && errno == ERANGE) || *endp != '\0') | |
1264 | goto invalid_range; | |
19bc17a9 | 1265 | |
4b10dd6c UD |
1266 | to = strtoul (nowstr + (cp - last_str), &endp, base); |
1267 | if ((to == UINT_MAX && errno == ERANGE) || *endp != '\0' || from >= to) | |
1268 | goto invalid_range; | |
19bc17a9 | 1269 | |
4b10dd6c UD |
1270 | /* OK, we have a range FROM - TO. Now we can create the symbolic names. */ |
1271 | if (!ignore_content) | |
1272 | { | |
1273 | now->val.str.startmb = tmp; | |
1274 | while (++from <= to) | |
1275 | { | |
1276 | struct charseq *seq; | |
1277 | uint32_t wch; | |
19bc17a9 | 1278 | |
4b10dd6c UD |
1279 | sprintf (tmp, (base == 10 ? "%.*s%0*d" : "%.*s%0*X"), cp - last_str, |
1280 | last_str, now->val.str.lenmb - (cp - last_str), from); | |
19bc17a9 | 1281 | |
4b10dd6c UD |
1282 | get_character (now, charmap, repertoire, &seq, &wch); |
1283 | ||
1284 | if (seq != NULL && seq->nbytes == 1) | |
1285 | /* Yep, we can store information about this byte sequence. */ | |
1286 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
19bc17a9 | 1287 | |
4b10dd6c UD |
1288 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) |
1289 | /* We have the UCS4 position. */ | |
1290 | *find_idx (ctype, &ctype->class_collection, | |
1291 | &ctype->class_collection_max, | |
1292 | &ctype->class_collection_act, wch) |= class_bit; | |
19bc17a9 | 1293 | |
4b10dd6c UD |
1294 | if (handle_digits == 1) |
1295 | { | |
1296 | /* We must store the digit values. */ | |
1297 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1298 | { | |
1299 | ctype->mbdigits_max *= 2; | |
1300 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1301 | (ctype->mbdigits_max | |
1302 | * sizeof (char *))); | |
1303 | ctype->wcdigits_max *= 2; | |
1304 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1305 | (ctype->wcdigits_max | |
1306 | * sizeof (uint32_t))); | |
1307 | } | |
1308 | ||
1309 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1310 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1311 | } | |
1312 | else if (handle_digits == 2) | |
1313 | { | |
1314 | /* We must store the digit values. */ | |
1315 | if (ctype->outdigits_act >= 10) | |
1316 | { | |
1317 | lr_error (ldfile, _("\ | |
1318 | %s: field `%s' does not contain exactly ten entries"), | |
1319 | "LC_CTYPE", "outdigit"); | |
1320 | return; | |
1321 | } | |
1322 | ||
1323 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1324 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1325 | ++ctype->outdigits_act; | |
1326 | } | |
1327 | } | |
1328 | } | |
19bc17a9 RM |
1329 | } |
1330 | ||
1331 | ||
4b10dd6c UD |
1332 | /* Ellipsis like in `<U1234>..<U2345>'. */ |
1333 | static void | |
1334 | charclass_ucs4_ellipsis (struct linereader *ldfile, | |
1335 | struct locale_ctype_t *ctype, | |
1336 | struct charmap_t *charmap, | |
1337 | struct repertoire_t *repertoire, | |
1338 | struct token *now, uint32_t last_wch, | |
1339 | unsigned long int class256_bit, | |
1340 | unsigned long int class_bit, int ignore_content, | |
1341 | int handle_digits) | |
19bc17a9 | 1342 | { |
4b10dd6c | 1343 | if (last_wch > now->val.ucs4) |
19bc17a9 | 1344 | { |
4b10dd6c UD |
1345 | lr_error (ldfile, _("\ |
1346 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1347 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, now->val.ucs4, | |
1348 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, last_wch); | |
19bc17a9 RM |
1349 | return; |
1350 | } | |
1351 | ||
4b10dd6c UD |
1352 | if (!ignore_content) |
1353 | while (++last_wch <= now->val.ucs4) | |
1354 | { | |
1355 | /* We have to find out whether there is a byte sequence corresponding | |
1356 | to this UCS4 value. */ | |
1357 | struct charseq *seq = repertoire_find_seq (repertoire, last_wch); | |
19bc17a9 | 1358 | |
4b10dd6c UD |
1359 | /* If this is the first time we look for this sequence create a new |
1360 | entry. */ | |
1361 | if (seq == NULL) | |
1362 | { | |
1363 | /* Find the symbolic name for this UCS4 value. */ | |
1364 | const char *symbol = repertoire_find_symbol (repertoire, last_wch); | |
1365 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, 4); | |
1366 | *newp = last_wch; | |
19bc17a9 | 1367 | |
4b10dd6c UD |
1368 | if (symbol != NULL) |
1369 | /* We have a name, now search the multibyte value. */ | |
1370 | seq = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1371 | |
4b10dd6c UD |
1372 | if (seq == NULL) |
1373 | { | |
1374 | /* We have to create a fake entry. */ | |
1375 | static const struct charseq negative | |
1376 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1377 | seq = (struct charseq *) &negative; | |
1378 | } | |
1379 | else | |
1380 | seq->ucs4 = last_wch; | |
1381 | ||
1382 | insert_entry (&repertoire->seq_table, newp, 4, seq); | |
1383 | } | |
1384 | ||
1385 | /* We have a name, now search the multibyte value. */ | |
1386 | if (seq->ucs4 == last_wch && seq->nbytes == 1) | |
1387 | /* Yep, we can store information about this byte sequence. */ | |
1388 | ctype->class256_collection[(size_t) seq->bytes[0]] | |
1389 | |= class256_bit; | |
1390 | ||
1391 | /* And of course we have the UCS4 position. */ | |
1392 | if (class_bit != 0 && class_bit != 0) | |
1393 | *find_idx (ctype, &ctype->class_collection, | |
1394 | &ctype->class_collection_max, | |
1395 | &ctype->class_collection_act, last_wch) |= class_bit; | |
1396 | ||
1397 | if (handle_digits == 1) | |
1398 | { | |
1399 | /* We must store the digit values. */ | |
1400 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1401 | { | |
1402 | ctype->mbdigits_max *= 2; | |
1403 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1404 | (ctype->mbdigits_max | |
1405 | * sizeof (char *))); | |
1406 | ctype->wcdigits_max *= 2; | |
1407 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1408 | (ctype->wcdigits_max | |
1409 | * sizeof (uint32_t))); | |
1410 | } | |
1411 | ||
1412 | ctype->mbdigits[ctype->mbdigits_act++] = (seq->ucs4 == last_wch | |
1413 | ? seq : NULL); | |
1414 | ctype->wcdigits[ctype->wcdigits_act++] = last_wch; | |
1415 | } | |
1416 | else if (handle_digits == 2) | |
1417 | { | |
1418 | /* We must store the digit values. */ | |
1419 | if (ctype->outdigits_act >= 10) | |
1420 | { | |
1421 | lr_error (ldfile, _("\ | |
1422 | %s: field `%s' does not contain exactly ten entries"), | |
1423 | "LC_CTYPE", "outdigit"); | |
1424 | return; | |
1425 | } | |
19bc17a9 | 1426 | |
4b10dd6c UD |
1427 | ctype->mboutdigits[ctype->outdigits_act] = (seq->ucs4 == last_wch |
1428 | ? seq : NULL); | |
1429 | ctype->wcoutdigits[ctype->outdigits_act] = last_wch; | |
1430 | ++ctype->outdigits_act; | |
1431 | } | |
1432 | } | |
19bc17a9 RM |
1433 | } |
1434 | ||
1435 | ||
4b10dd6c | 1436 | /* Ellipsis as in `/xea/x12.../xea/x34'. */ |
19bc17a9 | 1437 | static void |
4b10dd6c UD |
1438 | charclass_charcode_ellipsis (struct linereader *ldfile, |
1439 | struct locale_ctype_t *ctype, | |
1440 | struct charmap_t *charmap, | |
1441 | struct repertoire_t *repertoire, | |
1442 | struct token *now, char *last_charcode, | |
1443 | uint32_t last_charcode_len, | |
1444 | unsigned long int class256_bit, | |
1445 | unsigned long int class_bit, int ignore_content, | |
1446 | int handle_digits) | |
19bc17a9 | 1447 | { |
4b10dd6c UD |
1448 | /* First check whether the to-value is larger. */ |
1449 | if (now->val.charcode.nbytes != last_charcode_len) | |
1450 | { | |
1451 | lr_error (ldfile, _("\ | |
1452 | start end end character sequence of range must have the same length")); | |
1453 | return; | |
1454 | } | |
19bc17a9 | 1455 | |
4b10dd6c | 1456 | if (memcmp (last_charcode, now->val.charcode.bytes, last_charcode_len) > 0) |
19bc17a9 | 1457 | { |
4b10dd6c UD |
1458 | lr_error (ldfile, _("\ |
1459 | to-value character sequence is smaller than from-value sequence")); | |
19bc17a9 RM |
1460 | return; |
1461 | } | |
1462 | ||
4b10dd6c UD |
1463 | if (!ignore_content) |
1464 | { | |
1465 | do | |
1466 | { | |
1467 | /* Increment the byte sequence value. */ | |
1468 | struct charseq *seq; | |
1469 | uint32_t wch; | |
1470 | int i; | |
1471 | ||
1472 | for (i = last_charcode_len - 1; i >= 0; --i) | |
1473 | if (++last_charcode[i] != 0) | |
1474 | break; | |
1475 | ||
1476 | if (last_charcode_len == 1) | |
1477 | /* Of course we have the charcode value. */ | |
1478 | ctype->class256_collection[(size_t) last_charcode[0]] | |
1479 | |= class256_bit; | |
1480 | ||
1481 | /* Find the symbolic name. */ | |
1482 | seq = charmap_find_symbol (charmap, last_charcode, | |
1483 | last_charcode_len); | |
1484 | if (seq != NULL) | |
1485 | { | |
1486 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1487 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1488 | strlen (seq->name)); | |
1489 | wch = seq->ucs4; | |
1490 | ||
1491 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) | |
1492 | *find_idx (ctype, &ctype->class_collection, | |
1493 | &ctype->class_collection_max, | |
1494 | &ctype->class_collection_act, wch) |= class_bit; | |
1495 | } | |
1496 | else | |
1497 | wch = ILLEGAL_CHAR_VALUE; | |
19bc17a9 | 1498 | |
4b10dd6c UD |
1499 | if (handle_digits == 1) |
1500 | { | |
1501 | /* We must store the digit values. */ | |
1502 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1503 | { | |
1504 | ctype->mbdigits_max *= 2; | |
1505 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1506 | (ctype->mbdigits_max | |
1507 | * sizeof (char *))); | |
1508 | ctype->wcdigits_max *= 2; | |
1509 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1510 | (ctype->wcdigits_max | |
1511 | * sizeof (uint32_t))); | |
1512 | } | |
1513 | ||
1514 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1515 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1516 | seq->nbytes = last_charcode_len; | |
1517 | ||
1518 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1519 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1520 | } | |
1521 | else if (handle_digits == 2) | |
1522 | { | |
1523 | struct charseq *seq; | |
1524 | /* We must store the digit values. */ | |
1525 | if (ctype->outdigits_act >= 10) | |
1526 | { | |
1527 | lr_error (ldfile, _("\ | |
1528 | %s: field `%s' does not contain exactly ten entries"), | |
1529 | "LC_CTYPE", "outdigit"); | |
1530 | return; | |
1531 | } | |
1532 | ||
1533 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1534 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1535 | seq->nbytes = last_charcode_len; | |
1536 | ||
1537 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1538 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1539 | ++ctype->outdigits_act; | |
1540 | } | |
1541 | } | |
1542 | while (memcmp (last_charcode, now->val.charcode.bytes, | |
1543 | last_charcode_len) != 0); | |
1544 | } | |
19bc17a9 RM |
1545 | } |
1546 | ||
1547 | ||
4b10dd6c UD |
1548 | /* Read one transliteration entry. */ |
1549 | static uint32_t * | |
1550 | read_widestring (struct linereader *ldfile, struct token *now, | |
1551 | struct charmap_t *charmap, struct repertoire_t *repertoire) | |
19bc17a9 | 1552 | { |
4b10dd6c | 1553 | uint32_t *wstr; |
19bc17a9 | 1554 | |
4b10dd6c UD |
1555 | if (now->tok == tok_default_missing) |
1556 | /* The special name "" will denote this case. */ | |
1557 | wstr = (uint32_t *) L""; | |
1558 | else if (now->tok == tok_bsymbol) | |
19bc17a9 | 1559 | { |
4b10dd6c UD |
1560 | /* Get the value from the repertoire. */ |
1561 | wstr = xmalloc (2 * sizeof (uint32_t)); | |
1562 | wstr[0] = repertoire_find_value (repertoire, now->val.str.startmb, | |
1563 | now->val.str.lenmb); | |
1564 | if (wstr[0] == ILLEGAL_CHAR_VALUE) | |
1565 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1566 | return NULL; | |
1567 | ||
1568 | wstr[1] = 0; | |
19bc17a9 | 1569 | } |
4b10dd6c | 1570 | else if (now->tok == tok_ucs4) |
19bc17a9 | 1571 | { |
4b10dd6c UD |
1572 | wstr = xmalloc (2 * sizeof (uint32_t)); |
1573 | wstr[0] = now->val.ucs4; | |
1574 | wstr[1] = 0; | |
1575 | } | |
1576 | else if (now->tok == tok_charcode) | |
1577 | { | |
1578 | /* Argh, we have to convert to the symbol name first and then to the | |
1579 | UCS4 value. */ | |
1580 | struct charseq *seq = charmap_find_symbol (charmap, | |
1581 | now->val.str.startmb, | |
1582 | now->val.str.lenmb); | |
1583 | if (seq == NULL) | |
1584 | /* Cannot find the UCS4 value. */ | |
1585 | return NULL; | |
1586 | ||
1587 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1588 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1589 | strlen (seq->name)); | |
1590 | if (seq->ucs4 == ILLEGAL_CHAR_VALUE) | |
1591 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1592 | return NULL; | |
1593 | ||
1594 | wstr = xmalloc (2 * sizeof (uint32_t)); | |
1595 | wstr[0] = seq->ucs4; | |
1596 | wstr[1] = 0; | |
1597 | } | |
1598 | else if (now->tok == tok_string) | |
1599 | { | |
1600 | wstr = now->val.str.startwc; | |
1601 | if (wstr[0] == 0) | |
1602 | return NULL; | |
1603 | } | |
1604 | else | |
1605 | { | |
1606 | if (now->tok != tok_eol && now->tok != tok_eof) | |
1607 | lr_ignore_rest (ldfile, 0); | |
1608 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
1609 | return (uint32_t *) -1l; | |
19bc17a9 RM |
1610 | } |
1611 | ||
4b10dd6c UD |
1612 | return wstr; |
1613 | } | |
19bc17a9 | 1614 | |
19bc17a9 | 1615 | |
4b10dd6c UD |
1616 | static void |
1617 | read_translit_entry (struct linereader *ldfile, struct locale_ctype_t *ctype, | |
1618 | struct token *now, struct charmap_t *charmap, | |
1619 | struct repertoire_t *repertoire) | |
1620 | { | |
1621 | uint32_t *from_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1622 | struct translit_t *result; | |
1623 | struct translit_to_t **top; | |
1624 | struct obstack *ob = &ctype->mem_pool; | |
1625 | int first; | |
1626 | int ignore; | |
1627 | ||
1628 | if (from_wstr == NULL) | |
1629 | /* There is no valid from string. */ | |
1630 | return; | |
19bc17a9 | 1631 | |
4b10dd6c UD |
1632 | result = (struct translit_t *) obstack_alloc (ob, |
1633 | sizeof (struct translit_t)); | |
1634 | result->from = from_wstr; | |
1635 | result->next = NULL; | |
1636 | result->to = NULL; | |
1637 | top = &result->to; | |
1638 | first = 1; | |
1639 | ignore = 0; | |
1640 | ||
1641 | while (1) | |
1642 | { | |
1643 | uint32_t *to_wstr; | |
1644 | ||
1645 | /* Next we have one or more transliterations. They are | |
1646 | separated by semicolons. */ | |
1647 | now = lr_token (ldfile, charmap, repertoire); | |
1648 | ||
1649 | if (!first && (now->tok == tok_semicolon || now->tok == tok_eol)) | |
1650 | { | |
1651 | /* One string read. */ | |
1652 | const uint32_t zero = 0; | |
1653 | ||
1654 | if (!ignore) | |
1655 | { | |
1656 | obstack_grow (ob, &zero, 4); | |
1657 | to_wstr = obstack_finish (ob); | |
1658 | ||
1659 | *top = obstack_alloc (ob, sizeof (struct translit_to_t)); | |
1660 | (*top)->str = to_wstr; | |
1661 | (*top)->next = NULL; | |
1662 | } | |
1663 | ||
1664 | if (now->tok == tok_eol) | |
1665 | { | |
1666 | result->next = ctype->translit; | |
1667 | ctype->translit = result; | |
1668 | return; | |
1669 | } | |
1670 | ||
1671 | if (!ignore) | |
1672 | top = &(*top)->next; | |
1673 | ignore = 0; | |
1674 | } | |
1675 | else | |
1676 | { | |
1677 | to_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1678 | if (to_wstr == (uint32_t *) -1l) | |
1679 | { | |
1680 | /* An error occurred. */ | |
1681 | obstack_free (ob, result); | |
1682 | return; | |
1683 | } | |
1684 | ||
1685 | if (to_wstr == NULL) | |
1686 | ignore = 1; | |
1687 | else | |
1688 | /* This value is usable. */ | |
1689 | obstack_grow (ob, to_wstr, wcslen ((wchar_t *) to_wstr) * 4); | |
19bc17a9 | 1690 | |
4b10dd6c UD |
1691 | first = 0; |
1692 | } | |
1693 | } | |
19bc17a9 RM |
1694 | } |
1695 | ||
1696 | ||
4b10dd6c UD |
1697 | /* The parser for the LC_CTYPE section of the locale definition. */ |
1698 | void | |
1699 | ctype_read (struct linereader *ldfile, struct localedef_t *result, | |
1700 | struct charmap_t *charmap, const char *repertoire_name, | |
1701 | int ignore_content) | |
19bc17a9 | 1702 | { |
4b10dd6c UD |
1703 | struct repertoire_t *repertoire = NULL; |
1704 | struct locale_ctype_t *ctype; | |
1705 | struct token *now; | |
1706 | enum token_t nowtok; | |
19bc17a9 | 1707 | size_t cnt; |
4b10dd6c UD |
1708 | struct charseq *last_seq; |
1709 | uint32_t last_wch = 0; | |
1710 | enum token_t last_token; | |
1711 | enum token_t ellipsis_token; | |
1712 | char last_charcode[16]; | |
1713 | size_t last_charcode_len = 0; | |
1714 | const char *last_str = NULL; | |
1715 | int mapidx; | |
19bc17a9 | 1716 | |
4b10dd6c UD |
1717 | /* Get the repertoire we have to use. */ |
1718 | if (repertoire_name != NULL) | |
1719 | repertoire = repertoire_read (repertoire_name); | |
19bc17a9 | 1720 | |
4b10dd6c UD |
1721 | /* The rest of the line containing `LC_CTYPE' must be free. */ |
1722 | lr_ignore_rest (ldfile, 1); | |
19bc17a9 | 1723 | |
4b10dd6c UD |
1724 | |
1725 | do | |
19bc17a9 | 1726 | { |
4b10dd6c UD |
1727 | now = lr_token (ldfile, charmap, NULL); |
1728 | nowtok = now->tok; | |
19bc17a9 | 1729 | } |
4b10dd6c | 1730 | while (nowtok == tok_eol); |
19bc17a9 | 1731 | |
4b10dd6c UD |
1732 | /* If we see `copy' now we are almost done. */ |
1733 | if (nowtok == tok_copy) | |
1734 | { | |
1735 | handle_copy (ldfile, charmap, repertoire, tok_lc_ctype, LC_CTYPE, | |
1736 | "LC_CTYPE", ignore_content); | |
1737 | return; | |
1738 | } | |
75cd5204 | 1739 | |
4b10dd6c UD |
1740 | /* Prepare the data structures. */ |
1741 | ctype_startup (ldfile, result, charmap, ignore_content); | |
1742 | ctype = result->categories[LC_CTYPE].ctype; | |
1743 | ||
1744 | /* Remember the repertoire we use. */ | |
1745 | if (!ignore_content) | |
1746 | ctype->repertoire = repertoire; | |
1747 | ||
1748 | while (1) | |
19bc17a9 | 1749 | { |
4b10dd6c UD |
1750 | unsigned long int class_bit = 0; |
1751 | unsigned long int class256_bit = 0; | |
1752 | int handle_digits = 0; | |
1753 | ||
1754 | /* Of course we don't proceed beyond the end of file. */ | |
1755 | if (nowtok == tok_eof) | |
1756 | break; | |
1757 | ||
1758 | /* Ingore empty lines. */ | |
1759 | if (nowtok == tok_eol) | |
19bc17a9 | 1760 | { |
4b10dd6c UD |
1761 | now = lr_token (ldfile, charmap, NULL); |
1762 | nowtok = now->tok; | |
1763 | continue; | |
1764 | } | |
19bc17a9 | 1765 | |
4b10dd6c UD |
1766 | switch (nowtok) |
1767 | { | |
1768 | case tok_class: | |
1769 | /* We simply forget the `class' keyword and use the following | |
1770 | operand to determine the bit. */ | |
1771 | now = lr_token (ldfile, charmap, NULL); | |
1772 | if (now->tok == tok_ident || now->tok == tok_string) | |
1773 | { | |
1774 | /* Must be one of the predefined class names. */ | |
1775 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
1776 | if (strcmp (ctype->classnames[cnt], now->val.str.startmb) == 0) | |
1777 | break; | |
1778 | if (cnt >= ctype->nr_charclass) | |
1779 | { | |
1780 | if (now->val.str.lenmb == 8 | |
1781 | && memcmp ("special1", now->val.str.startmb, 8) == 0) | |
1782 | class_bit = _ISwspecial1; | |
1783 | else if (now->val.str.lenmb == 8 | |
1784 | && memcmp ("special2", now->val.str.startmb, 8) == 0) | |
1785 | class_bit = _ISwspecial2; | |
1786 | else if (now->val.str.lenmb == 8 | |
1787 | && memcmp ("special3", now->val.str.startmb, 8) == 0) | |
1788 | class_bit = _ISwspecial3; | |
1789 | else | |
1790 | { | |
1791 | lr_error (ldfile, _("\ | |
1792 | unknown character class `%s' in category `LC_CTYPE'"), | |
1793 | now->val.str.startmb); | |
1794 | free (now->val.str.startmb); | |
1795 | ||
1796 | lr_ignore_rest (ldfile, 0); | |
1797 | continue; | |
1798 | } | |
1799 | } | |
1800 | else | |
1801 | class_bit = _ISwbit (cnt); | |
1802 | ||
1803 | free (now->val.str.startmb); | |
1804 | } | |
1805 | else if (now->tok == tok_digit) | |
1806 | goto handle_tok_digit; | |
1807 | else if (now->tok < tok_upper || now->tok > tok_blank) | |
1808 | goto err_label; | |
1809 | else | |
1810 | { | |
1811 | class_bit = BITw (now->tok); | |
1812 | class256_bit = BIT (now->tok); | |
1813 | } | |
1814 | ||
1815 | /* The next character must be a semicolon. */ | |
1816 | now = lr_token (ldfile, charmap, NULL); | |
1817 | if (now->tok != tok_semicolon) | |
1818 | goto err_label; | |
1819 | goto read_charclass; | |
1820 | ||
1821 | case tok_upper: | |
1822 | case tok_lower: | |
1823 | case tok_alpha: | |
1824 | case tok_alnum: | |
1825 | case tok_space: | |
1826 | case tok_cntrl: | |
1827 | case tok_punct: | |
1828 | case tok_graph: | |
1829 | case tok_print: | |
1830 | case tok_xdigit: | |
1831 | case tok_blank: | |
1832 | class_bit = BITw (now->tok); | |
1833 | class256_bit = BIT (now->tok); | |
1834 | handle_digits = 0; | |
1835 | read_charclass: | |
1836 | ctype->class_done |= class_bit; | |
1837 | last_token = tok_none; | |
1838 | ellipsis_token = tok_none; | |
1839 | now = lr_token (ldfile, charmap, NULL); | |
1840 | while (now->tok != tok_eol && now->tok != tok_eof) | |
1841 | { | |
1842 | uint32_t wch; | |
1843 | struct charseq *seq; | |
1844 | ||
1845 | if (ellipsis_token == tok_none) | |
1846 | { | |
1847 | if (get_character (now, charmap, repertoire, &seq, &wch)) | |
1848 | goto err_label; | |
1849 | ||
1850 | if (!ignore_content && seq != NULL && seq->nbytes == 1) | |
1851 | /* Yep, we can store information about this byte | |
1852 | sequence. */ | |
1853 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
1854 | ||
1855 | if (!ignore_content && wch != ILLEGAL_CHAR_VALUE | |
1856 | && class_bit != 0) | |
1857 | /* We have the UCS4 position. */ | |
1858 | *find_idx (ctype, &ctype->class_collection, | |
1859 | &ctype->class_collection_max, | |
1860 | &ctype->class_collection_act, wch) |= class_bit; | |
1861 | ||
1862 | last_token = now->tok; | |
1863 | last_str = now->val.str.startmb; | |
1864 | last_seq = seq; | |
1865 | last_wch = wch; | |
1866 | memcpy (last_charcode, now->val.charcode.bytes, 16); | |
1867 | last_charcode_len = now->val.charcode.nbytes; | |
1868 | ||
1869 | if (!ignore_content && handle_digits == 1) | |
1870 | { | |
1871 | /* We must store the digit values. */ | |
1872 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1873 | { | |
1874 | ctype->mbdigits_max *= 2; | |
1875 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1876 | (ctype->mbdigits_max | |
1877 | * sizeof (char *))); | |
1878 | ctype->wcdigits_max *= 2; | |
1879 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1880 | (ctype->wcdigits_max | |
1881 | * sizeof (uint32_t))); | |
1882 | } | |
1883 | ||
1884 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1885 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1886 | } | |
1887 | else if (!ignore_content && handle_digits == 2) | |
1888 | { | |
1889 | /* We must store the digit values. */ | |
1890 | if (ctype->outdigits_act >= 10) | |
1891 | { | |
1892 | lr_error (ldfile, _("\ | |
1893 | %s: field `%s' does not contain exactly ten entries"), | |
1894 | "LC_CTYPE", "outdigit"); | |
1895 | goto err_label; | |
1896 | } | |
1897 | ||
1898 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1899 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1900 | ++ctype->outdigits_act; | |
1901 | } | |
1902 | } | |
1903 | else | |
1904 | { | |
1905 | /* Now it gets complicated. We have to resolve the | |
1906 | ellipsis problem. First we must distinguish between | |
1907 | the different kind of ellipsis and this must match the | |
1908 | tokens we have seen. */ | |
1909 | assert (last_token != tok_none); | |
1910 | ||
1911 | if (last_token != now->tok) | |
1912 | { | |
1913 | lr_error (ldfile, _("\ | |
1914 | ellipsis range must be marked by two operands of same type")); | |
1915 | lr_ignore_rest (ldfile, 0); | |
1916 | break; | |
1917 | } | |
1918 | ||
1919 | if (last_token == tok_bsymbol) | |
1920 | { | |
1921 | if (ellipsis_token == tok_ellipsis3) | |
1922 | lr_error (ldfile, _("with symbolic name range values \ | |
1923 | the absolute ellipsis `...' must not be used")); | |
1924 | ||
1925 | charclass_symbolic_ellipsis (ldfile, ctype, charmap, | |
1926 | repertoire, now, last_str, | |
1927 | class256_bit, class_bit, | |
1928 | (ellipsis_token | |
1929 | == tok_ellipsis4 | |
1930 | ? 10 : 16), | |
1931 | ignore_content, | |
1932 | handle_digits); | |
1933 | } | |
1934 | else if (last_token == tok_ucs4) | |
1935 | { | |
1936 | if (ellipsis_token != tok_ellipsis2) | |
1937 | lr_error (ldfile, _("\ | |
1938 | with UCS range values one must use the hexadecimal symbolic ellipsis `..'")); | |
1939 | ||
1940 | charclass_ucs4_ellipsis (ldfile, ctype, charmap, | |
1941 | repertoire, now, last_wch, | |
1942 | class256_bit, class_bit, | |
1943 | ignore_content, handle_digits); | |
1944 | } | |
1945 | else | |
1946 | { | |
1947 | assert (last_token == tok_charcode); | |
1948 | ||
1949 | if (ellipsis_token != tok_ellipsis3) | |
1950 | lr_error (ldfile, _("\ | |
1951 | with character code range values one must use the absolute ellipsis `...'")); | |
1952 | ||
1953 | charclass_charcode_ellipsis (ldfile, ctype, charmap, | |
1954 | repertoire, now, | |
1955 | last_charcode, | |
1956 | last_charcode_len, | |
1957 | class256_bit, class_bit, | |
1958 | ignore_content, | |
1959 | handle_digits); | |
1960 | } | |
1961 | ||
1962 | /* Now we have used the last value. */ | |
1963 | last_token = tok_none; | |
1964 | } | |
1965 | ||
1966 | /* Next we expect a semicolon or the end of the line. */ | |
1967 | now = lr_token (ldfile, charmap, NULL); | |
1968 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1969 | break; | |
1970 | ||
1971 | if (last_token != tok_none | |
1972 | && now->tok >= tok_ellipsis2 && now->tok <= tok_ellipsis4) | |
1973 | { | |
1974 | ellipsis_token = now->tok; | |
1975 | now = lr_token (ldfile, charmap, NULL); | |
1976 | continue; | |
1977 | } | |
1978 | ||
1979 | if (now->tok != tok_semicolon) | |
1980 | goto err_label; | |
1981 | ||
1982 | /* And get the next character. */ | |
1983 | now = lr_token (ldfile, charmap, NULL); | |
1984 | ||
1985 | ellipsis_token = tok_none; | |
1986 | } | |
1987 | break; | |
1988 | ||
1989 | case tok_digit: | |
1990 | handle_tok_digit: | |
1991 | class_bit = _ISwdigit; | |
1992 | class256_bit = _ISdigit; | |
1993 | handle_digits = 1; | |
1994 | goto read_charclass; | |
1995 | ||
1996 | case tok_outdigit: | |
1997 | if (ctype->outdigits_act != 0) | |
1998 | lr_error (ldfile, _("\ | |
1999 | %s: field `%s' declared more than once"), | |
2000 | "LC_CTYPE", "outdigit"); | |
2001 | class_bit = 0; | |
2002 | class256_bit = 0; | |
2003 | handle_digits = 2; | |
2004 | goto read_charclass; | |
2005 | ||
2006 | case tok_toupper: | |
2007 | mapidx = 0; | |
2008 | goto read_mapping; | |
2009 | ||
2010 | case tok_tolower: | |
2011 | mapidx = 1; | |
2012 | goto read_mapping; | |
2013 | ||
2014 | case tok_map: | |
2015 | /* We simply forget the `map' keyword and use the following | |
2016 | operand to determine the mapping. */ | |
2017 | now = lr_token (ldfile, charmap, NULL); | |
2018 | if (now->tok == tok_ident || now->tok == tok_string) | |
2019 | { | |
2020 | size_t cnt; | |
2021 | ||
2022 | for (cnt = 2; cnt < ctype->map_collection_nr; ++cnt) | |
2023 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2024 | break; | |
2025 | ||
2026 | if (cnt < ctype->map_collection_nr) | |
2027 | mapidx = cnt; | |
2028 | else | |
2029 | { | |
2030 | lr_error (ldfile, _("unknown map `%s'"), | |
2031 | now->val.str.startmb); | |
2032 | lr_ignore_rest (ldfile, 0); | |
2033 | break; | |
2034 | } | |
2035 | } | |
2036 | else if (now->tok < tok_toupper || now->tok > tok_tolower) | |
2037 | goto err_label; | |
2038 | else | |
2039 | mapidx = now->tok - tok_toupper; | |
2040 | ||
2041 | now = lr_token (ldfile, charmap, NULL); | |
2042 | /* This better should be a semicolon. */ | |
2043 | if (now->tok != tok_semicolon) | |
2044 | goto err_label; | |
2045 | ||
2046 | read_mapping: | |
2047 | /* Test whether this mapping was already defined. */ | |
2048 | if (ctype->tomap_done[mapidx]) | |
2049 | { | |
2050 | lr_error (ldfile, _("duplicated definition for mapping `%s'"), | |
2051 | ctype->mapnames[mapidx]); | |
2052 | lr_ignore_rest (ldfile, 0); | |
2053 | break; | |
2054 | } | |
2055 | ctype->tomap_done[mapidx] = 1; | |
2056 | ||
2057 | now = lr_token (ldfile, charmap, NULL); | |
2058 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2059 | { | |
2060 | struct charseq *from_seq; | |
2061 | uint32_t from_wch; | |
2062 | struct charseq *to_seq; | |
2063 | uint32_t to_wch; | |
2064 | ||
2065 | /* Every pair starts with an opening brace. */ | |
2066 | if (now->tok != tok_open_brace) | |
2067 | goto err_label; | |
2068 | ||
2069 | /* Next comes the from-value. */ | |
2070 | now = lr_token (ldfile, charmap, NULL); | |
2071 | if (get_character (now, charmap, repertoire, &from_seq, | |
2072 | &from_wch) != 0) | |
2073 | goto err_label; | |
2074 | ||
2075 | /* The next is a comma. */ | |
2076 | now = lr_token (ldfile, charmap, NULL); | |
2077 | if (now->tok != tok_comma) | |
2078 | goto err_label; | |
2079 | ||
2080 | /* And the other value. */ | |
2081 | now = lr_token (ldfile, charmap, NULL); | |
2082 | if (get_character (now, charmap, repertoire, &to_seq, | |
2083 | &to_wch) != 0) | |
2084 | goto err_label; | |
2085 | ||
2086 | /* And the last thing is the closing brace. */ | |
2087 | now = lr_token (ldfile, charmap, NULL); | |
2088 | if (now->tok != tok_close_brace) | |
2089 | goto err_label; | |
2090 | ||
2091 | if (!ignore_content) | |
2092 | { | |
2093 | if (mapidx < 2 && from_seq != NULL && to_seq != NULL | |
2094 | && from_seq->nbytes == 1 && to_seq->nbytes == 1) | |
2095 | /* We can use this value. */ | |
2096 | ctype->map256_collection[mapidx][from_seq->bytes[0]] | |
2097 | = to_seq->bytes[0]; | |
2098 | ||
2099 | if (from_wch != ILLEGAL_CHAR_VALUE | |
2100 | && to_wch != ILLEGAL_CHAR_VALUE) | |
2101 | /* Both correct values. */ | |
2102 | *find_idx (ctype, &ctype->map_collection[mapidx], | |
2103 | &ctype->map_collection_max[mapidx], | |
2104 | &ctype->map_collection_act[mapidx], | |
2105 | from_wch) = to_wch; | |
2106 | } | |
2107 | ||
2108 | /* Now comes a semicolon or the end of the line/file. */ | |
2109 | now = lr_token (ldfile, charmap, NULL); | |
2110 | if (now->tok == tok_semicolon) | |
2111 | now = lr_token (ldfile, charmap, NULL); | |
2112 | } | |
2113 | break; | |
2114 | ||
2115 | case tok_translit_start: | |
2116 | /* The rest of the line better should be empty. */ | |
2117 | lr_ignore_rest (ldfile, 1); | |
2118 | ||
2119 | /* We count here the number of allocated entries in the `translit' | |
2120 | array. */ | |
2121 | cnt = 0; | |
2122 | ||
2123 | /* We proceed until we see the `translit_end' token. */ | |
2124 | while (now = lr_token (ldfile, charmap, repertoire), | |
2125 | now->tok != tok_translit_end && now->tok != tok_eof) | |
2126 | { | |
2127 | if (now->tok == tok_eol) | |
2128 | /* Ignore empty lines. */ | |
2129 | continue; | |
2130 | ||
2131 | if (now->tok == tok_translit_end) | |
2132 | { | |
2133 | lr_ignore_rest (ldfile, 0); | |
2134 | break; | |
2135 | } | |
2136 | ||
2137 | if (now->tok == tok_include) | |
2138 | { | |
2139 | /* We have to include locale. */ | |
2140 | const char *locale_name; | |
2141 | const char *repertoire_name; | |
2142 | ||
2143 | now = lr_token (ldfile, charmap, NULL); | |
2144 | /* This should be a string or an identifier. In any | |
2145 | case something to name a locale. */ | |
2146 | if (now->tok != tok_string && now->tok != tok_ident) | |
2147 | { | |
2148 | translit_syntax: | |
2149 | lr_error (ldfile, _("%s: syntax error"), "LC_CTYPE"); | |
2150 | lr_ignore_rest (ldfile, 0); | |
2151 | continue; | |
2152 | } | |
2153 | locale_name = now->val.str.startmb; | |
2154 | ||
2155 | /* Next should be a semicolon. */ | |
2156 | now = lr_token (ldfile, charmap, NULL); | |
2157 | if (now->tok != tok_semicolon) | |
2158 | goto translit_syntax; | |
2159 | ||
2160 | /* Now the repertoire name. */ | |
2161 | now = lr_token (ldfile, charmap, NULL); | |
2162 | if ((now->tok != tok_string && now->tok != tok_ident) | |
2163 | || now->val.str.startmb == NULL) | |
2164 | goto translit_syntax; | |
2165 | repertoire_name = now->val.str.startmb; | |
2166 | ||
2167 | /* We must not have more than one `include'. */ | |
2168 | if (ctype->translit_copy_locale != NULL) | |
2169 | { | |
2170 | lr_error (ldfile, _("\ | |
2171 | %s: only one `include' instruction allowed"), "LC_CTYPE"); | |
2172 | lr_ignore_rest (ldfile, 0); | |
2173 | continue; | |
2174 | } | |
2175 | ||
2176 | ctype->translit_copy_locale = locale_name; | |
2177 | ctype->translit_copy_repertoire = repertoire_name; | |
2178 | ||
2179 | /* The rest of the line must be empty. */ | |
2180 | lr_ignore_rest (ldfile, 1); | |
2181 | continue; | |
2182 | } | |
2183 | ||
2184 | read_translit_entry (ldfile, ctype, now, charmap, repertoire); | |
2185 | } | |
2186 | break; | |
2187 | ||
2188 | case tok_ident: | |
2189 | /* This could mean one of several things. First test whether | |
2190 | it's a character class name. */ | |
2191 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
2192 | if (strcmp (now->val.str.startmb, ctype->classnames[cnt]) == 0) | |
2193 | break; | |
2194 | if (cnt < ctype->nr_charclass) | |
2195 | { | |
2196 | class_bit = _ISwbit (cnt); | |
2197 | class256_bit = cnt <= 11 ? _ISbit (cnt) : 0; | |
2198 | free (now->val.str.startmb); | |
2199 | goto read_charclass; | |
2200 | } | |
2201 | if (strcmp (now->val.str.startmb, "special1") == 0) | |
2202 | { | |
2203 | class_bit = _ISwspecial1; | |
2204 | free (now->val.str.startmb); | |
2205 | goto read_charclass; | |
2206 | } | |
2207 | if (strcmp (now->val.str.startmb, "special2") == 0) | |
2208 | { | |
2209 | class_bit = _ISwspecial2; | |
2210 | free (now->val.str.startmb); | |
2211 | goto read_charclass; | |
2212 | } | |
2213 | if (strcmp (now->val.str.startmb, "special3") == 0) | |
2214 | { | |
2215 | class_bit = _ISwspecial3; | |
2216 | free (now->val.str.startmb); | |
2217 | goto read_charclass; | |
2218 | } | |
2219 | if (strcmp (now->val.str.startmb, "tosymmetric") == 0) | |
2220 | { | |
2221 | mapidx = 2; | |
2222 | goto read_mapping; | |
2223 | } | |
2224 | break; | |
2225 | ||
2226 | case tok_end: | |
2227 | /* Next we assume `LC_CTYPE'. */ | |
2228 | now = lr_token (ldfile, charmap, NULL); | |
2229 | if (now->tok == tok_eof) | |
2230 | break; | |
2231 | if (now->tok == tok_eol) | |
2232 | lr_error (ldfile, _("%s: incomplete `END' line"), | |
2233 | "LC_CTYPE"); | |
2234 | else if (now->tok != tok_lc_ctype) | |
2235 | lr_error (ldfile, _("\ | |
2236 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2237 | lr_ignore_rest (ldfile, now->tok == tok_lc_ctype); | |
2238 | return; | |
2239 | ||
2240 | default: | |
2241 | err_label: | |
2242 | if (now->tok != tok_eof) | |
2243 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
19bc17a9 RM |
2244 | } |
2245 | ||
4b10dd6c UD |
2246 | /* Prepare for the next round. */ |
2247 | now = lr_token (ldfile, charmap, NULL); | |
2248 | nowtok = now->tok; | |
19bc17a9 RM |
2249 | } |
2250 | ||
4b10dd6c UD |
2251 | /* When we come here we reached the end of the file. */ |
2252 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); | |
19bc17a9 RM |
2253 | } |
2254 | ||
2255 | ||
2256 | static void | |
4b10dd6c UD |
2257 | set_class_defaults (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2258 | struct repertoire_t *repertoire) | |
19bc17a9 | 2259 | { |
4b10dd6c UD |
2260 | size_t cnt; |
2261 | ||
19bc17a9 RM |
2262 | /* These function defines the default values for the classes and conversions |
2263 | according to POSIX.2 2.5.2.1. | |
2264 | It may seem that the order of these if-blocks is arbitrary but it is NOT. | |
2265 | Don't move them unless you know what you do! */ | |
2266 | ||
4b10dd6c | 2267 | void set_default (int bitpos, int from, int to) |
19bc17a9 RM |
2268 | { |
2269 | char tmp[2]; | |
2270 | int ch; | |
4b10dd6c UD |
2271 | int bit = _ISbit (bitpos); |
2272 | int bitw = _ISwbit (bitpos); | |
19bc17a9 RM |
2273 | /* Define string. */ |
2274 | strcpy (tmp, "?"); | |
2275 | ||
2276 | for (ch = from; ch <= to; ++ch) | |
2277 | { | |
4b10dd6c UD |
2278 | uint32_t value; |
2279 | struct charseq *seq; | |
19bc17a9 RM |
2280 | tmp[0] = ch; |
2281 | ||
4b10dd6c UD |
2282 | value = repertoire_find_value (repertoire, tmp, 1); |
2283 | if (value == ILLEGAL_CHAR_VALUE) | |
19bc17a9 | 2284 | { |
880f421f UD |
2285 | if (!be_quiet) |
2286 | error (0, 0, _("\ | |
4b10dd6c UD |
2287 | %s: character `%s' not defined in repertoire while needed as default value"), |
2288 | "LC_CTYPE", tmp); | |
2289 | } | |
2290 | else | |
2291 | ELEM (ctype, class_collection, , value) |= bitw; | |
2292 | ||
2293 | seq = charmap_find_value (charmap, tmp, 1); | |
2294 | if (seq == NULL) | |
2295 | { | |
2296 | if (!be_quiet) | |
2297 | error (0, 0, _("\ | |
2298 | %s: character `%s' not defined in charmap while needed as default value"), | |
2299 | "LC_CTYPE", tmp); | |
19bc17a9 | 2300 | } |
4b10dd6c UD |
2301 | else if (seq->nbytes != 1) |
2302 | error (0, 0, _("\ | |
2303 | %s: character `%s' in charmap not representable with one byte"), | |
2304 | "LC_CTYPE", tmp); | |
19bc17a9 | 2305 | else |
4b10dd6c | 2306 | ctype->class256_collection[seq->bytes[0]] |= bit; |
19bc17a9 RM |
2307 | } |
2308 | } | |
2309 | ||
2310 | /* Set default values if keyword was not present. */ | |
4b10dd6c | 2311 | if ((ctype->class_done & BITw (tok_upper)) == 0) |
19bc17a9 RM |
2312 | /* "If this keyword [lower] is not specified, the lowercase letters |
2313 | `A' through `Z', ..., shall automatically belong to this class, | |
2314 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2315 | set_default (BITPOS (tok_upper), 'A', 'Z'); |
19bc17a9 | 2316 | |
4b10dd6c | 2317 | if ((ctype->class_done & BITw (tok_lower)) == 0) |
19bc17a9 RM |
2318 | /* "If this keyword [lower] is not specified, the lowercase letters |
2319 | `a' through `z', ..., shall automatically belong to this class, | |
2320 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2321 | set_default (BITPOS (tok_lower), 'a', 'z'); |
19bc17a9 | 2322 | |
4b10dd6c | 2323 | if ((ctype->class_done & BITw (tok_alpha)) == 0) |
19bc17a9 RM |
2324 | { |
2325 | /* Table 2-6 in P1003.2 says that characters in class `upper' or | |
2326 | class `lower' *must* be in class `alpha'. */ | |
2327 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower); | |
19bc17a9 RM |
2328 | |
2329 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2330 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2331 | ctype->class_collection[cnt] |= BIT (tok_alpha); | |
2332 | } | |
2333 | ||
4b10dd6c | 2334 | if ((ctype->class_done & BITw (tok_digit)) == 0) |
19bc17a9 RM |
2335 | /* "If this keyword [digit] is not specified, the digits `0' through |
2336 | `9', ..., shall automatically belong to this class, with | |
2337 | implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2338 | set_default (BITPOS (tok_digit), '0', '9'); |
19bc17a9 RM |
2339 | |
2340 | /* "Only characters specified for the `alpha' and `digit' keyword | |
2341 | shall be specified. Characters specified for the keyword `alpha' | |
2342 | and `digit' are automatically included in this class. */ | |
2343 | { | |
2344 | unsigned long int mask = BIT (tok_alpha) | BIT (tok_digit); | |
19bc17a9 RM |
2345 | |
2346 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2347 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2348 | ctype->class_collection[cnt] |= BIT (tok_alnum); | |
2349 | } | |
2350 | ||
4b10dd6c | 2351 | if ((ctype->class_done & BITw (tok_space)) == 0) |
19bc17a9 RM |
2352 | /* "If this keyword [space] is not specified, the characters <space>, |
2353 | <form-feed>, <newline>, <carriage-return>, <tab>, and | |
2354 | <vertical-tab>, ..., shall automatically belong to this class, | |
2355 | with implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
2356 | { | |
4b10dd6c UD |
2357 | uint32_t value; |
2358 | struct charseq *seq; | |
19bc17a9 | 2359 | |
4b10dd6c UD |
2360 | value = repertoire_find_value (repertoire, "space", 5); |
2361 | if (value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
2362 | { |
2363 | if (!be_quiet) | |
2364 | error (0, 0, _("\ | |
4b10dd6c UD |
2365 | %s: character `%s' not defined while needed as default value"), |
2366 | "LC_CTYPE", "<space>"); | |
880f421f | 2367 | } |
19bc17a9 RM |
2368 | else |
2369 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2370 | ||
4b10dd6c UD |
2371 | seq = charmap_find_value (charmap, "space", 5); |
2372 | if (seq == NULL) | |
880f421f UD |
2373 | { |
2374 | if (!be_quiet) | |
2375 | error (0, 0, _("\ | |
4b10dd6c UD |
2376 | %s: character `%s' not defined while needed as default value"), |
2377 | "LC_CTYPE", "<space>"); | |
2378 | } | |
2379 | else if (seq->nbytes != 1) | |
2380 | error (0, 0, _("\ | |
2381 | %s: character `%s' in charmap not representable with one byte"), | |
2382 | "LC_CTYPE", "<space>"); | |
2383 | else | |
2384 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2385 | ||
2386 | ||
2387 | value = repertoire_find_value (repertoire, "form-feed", 9); | |
2388 | if (value == ILLEGAL_CHAR_VALUE) | |
2389 | { | |
2390 | if (!be_quiet) | |
2391 | error (0, 0, _("\ | |
2392 | %s: character `%s' not defined while needed as default value"), | |
2393 | "LC_CTYPE", "<form-feed>"); | |
880f421f | 2394 | } |
19bc17a9 RM |
2395 | else |
2396 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2397 | ||
4b10dd6c UD |
2398 | seq = charmap_find_value (charmap, "form-feed", 9); |
2399 | if (seq == NULL) | |
880f421f UD |
2400 | { |
2401 | if (!be_quiet) | |
2402 | error (0, 0, _("\ | |
4b10dd6c UD |
2403 | %s: character `%s' not defined while needed as default value"), |
2404 | "LC_CTYPE", "<form-feed>"); | |
2405 | } | |
2406 | else if (seq->nbytes != 1) | |
2407 | error (0, 0, _("\ | |
2408 | %s: character `%s' in charmap not representable with one byte"), | |
2409 | "LC_CTYPE", "<form-feed>"); | |
2410 | else | |
2411 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2412 | ||
2413 | ||
2414 | value = repertoire_find_value (repertoire, "newline", 7); | |
2415 | if (value == ILLEGAL_CHAR_VALUE) | |
2416 | { | |
2417 | if (!be_quiet) | |
2418 | error (0, 0, _("\ | |
2419 | %s: character `%s' not defined while needed as default value"), | |
2420 | "LC_CTYPE", "<newline>"); | |
880f421f | 2421 | } |
19bc17a9 RM |
2422 | else |
2423 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2424 | ||
4b10dd6c UD |
2425 | seq = charmap_find_value (charmap, "newline", 7); |
2426 | if (seq == NULL) | |
880f421f UD |
2427 | { |
2428 | if (!be_quiet) | |
2429 | error (0, 0, _("\ | |
19bc17a9 | 2430 | character `%s' not defined while needed as default value"), |
4b10dd6c UD |
2431 | "<newline>"); |
2432 | } | |
2433 | else if (seq->nbytes != 1) | |
2434 | error (0, 0, _("\ | |
2435 | %s: character `%s' in charmap not representable with one byte"), | |
2436 | "LC_CTYPE", "<newline>"); | |
2437 | else | |
2438 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2439 | ||
2440 | ||
2441 | value = repertoire_find_value (repertoire, "carriage-return", 15); | |
2442 | if (value == ILLEGAL_CHAR_VALUE) | |
2443 | { | |
2444 | if (!be_quiet) | |
2445 | error (0, 0, _("\ | |
2446 | %s: character `%s' not defined while needed as default value"), | |
2447 | "LC_CTYPE", "<carriage-return>"); | |
880f421f | 2448 | } |
19bc17a9 RM |
2449 | else |
2450 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2451 | ||
4b10dd6c UD |
2452 | seq = charmap_find_value (charmap, "carriage-return", 15); |
2453 | if (seq == NULL) | |
880f421f UD |
2454 | { |
2455 | if (!be_quiet) | |
2456 | error (0, 0, _("\ | |
4b10dd6c UD |
2457 | %s: character `%s' not defined while needed as default value"), |
2458 | "LC_CTYPE", "<carriage-return>"); | |
2459 | } | |
2460 | else if (seq->nbytes != 1) | |
2461 | error (0, 0, _("\ | |
2462 | %s: character `%s' in charmap not representable with one byte"), | |
2463 | "LC_CTYPE", "<carriage-return>"); | |
2464 | else | |
2465 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2466 | ||
2467 | ||
2468 | value = repertoire_find_value (repertoire, "tab", 3); | |
2469 | if (value == ILLEGAL_CHAR_VALUE) | |
2470 | { | |
2471 | if (!be_quiet) | |
2472 | error (0, 0, _("\ | |
2473 | %s: character `%s' not defined while needed as default value"), | |
2474 | "LC_CTYPE", "<tab>"); | |
880f421f | 2475 | } |
19bc17a9 RM |
2476 | else |
2477 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
2478 | ||
4b10dd6c UD |
2479 | seq = charmap_find_value (charmap, "tab", 3); |
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", "<tab>"); | |
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", "<tab>"); | |
2491 | else | |
2492 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2493 | ||
2494 | ||
2495 | value = repertoire_find_value (repertoire, "vertical-tab", 12); | |
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", "<vertical-tab>"); | |
880f421f | 2502 | } |
19bc17a9 RM |
2503 | else |
2504 | ELEM (ctype, class_collection, , value) |= BIT (tok_space); | |
4b10dd6c UD |
2505 | |
2506 | seq = charmap_find_value (charmap, "vertical-tab", 12); | |
2507 | if (seq == NULL) | |
2508 | { | |
2509 | if (!be_quiet) | |
2510 | error (0, 0, _("\ | |
2511 | %s: character `%s' not defined while needed as default value"), | |
2512 | "LC_CTYPE", "<vertical-tab>"); | |
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", "<vertical-tab>"); | |
2518 | else | |
2519 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
19bc17a9 RM |
2520 | } |
2521 | ||
4b10dd6c | 2522 | if ((ctype->class_done & BITw (tok_xdigit)) == 0) |
19bc17a9 RM |
2523 | /* "If this keyword is not specified, the digits `0' to `9', the |
2524 | uppercase letters `A' through `F', and the lowercase letters `a' | |
2525 | through `f', ..., shell automatically belong to this class, with | |
2526 | implementation defined character values." [P1003.2, 2.5.2.1] */ | |
2527 | { | |
4b10dd6c UD |
2528 | set_default (BITPOS (tok_xdigit), '0', '9'); |
2529 | set_default (BITPOS (tok_xdigit), 'A', 'F'); | |
2530 | set_default (BITPOS (tok_xdigit), 'a', 'f'); | |
19bc17a9 RM |
2531 | } |
2532 | ||
4b10dd6c | 2533 | if ((ctype->class_done & BITw (tok_blank)) == 0) |
19bc17a9 RM |
2534 | /* "If this keyword [blank] is unspecified, the characters <space> and |
2535 | <tab> shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
2536 | { | |
4b10dd6c UD |
2537 | uint32_t value; |
2538 | struct charseq *seq; | |
19bc17a9 | 2539 | |
4b10dd6c UD |
2540 | value = repertoire_find_value (repertoire, "space", 5); |
2541 | if (value == ILLEGAL_CHAR_VALUE) | |
880f421f UD |
2542 | { |
2543 | if (!be_quiet) | |
2544 | error (0, 0, _("\ | |
4b10dd6c UD |
2545 | %s: character `%s' not defined while needed as default value"), |
2546 | "LC_CTYPE", "<space>"); | |
880f421f | 2547 | } |
19bc17a9 RM |
2548 | else |
2549 | ELEM (ctype, class_collection, , value) |= BIT (tok_blank); | |
2550 | ||
4b10dd6c UD |
2551 | seq = charmap_find_value (charmap, "space", 5); |
2552 | if (seq == NULL) | |
880f421f UD |
2553 | { |
2554 | if (!be_quiet) | |
2555 | error (0, 0, _("\ | |
4b10dd6c UD |
2556 | %s: character `%s' not defined while needed as default value"), |
2557 | "LC_CTYPE", "<space>"); | |
2558 | } | |
2559 | else if (seq->nbytes != 1) | |
2560 | error (0, 0, _("\ | |
2561 | %s: character `%s' in charmap not representable with one byte"), | |
2562 | "LC_CTYPE", "<space>"); | |
2563 | else | |
2564 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
2565 | ||
2566 | ||
2567 | value = repertoire_find_value (repertoire, "tab", 3); | |
2568 | if (value == ILLEGAL_CHAR_VALUE) | |
2569 | { | |
2570 | if (!be_quiet) | |
2571 | error (0, 0, _("\ | |
2572 | %s: character `%s' not defined while needed as default value"), | |
2573 | "LC_CTYPE", "<tab>"); | |
880f421f | 2574 | } |
19bc17a9 RM |
2575 | else |
2576 | ELEM (ctype, class_collection, , value) |= BIT (tok_blank); | |
4b10dd6c UD |
2577 | |
2578 | seq = charmap_find_value (charmap, "tab", 3); | |
2579 | if (seq == NULL) | |
2580 | { | |
2581 | if (!be_quiet) | |
2582 | error (0, 0, _("\ | |
2583 | %s: character `%s' not defined while needed as default value"), | |
2584 | "LC_CTYPE", "<tab>"); | |
2585 | } | |
2586 | else if (seq->nbytes != 1) | |
2587 | error (0, 0, _("\ | |
2588 | %s: character `%s' in charmap not representable with one byte"), | |
2589 | "LC_CTYPE", "<tab>"); | |
2590 | else | |
2591 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
19bc17a9 RM |
2592 | } |
2593 | ||
4b10dd6c | 2594 | if ((ctype->class_done & BITw (tok_graph)) == 0) |
19bc17a9 RM |
2595 | /* "If this keyword [graph] is not specified, characters specified for |
2596 | the keywords `upper', `lower', `alpha', `digit', `xdigit' and `punct', | |
2597 | shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
2598 | { | |
2599 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
2600 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
2601 | size_t cnt; | |
2602 | ||
2603 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2604 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2605 | ctype->class_collection[cnt] |= BIT (tok_graph); | |
4b10dd6c UD |
2606 | |
2607 | for (cnt = 0; cnt < 256; ++cnt) | |
2608 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2609 | ctype->class256_collection[cnt] |= BIT (tok_graph); | |
19bc17a9 RM |
2610 | } |
2611 | ||
4b10dd6c | 2612 | if ((ctype->class_done & BITw (tok_print)) == 0) |
19bc17a9 RM |
2613 | /* "If this keyword [print] is not provided, characters specified for |
2614 | the keywords `upper', `lower', `alpha', `digit', `xdigit', `punct', | |
2615 | and the <space> character shall belong to this character class." | |
2616 | [P1003.2, 2.5.2.1] */ | |
2617 | { | |
2618 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
2619 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
2620 | size_t cnt; | |
4b10dd6c UD |
2621 | uint32_t space; |
2622 | struct charseq *seq; | |
19bc17a9 RM |
2623 | |
2624 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
2625 | if ((ctype->class_collection[cnt] & mask) != 0) | |
2626 | ctype->class_collection[cnt] |= BIT (tok_print); | |
2627 | ||
4b10dd6c UD |
2628 | for (cnt = 0; cnt < 256; ++cnt) |
2629 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2630 | ctype->class256_collection[cnt] |= BIT (tok_print); | |
2631 | ||
2632 | ||
2633 | space = repertoire_find_value (repertoire, "space", 5); | |
880f421f UD |
2634 | if (space == ILLEGAL_CHAR_VALUE) |
2635 | { | |
2636 | if (!be_quiet) | |
2637 | error (0, 0, _("\ | |
4b10dd6c UD |
2638 | %s: character `%s' not defined while needed as default value"), |
2639 | "LC_CTYPE", "<space>"); | |
880f421f | 2640 | } |
19bc17a9 RM |
2641 | else |
2642 | ELEM (ctype, class_collection, , space) |= BIT (tok_print); | |
4b10dd6c UD |
2643 | |
2644 | seq = charmap_find_value (charmap, "space", 5); | |
2645 | if (seq == NULL) | |
2646 | { | |
2647 | if (!be_quiet) | |
2648 | error (0, 0, _("\ | |
2649 | %s: character `%s' not defined while needed as default value"), | |
2650 | "LC_CTYPE", "<space>"); | |
2651 | } | |
2652 | else if (seq->nbytes != 1) | |
2653 | error (0, 0, _("\ | |
2654 | %s: character `%s' in charmap not representable with one byte"), | |
2655 | "LC_CTYPE", "<space>"); | |
2656 | else | |
2657 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_print); | |
19bc17a9 RM |
2658 | } |
2659 | ||
4b10dd6c | 2660 | if (ctype->tomap_done[0] == 0) |
6d52618b | 2661 | /* "If this keyword [toupper] is not specified, the lowercase letters |
19bc17a9 RM |
2662 | `a' through `z', and their corresponding uppercase letters `A' to |
2663 | `Z', ..., shall automatically be included, with implementation- | |
2664 | defined character values." [P1003.2, 2.5.2.1] */ | |
2665 | { | |
2666 | char tmp[4]; | |
2667 | int ch; | |
2668 | ||
2669 | strcpy (tmp, "<?>"); | |
2670 | ||
2671 | for (ch = 'a'; ch <= 'z'; ++ch) | |
2672 | { | |
4b10dd6c UD |
2673 | uint32_t value_from, value_to; |
2674 | struct charseq *seq_from, *seq_to; | |
19bc17a9 RM |
2675 | |
2676 | tmp[1] = (char) ch; | |
2677 | ||
4b10dd6c UD |
2678 | value_from = repertoire_find_value (repertoire, &tmp[1], 1); |
2679 | if (value_from == ILLEGAL_CHAR_VALUE) | |
19bc17a9 | 2680 | { |
880f421f UD |
2681 | if (!be_quiet) |
2682 | error (0, 0, _("\ | |
4b10dd6c UD |
2683 | %s: character `%s' not defined while needed as default value"), |
2684 | "LC_CTYPE", tmp); | |
2685 | } | |
2686 | else | |
2687 | { | |
2688 | /* This conversion is implementation defined. */ | |
2689 | tmp[1] = (char) (ch + ('A' - 'a')); | |
2690 | value_to = repertoire_find_value (repertoire, &tmp[1], 1); | |
2691 | if (value_to == ILLEGAL_CHAR_VALUE) | |
2692 | { | |
2693 | if (!be_quiet) | |
2694 | error (0, 0, _("\ | |
2695 | %s: character `%s' not defined while needed as default value"), | |
2696 | "LC_CTYPE", tmp); | |
2697 | } | |
2698 | else | |
2699 | /* The index [0] is determined by the order of the | |
2700 | `ctype_map_newP' calls in `ctype_startup'. */ | |
2701 | ELEM (ctype, map_collection, [0], value_from) = value_to; | |
19bc17a9 RM |
2702 | } |
2703 | ||
4b10dd6c UD |
2704 | seq_from = charmap_find_value (charmap, &tmp[1], 1); |
2705 | if (seq_from == NULL) | |
19bc17a9 | 2706 | { |
880f421f UD |
2707 | if (!be_quiet) |
2708 | error (0, 0, _("\ | |
4b10dd6c UD |
2709 | %s: character `%s' not defined while needed as default value"), |
2710 | "LC_CTYPE", tmp); | |
2711 | } | |
2712 | else if (seq_from->nbytes != 1) | |
2713 | { | |
2714 | if (!be_quiet) | |
2715 | error (0, 0, _("\ | |
2716 | %s: character `%s' needed as default value not representable with one byte"), | |
2717 | "LC_CTYPE", tmp); | |
2718 | } | |
2719 | else | |
2720 | { | |
2721 | /* This conversion is implementation defined. */ | |
2722 | tmp[1] = (char) (ch + ('A' - 'a')); | |
2723 | seq_to = charmap_find_value (charmap, &tmp[1], 1); | |
2724 | if (seq_to == NULL) | |
2725 | { | |
2726 | if (!be_quiet) | |
2727 | error (0, 0, _("\ | |
2728 | %s: character `%s' not defined while needed as default value"), | |
2729 | "LC_CTYPE", tmp); | |
2730 | } | |
2731 | else if (seq_to->nbytes != 1) | |
2732 | { | |
2733 | if (!be_quiet) | |
2734 | error (0, 0, _("\ | |
2735 | %s: character `%s' needed as default value not representable with one byte"), | |
2736 | "LC_CTYPE", tmp); | |
2737 | } | |
2738 | else | |
2739 | /* The index [0] is determined by the order of the | |
2740 | `ctype_map_newP' calls in `ctype_startup'. */ | |
2741 | ctype->map256_collection[0][seq_from->bytes[0]] | |
2742 | = seq_to->bytes[0]; | |
19bc17a9 | 2743 | } |
19bc17a9 RM |
2744 | } |
2745 | } | |
2746 | ||
4b10dd6c | 2747 | if (ctype->tomap_done[1] == 0) |
19bc17a9 RM |
2748 | /* "If this keyword [tolower] is not specified, the mapping shall be |
2749 | the reverse mapping of the one specified to `toupper'." [P1003.2] */ | |
2750 | { | |
19bc17a9 RM |
2751 | for (cnt = 0; cnt < ctype->map_collection_act[0]; ++cnt) |
2752 | if (ctype->map_collection[0][cnt] != 0) | |
2753 | ELEM (ctype, map_collection, [1], | |
2754 | ctype->map_collection[0][cnt]) | |
2755 | = ctype->charnames[cnt]; | |
4b10dd6c UD |
2756 | |
2757 | for (cnt = 0; cnt < 256; ++cnt) | |
2758 | if (ctype->map256_collection[0][cnt] != 0) | |
2759 | ctype->map_collection[1][ctype->map_collection[0][cnt]] | |
2760 | = ctype->charnames[cnt]; | |
2761 | } | |
2762 | ||
2763 | if (ctype->outdigits_act == 0) | |
2764 | { | |
2765 | for (cnt = 0; cnt < 10; ++cnt) | |
2766 | { | |
2767 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
2768 | digits + cnt, 1); | |
2769 | ||
2770 | if (ctype->mboutdigits[cnt] == NULL) | |
2771 | { | |
2772 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
2773 | longnames[cnt], | |
2774 | strlen (longnames[cnt])); | |
2775 | ||
2776 | if (ctype->mboutdigits[cnt] == NULL) | |
2777 | { | |
2778 | /* Provide a replacement. */ | |
2779 | error (0, 0, _("\ | |
2780 | no output digits defined and none of the standard names in the charmap")); | |
2781 | ||
2782 | ctype->mboutdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
2783 | sizeof (struct charseq) + 1); | |
2784 | ||
2785 | /* This is better than nothing. */ | |
2786 | ctype->mboutdigits[cnt]->bytes[0] = digits[cnt]; | |
2787 | ctype->mboutdigits[cnt]->nbytes = 1; | |
2788 | } | |
2789 | } | |
2790 | } | |
2791 | ||
2792 | ctype->outdigits_act = 10; | |
19bc17a9 RM |
2793 | } |
2794 | } | |
2795 | ||
2796 | ||
2797 | static void | |
4b10dd6c UD |
2798 | allocate_arrays (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2799 | struct repertoire_t *repertoire) | |
19bc17a9 RM |
2800 | { |
2801 | size_t idx; | |
2802 | ||
6d52618b UD |
2803 | /* First we have to decide how we organize the arrays. It is easy |
2804 | for a one-byte character set. But multi-byte character set | |
2805 | cannot be stored flat because the chars might be sparsely used. | |
2806 | So we determine an optimal hashing function for the used | |
2807 | characters. | |
2808 | ||
2809 | We use a very trivial hashing function to store the sparse | |
2810 | table. CH % TABSIZE is used as an index. To solve multiple hits | |
2811 | we have N planes. This guarantees a fixed search time for a | |
2812 | character [N / 2]. In the following code we determine the minmum | |
2813 | value for TABSIZE * N, where TABSIZE >= 256. */ | |
19bc17a9 RM |
2814 | size_t min_total = UINT_MAX; |
2815 | size_t act_size = 256; | |
2816 | ||
c84142e8 UD |
2817 | if (!be_quiet) |
2818 | fputs (_("\ | |
19bc17a9 | 2819 | Computing table size for character classes might take a while..."), |
c84142e8 | 2820 | stderr); |
19bc17a9 RM |
2821 | |
2822 | while (act_size < min_total) | |
2823 | { | |
2824 | size_t cnt[act_size]; | |
2825 | size_t act_planes = 1; | |
2826 | ||
2827 | memset (cnt, '\0', sizeof cnt); | |
2828 | ||
2829 | for (idx = 0; idx < 256; ++idx) | |
2830 | cnt[idx] = 1; | |
2831 | ||
2832 | for (idx = 0; idx < ctype->charnames_act; ++idx) | |
2833 | if (ctype->charnames[idx] >= 256) | |
2834 | { | |
2835 | size_t nr = ctype->charnames[idx] % act_size; | |
2836 | ||
2837 | if (++cnt[nr] > act_planes) | |
2838 | { | |
2839 | act_planes = cnt[nr]; | |
2840 | if (act_size * act_planes >= min_total) | |
2841 | break; | |
2842 | } | |
2843 | } | |
2844 | ||
2845 | if (act_size * act_planes < min_total) | |
2846 | { | |
2847 | min_total = act_size * act_planes; | |
2848 | ctype->plane_size = act_size; | |
2849 | ctype->plane_cnt = act_planes; | |
2850 | } | |
2851 | ||
2852 | ++act_size; | |
2853 | } | |
2854 | ||
c84142e8 UD |
2855 | if (!be_quiet) |
2856 | fputs (_(" done\n"), stderr); | |
19bc17a9 | 2857 | |
75cd5204 | 2858 | |
19bc17a9 RM |
2859 | #if __BYTE_ORDER == __LITTLE_ENDIAN |
2860 | # define NAMES_B1 ctype->names_el | |
2861 | # define NAMES_B2 ctype->names_eb | |
2862 | #else | |
2863 | # define NAMES_B1 ctype->names_eb | |
2864 | # define NAMES_B2 ctype->names_el | |
2865 | #endif | |
2866 | ||
4b10dd6c UD |
2867 | ctype->names_eb = (uint32_t *) xcalloc (ctype->plane_size |
2868 | * ctype->plane_cnt, | |
2869 | sizeof (uint32_t)); | |
2870 | ctype->names_el = (uint32_t *) xcalloc (ctype->plane_size | |
2871 | * ctype->plane_cnt, | |
2872 | sizeof (uint32_t)); | |
19bc17a9 RM |
2873 | |
2874 | for (idx = 1; idx < 256; ++idx) | |
2875 | NAMES_B1[idx] = idx; | |
2876 | ||
2877 | /* Trick: change the 0th entry's name to 1 to mark the cell occupied. */ | |
2878 | NAMES_B1[0] = 1; | |
2879 | ||
2880 | for (idx = 256; idx < ctype->charnames_act; ++idx) | |
2881 | { | |
2882 | size_t nr = (ctype->charnames[idx] % ctype->plane_size); | |
2883 | size_t depth = 0; | |
2884 | ||
2885 | while (NAMES_B1[nr + depth * ctype->plane_size]) | |
2886 | ++depth; | |
2887 | assert (depth < ctype->plane_cnt); | |
2888 | ||
2889 | NAMES_B1[nr + depth * ctype->plane_size] = ctype->charnames[idx]; | |
2890 | ||
2891 | /* Now for faster access remember the index in the NAMES_B array. */ | |
2892 | ctype->charnames[idx] = nr + depth * ctype->plane_size; | |
2893 | } | |
2894 | NAMES_B1[0] = 0; | |
2895 | ||
2896 | for (idx = 0; idx < ctype->plane_size * ctype->plane_cnt; ++idx) | |
4b10dd6c | 2897 | NAMES_B2[idx] = bswap_32 (NAMES_B1[idx]); |
19bc17a9 RM |
2898 | |
2899 | ||
2900 | /* You wonder about this amount of memory? This is only because some | |
2901 | users do not manage to address the array with unsigned values or | |
2902 | data types with range >= 256. '\200' would result in the array | |
2903 | index -128. To help these poor people we duplicate the entries for | |
2904 | 128 up to 255 below the entry for \0. */ | |
2905 | ctype->ctype_b = (char_class_t *) xcalloc (256 + 128, | |
2906 | sizeof (char_class_t)); | |
2907 | ctype->ctype32_b = (char_class32_t *) xcalloc (ctype->plane_size | |
2908 | * ctype->plane_cnt, | |
2909 | sizeof (char_class32_t)); | |
2910 | ||
2911 | /* Fill in the character class information. */ | |
2912 | #if __BYTE_ORDER == __LITTLE_ENDIAN | |
2913 | # define TRANS(w) CHAR_CLASS_TRANS (w) | |
2914 | # define TRANS32(w) CHAR_CLASS32_TRANS (w) | |
2915 | #else | |
2916 | # define TRANS(w) (w) | |
2917 | # define TRANS32(w) (w) | |
2918 | #endif | |
2919 | ||
4b10dd6c UD |
2920 | /* This is the array accessed usig the multibyte string elements. */ |
2921 | for (idx = 0; idx < 256; ++idx) | |
2922 | ctype->ctype_b[128 + idx] = TRANS (ctype->class256_collection[idx]); | |
19bc17a9 | 2923 | |
75cd5204 RM |
2924 | /* Mirror first 127 entries. We must take care that entry -1 is not |
2925 | mirrored because EOF == -1. */ | |
2926 | for (idx = 0; idx < 127; ++idx) | |
19bc17a9 RM |
2927 | ctype->ctype_b[idx] = ctype->ctype_b[256 + idx]; |
2928 | ||
2929 | /* The 32 bit array contains all characters. */ | |
2930 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
2931 | ctype->ctype32_b[ctype->charnames[idx]] | |
2932 | = TRANS32 (ctype->class_collection[idx]); | |
2933 | ||
2934 | /* Room for table of mappings. */ | |
4b10dd6c UD |
2935 | ctype->map_eb = (uint32_t **) xmalloc (ctype->map_collection_nr |
2936 | * sizeof (uint32_t *)); | |
2937 | ctype->map_el = (uint32_t **) xmalloc (ctype->map_collection_nr | |
2938 | * sizeof (uint32_t *)); | |
19bc17a9 RM |
2939 | |
2940 | /* Fill in all mappings. */ | |
2941 | for (idx = 0; idx < ctype->map_collection_nr; ++idx) | |
2942 | { | |
2943 | unsigned int idx2; | |
2944 | ||
2945 | /* Allocate table. */ | |
4b10dd6c UD |
2946 | ctype->map_eb[idx] = (uint32_t *) xmalloc ((ctype->plane_size |
2947 | * ctype->plane_cnt + 128) | |
2948 | * sizeof (uint32_t)); | |
2949 | ctype->map_el[idx] = (uint32_t *) xmalloc ((ctype->plane_size | |
2950 | * ctype->plane_cnt + 128) | |
2951 | * sizeof (uint32_t)); | |
19bc17a9 RM |
2952 | |
2953 | #if __BYTE_ORDER == __LITTLE_ENDIAN | |
2954 | # define MAP_B1 ctype->map_el | |
2955 | # define MAP_B2 ctype->map_eb | |
2956 | #else | |
2957 | # define MAP_B1 ctype->map_eb | |
2958 | # define MAP_B2 ctype->map_el | |
2959 | #endif | |
2960 | ||
2961 | /* Copy default value (identity mapping). */ | |
2962 | memcpy (&MAP_B1[idx][128], NAMES_B1, | |
4b10dd6c | 2963 | ctype->plane_size * ctype->plane_cnt * sizeof (uint32_t)); |
19bc17a9 RM |
2964 | |
2965 | /* Copy values from collection. */ | |
4b10dd6c UD |
2966 | for (idx2 = 0; idx2 < 256; ++idx2) |
2967 | MAP_B1[idx][128 + idx2] = ctype->map256_collection[idx][idx2]; | |
19bc17a9 | 2968 | |
75cd5204 RM |
2969 | /* Mirror first 127 entries. We must take care not to map entry |
2970 | -1 because EOF == -1. */ | |
2971 | for (idx2 = 0; idx2 < 127; ++idx2) | |
19bc17a9 RM |
2972 | MAP_B1[idx][idx2] = MAP_B1[idx][256 + idx2]; |
2973 | ||
75cd5204 RM |
2974 | /* EOF must map to EOF. */ |
2975 | MAP_B1[idx][127] = EOF; | |
19bc17a9 RM |
2976 | |
2977 | /* And now the other byte order. */ | |
2978 | for (idx2 = 0; idx2 < ctype->plane_size * ctype->plane_cnt + 128; ++idx2) | |
4b10dd6c | 2979 | MAP_B2[idx][idx2] = bswap_32 (MAP_B1[idx][idx2]); |
19bc17a9 RM |
2980 | } |
2981 | ||
2982 | /* Extra array for class and map names. */ | |
4b10dd6c UD |
2983 | ctype->class_name_ptr = (uint32_t *) xmalloc (ctype->nr_charclass |
2984 | * sizeof (uint32_t)); | |
2985 | ctype->map_name_ptr = (uint32_t *) xmalloc (ctype->map_collection_nr | |
2986 | * sizeof (uint32_t)); | |
75cd5204 RM |
2987 | |
2988 | /* Array for width information. Because the expected width are very | |
2989 | small we use only one single byte. This save space and we need | |
2990 | not provide the information twice with both endianesses. */ | |
2991 | ctype->width = (unsigned char *) xmalloc (ctype->plane_size | |
2992 | * ctype->plane_cnt); | |
2993 | /* Initialize with default width value. */ | |
4b10dd6c | 2994 | memset (ctype->width, charmap->width_default, |
75cd5204 | 2995 | ctype->plane_size * ctype->plane_cnt); |
4b10dd6c | 2996 | if (charmap->width_rules != NULL) |
75cd5204 | 2997 | { |
4b10dd6c | 2998 | #if 0 |
75cd5204 RM |
2999 | size_t cnt; |
3000 | ||
4b10dd6c UD |
3001 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
3002 | if (charmap->width_rules[cnt].width != charmap->width_default) | |
3003 | for (idx = charmap->width_rules[cnt].from; | |
3004 | idx <= charmap->width_rules[cnt].to; ++idx) | |
75cd5204 RM |
3005 | { |
3006 | size_t nr = idx % ctype->plane_size; | |
3007 | size_t depth = 0; | |
3008 | ||
3009 | while (NAMES_B1[nr + depth * ctype->plane_size] != nr) | |
3010 | ++depth; | |
3011 | assert (depth < ctype->plane_cnt); | |
3012 | ||
3013 | ctype->width[nr + depth * ctype->plane_size] | |
4b10dd6c | 3014 | = charmap->width_rules[cnt].width; |
75cd5204 | 3015 | } |
4b10dd6c UD |
3016 | #else |
3017 | abort (); | |
3018 | #endif | |
75cd5204 | 3019 | } |
0200214b | 3020 | |
4b10dd6c UD |
3021 | /* Set MB_CUR_MAX. */ |
3022 | ctype->mb_cur_max = charmap->mb_cur_max; | |
6990326c RM |
3023 | |
3024 | /* We need the name of the currently used 8-bit character set to | |
3025 | make correct conversion between this 8-bit representation and the | |
3026 | ISO 10646 character set used internally for wide characters. */ | |
4b10dd6c UD |
3027 | ctype->codeset_name = charmap->code_set_name; |
3028 | ||
3029 | /* Now determine the table for the transliteration information. | |
3030 | ||
3031 | XXX It is not yet clear to me whether it is worth implementing a | |
3032 | complicated algorithm which uses a hash table to locate the entries. | |
3033 | For now I'll use a simple array which can be searching using binary | |
3034 | search. */ | |
3035 | if (ctype->translit_copy_locale != NULL) | |
3036 | { | |
3037 | /* Fold in the transliteration information from the locale mentioned | |
3038 | in the `include' statement. */ | |
3039 | struct locale_ctype_t *here = ctype; | |
3040 | ||
3041 | do | |
3042 | { | |
3043 | struct localedef_t *other = find_locale (LC_CTYPE, | |
3044 | here->translit_copy_locale, | |
3045 | repertoire->name, charmap); | |
3046 | ||
3047 | if (other == NULL) | |
3048 | { | |
3049 | error (0, 0, _("\ | |
3050 | %s: transliteration data from locale `%s' not available"), | |
3051 | "LC_CTYPE", here->translit_copy_locale); | |
3052 | break; | |
3053 | } | |
3054 | ||
3055 | here = other->categories[LC_CTYPE].ctype; | |
3056 | ||
3057 | /* Enqueue the information if necessary. */ | |
3058 | if (here->translit != NULL) | |
3059 | { | |
3060 | struct translit_t *endp = here->translit; | |
3061 | while (endp->next != NULL) | |
3062 | endp = endp->next; | |
3063 | ||
3064 | endp->next = ctype->translit; | |
3065 | ctype->translit = here->translit; | |
3066 | } | |
3067 | } | |
3068 | while (here->translit_copy_locale != NULL); | |
3069 | } | |
3070 | ||
3071 | if (ctype->translit != NULL) | |
3072 | { | |
3073 | /* First count how many entries we have. This is the upper limit | |
3074 | since some entries from the included files might be overwritten. */ | |
3075 | size_t number = 0; | |
3076 | size_t cnt; | |
3077 | struct translit_t *runp = ctype->translit; | |
3078 | struct translit_t **sorted; | |
3079 | size_t from_len, to_len; | |
3080 | ||
3081 | while (runp != NULL) | |
3082 | { | |
3083 | ++number; | |
3084 | runp = runp->next; | |
3085 | } | |
3086 | ||
3087 | /* Next we allocate an array large enough and fill in the values. */ | |
3088 | sorted = alloca (number * sizeof (struct translit_t **)); | |
3089 | runp = ctype->translit; | |
3090 | number = 0; | |
3091 | do | |
3092 | { | |
3093 | /* Search for the place where to insert this string. | |
3094 | XXX Better use a real sorting algorithm later. */ | |
3095 | size_t idx = 0; | |
3096 | int replace = 0; | |
3097 | ||
3098 | while (idx < number) | |
3099 | { | |
3100 | int res = wcscmp ((const wchar_t *) sorted[idx]->from, | |
3101 | (const wchar_t *) runp->from); | |
3102 | if (res == 0) | |
3103 | { | |
3104 | replace = 1; | |
3105 | break; | |
3106 | } | |
3107 | if (res > 0) | |
3108 | break; | |
3109 | ++idx; | |
3110 | } | |
3111 | ||
3112 | if (replace) | |
3113 | sorted[idx] = runp; | |
3114 | else | |
3115 | { | |
3116 | memmove (&sorted[idx + 1], &sorted[idx], | |
3117 | (number - idx) * sizeof (struct translit_t *)); | |
3118 | sorted[idx] = runp; | |
3119 | ++number; | |
3120 | } | |
3121 | ||
3122 | runp = runp->next; | |
3123 | } | |
3124 | while (runp != NULL); | |
3125 | ||
3126 | /* The next step is putting all the possible transliteration | |
3127 | strings in one memory block so that we can write it out. | |
3128 | We need several different blocks: | |
3129 | - index to the tfromstring array | |
3130 | - from-string array | |
3131 | - index to the to-string array | |
3132 | - to-string array. | |
3133 | And this all must be available for both endianes variants. | |
3134 | */ | |
3135 | from_len = to_len = 0; | |
3136 | for (cnt = 0; cnt < number; ++cnt) | |
3137 | { | |
3138 | struct translit_to_t *srunp; | |
3139 | from_len += wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
3140 | srunp = sorted[cnt]->to; | |
3141 | while (srunp != NULL) | |
3142 | { | |
3143 | to_len += wcslen ((const wchar_t *) srunp->str) + 1; | |
3144 | srunp = srunp->next; | |
3145 | } | |
3146 | /* Plus one for the extra NUL character marking the end of | |
3147 | the list for the current entry. */ | |
3148 | ++to_len; | |
3149 | } | |
3150 | ||
3151 | /* We can allocate the arrays for the results. */ | |
3152 | #if BYTE_ORDER == LITTLE_ENDIAN | |
3153 | # define from_idx translit_from_idx_el | |
3154 | # define from_tbl translit_from_tbl_el | |
3155 | # define to_idx translit_to_idx_el | |
3156 | # define to_tbl translit_to_tbl_el | |
3157 | # define from_idx_ob translit_from_idx_eb | |
3158 | # define from_tbl_ob translit_from_tbl_eb | |
3159 | # define to_idx_ob translit_to_idx_eb | |
3160 | # define to_tbl_ob translit_to_tbl_eb | |
3161 | #else | |
3162 | # define from_idx translit_from_idx_eb | |
3163 | # define from_tbl translit_from_tbl_eb | |
3164 | # define to_idx translit_to_idx_eb | |
3165 | # define to_tbl translit_to_tbl_eb | |
3166 | # define from_idx_ob translit_from_idx_el | |
3167 | # define from_tbl_ob translit_from_tbl_el | |
3168 | # define to_idx_ob translit_to_idx_el | |
3169 | # define to_tbl_ob translit_to_tbl_el | |
3170 | #endif | |
3171 | ctype->from_idx = xmalloc (number * sizeof (uint32_t)); | |
3172 | ctype->from_idx_ob = xmalloc (number * sizeof (uint32_t)); | |
3173 | ctype->from_tbl = xmalloc (from_len * sizeof (uint32_t)); | |
3174 | ctype->from_tbl_ob = xmalloc (from_len * sizeof (uint32_t)); | |
3175 | ctype->to_idx = xmalloc (number * sizeof (uint32_t)); | |
3176 | ctype->to_idx_ob = xmalloc (number * sizeof (uint32_t)); | |
3177 | ctype->to_tbl = xmalloc (to_len * sizeof (uint32_t)); | |
3178 | ctype->to_tbl_ob = xmalloc (to_len * sizeof (uint32_t)); | |
3179 | ||
3180 | from_len = 0; | |
3181 | to_len = 0; | |
3182 | for (cnt = 0; cnt < number; ++cnt) | |
3183 | { | |
3184 | size_t len; | |
3185 | struct translit_to_t *srunp; | |
3186 | ||
3187 | ctype->from_idx[cnt] = from_len; | |
3188 | ctype->to_idx[cnt] = to_len; | |
3189 | ||
3190 | len = wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
3191 | wmemcpy ((wchar_t *) &ctype->from_tbl[from_len], | |
3192 | (const wchar_t *) sorted[cnt]->from, len); | |
3193 | from_len += len; | |
3194 | ||
3195 | ctype->to_idx[cnt] = to_len; | |
3196 | srunp = sorted[cnt]->to; | |
3197 | while (srunp != NULL) | |
3198 | { | |
3199 | len = wcslen ((const wchar_t *) srunp->str) + 1; | |
3200 | wmemcpy ((wchar_t *) &ctype->to_tbl[to_len], | |
3201 | (const wchar_t *) srunp->str, len); | |
3202 | to_len += len; | |
3203 | srunp = srunp->next; | |
3204 | } | |
3205 | ctype->to_tbl[to_len++] = L'\0'; | |
3206 | } | |
3207 | ||
3208 | /* Now create the tables for the other endianess. */ | |
3209 | for (cnt = 0; cnt < number; ++cnt) | |
3210 | { | |
3211 | ctype->from_idx_ob[cnt] = bswap_32 (ctype->from_idx[cnt]); | |
3212 | ctype->to_idx_ob[cnt] = bswap_32 (ctype->to_idx[cnt]); | |
3213 | } | |
3214 | for (cnt = 0; cnt < from_len; ++cnt) | |
3215 | ctype->from_tbl[cnt] = bswap_32 (ctype->from_tbl_ob[cnt]); | |
3216 | for (cnt = 0; cnt < to_len; ++cnt) | |
3217 | ctype->to_tbl[cnt] = bswap_32 (ctype->to_tbl_ob[cnt]); | |
3218 | ||
3219 | /* Store the information about the length. */ | |
3220 | ctype->translit_idx_size = number * sizeof (uint32_t); | |
3221 | ctype->translit_from_tbl_size = from_len * sizeof (uint32_t); | |
3222 | ctype->translit_to_tbl_size = to_len * sizeof (uint32_t); | |
3223 | } | |
3224 | else | |
3225 | { | |
3226 | /* Provide some dummy pointers since we have nothing to write out. */ | |
3227 | static uint32_t no_str = { 0 }; | |
3228 | ||
3229 | ctype->translit_from_idx_el = &no_str; | |
3230 | ctype->translit_from_idx_eb = &no_str; | |
3231 | ctype->translit_from_tbl_el = &no_str; | |
3232 | ctype->translit_from_tbl_eb = &no_str; | |
3233 | ctype->translit_to_tbl_el = &no_str; | |
3234 | ctype->translit_to_tbl_eb = &no_str; | |
3235 | ctype->translit_idx_size = 0; | |
3236 | ctype->translit_from_tbl_size = 0; | |
3237 | ctype->translit_to_tbl_size = 0; | |
3238 | } | |
19bc17a9 | 3239 | } |