]>
Commit | Line | Data |
---|---|---|
568035b7 | 1 | /* Copyright (C) 1995-2013 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 | |
43bc8ac6 | 5 | This program is free software; you can redistribute it and/or modify |
2e2efe65 RM |
6 | it under the terms of the GNU General Public License as published |
7 | by the Free Software Foundation; version 2 of the License, or | |
8 | (at your option) any later version. | |
19bc17a9 | 9 | |
43bc8ac6 | 10 | This program is distributed in the hope that it will be useful, |
c84142e8 | 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
43bc8ac6 UD |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | GNU General Public License for more details. | |
19bc17a9 | 14 | |
43bc8ac6 | 15 | You should have received a copy of the GNU General Public License |
59ba27a6 | 16 | along with this program; if not, see <http://www.gnu.org/licenses/>. */ |
19bc17a9 RM |
17 | |
18 | #ifdef HAVE_CONFIG_H | |
19 | # include <config.h> | |
20 | #endif | |
21 | ||
a68b0d31 | 22 | #include <alloca.h> |
4b10dd6c | 23 | #include <byteswap.h> |
19bc17a9 | 24 | #include <endian.h> |
4b10dd6c | 25 | #include <errno.h> |
19bc17a9 | 26 | #include <limits.h> |
4b10dd6c UD |
27 | #include <obstack.h> |
28 | #include <stdlib.h> | |
19bc17a9 | 29 | #include <string.h> |
4b10dd6c UD |
30 | #include <wchar.h> |
31 | #include <wctype.h> | |
e054f494 | 32 | #include <stdint.h> |
4b10dd6c | 33 | #include <sys/uio.h> |
19bc17a9 | 34 | |
f2b98f97 | 35 | #include "localedef.h" |
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 | 41 | #include "locfile.h" |
19bc17a9 | 42 | |
19bc17a9 RM |
43 | #include <assert.h> |
44 | ||
45 | ||
011ebfab | 46 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
47 | /* These are the extra bits not in wctype.h since these are not preallocated |
48 | classes. */ | |
011ebfab UD |
49 | # define _ISwspecial1 (1 << 29) |
50 | # define _ISwspecial2 (1 << 30) | |
51 | # define _ISwspecial3 (1 << 31) | |
52 | #endif | |
19bc17a9 RM |
53 | |
54 | ||
55 | /* The bit used for representing a special class. */ | |
56 | #define BITPOS(class) ((class) - tok_upper) | |
4b10dd6c UD |
57 | #define BIT(class) (_ISbit (BITPOS (class))) |
58 | #define BITw(class) (_ISwbit (BITPOS (class))) | |
19bc17a9 RM |
59 | |
60 | #define ELEM(ctype, collection, idx, value) \ | |
61 | *find_idx (ctype, &ctype->collection idx, &ctype->collection##_max idx, \ | |
62 | &ctype->collection##_act idx, value) | |
63 | ||
19bc17a9 RM |
64 | |
65 | /* To be compatible with former implementations we for now restrict | |
66 | the number of bits for character classes to 16. When compatibility | |
67 | is not necessary anymore increase the number to 32. */ | |
4b10dd6c | 68 | #define char_class_t uint16_t |
4b10dd6c | 69 | #define char_class32_t uint32_t |
4b10dd6c UD |
70 | |
71 | ||
72 | /* Type to describe a transliteration action. We have a possibly | |
73 | multiple character from-string and a set of multiple character | |
74 | to-strings. All are 32bit values since this is what is used in | |
75 | the gconv functions. */ | |
76 | struct translit_to_t | |
77 | { | |
78 | uint32_t *str; | |
79 | ||
80 | struct translit_to_t *next; | |
81 | }; | |
82 | ||
83 | struct translit_t | |
84 | { | |
85 | uint32_t *from; | |
86 | ||
a673fbcb UD |
87 | const char *fname; |
88 | size_t lineno; | |
89 | ||
4b10dd6c UD |
90 | struct translit_to_t *to; |
91 | ||
92 | struct translit_t *next; | |
93 | }; | |
19bc17a9 | 94 | |
a673fbcb UD |
95 | struct translit_ignore_t |
96 | { | |
97 | uint32_t from; | |
98 | uint32_t to; | |
a0dc5206 | 99 | uint32_t step; |
a673fbcb UD |
100 | |
101 | const char *fname; | |
102 | size_t lineno; | |
103 | ||
104 | struct translit_ignore_t *next; | |
105 | }; | |
106 | ||
19bc17a9 | 107 | |
02fb3d17 | 108 | /* Type to describe a transliteration include statement. */ |
0a12bf88 | 109 | struct translit_include_t |
02fb3d17 UD |
110 | { |
111 | const char *copy_locale; | |
112 | const char *copy_repertoire; | |
113 | ||
114 | struct translit_include_t *next; | |
115 | }; | |
116 | ||
117 | ||
601d2942 UD |
118 | /* Sparse table of uint32_t. */ |
119 | #define TABLE idx_table | |
120 | #define ELEMENT uint32_t | |
bd75759f | 121 | #define DEFAULT ((uint32_t) ~0) |
1ecbb381 | 122 | #define NO_ADD_LOCALE |
601d2942 UD |
123 | #include "3level.h" |
124 | ||
1ecbb381 RS |
125 | #define TABLE wcwidth_table |
126 | #define ELEMENT uint8_t | |
127 | #define DEFAULT 0xff | |
128 | #include "3level.h" | |
129 | ||
130 | #define TABLE wctrans_table | |
131 | #define ELEMENT int32_t | |
132 | #define DEFAULT 0 | |
133 | #define wctrans_table_add wctrans_table_add_internal | |
134 | #include "3level.h" | |
135 | #undef wctrans_table_add | |
136 | /* The wctrans_table must actually store the difference between the | |
137 | desired result and the argument. */ | |
138 | static inline void | |
139 | wctrans_table_add (struct wctrans_table *t, uint32_t wc, uint32_t mapped_wc) | |
140 | { | |
141 | wctrans_table_add_internal (t, wc, mapped_wc - wc); | |
142 | } | |
143 | ||
144 | /* Construction of sparse 3-level tables. | |
145 | See wchar-lookup.h for their structure and the meaning of p and q. */ | |
146 | ||
147 | struct wctype_table | |
148 | { | |
149 | /* Parameters. */ | |
150 | unsigned int p; | |
151 | unsigned int q; | |
152 | /* Working representation. */ | |
153 | size_t level1_alloc; | |
154 | size_t level1_size; | |
155 | uint32_t *level1; | |
156 | size_t level2_alloc; | |
157 | size_t level2_size; | |
158 | uint32_t *level2; | |
159 | size_t level3_alloc; | |
160 | size_t level3_size; | |
161 | uint32_t *level3; | |
162 | size_t result_size; | |
163 | }; | |
164 | ||
165 | static void add_locale_wctype_table (struct locale_file *file, | |
166 | struct wctype_table *t); | |
601d2942 | 167 | |
19bc17a9 RM |
168 | /* The real definition of the struct for the LC_CTYPE locale. */ |
169 | struct locale_ctype_t | |
170 | { | |
4b10dd6c | 171 | uint32_t *charnames; |
19bc17a9 RM |
172 | size_t charnames_max; |
173 | size_t charnames_act; | |
04ea3b0f | 174 | /* An index lookup table, to speedup find_idx. */ |
601d2942 | 175 | struct idx_table charnames_idx; |
19bc17a9 | 176 | |
4b10dd6c UD |
177 | struct repertoire_t *repertoire; |
178 | ||
179 | /* We will allow up to 8 * sizeof (uint32_t) character classes. */ | |
180 | #define MAX_NR_CHARCLASS (8 * sizeof (uint32_t)) | |
ba1ffaa1 | 181 | size_t nr_charclass; |
19bc17a9 | 182 | const char *classnames[MAX_NR_CHARCLASS]; |
4b10dd6c UD |
183 | uint32_t last_class_char; |
184 | uint32_t class256_collection[256]; | |
185 | uint32_t *class_collection; | |
19bc17a9 RM |
186 | size_t class_collection_max; |
187 | size_t class_collection_act; | |
4b10dd6c | 188 | uint32_t class_done; |
ef446144 | 189 | uint32_t class_offset; |
4b10dd6c UD |
190 | |
191 | struct charseq **mbdigits; | |
192 | size_t mbdigits_act; | |
193 | size_t mbdigits_max; | |
194 | uint32_t *wcdigits; | |
195 | size_t wcdigits_act; | |
196 | size_t wcdigits_max; | |
197 | ||
198 | struct charseq *mboutdigits[10]; | |
199 | uint32_t wcoutdigits[10]; | |
200 | size_t outdigits_act; | |
19bc17a9 RM |
201 | |
202 | /* If the following number ever turns out to be too small simply | |
203 | increase it. But I doubt it will. --drepper@gnu */ | |
204 | #define MAX_NR_CHARMAP 16 | |
205 | const char *mapnames[MAX_NR_CHARMAP]; | |
4b10dd6c UD |
206 | uint32_t *map_collection[MAX_NR_CHARMAP]; |
207 | uint32_t map256_collection[2][256]; | |
9a0a462c UD |
208 | size_t map_collection_max[MAX_NR_CHARMAP]; |
209 | size_t map_collection_act[MAX_NR_CHARMAP]; | |
19bc17a9 RM |
210 | size_t map_collection_nr; |
211 | size_t last_map_idx; | |
4b10dd6c | 212 | int tomap_done[MAX_NR_CHARMAP]; |
ef446144 | 213 | uint32_t map_offset; |
4b10dd6c UD |
214 | |
215 | /* Transliteration information. */ | |
02fb3d17 | 216 | struct translit_include_t *translit_include; |
4b10dd6c | 217 | struct translit_t *translit; |
a673fbcb | 218 | struct translit_ignore_t *translit_ignore; |
a8e4c924 | 219 | uint32_t ntranslit_ignore; |
a673fbcb UD |
220 | |
221 | uint32_t *default_missing; | |
222 | const char *default_missing_file; | |
223 | size_t default_missing_lineno; | |
19bc17a9 | 224 | |
f0c7c524 | 225 | uint32_t to_nonascii; |
8a449450 | 226 | uint32_t nonascii_case; |
f0c7c524 | 227 | |
19bc17a9 | 228 | /* The arrays for the binary representation. */ |
19bc17a9 RM |
229 | char_class_t *ctype_b; |
230 | char_class32_t *ctype32_b; | |
4c7d276e UD |
231 | uint32_t **map_b; |
232 | uint32_t **map32_b; | |
8fb81470 | 233 | uint32_t **class_b; |
1ecbb381 RS |
234 | struct wctype_table *class_3level; |
235 | struct wctrans_table *map_3level; | |
4b10dd6c UD |
236 | uint32_t *class_name_ptr; |
237 | uint32_t *map_name_ptr; | |
1ecbb381 | 238 | struct wcwidth_table width; |
4b10dd6c | 239 | uint32_t mb_cur_max; |
6990326c | 240 | const char *codeset_name; |
4a33c2f5 UD |
241 | uint32_t *translit_from_idx; |
242 | uint32_t *translit_from_tbl; | |
243 | uint32_t *translit_to_idx; | |
244 | uint32_t *translit_to_tbl; | |
04fbc779 | 245 | uint32_t translit_idx_size; |
4b10dd6c UD |
246 | size_t translit_from_tbl_size; |
247 | size_t translit_to_tbl_size; | |
248 | ||
a673fbcb | 249 | struct obstack mempool; |
19bc17a9 RM |
250 | }; |
251 | ||
47e8b443 | 252 | |
bd75759f AJ |
253 | /* Marker for an empty slot. This has the value 0xFFFFFFFF, regardless |
254 | whether 'int' is 16 bit, 32 bit, or 64 bit. */ | |
255 | #define EMPTY ((uint32_t) ~0) | |
256 | ||
19bc17a9 | 257 | |
4b10dd6c UD |
258 | #define obstack_chunk_alloc xmalloc |
259 | #define obstack_chunk_free free | |
260 | ||
261 | ||
19bc17a9 | 262 | /* Prototypes for local functions. */ |
4b10dd6c | 263 | static void ctype_startup (struct linereader *lr, struct localedef_t *locale, |
47e8b443 | 264 | const struct charmap_t *charmap, |
a6bd56c7 UD |
265 | struct localedef_t *copy_locale, |
266 | int ignore_content); | |
4b10dd6c UD |
267 | static void ctype_class_new (struct linereader *lr, |
268 | struct locale_ctype_t *ctype, const char *name); | |
269 | static void ctype_map_new (struct linereader *lr, | |
270 | struct locale_ctype_t *ctype, | |
47e8b443 | 271 | const char *name, const struct charmap_t *charmap); |
4b10dd6c | 272 | static uint32_t *find_idx (struct locale_ctype_t *ctype, uint32_t **table, |
08ffcf34 | 273 | size_t *max, size_t *act, uint32_t idx); |
19bc17a9 | 274 | static void set_class_defaults (struct locale_ctype_t *ctype, |
47e8b443 | 275 | const struct charmap_t *charmap, |
4b10dd6c | 276 | struct repertoire_t *repertoire); |
75cd5204 | 277 | static void allocate_arrays (struct locale_ctype_t *ctype, |
47e8b443 | 278 | const struct charmap_t *charmap, |
4b10dd6c | 279 | struct repertoire_t *repertoire); |
19bc17a9 RM |
280 | |
281 | ||
4b10dd6c UD |
282 | static const char *longnames[] = |
283 | { | |
284 | "zero", "one", "two", "three", "four", | |
285 | "five", "six", "seven", "eight", "nine" | |
286 | }; | |
1b97149d UD |
287 | static const char *uninames[] = |
288 | { | |
289 | "U00000030", "U00000031", "U00000032", "U00000033", "U00000034", | |
290 | "U00000035", "U00000036", "U00000037", "U00000038", "U00000039" | |
291 | }; | |
4b10dd6c UD |
292 | static const unsigned char digits[] = "0123456789"; |
293 | ||
294 | ||
295 | static void | |
19bc17a9 | 296 | ctype_startup (struct linereader *lr, struct localedef_t *locale, |
47e8b443 UD |
297 | const struct charmap_t *charmap, |
298 | struct localedef_t *copy_locale, int ignore_content) | |
19bc17a9 RM |
299 | { |
300 | unsigned int cnt; | |
301 | struct locale_ctype_t *ctype; | |
302 | ||
a6bd56c7 | 303 | if (!ignore_content && locale->categories[LC_CTYPE].ctype == NULL) |
19bc17a9 | 304 | { |
a6bd56c7 UD |
305 | if (copy_locale == NULL) |
306 | { | |
307 | /* Allocate the needed room. */ | |
308 | locale->categories[LC_CTYPE].ctype = ctype = | |
309 | (struct locale_ctype_t *) xcalloc (1, | |
310 | sizeof (struct locale_ctype_t)); | |
311 | ||
312 | /* We have seen no names yet. */ | |
313 | ctype->charnames_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
08ffcf34 RS |
314 | ctype->charnames = (uint32_t *) xmalloc (ctype->charnames_max |
315 | * sizeof (uint32_t)); | |
a6bd56c7 UD |
316 | for (cnt = 0; cnt < 256; ++cnt) |
317 | ctype->charnames[cnt] = cnt; | |
318 | ctype->charnames_act = 256; | |
601d2942 | 319 | idx_table_init (&ctype->charnames_idx); |
a6bd56c7 UD |
320 | |
321 | /* Fill character class information. */ | |
322 | ctype->last_class_char = ILLEGAL_CHAR_VALUE; | |
323 | /* The order of the following instructions determines the bit | |
324 | positions! */ | |
325 | ctype_class_new (lr, ctype, "upper"); | |
326 | ctype_class_new (lr, ctype, "lower"); | |
327 | ctype_class_new (lr, ctype, "alpha"); | |
328 | ctype_class_new (lr, ctype, "digit"); | |
329 | ctype_class_new (lr, ctype, "xdigit"); | |
330 | ctype_class_new (lr, ctype, "space"); | |
331 | ctype_class_new (lr, ctype, "print"); | |
332 | ctype_class_new (lr, ctype, "graph"); | |
333 | ctype_class_new (lr, ctype, "blank"); | |
334 | ctype_class_new (lr, ctype, "cntrl"); | |
335 | ctype_class_new (lr, ctype, "punct"); | |
336 | ctype_class_new (lr, ctype, "alnum"); | |
011ebfab | 337 | #ifdef PREDEFINED_CLASSES |
a6bd56c7 UD |
338 | /* The following are extensions from ISO 14652. */ |
339 | ctype_class_new (lr, ctype, "left_to_right"); | |
340 | ctype_class_new (lr, ctype, "right_to_left"); | |
341 | ctype_class_new (lr, ctype, "num_terminator"); | |
342 | ctype_class_new (lr, ctype, "num_separator"); | |
343 | ctype_class_new (lr, ctype, "segment_separator"); | |
344 | ctype_class_new (lr, ctype, "block_separator"); | |
345 | ctype_class_new (lr, ctype, "direction_control"); | |
346 | ctype_class_new (lr, ctype, "sym_swap_layout"); | |
347 | ctype_class_new (lr, ctype, "char_shape_selector"); | |
348 | ctype_class_new (lr, ctype, "num_shape_selector"); | |
349 | ctype_class_new (lr, ctype, "non_spacing"); | |
350 | ctype_class_new (lr, ctype, "non_spacing_level3"); | |
351 | ctype_class_new (lr, ctype, "normal_connect"); | |
352 | ctype_class_new (lr, ctype, "r_connect"); | |
353 | ctype_class_new (lr, ctype, "no_connect"); | |
354 | ctype_class_new (lr, ctype, "no_connect-space"); | |
355 | ctype_class_new (lr, ctype, "vowel_connect"); | |
011ebfab | 356 | #endif |
4b10dd6c | 357 | |
a6bd56c7 UD |
358 | ctype->class_collection_max = charmap->mb_cur_max == 1 ? 256 : 512; |
359 | ctype->class_collection | |
360 | = (uint32_t *) xcalloc (sizeof (unsigned long int), | |
361 | ctype->class_collection_max); | |
362 | ctype->class_collection_act = 256; | |
4b10dd6c | 363 | |
a6bd56c7 UD |
364 | /* Fill character map information. */ |
365 | ctype->last_map_idx = MAX_NR_CHARMAP; | |
366 | ctype_map_new (lr, ctype, "toupper", charmap); | |
367 | ctype_map_new (lr, ctype, "tolower", charmap); | |
011ebfab | 368 | #ifdef PREDEFINED_CLASSES |
a6bd56c7 | 369 | ctype_map_new (lr, ctype, "tosymmetric", charmap); |
011ebfab | 370 | #endif |
4b10dd6c | 371 | |
a6bd56c7 UD |
372 | /* Fill first 256 entries in `toXXX' arrays. */ |
373 | for (cnt = 0; cnt < 256; ++cnt) | |
374 | { | |
375 | ctype->map_collection[0][cnt] = cnt; | |
376 | ctype->map_collection[1][cnt] = cnt; | |
9e2b7438 | 377 | #ifdef PREDEFINED_CLASSES |
a6bd56c7 | 378 | ctype->map_collection[2][cnt] = cnt; |
9e2b7438 | 379 | #endif |
a6bd56c7 UD |
380 | ctype->map256_collection[0][cnt] = cnt; |
381 | ctype->map256_collection[1][cnt] = cnt; | |
382 | } | |
4b10dd6c | 383 | |
cb2eab1f UD |
384 | if (enc_not_ascii_compatible) |
385 | ctype->to_nonascii = 1; | |
386 | ||
a6bd56c7 UD |
387 | obstack_init (&ctype->mempool); |
388 | } | |
389 | else | |
390 | ctype = locale->categories[LC_CTYPE].ctype = | |
391 | copy_locale->categories[LC_CTYPE].ctype; | |
19bc17a9 RM |
392 | } |
393 | } | |
394 | ||
395 | ||
396 | void | |
47e8b443 | 397 | ctype_finish (struct localedef_t *locale, const struct charmap_t *charmap) |
19bc17a9 RM |
398 | { |
399 | /* See POSIX.2, table 2-6 for the meaning of the following table. */ | |
400 | #define NCLASS 12 | |
401 | static const struct | |
402 | { | |
403 | const char *name; | |
404 | const char allow[NCLASS]; | |
405 | } | |
406 | valid_table[NCLASS] = | |
407 | { | |
408 | /* The order is important. See token.h for more information. | |
409 | M = Always, D = Default, - = Permitted, X = Mutually exclusive */ | |
410 | { "upper", "--MX-XDDXXX-" }, | |
411 | { "lower", "--MX-XDDXXX-" }, | |
412 | { "alpha", "---X-XDDXXX-" }, | |
413 | { "digit", "XXX--XDDXXX-" }, | |
414 | { "xdigit", "-----XDDXXX-" }, | |
415 | { "space", "XXXXX------X" }, | |
416 | { "print", "---------X--" }, | |
417 | { "graph", "---------X--" }, | |
418 | { "blank", "XXXXXM-----X" }, | |
419 | { "cntrl", "XXXXX-XX--XX" }, | |
420 | { "punct", "XXXXX-DD-X-X" }, | |
421 | { "alnum", "-----XDDXXX-" } | |
422 | }; | |
423 | size_t cnt; | |
424 | int cls1, cls2; | |
4b10dd6c UD |
425 | uint32_t space_value; |
426 | struct charseq *space_seq; | |
19bc17a9 | 427 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
4b10dd6c | 428 | int warned; |
0e16ecfa UD |
429 | const void *key; |
430 | size_t len; | |
431 | void *vdata; | |
432 | void *curs; | |
19bc17a9 | 433 | |
b9eb05d6 UD |
434 | /* Now resolve copying and also handle completely missing definitions. */ |
435 | if (ctype == NULL) | |
436 | { | |
70e51ab9 UD |
437 | const char *repertoire_name; |
438 | ||
b9eb05d6 UD |
439 | /* First see whether we were supposed to copy. If yes, find the |
440 | actual definition. */ | |
441 | if (locale->copy_name[LC_CTYPE] != NULL) | |
442 | { | |
443 | /* Find the copying locale. This has to happen transitively since | |
444 | the locale we are copying from might also copying another one. */ | |
445 | struct localedef_t *from = locale; | |
446 | ||
447 | do | |
448 | from = find_locale (LC_CTYPE, from->copy_name[LC_CTYPE], | |
449 | from->repertoire_name, charmap); | |
450 | while (from->categories[LC_CTYPE].ctype == NULL | |
451 | && from->copy_name[LC_CTYPE] != NULL); | |
452 | ||
453 | ctype = locale->categories[LC_CTYPE].ctype | |
454 | = from->categories[LC_CTYPE].ctype; | |
455 | } | |
456 | ||
457 | /* If there is still no definition issue an warning and create an | |
458 | empty one. */ | |
459 | if (ctype == NULL) | |
460 | { | |
f6ada7ad | 461 | if (! be_quiet) |
f2b98f97 UD |
462 | WITH_CUR_LOCALE (error (0, 0, _("\ |
463 | No definition for %s category found"), "LC_CTYPE")); | |
a6bd56c7 | 464 | ctype_startup (NULL, locale, charmap, NULL, 0); |
b9eb05d6 UD |
465 | ctype = locale->categories[LC_CTYPE].ctype; |
466 | } | |
70e51ab9 UD |
467 | |
468 | /* Get the repertoire we have to use. */ | |
469 | repertoire_name = locale->repertoire_name ?: repertoire_global; | |
470 | if (repertoire_name != NULL) | |
471 | ctype->repertoire = repertoire_read (repertoire_name); | |
b9eb05d6 UD |
472 | } |
473 | ||
db76d943 UD |
474 | /* We need the name of the currently used 8-bit character set to |
475 | make correct conversion between this 8-bit representation and the | |
476 | ISO 10646 character set used internally for wide characters. */ | |
477 | ctype->codeset_name = charmap->code_set_name; | |
478 | if (ctype->codeset_name == NULL) | |
479 | { | |
480 | if (! be_quiet) | |
f2b98f97 UD |
481 | WITH_CUR_LOCALE (error (0, 0, _("\ |
482 | No character set name specified in charmap"))); | |
db76d943 UD |
483 | ctype->codeset_name = "//UNKNOWN//"; |
484 | } | |
485 | ||
19bc17a9 | 486 | /* Set default value for classes not specified. */ |
4b10dd6c | 487 | set_class_defaults (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
488 | |
489 | /* Check according to table. */ | |
42d7c593 | 490 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) |
19bc17a9 | 491 | { |
4b10dd6c | 492 | uint32_t tmp = ctype->class_collection[cnt]; |
19bc17a9 | 493 | |
4b10dd6c UD |
494 | if (tmp != 0) |
495 | { | |
496 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
497 | if ((tmp & _ISwbit (cls1)) != 0) | |
498 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
499 | if (valid_table[cls1].allow[cls2] != '-') | |
19bc17a9 | 500 | { |
4b10dd6c UD |
501 | int eq = (tmp & _ISwbit (cls2)) != 0; |
502 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 503 | { |
4b10dd6c UD |
504 | case 'M': |
505 | if (!eq) | |
506 | { | |
507 | uint32_t value = ctype->charnames[cnt]; | |
508 | ||
509 | if (!be_quiet) | |
f2b98f97 | 510 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 511 | character L'\\u%0*x' in class `%s' must be in class `%s'"), |
f2b98f97 UD |
512 | value > 0xffff ? 8 : 4, |
513 | value, | |
514 | valid_table[cls1].name, | |
515 | valid_table[cls2].name)); | |
4b10dd6c UD |
516 | } |
517 | break; | |
518 | ||
519 | case 'X': | |
520 | if (eq) | |
521 | { | |
522 | uint32_t value = ctype->charnames[cnt]; | |
523 | ||
524 | if (!be_quiet) | |
f2b98f97 | 525 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 526 | character L'\\u%0*x' in class `%s' must not be in class `%s'"), |
f2b98f97 UD |
527 | value > 0xffff ? 8 : 4, |
528 | value, | |
529 | valid_table[cls1].name, | |
530 | valid_table[cls2].name)); | |
4b10dd6c UD |
531 | } |
532 | break; | |
533 | ||
534 | case 'D': | |
535 | ctype->class_collection[cnt] |= _ISwbit (cls2); | |
536 | break; | |
537 | ||
538 | default: | |
f2b98f97 UD |
539 | WITH_CUR_LOCALE (error (5, 0, _("\ |
540 | internal error in %s, line %u"), __FUNCTION__, __LINE__)); | |
19bc17a9 | 541 | } |
4b10dd6c UD |
542 | } |
543 | } | |
544 | } | |
545 | ||
546 | for (cnt = 0; cnt < 256; ++cnt) | |
547 | { | |
548 | uint32_t tmp = ctype->class256_collection[cnt]; | |
19bc17a9 | 549 | |
4b10dd6c UD |
550 | if (tmp != 0) |
551 | { | |
552 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
553 | if ((tmp & _ISbit (cls1)) != 0) | |
554 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
555 | if (valid_table[cls1].allow[cls2] != '-') | |
556 | { | |
557 | int eq = (tmp & _ISbit (cls2)) != 0; | |
558 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 559 | { |
4b10dd6c UD |
560 | case 'M': |
561 | if (!eq) | |
562 | { | |
563 | char buf[17]; | |
564 | ||
5d431a3e | 565 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
566 | |
567 | if (!be_quiet) | |
f2b98f97 | 568 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 569 | character '%s' in class `%s' must be in class `%s'"), |
f2b98f97 UD |
570 | buf, |
571 | valid_table[cls1].name, | |
572 | valid_table[cls2].name)); | |
4b10dd6c UD |
573 | } |
574 | break; | |
575 | ||
576 | case 'X': | |
577 | if (eq) | |
578 | { | |
579 | char buf[17]; | |
580 | ||
5d431a3e | 581 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
582 | |
583 | if (!be_quiet) | |
f2b98f97 | 584 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 585 | character '%s' in class `%s' must not be in class `%s'"), |
f2b98f97 UD |
586 | buf, |
587 | valid_table[cls1].name, | |
588 | valid_table[cls2].name)); | |
4b10dd6c UD |
589 | } |
590 | break; | |
591 | ||
592 | case 'D': | |
593 | ctype->class256_collection[cnt] |= _ISbit (cls2); | |
594 | break; | |
595 | ||
596 | default: | |
f2b98f97 UD |
597 | WITH_CUR_LOCALE (error (5, 0, _("\ |
598 | internal error in %s, line %u"), __FUNCTION__, __LINE__)); | |
19bc17a9 | 599 | } |
4b10dd6c UD |
600 | } |
601 | } | |
19bc17a9 RM |
602 | } |
603 | ||
604 | /* ... and now test <SP> as a special case. */ | |
a0dc5206 UD |
605 | space_value = 32; |
606 | if (((cnt = BITPOS (tok_space), | |
607 | (ELEM (ctype, class_collection, , space_value) | |
608 | & BITw (tok_space)) == 0) | |
609 | || (cnt = BITPOS (tok_blank), | |
610 | (ELEM (ctype, class_collection, , space_value) | |
611 | & BITw (tok_blank)) == 0))) | |
880f421f UD |
612 | { |
613 | if (!be_quiet) | |
f2b98f97 UD |
614 | WITH_CUR_LOCALE (error (0, 0, _("<SP> character not in class `%s'"), |
615 | valid_table[cnt].name)); | |
880f421f | 616 | } |
c84142e8 UD |
617 | else if (((cnt = BITPOS (tok_punct), |
618 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 619 | & BITw (tok_punct)) != 0) |
c84142e8 UD |
620 | || (cnt = BITPOS (tok_graph), |
621 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 622 | & BITw (tok_graph)) |
880f421f UD |
623 | != 0))) |
624 | { | |
625 | if (!be_quiet) | |
f2b98f97 UD |
626 | WITH_CUR_LOCALE (error (0, 0, _("\ |
627 | <SP> character must not be in class `%s'"), | |
628 | valid_table[cnt].name)); | |
880f421f | 629 | } |
19bc17a9 | 630 | else |
4b10dd6c UD |
631 | ELEM (ctype, class_collection, , space_value) |= BITw (tok_print); |
632 | ||
633 | space_seq = charmap_find_value (charmap, "SP", 2); | |
ce177a84 | 634 | if (space_seq == NULL) |
45c95239 UD |
635 | space_seq = charmap_find_value (charmap, "space", 5); |
636 | if (space_seq == NULL) | |
1b97149d | 637 | space_seq = charmap_find_value (charmap, "U00000020", 9); |
4b10dd6c UD |
638 | if (space_seq == NULL || space_seq->nbytes != 1) |
639 | { | |
640 | if (!be_quiet) | |
f2b98f97 UD |
641 | WITH_CUR_LOCALE (error (0, 0, _("\ |
642 | character <SP> not defined in character map"))); | |
4b10dd6c UD |
643 | } |
644 | else if (((cnt = BITPOS (tok_space), | |
645 | (ctype->class256_collection[space_seq->bytes[0]] | |
646 | & BIT (tok_space)) == 0) | |
647 | || (cnt = BITPOS (tok_blank), | |
648 | (ctype->class256_collection[space_seq->bytes[0]] | |
649 | & BIT (tok_blank)) == 0))) | |
650 | { | |
651 | if (!be_quiet) | |
f2b98f97 UD |
652 | WITH_CUR_LOCALE (error (0, 0, _("<SP> character not in class `%s'"), |
653 | valid_table[cnt].name)); | |
4b10dd6c UD |
654 | } |
655 | else if (((cnt = BITPOS (tok_punct), | |
656 | (ctype->class256_collection[space_seq->bytes[0]] | |
657 | & BIT (tok_punct)) != 0) | |
658 | || (cnt = BITPOS (tok_graph), | |
659 | (ctype->class256_collection[space_seq->bytes[0]] | |
660 | & BIT (tok_graph)) != 0))) | |
661 | { | |
662 | if (!be_quiet) | |
f2b98f97 UD |
663 | WITH_CUR_LOCALE (error (0, 0, _("\ |
664 | <SP> character must not be in class `%s'"), | |
665 | valid_table[cnt].name)); | |
4b10dd6c UD |
666 | } |
667 | else | |
668 | ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print); | |
75cd5204 | 669 | |
8a449450 UD |
670 | /* Check whether all single-byte characters make to their upper/lowercase |
671 | equivalent according to the ASCII rules. */ | |
672 | for (cnt = 'A'; cnt <= 'Z'; ++cnt) | |
673 | { | |
674 | uint32_t uppval = ctype->map256_collection[0][cnt]; | |
675 | uint32_t lowval = ctype->map256_collection[1][cnt]; | |
676 | uint32_t lowuppval = ctype->map256_collection[0][lowval]; | |
677 | uint32_t lowlowval = ctype->map256_collection[1][lowval]; | |
678 | ||
679 | if (uppval != cnt | |
680 | || lowval != cnt + 0x20 | |
681 | || lowuppval != cnt | |
682 | || lowlowval != cnt + 0x20) | |
683 | ctype->nonascii_case = 1; | |
684 | } | |
685 | for (cnt = 0; cnt < 256; ++cnt) | |
686 | if (cnt < 'A' || (cnt > 'Z' && cnt < 'a') || cnt > 'z') | |
687 | if (ctype->map256_collection[0][cnt] != cnt | |
688 | || ctype->map256_collection[1][cnt] != cnt) | |
689 | ctype->nonascii_case = 1; | |
690 | ||
75cd5204 RM |
691 | /* Now that the tests are done make sure the name array contains all |
692 | characters which are handled in the WIDTH section of the | |
693 | character set definition file. */ | |
4b10dd6c UD |
694 | if (charmap->width_rules != NULL) |
695 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) | |
75cd5204 | 696 | { |
827ff758 UD |
697 | unsigned char bytes[charmap->mb_cur_max]; |
698 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
699 | ||
700 | /* We have the range of character for which the width is | |
701 | specified described using byte sequences of the multibyte | |
702 | charset. We have to convert this to UCS4 now. And we | |
703 | cannot simply convert the beginning and the end of the | |
704 | sequence, we have to iterate over the byte sequence and | |
705 | convert it for every single character. */ | |
706 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
707 | ||
708 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
709 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
710 | nbytes) <= 0) | |
711 | { | |
712 | /* Find the UCS value for `bytes'. */ | |
827ff758 | 713 | int inner; |
76e680a8 | 714 | uint32_t wch; |
701666b7 UD |
715 | struct charseq *seq |
716 | = charmap_find_symbol (charmap, (char *) bytes, nbytes); | |
76e680a8 UD |
717 | |
718 | if (seq == NULL) | |
719 | wch = ILLEGAL_CHAR_VALUE; | |
720 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
721 | wch = seq->ucs4; | |
722 | else | |
723 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
724 | strlen (seq->name)); | |
827ff758 UD |
725 | |
726 | if (wch != ILLEGAL_CHAR_VALUE) | |
727 | /* We are only interested in the side-effects of the | |
728 | `find_idx' call. It will add appropriate entries in | |
729 | the name array if this is necessary. */ | |
730 | (void) find_idx (ctype, NULL, NULL, NULL, wch); | |
731 | ||
732 | /* "Increment" the bytes sequence. */ | |
733 | inner = nbytes - 1; | |
734 | while (inner >= 0 && bytes[inner] == 0xff) | |
735 | --inner; | |
736 | ||
737 | if (inner < 0) | |
738 | { | |
739 | /* We have to extend the byte sequence. */ | |
740 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
741 | break; | |
742 | ||
743 | bytes[0] = 1; | |
744 | memset (&bytes[1], 0, nbytes); | |
745 | ++nbytes; | |
746 | } | |
747 | else | |
748 | { | |
749 | ++bytes[inner]; | |
750 | while (++inner < nbytes) | |
751 | bytes[inner] = 0; | |
752 | } | |
753 | } | |
4b10dd6c UD |
754 | } |
755 | ||
0e16ecfa UD |
756 | /* Now set all the other characters of the character set to the |
757 | default width. */ | |
758 | curs = NULL; | |
759 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
760 | { | |
761 | struct charseq *data = (struct charseq *) vdata; | |
762 | ||
763 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
764 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
765 | data->name, len); | |
766 | ||
767 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) | |
768 | (void) find_idx (ctype, NULL, NULL, NULL, data->ucs4); | |
769 | } | |
770 | ||
4b10dd6c UD |
771 | /* There must be a multiple of 10 digits. */ |
772 | if (ctype->mbdigits_act % 10 != 0) | |
773 | { | |
774 | assert (ctype->mbdigits_act == ctype->wcdigits_act); | |
775 | ctype->wcdigits_act -= ctype->mbdigits_act % 10; | |
776 | ctype->mbdigits_act -= ctype->mbdigits_act % 10; | |
f2b98f97 UD |
777 | WITH_CUR_LOCALE (error (0, 0, _("\ |
778 | `digit' category has not entries in groups of ten"))); | |
4b10dd6c UD |
779 | } |
780 | ||
781 | /* Check the input digits. There must be a multiple of ten available. | |
42d7c593 | 782 | In each group it could be that one or the other character is missing. |
4b10dd6c UD |
783 | In this case the whole group must be removed. */ |
784 | cnt = 0; | |
785 | while (cnt < ctype->mbdigits_act) | |
786 | { | |
787 | size_t inner; | |
788 | for (inner = 0; inner < 10; ++inner) | |
789 | if (ctype->mbdigits[cnt + inner] == NULL) | |
790 | break; | |
791 | ||
792 | if (inner == 10) | |
793 | cnt += 10; | |
794 | else | |
795 | { | |
796 | /* Remove the group. */ | |
797 | memmove (&ctype->mbdigits[cnt], &ctype->mbdigits[cnt + 10], | |
798 | ((ctype->wcdigits_act - cnt - 10) | |
799 | * sizeof (ctype->mbdigits[0]))); | |
800 | ctype->mbdigits_act -= 10; | |
801 | } | |
802 | } | |
803 | ||
804 | /* If no input digits are given use the default. */ | |
805 | if (ctype->mbdigits_act == 0) | |
806 | { | |
807 | if (ctype->mbdigits_max == 0) | |
808 | { | |
47e8b443 | 809 | ctype->mbdigits = obstack_alloc (&((struct charmap_t *) charmap)->mem_pool, |
4b10dd6c UD |
810 | 10 * sizeof (struct charseq *)); |
811 | ctype->mbdigits_max = 10; | |
812 | } | |
813 | ||
814 | for (cnt = 0; cnt < 10; ++cnt) | |
815 | { | |
816 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
701666b7 | 817 | (char *) digits + cnt, 1); |
4b10dd6c UD |
818 | if (ctype->mbdigits[cnt] == NULL) |
819 | { | |
820 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
821 | longnames[cnt], | |
822 | strlen (longnames[cnt])); | |
823 | if (ctype->mbdigits[cnt] == NULL) | |
824 | { | |
825 | /* Hum, this ain't good. */ | |
f2b98f97 UD |
826 | WITH_CUR_LOCALE (error (0, 0, _("\ |
827 | no input digits defined and none of the standard names in the charmap"))); | |
4b10dd6c | 828 | |
47e8b443 | 829 | ctype->mbdigits[cnt] = obstack_alloc (&((struct charmap_t *) charmap)->mem_pool, |
4b10dd6c UD |
830 | sizeof (struct charseq) + 1); |
831 | ||
832 | /* This is better than nothing. */ | |
833 | ctype->mbdigits[cnt]->bytes[0] = digits[cnt]; | |
834 | ctype->mbdigits[cnt]->nbytes = 1; | |
835 | } | |
836 | } | |
837 | } | |
838 | ||
839 | ctype->mbdigits_act = 10; | |
840 | } | |
841 | ||
842 | /* Check the wide character input digits. There must be a multiple | |
42d7c593 | 843 | of ten available. In each group it could be that one or the other |
4b10dd6c UD |
844 | character is missing. In this case the whole group must be |
845 | removed. */ | |
846 | cnt = 0; | |
847 | while (cnt < ctype->wcdigits_act) | |
848 | { | |
849 | size_t inner; | |
850 | for (inner = 0; inner < 10; ++inner) | |
851 | if (ctype->wcdigits[cnt + inner] == ILLEGAL_CHAR_VALUE) | |
852 | break; | |
853 | ||
854 | if (inner == 10) | |
855 | cnt += 10; | |
856 | else | |
857 | { | |
858 | /* Remove the group. */ | |
859 | memmove (&ctype->wcdigits[cnt], &ctype->wcdigits[cnt + 10], | |
860 | ((ctype->wcdigits_act - cnt - 10) | |
861 | * sizeof (ctype->wcdigits[0]))); | |
862 | ctype->wcdigits_act -= 10; | |
863 | } | |
864 | } | |
865 | ||
866 | /* If no input digits are given use the default. */ | |
867 | if (ctype->wcdigits_act == 0) | |
868 | { | |
869 | if (ctype->wcdigits_max == 0) | |
870 | { | |
47e8b443 | 871 | ctype->wcdigits = obstack_alloc (&((struct charmap_t *) charmap)->mem_pool, |
4b10dd6c UD |
872 | 10 * sizeof (uint32_t)); |
873 | ctype->wcdigits_max = 10; | |
874 | } | |
875 | ||
876 | for (cnt = 0; cnt < 10; ++cnt) | |
877 | ctype->wcdigits[cnt] = L'0' + cnt; | |
878 | ||
879 | ctype->mbdigits_act = 10; | |
880 | } | |
881 | ||
882 | /* Check the outdigits. */ | |
883 | warned = 0; | |
884 | for (cnt = 0; cnt < 10; ++cnt) | |
885 | if (ctype->mboutdigits[cnt] == NULL) | |
886 | { | |
887 | static struct charseq replace[2]; | |
888 | ||
889 | if (!warned) | |
890 | { | |
f2b98f97 UD |
891 | WITH_CUR_LOCALE (error (0, 0, _("\ |
892 | not all characters used in `outdigit' are available in the charmap"))); | |
4b10dd6c UD |
893 | warned = 1; |
894 | } | |
895 | ||
896 | replace[0].nbytes = 1; | |
897 | replace[0].bytes[0] = '?'; | |
898 | replace[0].bytes[1] = '\0'; | |
899 | ctype->mboutdigits[cnt] = &replace[0]; | |
900 | } | |
901 | ||
902 | warned = 0; | |
903 | for (cnt = 0; cnt < 10; ++cnt) | |
904 | if (ctype->wcoutdigits[cnt] == 0) | |
905 | { | |
906 | if (!warned) | |
907 | { | |
f2b98f97 UD |
908 | WITH_CUR_LOCALE (error (0, 0, _("\ |
909 | not all characters used in `outdigit' are available in the repertoire"))); | |
4b10dd6c UD |
910 | warned = 1; |
911 | } | |
912 | ||
913 | ctype->wcoutdigits[cnt] = L'?'; | |
75cd5204 | 914 | } |
a8e4c924 UD |
915 | |
916 | /* Sort the entries in the translit_ignore list. */ | |
917 | if (ctype->translit_ignore != NULL) | |
918 | { | |
919 | struct translit_ignore_t *firstp = ctype->translit_ignore; | |
920 | struct translit_ignore_t *runp; | |
921 | ||
922 | ctype->ntranslit_ignore = 1; | |
923 | ||
924 | for (runp = firstp->next; runp != NULL; runp = runp->next) | |
925 | { | |
926 | struct translit_ignore_t *lastp = NULL; | |
927 | struct translit_ignore_t *cmpp; | |
928 | ||
929 | ++ctype->ntranslit_ignore; | |
930 | ||
931 | for (cmpp = firstp; cmpp != NULL; lastp = cmpp, cmpp = cmpp->next) | |
932 | if (runp->from < cmpp->from) | |
933 | break; | |
934 | ||
935 | runp->next = lastp; | |
936 | if (lastp == NULL) | |
937 | firstp = runp; | |
938 | } | |
939 | ||
940 | ctype->translit_ignore = firstp; | |
941 | } | |
19bc17a9 RM |
942 | } |
943 | ||
944 | ||
945 | void | |
47e8b443 | 946 | ctype_output (struct localedef_t *locale, const struct charmap_t *charmap, |
75cd5204 | 947 | const char *output_path) |
19bc17a9 RM |
948 | { |
949 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; | |
a9706118 | 950 | const size_t nelems = (_NL_ITEM_INDEX (_NL_CTYPE_EXTRA_MAP_1) |
4c7d276e | 951 | + ctype->nr_charclass + ctype->map_collection_nr); |
1ecbb381 | 952 | struct locale_file file; |
1d96d74d | 953 | uint32_t default_missing_len; |
1ecbb381 | 954 | size_t elem, cnt; |
19bc17a9 RM |
955 | |
956 | /* Now prepare the output: Find the sizes of the table we can use. */ | |
4b10dd6c | 957 | allocate_arrays (ctype, charmap, ctype->repertoire); |
19bc17a9 | 958 | |
1ecbb381 RS |
959 | default_missing_len = (ctype->default_missing |
960 | ? wcslen ((wchar_t *) ctype->default_missing) | |
961 | : 0); | |
19bc17a9 | 962 | |
1ecbb381 | 963 | init_locale_data (&file, nelems); |
19bc17a9 RM |
964 | for (elem = 0; elem < nelems; ++elem) |
965 | { | |
a9706118 | 966 | if (elem < _NL_ITEM_INDEX (_NL_CTYPE_EXTRA_MAP_1)) |
19bc17a9 RM |
967 | switch (elem) |
968 | { | |
c6df09ad UD |
969 | #define CTYPE_EMPTY(name) \ |
970 | case name: \ | |
1ecbb381 | 971 | add_locale_empty (&file); \ |
c6df09ad UD |
972 | break |
973 | ||
974 | CTYPE_EMPTY(_NL_CTYPE_GAP1); | |
975 | CTYPE_EMPTY(_NL_CTYPE_GAP2); | |
976 | CTYPE_EMPTY(_NL_CTYPE_GAP3); | |
4c7d276e UD |
977 | CTYPE_EMPTY(_NL_CTYPE_GAP4); |
978 | CTYPE_EMPTY(_NL_CTYPE_GAP5); | |
979 | CTYPE_EMPTY(_NL_CTYPE_GAP6); | |
c6df09ad | 980 | |
1ecbb381 | 981 | #define CTYPE_RAW_DATA(name, base, size) \ |
19bc17a9 | 982 | case _NL_ITEM_INDEX (name): \ |
1ecbb381 | 983 | add_locale_raw_data (&file, base, size); \ |
19bc17a9 RM |
984 | break |
985 | ||
1ecbb381 RS |
986 | CTYPE_RAW_DATA (_NL_CTYPE_CLASS, |
987 | ctype->ctype_b, | |
988 | (256 + 128) * sizeof (char_class_t)); | |
19bc17a9 | 989 | |
1ecbb381 RS |
990 | #define CTYPE_UINT32_ARRAY(name, base, n_elems) \ |
991 | case _NL_ITEM_INDEX (name): \ | |
992 | add_locale_uint32_array (&file, base, n_elems); \ | |
993 | break | |
ef446144 | 994 | |
1ecbb381 RS |
995 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TOUPPER, ctype->map_b[0], 256 + 128); |
996 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TOLOWER, ctype->map_b[1], 256 + 128); | |
997 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TOUPPER32, ctype->map32_b[0], 256); | |
998 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TOLOWER32, ctype->map32_b[1], 256); | |
999 | CTYPE_RAW_DATA (_NL_CTYPE_CLASS32, | |
1000 | ctype->ctype32_b, | |
1001 | 256 * sizeof (char_class32_t)); | |
ef446144 | 1002 | |
1ecbb381 RS |
1003 | #define CTYPE_UINT32(name, value) \ |
1004 | case _NL_ITEM_INDEX (name): \ | |
1005 | add_locale_uint32 (&file, value); \ | |
1006 | break | |
4a33c2f5 | 1007 | |
1ecbb381 RS |
1008 | CTYPE_UINT32 (_NL_CTYPE_CLASS_OFFSET, ctype->class_offset); |
1009 | CTYPE_UINT32 (_NL_CTYPE_MAP_OFFSET, ctype->map_offset); | |
1010 | CTYPE_UINT32 (_NL_CTYPE_TRANSLIT_TAB_SIZE, ctype->translit_idx_size); | |
4a33c2f5 | 1011 | |
1ecbb381 RS |
1012 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TRANSLIT_FROM_IDX, |
1013 | ctype->translit_from_idx, | |
1014 | ctype->translit_idx_size); | |
4b10dd6c | 1015 | |
1ecbb381 RS |
1016 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TRANSLIT_FROM_TBL, |
1017 | ctype->translit_from_tbl, | |
1018 | ctype->translit_from_tbl_size | |
1019 | / sizeof (uint32_t)); | |
4b10dd6c | 1020 | |
1ecbb381 RS |
1021 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TRANSLIT_TO_IDX, |
1022 | ctype->translit_to_idx, | |
1023 | ctype->translit_idx_size); | |
4b10dd6c | 1024 | |
1ecbb381 RS |
1025 | CTYPE_UINT32_ARRAY (_NL_CTYPE_TRANSLIT_TO_TBL, |
1026 | ctype->translit_to_tbl, | |
1027 | ctype->translit_to_tbl_size / sizeof (uint32_t)); | |
4b10dd6c | 1028 | |
75cd5204 RM |
1029 | case _NL_ITEM_INDEX (_NL_CTYPE_CLASS_NAMES): |
1030 | /* The class name array. */ | |
1ecbb381 RS |
1031 | start_locale_structure (&file); |
1032 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
1033 | add_locale_string (&file, ctype->classnames[cnt]); | |
1034 | add_locale_char (&file, 0); | |
7602d070 | 1035 | align_locale_data (&file, LOCFILE_ALIGN); |
1ecbb381 | 1036 | end_locale_structure (&file); |
75cd5204 RM |
1037 | break; |
1038 | ||
1039 | case _NL_ITEM_INDEX (_NL_CTYPE_MAP_NAMES): | |
1040 | /* The class name array. */ | |
1ecbb381 RS |
1041 | start_locale_structure (&file); |
1042 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) | |
1043 | add_locale_string (&file, ctype->mapnames[cnt]); | |
1044 | add_locale_char (&file, 0); | |
7602d070 | 1045 | align_locale_data (&file, LOCFILE_ALIGN); |
1ecbb381 | 1046 | end_locale_structure (&file); |
75cd5204 | 1047 | break; |
19bc17a9 | 1048 | |
1ecbb381 RS |
1049 | case _NL_ITEM_INDEX (_NL_CTYPE_WIDTH): |
1050 | add_locale_wcwidth_table (&file, &ctype->width); | |
1051 | break; | |
19bc17a9 | 1052 | |
1ecbb381 | 1053 | CTYPE_UINT32 (_NL_CTYPE_MB_CUR_MAX, ctype->mb_cur_max); |
0200214b | 1054 | |
ce7a5ef4 | 1055 | case _NL_ITEM_INDEX (_NL_CTYPE_CODESET_NAME): |
1ecbb381 | 1056 | add_locale_string (&file, ctype->codeset_name); |
4b10dd6c UD |
1057 | break; |
1058 | ||
1ecbb381 | 1059 | CTYPE_UINT32 (_NL_CTYPE_MAP_TO_NONASCII, ctype->to_nonascii); |
f0c7c524 | 1060 | |
1ecbb381 | 1061 | CTYPE_UINT32 (_NL_CTYPE_NONASCII_CASE, ctype->nonascii_case); |
8a449450 | 1062 | |
4a33c2f5 | 1063 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN): |
1ecbb381 | 1064 | add_locale_uint32 (&file, ctype->mbdigits_act / 10); |
4b10dd6c UD |
1065 | break; |
1066 | ||
4a33c2f5 | 1067 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN): |
1ecbb381 | 1068 | add_locale_uint32 (&file, ctype->wcdigits_act / 10); |
4b10dd6c UD |
1069 | break; |
1070 | ||
e43e0dd6 | 1071 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_MB): |
1ecbb381 | 1072 | start_locale_structure (&file); |
498b733e | 1073 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB); |
4b10dd6c UD |
1074 | cnt < ctype->mbdigits_act; cnt += 10) |
1075 | { | |
1ecbb381 RS |
1076 | add_locale_raw_data (&file, ctype->mbdigits[cnt]->bytes, |
1077 | ctype->mbdigits[cnt]->nbytes); | |
1078 | add_locale_char (&file, 0); | |
4b10dd6c | 1079 | } |
1ecbb381 | 1080 | end_locale_structure (&file); |
4b10dd6c UD |
1081 | break; |
1082 | ||
1083 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_MB): | |
1ecbb381 | 1084 | start_locale_structure (&file); |
498b733e | 1085 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB); |
1ecbb381 RS |
1086 | add_locale_raw_data (&file, ctype->mboutdigits[cnt]->bytes, |
1087 | ctype->mboutdigits[cnt]->nbytes); | |
1088 | add_locale_char (&file, 0); | |
1089 | end_locale_structure (&file); | |
4b10dd6c UD |
1090 | break; |
1091 | ||
e43e0dd6 | 1092 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC): |
1ecbb381 | 1093 | start_locale_structure (&file); |
498b733e | 1094 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC); |
4b10dd6c | 1095 | cnt < ctype->wcdigits_act; cnt += 10) |
1ecbb381 RS |
1096 | add_locale_uint32 (&file, ctype->wcdigits[cnt]); |
1097 | end_locale_structure (&file); | |
4b10dd6c UD |
1098 | break; |
1099 | ||
1ecbb381 | 1100 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC): |
498b733e | 1101 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC); |
1ecbb381 | 1102 | add_locale_uint32 (&file, ctype->wcoutdigits[cnt]); |
4b10dd6c UD |
1103 | break; |
1104 | ||
a8e4c924 | 1105 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING_LEN): |
1ecbb381 | 1106 | add_locale_uint32 (&file, default_missing_len); |
a8e4c924 UD |
1107 | break; |
1108 | ||
1d96d74d | 1109 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING): |
1ecbb381 RS |
1110 | add_locale_uint32_array (&file, ctype->default_missing, |
1111 | default_missing_len); | |
1d96d74d UD |
1112 | break; |
1113 | ||
a8e4c924 | 1114 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE_LEN): |
1ecbb381 | 1115 | add_locale_uint32 (&file, ctype->ntranslit_ignore); |
a8e4c924 UD |
1116 | break; |
1117 | ||
1118 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE): | |
1ecbb381 | 1119 | start_locale_structure (&file); |
a8e4c924 | 1120 | { |
a8e4c924 | 1121 | struct translit_ignore_t *runp; |
a8e4c924 UD |
1122 | for (runp = ctype->translit_ignore; runp != NULL; |
1123 | runp = runp->next) | |
1124 | { | |
1ecbb381 RS |
1125 | add_locale_uint32 (&file, runp->from); |
1126 | add_locale_uint32 (&file, runp->to); | |
1127 | add_locale_uint32 (&file, runp->step); | |
a8e4c924 UD |
1128 | } |
1129 | } | |
1ecbb381 | 1130 | end_locale_structure (&file); |
1d96d74d UD |
1131 | break; |
1132 | ||
19bc17a9 RM |
1133 | default: |
1134 | assert (! "unknown CTYPE element"); | |
1135 | } | |
1136 | else | |
1137 | { | |
1138 | /* Handle extra maps. */ | |
4c7d276e UD |
1139 | size_t nr = elem - _NL_ITEM_INDEX (_NL_CTYPE_EXTRA_MAP_1); |
1140 | if (nr < ctype->nr_charclass) | |
ef446144 | 1141 | { |
1ecbb381 RS |
1142 | start_locale_prelude (&file); |
1143 | add_locale_uint32_array (&file, ctype->class_b[nr], 256 / 32); | |
1144 | end_locale_prelude (&file); | |
1145 | add_locale_wctype_table (&file, &ctype->class_3level[nr]); | |
ef446144 UD |
1146 | } |
1147 | else | |
1148 | { | |
4c7d276e UD |
1149 | nr -= ctype->nr_charclass; |
1150 | assert (nr < ctype->map_collection_nr); | |
1ecbb381 | 1151 | add_locale_wctrans_table (&file, &ctype->map_3level[nr]); |
ef446144 | 1152 | } |
75cd5204 | 1153 | } |
19bc17a9 | 1154 | } |
19bc17a9 | 1155 | |
1ecbb381 | 1156 | write_locale_data (output_path, LC_CTYPE, "LC_CTYPE", &file); |
19bc17a9 RM |
1157 | } |
1158 | ||
1159 | ||
4b10dd6c UD |
1160 | /* Local functions. */ |
1161 | static void | |
1162 | ctype_class_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1163 | const char *name) | |
19bc17a9 | 1164 | { |
4b10dd6c | 1165 | size_t cnt; |
19bc17a9 | 1166 | |
4b10dd6c UD |
1167 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1168 | if (strcmp (ctype->classnames[cnt], name) == 0) | |
1169 | break; | |
19bc17a9 | 1170 | |
4b10dd6c UD |
1171 | if (cnt < ctype->nr_charclass) |
1172 | { | |
1173 | lr_error (lr, _("character class `%s' already defined"), name); | |
1174 | return; | |
1175 | } | |
19bc17a9 | 1176 | |
4b10dd6c UD |
1177 | if (ctype->nr_charclass == MAX_NR_CHARCLASS) |
1178 | /* Exit code 2 is prescribed in P1003.2b. */ | |
f2b98f97 | 1179 | WITH_CUR_LOCALE (error (2, 0, _("\ |
5d431a3e | 1180 | implementation limit: no more than %Zd character classes allowed"), |
f2b98f97 | 1181 | MAX_NR_CHARCLASS)); |
19bc17a9 | 1182 | |
4b10dd6c | 1183 | ctype->classnames[ctype->nr_charclass++] = name; |
19bc17a9 RM |
1184 | } |
1185 | ||
1186 | ||
4b10dd6c UD |
1187 | static void |
1188 | ctype_map_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
47e8b443 | 1189 | const char *name, const struct charmap_t *charmap) |
19bc17a9 | 1190 | { |
4b10dd6c | 1191 | size_t max_chars = 0; |
ba1ffaa1 | 1192 | size_t cnt; |
19bc17a9 | 1193 | |
4b10dd6c | 1194 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
19bc17a9 | 1195 | { |
4b10dd6c UD |
1196 | if (strcmp (ctype->mapnames[cnt], name) == 0) |
1197 | break; | |
1198 | ||
1199 | if (max_chars < ctype->map_collection_max[cnt]) | |
1200 | max_chars = ctype->map_collection_max[cnt]; | |
19bc17a9 RM |
1201 | } |
1202 | ||
4b10dd6c UD |
1203 | if (cnt < ctype->map_collection_nr) |
1204 | { | |
1205 | lr_error (lr, _("character map `%s' already defined"), name); | |
1206 | return; | |
1207 | } | |
19bc17a9 | 1208 | |
4b10dd6c UD |
1209 | if (ctype->map_collection_nr == MAX_NR_CHARMAP) |
1210 | /* Exit code 2 is prescribed in P1003.2b. */ | |
f2b98f97 | 1211 | WITH_CUR_LOCALE (error (2, 0, _("\ |
4b10dd6c | 1212 | implementation limit: no more than %d character maps allowed"), |
f2b98f97 | 1213 | MAX_NR_CHARMAP)); |
19bc17a9 | 1214 | |
4b10dd6c UD |
1215 | ctype->mapnames[cnt] = name; |
1216 | ||
1217 | if (max_chars == 0) | |
1218 | ctype->map_collection_max[cnt] = charmap->mb_cur_max == 1 ? 256 : 512; | |
1219 | else | |
1220 | ctype->map_collection_max[cnt] = max_chars; | |
1221 | ||
1222 | ctype->map_collection[cnt] = (uint32_t *) | |
5866b131 | 1223 | xcalloc (sizeof (uint32_t), ctype->map_collection_max[cnt]); |
4b10dd6c | 1224 | ctype->map_collection_act[cnt] = 256; |
19bc17a9 | 1225 | |
4b10dd6c | 1226 | ++ctype->map_collection_nr; |
19bc17a9 RM |
1227 | } |
1228 | ||
1229 | ||
4b10dd6c | 1230 | /* We have to be prepared that TABLE, MAX, and ACT can be NULL. This |
42d7c593 | 1231 | is possible if we only want to extend the name array. */ |
4b10dd6c UD |
1232 | static uint32_t * |
1233 | find_idx (struct locale_ctype_t *ctype, uint32_t **table, size_t *max, | |
1234 | size_t *act, uint32_t idx) | |
19bc17a9 | 1235 | { |
4b10dd6c | 1236 | size_t cnt; |
19bc17a9 | 1237 | |
4b10dd6c UD |
1238 | if (idx < 256) |
1239 | return table == NULL ? NULL : &(*table)[idx]; | |
19bc17a9 | 1240 | |
601d2942 UD |
1241 | /* Use the charnames_idx lookup table instead of the slow search loop. */ |
1242 | #if 1 | |
1243 | cnt = idx_table_get (&ctype->charnames_idx, idx); | |
bd75759f | 1244 | if (cnt == EMPTY) |
601d2942 UD |
1245 | /* Not found. */ |
1246 | cnt = ctype->charnames_act; | |
1247 | #else | |
1248 | for (cnt = 256; cnt < ctype->charnames_act; ++cnt) | |
1249 | if (ctype->charnames[cnt] == idx) | |
1250 | break; | |
1251 | #endif | |
19bc17a9 | 1252 | |
4b10dd6c UD |
1253 | /* We have to distinguish two cases: the name is found or not. */ |
1254 | if (cnt == ctype->charnames_act) | |
1255 | { | |
1256 | /* Extend the name array. */ | |
1257 | if (ctype->charnames_act == ctype->charnames_max) | |
1258 | { | |
1259 | ctype->charnames_max *= 2; | |
5866b131 | 1260 | ctype->charnames = (uint32_t *) |
4b10dd6c | 1261 | xrealloc (ctype->charnames, |
5866b131 | 1262 | sizeof (uint32_t) * ctype->charnames_max); |
4b10dd6c UD |
1263 | } |
1264 | ctype->charnames[ctype->charnames_act++] = idx; | |
601d2942 | 1265 | idx_table_add (&ctype->charnames_idx, idx, cnt); |
4b10dd6c | 1266 | } |
19bc17a9 | 1267 | |
4b10dd6c UD |
1268 | if (table == NULL) |
1269 | /* We have done everything we are asked to do. */ | |
1270 | return NULL; | |
19bc17a9 | 1271 | |
4a9dcff1 UD |
1272 | if (max == NULL) |
1273 | /* The caller does not want to extend the table. */ | |
1274 | return (cnt >= *act ? NULL : &(*table)[cnt]); | |
1275 | ||
4b10dd6c UD |
1276 | if (cnt >= *act) |
1277 | { | |
1278 | if (cnt >= *max) | |
1279 | { | |
1280 | size_t old_max = *max; | |
1281 | do | |
1282 | *max *= 2; | |
1283 | while (*max <= cnt); | |
19bc17a9 | 1284 | |
4b10dd6c | 1285 | *table = |
5866b131 | 1286 | (uint32_t *) xrealloc (*table, *max * sizeof (uint32_t)); |
4b10dd6c UD |
1287 | memset (&(*table)[old_max], '\0', |
1288 | (*max - old_max) * sizeof (uint32_t)); | |
1289 | } | |
19bc17a9 | 1290 | |
76e680a8 | 1291 | *act = cnt + 1; |
4b10dd6c | 1292 | } |
19bc17a9 | 1293 | |
4b10dd6c | 1294 | return &(*table)[cnt]; |
19bc17a9 RM |
1295 | } |
1296 | ||
1297 | ||
4b10dd6c | 1298 | static int |
47e8b443 | 1299 | get_character (struct token *now, const struct charmap_t *charmap, |
4b10dd6c UD |
1300 | struct repertoire_t *repertoire, |
1301 | struct charseq **seqp, uint32_t *wchp) | |
19bc17a9 | 1302 | { |
4b10dd6c UD |
1303 | if (now->tok == tok_bsymbol) |
1304 | { | |
1305 | /* This will hopefully be the normal case. */ | |
1306 | *wchp = repertoire_find_value (repertoire, now->val.str.startmb, | |
1307 | now->val.str.lenmb); | |
1308 | *seqp = charmap_find_value (charmap, now->val.str.startmb, | |
1309 | now->val.str.lenmb); | |
1310 | } | |
1311 | else if (now->tok == tok_ucs4) | |
1312 | { | |
f0a4b6b1 UD |
1313 | char utmp[10]; |
1314 | ||
1315 | snprintf (utmp, sizeof (utmp), "U%08X", now->val.ucs4); | |
1316 | *seqp = charmap_find_value (charmap, utmp, 9); | |
1317 | ||
1318 | if (*seqp == NULL) | |
1319 | *seqp = repertoire_find_seq (repertoire, now->val.ucs4); | |
19bc17a9 | 1320 | |
4b10dd6c UD |
1321 | if (*seqp == NULL) |
1322 | { | |
1323 | /* Compute the value in the charmap from the UCS value. */ | |
1324 | const char *symbol = repertoire_find_symbol (repertoire, | |
1325 | now->val.ucs4); | |
19bc17a9 | 1326 | |
4b10dd6c UD |
1327 | if (symbol == NULL) |
1328 | *seqp = NULL; | |
1329 | else | |
1330 | *seqp = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1331 | |
4b10dd6c UD |
1332 | if (*seqp == NULL) |
1333 | { | |
723faa38 UD |
1334 | if (repertoire != NULL) |
1335 | { | |
1336 | /* Insert a negative entry. */ | |
1337 | static const struct charseq negative | |
1338 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1339 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, | |
1340 | sizeof (uint32_t)); | |
1341 | *newp = now->val.ucs4; | |
1342 | ||
1343 | insert_entry (&repertoire->seq_table, newp, | |
1344 | sizeof (uint32_t), (void *) &negative); | |
1345 | } | |
4b10dd6c UD |
1346 | } |
1347 | else | |
1348 | (*seqp)->ucs4 = now->val.ucs4; | |
1349 | } | |
1350 | else if ((*seqp)->ucs4 != now->val.ucs4) | |
1351 | *seqp = NULL; | |
19bc17a9 | 1352 | |
4b10dd6c UD |
1353 | *wchp = now->val.ucs4; |
1354 | } | |
1355 | else if (now->tok == tok_charcode) | |
1356 | { | |
1357 | /* We must map from the byte code to UCS4. */ | |
1358 | *seqp = charmap_find_symbol (charmap, now->val.str.startmb, | |
1359 | now->val.str.lenmb); | |
19bc17a9 | 1360 | |
4b10dd6c UD |
1361 | if (*seqp == NULL) |
1362 | *wchp = ILLEGAL_CHAR_VALUE; | |
1363 | else | |
1364 | { | |
1365 | if ((*seqp)->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1366 | (*seqp)->ucs4 = repertoire_find_value (repertoire, (*seqp)->name, | |
1367 | strlen ((*seqp)->name)); | |
1368 | *wchp = (*seqp)->ucs4; | |
1369 | } | |
1370 | } | |
1371 | else | |
1372 | return 1; | |
19bc17a9 RM |
1373 | |
1374 | return 0; | |
1375 | } | |
1376 | ||
1377 | ||
a0dc5206 UD |
1378 | /* Ellipsis like in `<foo123>..<foo12a>' or `<j1234>....<j1245>' and |
1379 | the .(2). counterparts. */ | |
4b10dd6c UD |
1380 | static void |
1381 | charclass_symbolic_ellipsis (struct linereader *ldfile, | |
1382 | struct locale_ctype_t *ctype, | |
47e8b443 | 1383 | const struct charmap_t *charmap, |
4b10dd6c UD |
1384 | struct repertoire_t *repertoire, |
1385 | struct token *now, | |
1386 | const char *last_str, | |
1387 | unsigned long int class256_bit, | |
1388 | unsigned long int class_bit, int base, | |
a0dc5206 | 1389 | int ignore_content, int handle_digits, int step) |
19bc17a9 | 1390 | { |
4b10dd6c UD |
1391 | const char *nowstr = now->val.str.startmb; |
1392 | char tmp[now->val.str.lenmb + 1]; | |
1393 | const char *cp; | |
1394 | char *endp; | |
1395 | unsigned long int from; | |
1396 | unsigned long int to; | |
19bc17a9 | 1397 | |
4b10dd6c UD |
1398 | /* We have to compute the ellipsis values using the symbolic names. */ |
1399 | assert (last_str != NULL); | |
1400 | ||
1401 | if (strlen (last_str) != now->val.str.lenmb) | |
19bc17a9 | 1402 | { |
4b10dd6c UD |
1403 | invalid_range: |
1404 | lr_error (ldfile, | |
c69136ae | 1405 | _("`%s' and `%.*s' are not valid names for symbolic range"), |
f6ada7ad | 1406 | last_str, (int) now->val.str.lenmb, nowstr); |
4b10dd6c | 1407 | return; |
19bc17a9 RM |
1408 | } |
1409 | ||
4b10dd6c UD |
1410 | if (memcmp (last_str, nowstr, now->val.str.lenmb) == 0) |
1411 | /* Nothing to do, the names are the same. */ | |
1412 | return; | |
19bc17a9 | 1413 | |
4b10dd6c UD |
1414 | for (cp = last_str; *cp == *(nowstr + (cp - last_str)); ++cp) |
1415 | ; | |
19bc17a9 | 1416 | |
4b10dd6c UD |
1417 | errno = 0; |
1418 | from = strtoul (cp, &endp, base); | |
1419 | if ((from == UINT_MAX && errno == ERANGE) || *endp != '\0') | |
1420 | goto invalid_range; | |
19bc17a9 | 1421 | |
4b10dd6c | 1422 | to = strtoul (nowstr + (cp - last_str), &endp, base); |
549b3c3a UD |
1423 | if ((to == UINT_MAX && errno == ERANGE) |
1424 | || (endp - nowstr) != now->val.str.lenmb || from >= to) | |
4b10dd6c | 1425 | goto invalid_range; |
19bc17a9 | 1426 | |
4b10dd6c UD |
1427 | /* OK, we have a range FROM - TO. Now we can create the symbolic names. */ |
1428 | if (!ignore_content) | |
1429 | { | |
1430 | now->val.str.startmb = tmp; | |
a0dc5206 | 1431 | while ((from += step) <= to) |
4b10dd6c UD |
1432 | { |
1433 | struct charseq *seq; | |
1434 | uint32_t wch; | |
19bc17a9 | 1435 | |
9068de33 | 1436 | sprintf (tmp, (base == 10 ? "%.*s%0*ld" : "%.*s%0*lX"), |
4e8d52c9 AJ |
1437 | (int) (cp - last_str), last_str, |
1438 | (int) (now->val.str.lenmb - (cp - last_str)), | |
9068de33 | 1439 | from); |
19bc17a9 | 1440 | |
4b10dd6c UD |
1441 | get_character (now, charmap, repertoire, &seq, &wch); |
1442 | ||
1443 | if (seq != NULL && seq->nbytes == 1) | |
1444 | /* Yep, we can store information about this byte sequence. */ | |
1445 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
19bc17a9 | 1446 | |
4b10dd6c UD |
1447 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) |
1448 | /* We have the UCS4 position. */ | |
1449 | *find_idx (ctype, &ctype->class_collection, | |
1450 | &ctype->class_collection_max, | |
1451 | &ctype->class_collection_act, wch) |= class_bit; | |
19bc17a9 | 1452 | |
4b10dd6c UD |
1453 | if (handle_digits == 1) |
1454 | { | |
1455 | /* We must store the digit values. */ | |
1456 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1457 | { | |
1458 | ctype->mbdigits_max *= 2; | |
1459 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1460 | (ctype->mbdigits_max | |
1461 | * sizeof (char *))); | |
1462 | ctype->wcdigits_max *= 2; | |
1463 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1464 | (ctype->wcdigits_max | |
1465 | * sizeof (uint32_t))); | |
1466 | } | |
1467 | ||
1468 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1469 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1470 | } | |
1471 | else if (handle_digits == 2) | |
1472 | { | |
1473 | /* We must store the digit values. */ | |
1474 | if (ctype->outdigits_act >= 10) | |
1475 | { | |
1476 | lr_error (ldfile, _("\ | |
1477 | %s: field `%s' does not contain exactly ten entries"), | |
1478 | "LC_CTYPE", "outdigit"); | |
1479 | return; | |
1480 | } | |
1481 | ||
1482 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1483 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1484 | ++ctype->outdigits_act; | |
1485 | } | |
1486 | } | |
1487 | } | |
19bc17a9 RM |
1488 | } |
1489 | ||
1490 | ||
a0dc5206 | 1491 | /* Ellipsis like in `<U1234>..<U2345>' or `<U1234>..(2)..<U2345>'. */ |
4b10dd6c UD |
1492 | static void |
1493 | charclass_ucs4_ellipsis (struct linereader *ldfile, | |
1494 | struct locale_ctype_t *ctype, | |
47e8b443 | 1495 | const struct charmap_t *charmap, |
4b10dd6c UD |
1496 | struct repertoire_t *repertoire, |
1497 | struct token *now, uint32_t last_wch, | |
1498 | unsigned long int class256_bit, | |
1499 | unsigned long int class_bit, int ignore_content, | |
a0dc5206 | 1500 | int handle_digits, int step) |
19bc17a9 | 1501 | { |
4b10dd6c | 1502 | if (last_wch > now->val.ucs4) |
19bc17a9 | 1503 | { |
4b10dd6c UD |
1504 | lr_error (ldfile, _("\ |
1505 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1506 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, now->val.ucs4, | |
1507 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, last_wch); | |
19bc17a9 RM |
1508 | return; |
1509 | } | |
1510 | ||
4b10dd6c | 1511 | if (!ignore_content) |
a0dc5206 | 1512 | while ((last_wch += step) <= now->val.ucs4) |
4b10dd6c UD |
1513 | { |
1514 | /* We have to find out whether there is a byte sequence corresponding | |
1515 | to this UCS4 value. */ | |
f0a4b6b1 UD |
1516 | struct charseq *seq; |
1517 | char utmp[10]; | |
1518 | ||
1519 | snprintf (utmp, sizeof (utmp), "U%08X", last_wch); | |
1520 | seq = charmap_find_value (charmap, utmp, 9); | |
a0dc5206 UD |
1521 | if (seq == NULL) |
1522 | { | |
1523 | snprintf (utmp, sizeof (utmp), "U%04X", last_wch); | |
1524 | seq = charmap_find_value (charmap, utmp, 5); | |
1525 | } | |
f0a4b6b1 UD |
1526 | |
1527 | if (seq == NULL) | |
1528 | /* Try looking in the repertoire map. */ | |
1529 | seq = repertoire_find_seq (repertoire, last_wch); | |
19bc17a9 | 1530 | |
4b10dd6c UD |
1531 | /* If this is the first time we look for this sequence create a new |
1532 | entry. */ | |
1533 | if (seq == NULL) | |
1534 | { | |
f0a4b6b1 UD |
1535 | static const struct charseq negative |
1536 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
19bc17a9 | 1537 | |
f0a4b6b1 UD |
1538 | /* Find the symbolic name for this UCS4 value. */ |
1539 | if (repertoire != NULL) | |
4b10dd6c | 1540 | { |
f0a4b6b1 UD |
1541 | const char *symbol = repertoire_find_symbol (repertoire, |
1542 | last_wch); | |
5866b131 UD |
1543 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, |
1544 | sizeof (uint32_t)); | |
f0a4b6b1 UD |
1545 | *newp = last_wch; |
1546 | ||
1547 | if (symbol != NULL) | |
1548 | /* We have a name, now search the multibyte value. */ | |
1549 | seq = charmap_find_value (charmap, symbol, strlen (symbol)); | |
1550 | ||
1551 | if (seq == NULL) | |
1552 | /* We have to create a fake entry. */ | |
1553 | seq = (struct charseq *) &negative; | |
1554 | else | |
1555 | seq->ucs4 = last_wch; | |
1556 | ||
5866b131 UD |
1557 | insert_entry (&repertoire->seq_table, newp, sizeof (uint32_t), |
1558 | seq); | |
4b10dd6c UD |
1559 | } |
1560 | else | |
f0a4b6b1 UD |
1561 | /* We have to create a fake entry. */ |
1562 | seq = (struct charseq *) &negative; | |
4b10dd6c UD |
1563 | } |
1564 | ||
1565 | /* We have a name, now search the multibyte value. */ | |
1566 | if (seq->ucs4 == last_wch && seq->nbytes == 1) | |
1567 | /* Yep, we can store information about this byte sequence. */ | |
1568 | ctype->class256_collection[(size_t) seq->bytes[0]] | |
1569 | |= class256_bit; | |
1570 | ||
1571 | /* And of course we have the UCS4 position. */ | |
5866b131 | 1572 | if (class_bit != 0) |
4b10dd6c UD |
1573 | *find_idx (ctype, &ctype->class_collection, |
1574 | &ctype->class_collection_max, | |
1575 | &ctype->class_collection_act, last_wch) |= class_bit; | |
1576 | ||
1577 | if (handle_digits == 1) | |
1578 | { | |
1579 | /* We must store the digit values. */ | |
1580 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1581 | { | |
1582 | ctype->mbdigits_max *= 2; | |
1583 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1584 | (ctype->mbdigits_max | |
1585 | * sizeof (char *))); | |
1586 | ctype->wcdigits_max *= 2; | |
1587 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1588 | (ctype->wcdigits_max | |
1589 | * sizeof (uint32_t))); | |
1590 | } | |
1591 | ||
1592 | ctype->mbdigits[ctype->mbdigits_act++] = (seq->ucs4 == last_wch | |
1593 | ? seq : NULL); | |
1594 | ctype->wcdigits[ctype->wcdigits_act++] = last_wch; | |
1595 | } | |
1596 | else if (handle_digits == 2) | |
1597 | { | |
1598 | /* We must store the digit values. */ | |
1599 | if (ctype->outdigits_act >= 10) | |
1600 | { | |
1601 | lr_error (ldfile, _("\ | |
1602 | %s: field `%s' does not contain exactly ten entries"), | |
1603 | "LC_CTYPE", "outdigit"); | |
1604 | return; | |
1605 | } | |
19bc17a9 | 1606 | |
4b10dd6c UD |
1607 | ctype->mboutdigits[ctype->outdigits_act] = (seq->ucs4 == last_wch |
1608 | ? seq : NULL); | |
1609 | ctype->wcoutdigits[ctype->outdigits_act] = last_wch; | |
1610 | ++ctype->outdigits_act; | |
1611 | } | |
1612 | } | |
19bc17a9 RM |
1613 | } |
1614 | ||
1615 | ||
4b10dd6c | 1616 | /* Ellipsis as in `/xea/x12.../xea/x34'. */ |
19bc17a9 | 1617 | static void |
4b10dd6c UD |
1618 | charclass_charcode_ellipsis (struct linereader *ldfile, |
1619 | struct locale_ctype_t *ctype, | |
47e8b443 | 1620 | const struct charmap_t *charmap, |
4b10dd6c UD |
1621 | struct repertoire_t *repertoire, |
1622 | struct token *now, char *last_charcode, | |
1623 | uint32_t last_charcode_len, | |
1624 | unsigned long int class256_bit, | |
1625 | unsigned long int class_bit, int ignore_content, | |
1626 | int handle_digits) | |
19bc17a9 | 1627 | { |
4b10dd6c UD |
1628 | /* First check whether the to-value is larger. */ |
1629 | if (now->val.charcode.nbytes != last_charcode_len) | |
1630 | { | |
1631 | lr_error (ldfile, _("\ | |
379ed351 | 1632 | start and end character sequence of range must have the same length")); |
4b10dd6c UD |
1633 | return; |
1634 | } | |
19bc17a9 | 1635 | |
4b10dd6c | 1636 | if (memcmp (last_charcode, now->val.charcode.bytes, last_charcode_len) > 0) |
19bc17a9 | 1637 | { |
4b10dd6c UD |
1638 | lr_error (ldfile, _("\ |
1639 | to-value character sequence is smaller than from-value sequence")); | |
19bc17a9 RM |
1640 | return; |
1641 | } | |
1642 | ||
4b10dd6c UD |
1643 | if (!ignore_content) |
1644 | { | |
1645 | do | |
1646 | { | |
1647 | /* Increment the byte sequence value. */ | |
1648 | struct charseq *seq; | |
1649 | uint32_t wch; | |
1650 | int i; | |
1651 | ||
1652 | for (i = last_charcode_len - 1; i >= 0; --i) | |
1653 | if (++last_charcode[i] != 0) | |
1654 | break; | |
1655 | ||
1656 | if (last_charcode_len == 1) | |
1657 | /* Of course we have the charcode value. */ | |
1658 | ctype->class256_collection[(size_t) last_charcode[0]] | |
1659 | |= class256_bit; | |
1660 | ||
1661 | /* Find the symbolic name. */ | |
1662 | seq = charmap_find_symbol (charmap, last_charcode, | |
1663 | last_charcode_len); | |
1664 | if (seq != NULL) | |
1665 | { | |
1666 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1667 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1668 | strlen (seq->name)); | |
f0a4b6b1 | 1669 | wch = seq == NULL ? ILLEGAL_CHAR_VALUE : seq->ucs4; |
4b10dd6c UD |
1670 | |
1671 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) | |
1672 | *find_idx (ctype, &ctype->class_collection, | |
1673 | &ctype->class_collection_max, | |
1674 | &ctype->class_collection_act, wch) |= class_bit; | |
1675 | } | |
1676 | else | |
1677 | wch = ILLEGAL_CHAR_VALUE; | |
19bc17a9 | 1678 | |
4b10dd6c UD |
1679 | if (handle_digits == 1) |
1680 | { | |
1681 | /* We must store the digit values. */ | |
1682 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1683 | { | |
1684 | ctype->mbdigits_max *= 2; | |
1685 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1686 | (ctype->mbdigits_max | |
1687 | * sizeof (char *))); | |
1688 | ctype->wcdigits_max *= 2; | |
1689 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1690 | (ctype->wcdigits_max | |
1691 | * sizeof (uint32_t))); | |
1692 | } | |
1693 | ||
1694 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1695 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1696 | seq->nbytes = last_charcode_len; | |
1697 | ||
1698 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1699 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1700 | } | |
1701 | else if (handle_digits == 2) | |
1702 | { | |
1703 | struct charseq *seq; | |
1704 | /* We must store the digit values. */ | |
1705 | if (ctype->outdigits_act >= 10) | |
1706 | { | |
1707 | lr_error (ldfile, _("\ | |
1708 | %s: field `%s' does not contain exactly ten entries"), | |
1709 | "LC_CTYPE", "outdigit"); | |
1710 | return; | |
1711 | } | |
1712 | ||
1713 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1714 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1715 | seq->nbytes = last_charcode_len; | |
1716 | ||
1717 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1718 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1719 | ++ctype->outdigits_act; | |
1720 | } | |
1721 | } | |
1722 | while (memcmp (last_charcode, now->val.charcode.bytes, | |
1723 | last_charcode_len) != 0); | |
1724 | } | |
19bc17a9 RM |
1725 | } |
1726 | ||
1727 | ||
47e8b443 UD |
1728 | static uint32_t * |
1729 | find_translit2 (struct locale_ctype_t *ctype, const struct charmap_t *charmap, | |
1730 | uint32_t wch) | |
1731 | { | |
1732 | struct translit_t *trunp = ctype->translit; | |
1733 | struct translit_ignore_t *tirunp = ctype->translit_ignore; | |
1734 | ||
1735 | while (trunp != NULL) | |
1736 | { | |
1737 | /* XXX We simplify things here. The transliterations we look | |
1738 | for are only allowed to have one character. */ | |
1739 | if (trunp->from[0] == wch && trunp->from[1] == 0) | |
1740 | { | |
1741 | /* Found it. Now look for a transliteration which can be | |
1742 | represented with the character set. */ | |
1743 | struct translit_to_t *torunp = trunp->to; | |
1744 | ||
1745 | while (torunp != NULL) | |
1746 | { | |
1747 | int i; | |
1748 | ||
1749 | for (i = 0; torunp->str[i] != 0; ++i) | |
1750 | { | |
1751 | char utmp[10]; | |
1752 | ||
1753 | snprintf (utmp, sizeof (utmp), "U%08X", torunp->str[i]); | |
1754 | if (charmap_find_value (charmap, utmp, 9) == NULL) | |
1755 | /* This character cannot be represented. */ | |
1756 | break; | |
1757 | } | |
1758 | ||
1759 | if (torunp->str[i] == 0) | |
1760 | return torunp->str; | |
1761 | ||
1762 | torunp = torunp->next; | |
1763 | } | |
1764 | ||
1765 | break; | |
1766 | } | |
1767 | ||
1768 | trunp = trunp->next; | |
1769 | } | |
1770 | ||
1771 | /* Check for ignored chars. */ | |
1772 | while (tirunp != NULL) | |
1773 | { | |
1774 | if (tirunp->from <= wch && tirunp->to >= wch) | |
1775 | { | |
1776 | uint32_t wi; | |
1777 | ||
1778 | for (wi = tirunp->from; wi <= wch; wi += tirunp->step) | |
1779 | if (wi == wch) | |
1780 | return (uint32_t []) { 0 }; | |
1781 | } | |
1782 | } | |
1783 | ||
1784 | /* Nothing found. */ | |
1785 | return NULL; | |
1786 | } | |
1787 | ||
1788 | ||
1789 | uint32_t * | |
1790 | find_translit (struct localedef_t *locale, const struct charmap_t *charmap, | |
1791 | uint32_t wch) | |
1792 | { | |
1793 | struct locale_ctype_t *ctype; | |
1794 | uint32_t *result = NULL; | |
1795 | ||
1796 | assert (locale != NULL); | |
1797 | ctype = locale->categories[LC_CTYPE].ctype; | |
1798 | ||
b037a293 UD |
1799 | if (ctype == NULL) |
1800 | return NULL; | |
1801 | ||
47e8b443 UD |
1802 | if (ctype->translit != NULL) |
1803 | result = find_translit2 (ctype, charmap, wch); | |
1804 | ||
1805 | if (result == NULL) | |
1806 | { | |
1807 | struct translit_include_t *irunp = ctype->translit_include; | |
1808 | ||
1809 | while (irunp != NULL && result == NULL) | |
1810 | { | |
1811 | result = find_translit (find_locale (CTYPE_LOCALE, | |
1812 | irunp->copy_locale, | |
1813 | irunp->copy_repertoire, | |
1814 | charmap), | |
1815 | charmap, wch); | |
1816 | irunp = irunp->next; | |
1817 | } | |
1818 | } | |
1819 | ||
1820 | return result; | |
1821 | } | |
1822 | ||
1823 | ||
4b10dd6c UD |
1824 | /* Read one transliteration entry. */ |
1825 | static uint32_t * | |
1826 | read_widestring (struct linereader *ldfile, struct token *now, | |
47e8b443 UD |
1827 | const struct charmap_t *charmap, |
1828 | struct repertoire_t *repertoire) | |
19bc17a9 | 1829 | { |
4b10dd6c | 1830 | uint32_t *wstr; |
19bc17a9 | 1831 | |
4b10dd6c UD |
1832 | if (now->tok == tok_default_missing) |
1833 | /* The special name "" will denote this case. */ | |
5866b131 | 1834 | wstr = ((uint32_t *) { 0 }); |
4b10dd6c | 1835 | else if (now->tok == tok_bsymbol) |
19bc17a9 | 1836 | { |
4b10dd6c | 1837 | /* Get the value from the repertoire. */ |
a673fbcb | 1838 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1839 | wstr[0] = repertoire_find_value (repertoire, now->val.str.startmb, |
1840 | now->val.str.lenmb); | |
1841 | if (wstr[0] == ILLEGAL_CHAR_VALUE) | |
f0a4b6b1 UD |
1842 | { |
1843 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1844 | free (wstr); | |
1845 | return NULL; | |
1846 | } | |
4b10dd6c UD |
1847 | |
1848 | wstr[1] = 0; | |
19bc17a9 | 1849 | } |
4b10dd6c | 1850 | else if (now->tok == tok_ucs4) |
19bc17a9 | 1851 | { |
a673fbcb | 1852 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1853 | wstr[0] = now->val.ucs4; |
1854 | wstr[1] = 0; | |
1855 | } | |
1856 | else if (now->tok == tok_charcode) | |
1857 | { | |
1858 | /* Argh, we have to convert to the symbol name first and then to the | |
1859 | UCS4 value. */ | |
1860 | struct charseq *seq = charmap_find_symbol (charmap, | |
1861 | now->val.str.startmb, | |
1862 | now->val.str.lenmb); | |
1863 | if (seq == NULL) | |
1864 | /* Cannot find the UCS4 value. */ | |
1865 | return NULL; | |
1866 | ||
1867 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1868 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1869 | strlen (seq->name)); | |
1870 | if (seq->ucs4 == ILLEGAL_CHAR_VALUE) | |
1871 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1872 | return NULL; | |
1873 | ||
a673fbcb | 1874 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1875 | wstr[0] = seq->ucs4; |
1876 | wstr[1] = 0; | |
1877 | } | |
1878 | else if (now->tok == tok_string) | |
1879 | { | |
1880 | wstr = now->val.str.startwc; | |
a673fbcb | 1881 | if (wstr == NULL || wstr[0] == 0) |
4b10dd6c UD |
1882 | return NULL; |
1883 | } | |
1884 | else | |
1885 | { | |
1886 | if (now->tok != tok_eol && now->tok != tok_eof) | |
1887 | lr_ignore_rest (ldfile, 0); | |
1888 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
1889 | return (uint32_t *) -1l; | |
19bc17a9 RM |
1890 | } |
1891 | ||
4b10dd6c UD |
1892 | return wstr; |
1893 | } | |
19bc17a9 | 1894 | |
19bc17a9 | 1895 | |
4b10dd6c UD |
1896 | static void |
1897 | read_translit_entry (struct linereader *ldfile, struct locale_ctype_t *ctype, | |
47e8b443 | 1898 | struct token *now, const struct charmap_t *charmap, |
4b10dd6c UD |
1899 | struct repertoire_t *repertoire) |
1900 | { | |
1901 | uint32_t *from_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1902 | struct translit_t *result; | |
1903 | struct translit_to_t **top; | |
a673fbcb | 1904 | struct obstack *ob = &ctype->mempool; |
4b10dd6c UD |
1905 | int first; |
1906 | int ignore; | |
1907 | ||
1908 | if (from_wstr == NULL) | |
1909 | /* There is no valid from string. */ | |
1910 | return; | |
19bc17a9 | 1911 | |
4b10dd6c UD |
1912 | result = (struct translit_t *) obstack_alloc (ob, |
1913 | sizeof (struct translit_t)); | |
1914 | result->from = from_wstr; | |
a673fbcb UD |
1915 | result->fname = ldfile->fname; |
1916 | result->lineno = ldfile->lineno; | |
4b10dd6c UD |
1917 | result->next = NULL; |
1918 | result->to = NULL; | |
1919 | top = &result->to; | |
1920 | first = 1; | |
1921 | ignore = 0; | |
1922 | ||
1923 | while (1) | |
1924 | { | |
1925 | uint32_t *to_wstr; | |
1926 | ||
1927 | /* Next we have one or more transliterations. They are | |
1928 | separated by semicolons. */ | |
47e8b443 | 1929 | now = lr_token (ldfile, charmap, NULL, repertoire, verbose); |
4b10dd6c UD |
1930 | |
1931 | if (!first && (now->tok == tok_semicolon || now->tok == tok_eol)) | |
1932 | { | |
1933 | /* One string read. */ | |
1934 | const uint32_t zero = 0; | |
1935 | ||
1936 | if (!ignore) | |
1937 | { | |
1938 | obstack_grow (ob, &zero, 4); | |
1939 | to_wstr = obstack_finish (ob); | |
1940 | ||
1941 | *top = obstack_alloc (ob, sizeof (struct translit_to_t)); | |
1942 | (*top)->str = to_wstr; | |
1943 | (*top)->next = NULL; | |
1944 | } | |
1945 | ||
1946 | if (now->tok == tok_eol) | |
1947 | { | |
1948 | result->next = ctype->translit; | |
1949 | ctype->translit = result; | |
1950 | return; | |
1951 | } | |
1952 | ||
1953 | if (!ignore) | |
1954 | top = &(*top)->next; | |
1955 | ignore = 0; | |
1956 | } | |
1957 | else | |
1958 | { | |
1959 | to_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1960 | if (to_wstr == (uint32_t *) -1l) | |
1961 | { | |
1962 | /* An error occurred. */ | |
1963 | obstack_free (ob, result); | |
1964 | return; | |
1965 | } | |
1966 | ||
1967 | if (to_wstr == NULL) | |
1968 | ignore = 1; | |
1969 | else | |
1970 | /* This value is usable. */ | |
1971 | obstack_grow (ob, to_wstr, wcslen ((wchar_t *) to_wstr) * 4); | |
19bc17a9 | 1972 | |
4b10dd6c UD |
1973 | first = 0; |
1974 | } | |
1975 | } | |
19bc17a9 RM |
1976 | } |
1977 | ||
1978 | ||
a673fbcb UD |
1979 | static void |
1980 | read_translit_ignore_entry (struct linereader *ldfile, | |
1981 | struct locale_ctype_t *ctype, | |
47e8b443 | 1982 | const struct charmap_t *charmap, |
a673fbcb UD |
1983 | struct repertoire_t *repertoire) |
1984 | { | |
1985 | /* We expect a semicolon-separated list of characters we ignore. We are | |
1986 | only interested in the wide character definitions. These must be | |
1987 | single characters, possibly defining a range when an ellipsis is used. */ | |
1988 | while (1) | |
1989 | { | |
47e8b443 UD |
1990 | struct token *now = lr_token (ldfile, charmap, NULL, repertoire, |
1991 | verbose); | |
a673fbcb UD |
1992 | struct translit_ignore_t *newp; |
1993 | uint32_t from; | |
1994 | ||
1995 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1996 | { | |
1997 | lr_error (ldfile, | |
1998 | _("premature end of `translit_ignore' definition")); | |
1999 | return; | |
2000 | } | |
2001 | ||
2002 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
2003 | { | |
2004 | lr_error (ldfile, _("syntax error")); | |
2005 | lr_ignore_rest (ldfile, 0); | |
2006 | return; | |
2007 | } | |
2008 | ||
2009 | if (now->tok == tok_ucs4) | |
2010 | from = now->val.ucs4; | |
2011 | else | |
f0a4b6b1 UD |
2012 | /* Try to get the value. */ |
2013 | from = repertoire_find_value (repertoire, now->val.str.startmb, | |
2014 | now->val.str.lenmb); | |
a673fbcb UD |
2015 | |
2016 | if (from == ILLEGAL_CHAR_VALUE) | |
2017 | { | |
2018 | lr_error (ldfile, "invalid character name"); | |
2019 | newp = NULL; | |
2020 | } | |
2021 | else | |
2022 | { | |
2023 | newp = (struct translit_ignore_t *) | |
2024 | obstack_alloc (&ctype->mempool, sizeof (struct translit_ignore_t)); | |
2025 | newp->from = from; | |
2026 | newp->to = from; | |
a0dc5206 | 2027 | newp->step = 1; |
a673fbcb UD |
2028 | |
2029 | newp->next = ctype->translit_ignore; | |
2030 | ctype->translit_ignore = newp; | |
2031 | } | |
2032 | ||
2033 | /* Now we expect either a semicolon, an ellipsis, or the end of the | |
2034 | line. */ | |
47e8b443 | 2035 | now = lr_token (ldfile, charmap, NULL, repertoire, verbose); |
a673fbcb | 2036 | |
a0dc5206 | 2037 | if (now->tok == tok_ellipsis2 || now->tok == tok_ellipsis2_2) |
a673fbcb UD |
2038 | { |
2039 | /* XXX Should we bother implementing `....'? `...' certainly | |
2040 | will not be implemented. */ | |
2041 | uint32_t to; | |
a0dc5206 | 2042 | int step = now->tok == tok_ellipsis2_2 ? 2 : 1; |
a673fbcb | 2043 | |
47e8b443 | 2044 | now = lr_token (ldfile, charmap, NULL, repertoire, verbose); |
a673fbcb UD |
2045 | |
2046 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2047 | { | |
2048 | lr_error (ldfile, | |
2049 | _("premature end of `translit_ignore' definition")); | |
2050 | return; | |
2051 | } | |
2052 | ||
2053 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
2054 | { | |
2055 | lr_error (ldfile, _("syntax error")); | |
2056 | lr_ignore_rest (ldfile, 0); | |
2057 | return; | |
2058 | } | |
2059 | ||
2060 | if (now->tok == tok_ucs4) | |
2061 | to = now->val.ucs4; | |
2062 | else | |
f0a4b6b1 UD |
2063 | /* Try to get the value. */ |
2064 | to = repertoire_find_value (repertoire, now->val.str.startmb, | |
2065 | now->val.str.lenmb); | |
a673fbcb UD |
2066 | |
2067 | if (to == ILLEGAL_CHAR_VALUE) | |
2068 | lr_error (ldfile, "invalid character name"); | |
2069 | else | |
2070 | { | |
2071 | /* Make sure the `to'-value is larger. */ | |
2072 | if (to >= from) | |
a0dc5206 UD |
2073 | { |
2074 | newp->to = to; | |
2075 | newp->step = step; | |
2076 | } | |
a673fbcb UD |
2077 | else |
2078 | lr_error (ldfile, _("\ | |
2079 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
2080 | (to | from) < 65536 ? 4 : 8, to, | |
2081 | (to | from) < 65536 ? 4 : 8, from); | |
2082 | } | |
2083 | ||
2084 | /* And the next token. */ | |
47e8b443 | 2085 | now = lr_token (ldfile, charmap, NULL, repertoire, verbose); |
a673fbcb UD |
2086 | } |
2087 | ||
2088 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2089 | /* We are done. */ | |
2090 | return; | |
2091 | ||
2092 | if (now->tok == tok_semicolon) | |
2093 | /* Next round. */ | |
2094 | continue; | |
2095 | ||
2096 | /* If we come here something is wrong. */ | |
2097 | lr_error (ldfile, _("syntax error")); | |
2098 | lr_ignore_rest (ldfile, 0); | |
2099 | return; | |
2100 | } | |
2101 | } | |
2102 | ||
2103 | ||
4b10dd6c UD |
2104 | /* The parser for the LC_CTYPE section of the locale definition. */ |
2105 | void | |
2106 | ctype_read (struct linereader *ldfile, struct localedef_t *result, | |
47e8b443 | 2107 | const struct charmap_t *charmap, const char *repertoire_name, |
4b10dd6c | 2108 | int ignore_content) |
19bc17a9 | 2109 | { |
4b10dd6c UD |
2110 | struct repertoire_t *repertoire = NULL; |
2111 | struct locale_ctype_t *ctype; | |
2112 | struct token *now; | |
2113 | enum token_t nowtok; | |
19bc17a9 | 2114 | size_t cnt; |
4b10dd6c UD |
2115 | uint32_t last_wch = 0; |
2116 | enum token_t last_token; | |
2117 | enum token_t ellipsis_token; | |
a0dc5206 | 2118 | int step; |
4b10dd6c UD |
2119 | char last_charcode[16]; |
2120 | size_t last_charcode_len = 0; | |
2121 | const char *last_str = NULL; | |
2122 | int mapidx; | |
a6bd56c7 | 2123 | struct localedef_t *copy_locale = NULL; |
19bc17a9 | 2124 | |
4b10dd6c UD |
2125 | /* Get the repertoire we have to use. */ |
2126 | if (repertoire_name != NULL) | |
2127 | repertoire = repertoire_read (repertoire_name); | |
19bc17a9 | 2128 | |
4b10dd6c UD |
2129 | /* The rest of the line containing `LC_CTYPE' must be free. */ |
2130 | lr_ignore_rest (ldfile, 1); | |
19bc17a9 | 2131 | |
4b10dd6c UD |
2132 | |
2133 | do | |
19bc17a9 | 2134 | { |
47e8b443 | 2135 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c | 2136 | nowtok = now->tok; |
19bc17a9 | 2137 | } |
4b10dd6c | 2138 | while (nowtok == tok_eol); |
19bc17a9 | 2139 | |
4b10dd6c UD |
2140 | /* If we see `copy' now we are almost done. */ |
2141 | if (nowtok == tok_copy) | |
2142 | { | |
47e8b443 | 2143 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
a6bd56c7 UD |
2144 | if (now->tok != tok_string) |
2145 | { | |
2146 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
2147 | ||
2148 | skip_category: | |
2149 | do | |
47e8b443 | 2150 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
a6bd56c7 UD |
2151 | while (now->tok != tok_eof && now->tok != tok_end); |
2152 | ||
2153 | if (now->tok != tok_eof | |
47e8b443 | 2154 | || (now = lr_token (ldfile, charmap, NULL, NULL, verbose), |
93693c4d | 2155 | now->tok == tok_eof)) |
a6bd56c7 UD |
2156 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); |
2157 | else if (now->tok != tok_lc_ctype) | |
2158 | { | |
2159 | lr_error (ldfile, _("\ | |
2160 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2161 | lr_ignore_rest (ldfile, 0); | |
2162 | } | |
2163 | else | |
2164 | lr_ignore_rest (ldfile, 1); | |
2165 | ||
2166 | return; | |
2167 | } | |
2168 | ||
2169 | if (! ignore_content) | |
2170 | { | |
2171 | /* Get the locale definition. */ | |
2172 | copy_locale = load_locale (LC_CTYPE, now->val.str.startmb, | |
2173 | repertoire_name, charmap, NULL); | |
2174 | if ((copy_locale->avail & CTYPE_LOCALE) == 0) | |
2175 | { | |
2176 | /* Not yet loaded. So do it now. */ | |
2177 | if (locfile_read (copy_locale, charmap) != 0) | |
2178 | goto skip_category; | |
2179 | } | |
71663747 UD |
2180 | |
2181 | if (copy_locale->categories[LC_CTYPE].ctype == NULL) | |
2182 | return; | |
a6bd56c7 UD |
2183 | } |
2184 | ||
2185 | lr_ignore_rest (ldfile, 1); | |
2186 | ||
47e8b443 | 2187 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
a6bd56c7 | 2188 | nowtok = now->tok; |
4b10dd6c | 2189 | } |
75cd5204 | 2190 | |
4b10dd6c | 2191 | /* Prepare the data structures. */ |
a6bd56c7 | 2192 | ctype_startup (ldfile, result, charmap, copy_locale, ignore_content); |
4b10dd6c UD |
2193 | ctype = result->categories[LC_CTYPE].ctype; |
2194 | ||
2195 | /* Remember the repertoire we use. */ | |
2196 | if (!ignore_content) | |
2197 | ctype->repertoire = repertoire; | |
2198 | ||
2199 | while (1) | |
19bc17a9 | 2200 | { |
4b10dd6c UD |
2201 | unsigned long int class_bit = 0; |
2202 | unsigned long int class256_bit = 0; | |
2203 | int handle_digits = 0; | |
2204 | ||
2205 | /* Of course we don't proceed beyond the end of file. */ | |
2206 | if (nowtok == tok_eof) | |
2207 | break; | |
2208 | ||
2209 | /* Ingore empty lines. */ | |
2210 | if (nowtok == tok_eol) | |
19bc17a9 | 2211 | { |
47e8b443 | 2212 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2213 | nowtok = now->tok; |
2214 | continue; | |
2215 | } | |
19bc17a9 | 2216 | |
4b10dd6c UD |
2217 | switch (nowtok) |
2218 | { | |
5491da0d | 2219 | case tok_charclass: |
47e8b443 | 2220 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2221 | while (now->tok == tok_ident || now->tok == tok_string) |
2222 | { | |
2223 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
47e8b443 | 2224 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2225 | if (now->tok != tok_semicolon) |
2226 | break; | |
47e8b443 | 2227 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2228 | } |
2229 | if (now->tok != tok_eol) | |
2230 | SYNTAX_ERROR (_("\ | |
2231 | %s: syntax error in definition of new character class"), "LC_CTYPE"); | |
2232 | break; | |
2233 | ||
2234 | case tok_charconv: | |
47e8b443 | 2235 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2236 | while (now->tok == tok_ident || now->tok == tok_string) |
2237 | { | |
2238 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
47e8b443 | 2239 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2240 | if (now->tok != tok_semicolon) |
2241 | break; | |
47e8b443 | 2242 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
5491da0d UD |
2243 | } |
2244 | if (now->tok != tok_eol) | |
2245 | SYNTAX_ERROR (_("\ | |
2246 | %s: syntax error in definition of new character map"), "LC_CTYPE"); | |
2247 | break; | |
2248 | ||
4b10dd6c | 2249 | case tok_class: |
b9eb05d6 UD |
2250 | /* Ignore the rest of the line if we don't need the input of |
2251 | this line. */ | |
2252 | if (ignore_content) | |
2253 | { | |
2254 | lr_ignore_rest (ldfile, 0); | |
2255 | break; | |
2256 | } | |
2257 | ||
4b10dd6c UD |
2258 | /* We simply forget the `class' keyword and use the following |
2259 | operand to determine the bit. */ | |
47e8b443 | 2260 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2261 | if (now->tok == tok_ident || now->tok == tok_string) |
2262 | { | |
87372aa9 | 2263 | /* Must can be one of the predefined class names. */ |
4b10dd6c UD |
2264 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
2265 | if (strcmp (ctype->classnames[cnt], now->val.str.startmb) == 0) | |
2266 | break; | |
2267 | if (cnt >= ctype->nr_charclass) | |
2268 | { | |
011ebfab | 2269 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2270 | if (now->val.str.lenmb == 8 |
2271 | && memcmp ("special1", now->val.str.startmb, 8) == 0) | |
2272 | class_bit = _ISwspecial1; | |
2273 | else if (now->val.str.lenmb == 8 | |
2274 | && memcmp ("special2", now->val.str.startmb, 8) == 0) | |
2275 | class_bit = _ISwspecial2; | |
2276 | else if (now->val.str.lenmb == 8 | |
2277 | && memcmp ("special3", now->val.str.startmb, 8) == 0) | |
2278 | class_bit = _ISwspecial3; | |
2279 | else | |
011ebfab | 2280 | #endif |
4b10dd6c | 2281 | { |
87372aa9 UD |
2282 | /* OK, it's a new class. */ |
2283 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
4b10dd6c | 2284 | |
87372aa9 | 2285 | class_bit = _ISwbit (ctype->nr_charclass - 1); |
4b10dd6c UD |
2286 | } |
2287 | } | |
2288 | else | |
7f653277 UD |
2289 | { |
2290 | class_bit = _ISwbit (cnt); | |
4b10dd6c | 2291 | |
7f653277 UD |
2292 | free (now->val.str.startmb); |
2293 | } | |
4b10dd6c UD |
2294 | } |
2295 | else if (now->tok == tok_digit) | |
2296 | goto handle_tok_digit; | |
2297 | else if (now->tok < tok_upper || now->tok > tok_blank) | |
2298 | goto err_label; | |
2299 | else | |
2300 | { | |
2301 | class_bit = BITw (now->tok); | |
2302 | class256_bit = BIT (now->tok); | |
2303 | } | |
2304 | ||
2305 | /* The next character must be a semicolon. */ | |
47e8b443 | 2306 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2307 | if (now->tok != tok_semicolon) |
2308 | goto err_label; | |
2309 | goto read_charclass; | |
2310 | ||
2311 | case tok_upper: | |
2312 | case tok_lower: | |
2313 | case tok_alpha: | |
2314 | case tok_alnum: | |
2315 | case tok_space: | |
2316 | case tok_cntrl: | |
2317 | case tok_punct: | |
2318 | case tok_graph: | |
2319 | case tok_print: | |
2320 | case tok_xdigit: | |
2321 | case tok_blank: | |
b9eb05d6 UD |
2322 | /* Ignore the rest of the line if we don't need the input of |
2323 | this line. */ | |
2324 | if (ignore_content) | |
2325 | { | |
2326 | lr_ignore_rest (ldfile, 0); | |
2327 | break; | |
2328 | } | |
2329 | ||
4b10dd6c UD |
2330 | class_bit = BITw (now->tok); |
2331 | class256_bit = BIT (now->tok); | |
2332 | handle_digits = 0; | |
2333 | read_charclass: | |
2334 | ctype->class_done |= class_bit; | |
2335 | last_token = tok_none; | |
2336 | ellipsis_token = tok_none; | |
a0dc5206 | 2337 | step = 1; |
47e8b443 | 2338 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2339 | while (now->tok != tok_eol && now->tok != tok_eof) |
2340 | { | |
2341 | uint32_t wch; | |
2342 | struct charseq *seq; | |
2343 | ||
2344 | if (ellipsis_token == tok_none) | |
2345 | { | |
2346 | if (get_character (now, charmap, repertoire, &seq, &wch)) | |
2347 | goto err_label; | |
2348 | ||
2349 | if (!ignore_content && seq != NULL && seq->nbytes == 1) | |
2350 | /* Yep, we can store information about this byte | |
2351 | sequence. */ | |
2352 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
2353 | ||
2354 | if (!ignore_content && wch != ILLEGAL_CHAR_VALUE | |
2355 | && class_bit != 0) | |
2356 | /* We have the UCS4 position. */ | |
2357 | *find_idx (ctype, &ctype->class_collection, | |
2358 | &ctype->class_collection_max, | |
2359 | &ctype->class_collection_act, wch) |= class_bit; | |
2360 | ||
2361 | last_token = now->tok; | |
549b3c3a | 2362 | /* Terminate the string. */ |
9e2b7438 UD |
2363 | if (last_token == tok_bsymbol) |
2364 | { | |
2365 | now->val.str.startmb[now->val.str.lenmb] = '\0'; | |
2366 | last_str = now->val.str.startmb; | |
2367 | } | |
2368 | else | |
2369 | last_str = NULL; | |
4b10dd6c UD |
2370 | last_wch = wch; |
2371 | memcpy (last_charcode, now->val.charcode.bytes, 16); | |
2372 | last_charcode_len = now->val.charcode.nbytes; | |
2373 | ||
2374 | if (!ignore_content && handle_digits == 1) | |
2375 | { | |
2376 | /* We must store the digit values. */ | |
2377 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
2378 | { | |
b9eb05d6 | 2379 | ctype->mbdigits_max += 10; |
4b10dd6c UD |
2380 | ctype->mbdigits = xrealloc (ctype->mbdigits, |
2381 | (ctype->mbdigits_max | |
2382 | * sizeof (char *))); | |
b9eb05d6 | 2383 | ctype->wcdigits_max += 10; |
4b10dd6c UD |
2384 | ctype->wcdigits = xrealloc (ctype->wcdigits, |
2385 | (ctype->wcdigits_max | |
2386 | * sizeof (uint32_t))); | |
2387 | } | |
2388 | ||
2389 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
2390 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
2391 | } | |
2392 | else if (!ignore_content && handle_digits == 2) | |
2393 | { | |
2394 | /* We must store the digit values. */ | |
2395 | if (ctype->outdigits_act >= 10) | |
2396 | { | |
2397 | lr_error (ldfile, _("\ | |
2398 | %s: field `%s' does not contain exactly ten entries"), | |
2399 | "LC_CTYPE", "outdigit"); | |
69c69fe1 UD |
2400 | lr_ignore_rest (ldfile, 0); |
2401 | break; | |
4b10dd6c UD |
2402 | } |
2403 | ||
2404 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
2405 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
2406 | ++ctype->outdigits_act; | |
2407 | } | |
2408 | } | |
2409 | else | |
2410 | { | |
2411 | /* Now it gets complicated. We have to resolve the | |
2412 | ellipsis problem. First we must distinguish between | |
2413 | the different kind of ellipsis and this must match the | |
2414 | tokens we have seen. */ | |
2415 | assert (last_token != tok_none); | |
2416 | ||
2417 | if (last_token != now->tok) | |
2418 | { | |
2419 | lr_error (ldfile, _("\ | |
2420 | ellipsis range must be marked by two operands of same type")); | |
2421 | lr_ignore_rest (ldfile, 0); | |
2422 | break; | |
2423 | } | |
2424 | ||
2425 | if (last_token == tok_bsymbol) | |
2426 | { | |
2427 | if (ellipsis_token == tok_ellipsis3) | |
2428 | lr_error (ldfile, _("with symbolic name range values \ | |
2429 | the absolute ellipsis `...' must not be used")); | |
2430 | ||
2431 | charclass_symbolic_ellipsis (ldfile, ctype, charmap, | |
2432 | repertoire, now, last_str, | |
2433 | class256_bit, class_bit, | |
2434 | (ellipsis_token | |
2435 | == tok_ellipsis4 | |
2436 | ? 10 : 16), | |
2437 | ignore_content, | |
a0dc5206 | 2438 | handle_digits, step); |
4b10dd6c UD |
2439 | } |
2440 | else if (last_token == tok_ucs4) | |
2441 | { | |
2442 | if (ellipsis_token != tok_ellipsis2) | |
2443 | lr_error (ldfile, _("\ | |
2444 | with UCS range values one must use the hexadecimal symbolic ellipsis `..'")); | |
2445 | ||
2446 | charclass_ucs4_ellipsis (ldfile, ctype, charmap, | |
2447 | repertoire, now, last_wch, | |
2448 | class256_bit, class_bit, | |
a0dc5206 UD |
2449 | ignore_content, handle_digits, |
2450 | step); | |
4b10dd6c UD |
2451 | } |
2452 | else | |
2453 | { | |
2454 | assert (last_token == tok_charcode); | |
2455 | ||
2456 | if (ellipsis_token != tok_ellipsis3) | |
2457 | lr_error (ldfile, _("\ | |
2458 | with character code range values one must use the absolute ellipsis `...'")); | |
2459 | ||
2460 | charclass_charcode_ellipsis (ldfile, ctype, charmap, | |
2461 | repertoire, now, | |
2462 | last_charcode, | |
2463 | last_charcode_len, | |
2464 | class256_bit, class_bit, | |
2465 | ignore_content, | |
2466 | handle_digits); | |
2467 | } | |
2468 | ||
2469 | /* Now we have used the last value. */ | |
2470 | last_token = tok_none; | |
2471 | } | |
2472 | ||
2473 | /* Next we expect a semicolon or the end of the line. */ | |
47e8b443 | 2474 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2475 | if (now->tok == tok_eol || now->tok == tok_eof) |
2476 | break; | |
2477 | ||
2478 | if (last_token != tok_none | |
a0dc5206 | 2479 | && now->tok >= tok_ellipsis2 && now->tok <= tok_ellipsis4_2) |
4b10dd6c | 2480 | { |
a0dc5206 UD |
2481 | if (now->tok == tok_ellipsis2_2) |
2482 | { | |
2483 | now->tok = tok_ellipsis2; | |
2484 | step = 2; | |
2485 | } | |
2486 | else if (now->tok == tok_ellipsis4_2) | |
2487 | { | |
2488 | now->tok = tok_ellipsis4; | |
2489 | step = 2; | |
2490 | } | |
2491 | ||
4b10dd6c | 2492 | ellipsis_token = now->tok; |
a0dc5206 | 2493 | |
47e8b443 | 2494 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2495 | continue; |
2496 | } | |
2497 | ||
2498 | if (now->tok != tok_semicolon) | |
2499 | goto err_label; | |
2500 | ||
2501 | /* And get the next character. */ | |
47e8b443 | 2502 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2503 | |
2504 | ellipsis_token = tok_none; | |
a0dc5206 | 2505 | step = 1; |
4b10dd6c UD |
2506 | } |
2507 | break; | |
2508 | ||
2509 | case tok_digit: | |
b9eb05d6 UD |
2510 | /* Ignore the rest of the line if we don't need the input of |
2511 | this line. */ | |
2512 | if (ignore_content) | |
42d7c593 UD |
2513 | { |
2514 | lr_ignore_rest (ldfile, 0); | |
2515 | break; | |
2516 | } | |
b9eb05d6 | 2517 | |
4b10dd6c UD |
2518 | handle_tok_digit: |
2519 | class_bit = _ISwdigit; | |
2520 | class256_bit = _ISdigit; | |
2521 | handle_digits = 1; | |
2522 | goto read_charclass; | |
2523 | ||
2524 | case tok_outdigit: | |
b9eb05d6 UD |
2525 | /* Ignore the rest of the line if we don't need the input of |
2526 | this line. */ | |
2527 | if (ignore_content) | |
2528 | { | |
2529 | lr_ignore_rest (ldfile, 0); | |
2530 | break; | |
2531 | } | |
2532 | ||
4b10dd6c UD |
2533 | if (ctype->outdigits_act != 0) |
2534 | lr_error (ldfile, _("\ | |
2535 | %s: field `%s' declared more than once"), | |
2536 | "LC_CTYPE", "outdigit"); | |
2537 | class_bit = 0; | |
2538 | class256_bit = 0; | |
2539 | handle_digits = 2; | |
2540 | goto read_charclass; | |
2541 | ||
2542 | case tok_toupper: | |
b9eb05d6 UD |
2543 | /* Ignore the rest of the line if we don't need the input of |
2544 | this line. */ | |
2545 | if (ignore_content) | |
2546 | { | |
2547 | lr_ignore_rest (ldfile, 0); | |
2548 | break; | |
2549 | } | |
2550 | ||
4b10dd6c UD |
2551 | mapidx = 0; |
2552 | goto read_mapping; | |
2553 | ||
2554 | case tok_tolower: | |
b9eb05d6 UD |
2555 | /* Ignore the rest of the line if we don't need the input of |
2556 | this line. */ | |
2557 | if (ignore_content) | |
2558 | { | |
2559 | lr_ignore_rest (ldfile, 0); | |
2560 | break; | |
2561 | } | |
2562 | ||
4b10dd6c UD |
2563 | mapidx = 1; |
2564 | goto read_mapping; | |
2565 | ||
2566 | case tok_map: | |
b9eb05d6 UD |
2567 | /* Ignore the rest of the line if we don't need the input of |
2568 | this line. */ | |
2569 | if (ignore_content) | |
2570 | { | |
2571 | lr_ignore_rest (ldfile, 0); | |
2572 | break; | |
2573 | } | |
2574 | ||
4b10dd6c UD |
2575 | /* We simply forget the `map' keyword and use the following |
2576 | operand to determine the mapping. */ | |
47e8b443 | 2577 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2578 | if (now->tok == tok_ident || now->tok == tok_string) |
2579 | { | |
2580 | size_t cnt; | |
2581 | ||
2582 | for (cnt = 2; cnt < ctype->map_collection_nr; ++cnt) | |
2583 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2584 | break; | |
2585 | ||
7f653277 UD |
2586 | if (cnt < ctype->map_collection_nr) |
2587 | free (now->val.str.startmb); | |
2588 | else | |
87372aa9 UD |
2589 | /* OK, it's a new map. */ |
2590 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2591 | ||
2592 | mapidx = cnt; | |
4b10dd6c UD |
2593 | } |
2594 | else if (now->tok < tok_toupper || now->tok > tok_tolower) | |
2595 | goto err_label; | |
2596 | else | |
2597 | mapidx = now->tok - tok_toupper; | |
2598 | ||
47e8b443 | 2599 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2600 | /* This better should be a semicolon. */ |
2601 | if (now->tok != tok_semicolon) | |
2602 | goto err_label; | |
2603 | ||
2604 | read_mapping: | |
2605 | /* Test whether this mapping was already defined. */ | |
2606 | if (ctype->tomap_done[mapidx]) | |
2607 | { | |
2608 | lr_error (ldfile, _("duplicated definition for mapping `%s'"), | |
2609 | ctype->mapnames[mapidx]); | |
2610 | lr_ignore_rest (ldfile, 0); | |
2611 | break; | |
2612 | } | |
2613 | ctype->tomap_done[mapidx] = 1; | |
2614 | ||
47e8b443 | 2615 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2616 | while (now->tok != tok_eol && now->tok != tok_eof) |
2617 | { | |
2618 | struct charseq *from_seq; | |
2619 | uint32_t from_wch; | |
2620 | struct charseq *to_seq; | |
2621 | uint32_t to_wch; | |
2622 | ||
2623 | /* Every pair starts with an opening brace. */ | |
2624 | if (now->tok != tok_open_brace) | |
2625 | goto err_label; | |
2626 | ||
2627 | /* Next comes the from-value. */ | |
47e8b443 | 2628 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2629 | if (get_character (now, charmap, repertoire, &from_seq, |
2630 | &from_wch) != 0) | |
2631 | goto err_label; | |
2632 | ||
2633 | /* The next is a comma. */ | |
47e8b443 | 2634 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2635 | if (now->tok != tok_comma) |
2636 | goto err_label; | |
2637 | ||
2638 | /* And the other value. */ | |
47e8b443 | 2639 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2640 | if (get_character (now, charmap, repertoire, &to_seq, |
2641 | &to_wch) != 0) | |
2642 | goto err_label; | |
2643 | ||
2644 | /* And the last thing is the closing brace. */ | |
47e8b443 | 2645 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2646 | if (now->tok != tok_close_brace) |
2647 | goto err_label; | |
2648 | ||
2649 | if (!ignore_content) | |
2650 | { | |
f0c7c524 UD |
2651 | /* Check whether the mapping converts from an ASCII value |
2652 | to a non-ASCII value. */ | |
2653 | if (from_seq != NULL && from_seq->nbytes == 1 | |
2654 | && isascii (from_seq->bytes[0]) | |
2655 | && to_seq != NULL && (to_seq->nbytes != 1 | |
2656 | || !isascii (to_seq->bytes[0]))) | |
2657 | ctype->to_nonascii = 1; | |
2658 | ||
4b10dd6c UD |
2659 | if (mapidx < 2 && from_seq != NULL && to_seq != NULL |
2660 | && from_seq->nbytes == 1 && to_seq->nbytes == 1) | |
2661 | /* We can use this value. */ | |
2662 | ctype->map256_collection[mapidx][from_seq->bytes[0]] | |
2663 | = to_seq->bytes[0]; | |
2664 | ||
2665 | if (from_wch != ILLEGAL_CHAR_VALUE | |
2666 | && to_wch != ILLEGAL_CHAR_VALUE) | |
2667 | /* Both correct values. */ | |
2668 | *find_idx (ctype, &ctype->map_collection[mapidx], | |
2669 | &ctype->map_collection_max[mapidx], | |
2670 | &ctype->map_collection_act[mapidx], | |
2671 | from_wch) = to_wch; | |
2672 | } | |
2673 | ||
2674 | /* Now comes a semicolon or the end of the line/file. */ | |
47e8b443 | 2675 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c | 2676 | if (now->tok == tok_semicolon) |
47e8b443 | 2677 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2678 | } |
2679 | break; | |
2680 | ||
2681 | case tok_translit_start: | |
02fb3d17 UD |
2682 | /* Ignore the entire translit section with its peculiar syntax |
2683 | if we don't need the input. */ | |
b9eb05d6 UD |
2684 | if (ignore_content) |
2685 | { | |
02fb3d17 UD |
2686 | do |
2687 | { | |
2688 | lr_ignore_rest (ldfile, 0); | |
47e8b443 | 2689 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
02fb3d17 UD |
2690 | } |
2691 | while (now->tok != tok_translit_end && now->tok != tok_eof); | |
2692 | ||
2693 | if (now->tok == tok_eof) | |
2694 | lr_error (ldfile, _(\ | |
2695 | "%s: `translit_start' section does not end with `translit_end'"), | |
2696 | "LC_CTYPE"); | |
2697 | ||
b9eb05d6 UD |
2698 | break; |
2699 | } | |
2700 | ||
4b10dd6c UD |
2701 | /* The rest of the line better should be empty. */ |
2702 | lr_ignore_rest (ldfile, 1); | |
2703 | ||
2704 | /* We count here the number of allocated entries in the `translit' | |
2705 | array. */ | |
2706 | cnt = 0; | |
2707 | ||
4b156cb2 UD |
2708 | ldfile->translate_strings = 1; |
2709 | ldfile->return_widestr = 1; | |
2710 | ||
4b10dd6c | 2711 | /* We proceed until we see the `translit_end' token. */ |
47e8b443 | 2712 | while (now = lr_token (ldfile, charmap, NULL, repertoire, verbose), |
4b10dd6c UD |
2713 | now->tok != tok_translit_end && now->tok != tok_eof) |
2714 | { | |
2715 | if (now->tok == tok_eol) | |
2716 | /* Ignore empty lines. */ | |
2717 | continue; | |
2718 | ||
4b10dd6c UD |
2719 | if (now->tok == tok_include) |
2720 | { | |
2721 | /* We have to include locale. */ | |
2722 | const char *locale_name; | |
2723 | const char *repertoire_name; | |
02fb3d17 | 2724 | struct translit_include_t *include_stmt, **include_ptr; |
4b10dd6c | 2725 | |
47e8b443 | 2726 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2727 | /* This should be a string or an identifier. In any |
2728 | case something to name a locale. */ | |
2729 | if (now->tok != tok_string && now->tok != tok_ident) | |
2730 | { | |
2731 | translit_syntax: | |
2732 | lr_error (ldfile, _("%s: syntax error"), "LC_CTYPE"); | |
2733 | lr_ignore_rest (ldfile, 0); | |
2734 | continue; | |
2735 | } | |
2736 | locale_name = now->val.str.startmb; | |
2737 | ||
2738 | /* Next should be a semicolon. */ | |
47e8b443 | 2739 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2740 | if (now->tok != tok_semicolon) |
2741 | goto translit_syntax; | |
2742 | ||
2743 | /* Now the repertoire name. */ | |
47e8b443 | 2744 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2745 | if ((now->tok != tok_string && now->tok != tok_ident) |
2746 | || now->val.str.startmb == NULL) | |
2747 | goto translit_syntax; | |
2748 | repertoire_name = now->val.str.startmb; | |
d9cab009 RM |
2749 | if (repertoire_name[0] == '\0') |
2750 | /* Ignore the empty string. */ | |
2751 | repertoire_name = NULL; | |
4b10dd6c | 2752 | |
02fb3d17 UD |
2753 | /* Save the include statement for later processing. */ |
2754 | include_stmt = (struct translit_include_t *) | |
2755 | xmalloc (sizeof (struct translit_include_t)); | |
2756 | include_stmt->copy_locale = locale_name; | |
2757 | include_stmt->copy_repertoire = repertoire_name; | |
2758 | include_stmt->next = NULL; | |
4b10dd6c | 2759 | |
02fb3d17 UD |
2760 | include_ptr = &ctype->translit_include; |
2761 | while (*include_ptr != NULL) | |
2762 | include_ptr = &(*include_ptr)->next; | |
2763 | *include_ptr = include_stmt; | |
4b10dd6c UD |
2764 | |
2765 | /* The rest of the line must be empty. */ | |
2766 | lr_ignore_rest (ldfile, 1); | |
a673fbcb UD |
2767 | |
2768 | /* Make sure the locale is read. */ | |
02fb3d17 UD |
2769 | add_to_readlist (LC_CTYPE, locale_name, repertoire_name, |
2770 | 1, NULL); | |
a673fbcb UD |
2771 | continue; |
2772 | } | |
2773 | else if (now->tok == tok_default_missing) | |
2774 | { | |
2775 | uint32_t *wstr; | |
2776 | ||
c9f79e08 | 2777 | while (1) |
a673fbcb | 2778 | { |
c9f79e08 UD |
2779 | /* We expect a single character or string as the |
2780 | argument. */ | |
47e8b443 | 2781 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
c9f79e08 UD |
2782 | wstr = read_widestring (ldfile, now, charmap, |
2783 | repertoire); | |
2784 | ||
2785 | if (wstr != NULL) | |
a673fbcb | 2786 | { |
c9f79e08 UD |
2787 | if (ctype->default_missing != NULL) |
2788 | { | |
2789 | lr_error (ldfile, _("\ | |
a673fbcb | 2790 | %s: duplicate `default_missing' definition"), "LC_CTYPE"); |
f2b98f97 UD |
2791 | WITH_CUR_LOCALE (error_at_line (0, 0, |
2792 | ctype->default_missing_file, | |
2793 | ctype->default_missing_lineno, | |
2794 | _("\ | |
2795 | previous definition was here"))); | |
c9f79e08 UD |
2796 | } |
2797 | else | |
2798 | { | |
2799 | ctype->default_missing = wstr; | |
2800 | ctype->default_missing_file = ldfile->fname; | |
2801 | ctype->default_missing_lineno = ldfile->lineno; | |
2802 | } | |
4b156cb2 UD |
2803 | /* We can have more entries, ignore them. */ |
2804 | lr_ignore_rest (ldfile, 0); | |
c9f79e08 | 2805 | break; |
a673fbcb | 2806 | } |
c9f79e08 UD |
2807 | else if (wstr == (uint32_t *) -1l) |
2808 | /* This was an syntax error. */ | |
2809 | break; | |
2810 | ||
2811 | /* Maybe there is another replacement we can use. */ | |
47e8b443 | 2812 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
c9f79e08 | 2813 | if (now->tok == tok_eol || now->tok == tok_eof) |
a673fbcb | 2814 | { |
c9f79e08 UD |
2815 | /* Nothing found. We tell the user. */ |
2816 | lr_error (ldfile, _("\ | |
0232a3ae | 2817 | %s: no representable `default_missing' definition found"), "LC_CTYPE"); |
c9f79e08 | 2818 | break; |
a673fbcb | 2819 | } |
c9f79e08 UD |
2820 | if (now->tok != tok_semicolon) |
2821 | goto translit_syntax; | |
a673fbcb | 2822 | } |
c9f79e08 | 2823 | |
a673fbcb UD |
2824 | continue; |
2825 | } | |
2826 | else if (now->tok == tok_translit_ignore) | |
2827 | { | |
2828 | read_translit_ignore_entry (ldfile, ctype, charmap, | |
2829 | repertoire); | |
4b10dd6c UD |
2830 | continue; |
2831 | } | |
2832 | ||
2833 | read_translit_entry (ldfile, ctype, now, charmap, repertoire); | |
2834 | } | |
4b156cb2 | 2835 | ldfile->return_widestr = 0; |
02fb3d17 UD |
2836 | |
2837 | if (now->tok == tok_eof) | |
2838 | lr_error (ldfile, _(\ | |
2839 | "%s: `translit_start' section does not end with `translit_end'"), | |
2840 | "LC_CTYPE"); | |
2841 | ||
4b10dd6c UD |
2842 | break; |
2843 | ||
2844 | case tok_ident: | |
b9eb05d6 UD |
2845 | /* Ignore the rest of the line if we don't need the input of |
2846 | this line. */ | |
2847 | if (ignore_content) | |
2848 | { | |
2849 | lr_ignore_rest (ldfile, 0); | |
2850 | break; | |
2851 | } | |
2852 | ||
4b10dd6c UD |
2853 | /* This could mean one of several things. First test whether |
2854 | it's a character class name. */ | |
2855 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
2856 | if (strcmp (now->val.str.startmb, ctype->classnames[cnt]) == 0) | |
2857 | break; | |
2858 | if (cnt < ctype->nr_charclass) | |
2859 | { | |
2860 | class_bit = _ISwbit (cnt); | |
2861 | class256_bit = cnt <= 11 ? _ISbit (cnt) : 0; | |
2862 | free (now->val.str.startmb); | |
2863 | goto read_charclass; | |
2864 | } | |
5491da0d UD |
2865 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
2866 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2867 | break; | |
2868 | if (cnt < ctype->map_collection_nr) | |
2869 | { | |
2870 | mapidx = cnt; | |
2871 | free (now->val.str.startmb); | |
2872 | goto read_mapping; | |
2873 | } | |
011ebfab | 2874 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2875 | if (strcmp (now->val.str.startmb, "special1") == 0) |
2876 | { | |
2877 | class_bit = _ISwspecial1; | |
2878 | free (now->val.str.startmb); | |
2879 | goto read_charclass; | |
2880 | } | |
2881 | if (strcmp (now->val.str.startmb, "special2") == 0) | |
2882 | { | |
2883 | class_bit = _ISwspecial2; | |
2884 | free (now->val.str.startmb); | |
2885 | goto read_charclass; | |
2886 | } | |
2887 | if (strcmp (now->val.str.startmb, "special3") == 0) | |
2888 | { | |
2889 | class_bit = _ISwspecial3; | |
2890 | free (now->val.str.startmb); | |
2891 | goto read_charclass; | |
2892 | } | |
2893 | if (strcmp (now->val.str.startmb, "tosymmetric") == 0) | |
2894 | { | |
2895 | mapidx = 2; | |
2896 | goto read_mapping; | |
2897 | } | |
011ebfab | 2898 | #endif |
4b10dd6c UD |
2899 | break; |
2900 | ||
2901 | case tok_end: | |
2902 | /* Next we assume `LC_CTYPE'. */ | |
47e8b443 | 2903 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c UD |
2904 | if (now->tok == tok_eof) |
2905 | break; | |
2906 | if (now->tok == tok_eol) | |
2907 | lr_error (ldfile, _("%s: incomplete `END' line"), | |
2908 | "LC_CTYPE"); | |
2909 | else if (now->tok != tok_lc_ctype) | |
2910 | lr_error (ldfile, _("\ | |
2911 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2912 | lr_ignore_rest (ldfile, now->tok == tok_lc_ctype); | |
2913 | return; | |
2914 | ||
2915 | default: | |
2916 | err_label: | |
2917 | if (now->tok != tok_eof) | |
2918 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
19bc17a9 RM |
2919 | } |
2920 | ||
4b10dd6c | 2921 | /* Prepare for the next round. */ |
47e8b443 | 2922 | now = lr_token (ldfile, charmap, NULL, NULL, verbose); |
4b10dd6c | 2923 | nowtok = now->tok; |
19bc17a9 RM |
2924 | } |
2925 | ||
4b10dd6c UD |
2926 | /* When we come here we reached the end of the file. */ |
2927 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); | |
19bc17a9 RM |
2928 | } |
2929 | ||
2930 | ||
2931 | static void | |
47e8b443 UD |
2932 | set_class_defaults (struct locale_ctype_t *ctype, |
2933 | const struct charmap_t *charmap, | |
4b10dd6c | 2934 | struct repertoire_t *repertoire) |
19bc17a9 | 2935 | { |
4b10dd6c UD |
2936 | size_t cnt; |
2937 | ||
19bc17a9 RM |
2938 | /* These function defines the default values for the classes and conversions |
2939 | according to POSIX.2 2.5.2.1. | |
2940 | It may seem that the order of these if-blocks is arbitrary but it is NOT. | |
2941 | Don't move them unless you know what you do! */ | |
2942 | ||
938c669e AJ |
2943 | auto void set_default (int bitpos, int from, int to); |
2944 | ||
4b10dd6c | 2945 | void set_default (int bitpos, int from, int to) |
19bc17a9 RM |
2946 | { |
2947 | char tmp[2]; | |
2948 | int ch; | |
4b10dd6c UD |
2949 | int bit = _ISbit (bitpos); |
2950 | int bitw = _ISwbit (bitpos); | |
19bc17a9 RM |
2951 | /* Define string. */ |
2952 | strcpy (tmp, "?"); | |
2953 | ||
2954 | for (ch = from; ch <= to; ++ch) | |
2955 | { | |
4b10dd6c | 2956 | struct charseq *seq; |
19bc17a9 RM |
2957 | tmp[0] = ch; |
2958 | ||
4b10dd6c | 2959 | seq = charmap_find_value (charmap, tmp, 1); |
69c69fe1 UD |
2960 | if (seq == NULL) |
2961 | { | |
2962 | char buf[10]; | |
2963 | sprintf (buf, "U%08X", ch); | |
2964 | seq = charmap_find_value (charmap, buf, 9); | |
2965 | } | |
4b10dd6c UD |
2966 | if (seq == NULL) |
2967 | { | |
2968 | if (!be_quiet) | |
f2b98f97 | 2969 | WITH_CUR_LOCALE (error (0, 0, _("\ |
11bf311e | 2970 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 2971 | "LC_CTYPE", tmp)); |
19bc17a9 | 2972 | } |
4b10dd6c | 2973 | else if (seq->nbytes != 1) |
f2b98f97 | 2974 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 2975 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 2976 | "LC_CTYPE", tmp)); |
19bc17a9 | 2977 | else |
4b10dd6c | 2978 | ctype->class256_collection[seq->bytes[0]] |= bit; |
f0a4b6b1 UD |
2979 | |
2980 | /* No need to search here, the ASCII value is also the Unicode | |
2981 | value. */ | |
2982 | ELEM (ctype, class_collection, , ch) |= bitw; | |
19bc17a9 RM |
2983 | } |
2984 | } | |
2985 | ||
2986 | /* Set default values if keyword was not present. */ | |
4b10dd6c | 2987 | if ((ctype->class_done & BITw (tok_upper)) == 0) |
19bc17a9 RM |
2988 | /* "If this keyword [lower] is not specified, the lowercase letters |
2989 | `A' through `Z', ..., shall automatically belong to this class, | |
2990 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2991 | set_default (BITPOS (tok_upper), 'A', 'Z'); |
19bc17a9 | 2992 | |
4b10dd6c | 2993 | if ((ctype->class_done & BITw (tok_lower)) == 0) |
19bc17a9 RM |
2994 | /* "If this keyword [lower] is not specified, the lowercase letters |
2995 | `a' through `z', ..., shall automatically belong to this class, | |
2996 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2997 | set_default (BITPOS (tok_lower), 'a', 'z'); |
19bc17a9 | 2998 | |
4b10dd6c | 2999 | if ((ctype->class_done & BITw (tok_alpha)) == 0) |
19bc17a9 RM |
3000 | { |
3001 | /* Table 2-6 in P1003.2 says that characters in class `upper' or | |
3002 | class `lower' *must* be in class `alpha'. */ | |
3003 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower); | |
96f0d1f5 UD |
3004 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower); |
3005 | ||
3006 | for (cnt = 0; cnt < 256; ++cnt) | |
3007 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3008 | ctype->class256_collection[cnt] |= BIT (tok_alpha); | |
19bc17a9 RM |
3009 | |
3010 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
3011 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3012 | ctype->class_collection[cnt] |= BITw (tok_alpha); | |
19bc17a9 RM |
3013 | } |
3014 | ||
4b10dd6c | 3015 | if ((ctype->class_done & BITw (tok_digit)) == 0) |
19bc17a9 RM |
3016 | /* "If this keyword [digit] is not specified, the digits `0' through |
3017 | `9', ..., shall automatically belong to this class, with | |
3018 | implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 3019 | set_default (BITPOS (tok_digit), '0', '9'); |
19bc17a9 RM |
3020 | |
3021 | /* "Only characters specified for the `alpha' and `digit' keyword | |
3022 | shall be specified. Characters specified for the keyword `alpha' | |
3023 | and `digit' are automatically included in this class. */ | |
3024 | { | |
3025 | unsigned long int mask = BIT (tok_alpha) | BIT (tok_digit); | |
96f0d1f5 UD |
3026 | unsigned long int maskw = BITw (tok_alpha) | BITw (tok_digit); |
3027 | ||
3028 | for (cnt = 0; cnt < 256; ++cnt) | |
3029 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3030 | ctype->class256_collection[cnt] |= BIT (tok_alnum); | |
19bc17a9 RM |
3031 | |
3032 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
3033 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3034 | ctype->class_collection[cnt] |= BITw (tok_alnum); | |
19bc17a9 RM |
3035 | } |
3036 | ||
4b10dd6c | 3037 | if ((ctype->class_done & BITw (tok_space)) == 0) |
19bc17a9 RM |
3038 | /* "If this keyword [space] is not specified, the characters <space>, |
3039 | <form-feed>, <newline>, <carriage-return>, <tab>, and | |
3040 | <vertical-tab>, ..., shall automatically belong to this class, | |
3041 | with implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
3042 | { | |
4b10dd6c | 3043 | struct charseq *seq; |
19bc17a9 | 3044 | |
4b10dd6c | 3045 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3046 | if (seq == NULL) |
3047 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3048 | if (seq == NULL) |
3049 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 3050 | if (seq == NULL) |
880f421f UD |
3051 | { |
3052 | if (!be_quiet) | |
f2b98f97 | 3053 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3054 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3055 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3056 | } |
3057 | else if (seq->nbytes != 1) | |
f2b98f97 | 3058 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3059 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3060 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3061 | else |
3062 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
3063 | ||
f0a4b6b1 | 3064 | /* No need to search. */ |
ce177a84 | 3065 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_space); |
19bc17a9 | 3066 | |
4b10dd6c | 3067 | seq = charmap_find_value (charmap, "form-feed", 9); |
f0a4b6b1 UD |
3068 | if (seq == NULL) |
3069 | seq = charmap_find_value (charmap, "U0000000C", 9); | |
4b10dd6c | 3070 | if (seq == NULL) |
880f421f UD |
3071 | { |
3072 | if (!be_quiet) | |
f2b98f97 | 3073 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3074 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3075 | "LC_CTYPE", "<form-feed>")); |
4b10dd6c UD |
3076 | } |
3077 | else if (seq->nbytes != 1) | |
f2b98f97 | 3078 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3079 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3080 | "LC_CTYPE", "<form-feed>")); |
4b10dd6c UD |
3081 | else |
3082 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
3083 | ||
f0a4b6b1 | 3084 | /* No need to search. */ |
ce177a84 | 3085 | ELEM (ctype, class_collection, , L'\f') |= BITw (tok_space); |
4b10dd6c | 3086 | |
19bc17a9 | 3087 | |
4b10dd6c | 3088 | seq = charmap_find_value (charmap, "newline", 7); |
f0a4b6b1 UD |
3089 | if (seq == NULL) |
3090 | seq = charmap_find_value (charmap, "U0000000A", 9); | |
4b10dd6c | 3091 | if (seq == NULL) |
880f421f UD |
3092 | { |
3093 | if (!be_quiet) | |
f2b98f97 | 3094 | WITH_CUR_LOCALE (error (0, 0, _("\ |
11bf311e UD |
3095 | %s: character `%s' not defined while needed as default value"), |
3096 | "LC_CTYPE", "<newline>")); | |
4b10dd6c UD |
3097 | } |
3098 | else if (seq->nbytes != 1) | |
f2b98f97 | 3099 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3100 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3101 | "LC_CTYPE", "<newline>")); |
4b10dd6c UD |
3102 | else |
3103 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
3104 | ||
f0a4b6b1 | 3105 | /* No need to search. */ |
ce177a84 | 3106 | ELEM (ctype, class_collection, , L'\n') |= BITw (tok_space); |
4b10dd6c | 3107 | |
19bc17a9 | 3108 | |
4b10dd6c | 3109 | seq = charmap_find_value (charmap, "carriage-return", 15); |
f0a4b6b1 UD |
3110 | if (seq == NULL) |
3111 | seq = charmap_find_value (charmap, "U0000000D", 9); | |
4b10dd6c | 3112 | if (seq == NULL) |
880f421f UD |
3113 | { |
3114 | if (!be_quiet) | |
f2b98f97 | 3115 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3116 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3117 | "LC_CTYPE", "<carriage-return>")); |
4b10dd6c UD |
3118 | } |
3119 | else if (seq->nbytes != 1) | |
f2b98f97 | 3120 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3121 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3122 | "LC_CTYPE", "<carriage-return>")); |
4b10dd6c UD |
3123 | else |
3124 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
3125 | ||
f0a4b6b1 | 3126 | /* No need to search. */ |
ce177a84 | 3127 | ELEM (ctype, class_collection, , L'\r') |= BITw (tok_space); |
4b10dd6c | 3128 | |
19bc17a9 | 3129 | |
4b10dd6c | 3130 | seq = charmap_find_value (charmap, "tab", 3); |
f0a4b6b1 UD |
3131 | if (seq == NULL) |
3132 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c | 3133 | if (seq == NULL) |
880f421f UD |
3134 | { |
3135 | if (!be_quiet) | |
f2b98f97 | 3136 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3137 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3138 | "LC_CTYPE", "<tab>")); |
4b10dd6c UD |
3139 | } |
3140 | else if (seq->nbytes != 1) | |
f2b98f97 | 3141 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3142 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3143 | "LC_CTYPE", "<tab>")); |
4b10dd6c UD |
3144 | else |
3145 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
3146 | ||
f0a4b6b1 | 3147 | /* No need to search. */ |
ce177a84 | 3148 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_space); |
4b10dd6c | 3149 | |
4b10dd6c UD |
3150 | |
3151 | seq = charmap_find_value (charmap, "vertical-tab", 12); | |
f0a4b6b1 UD |
3152 | if (seq == NULL) |
3153 | seq = charmap_find_value (charmap, "U0000000B", 9); | |
4b10dd6c UD |
3154 | if (seq == NULL) |
3155 | { | |
3156 | if (!be_quiet) | |
f2b98f97 | 3157 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3158 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3159 | "LC_CTYPE", "<vertical-tab>")); |
4b10dd6c UD |
3160 | } |
3161 | else if (seq->nbytes != 1) | |
f2b98f97 | 3162 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3163 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3164 | "LC_CTYPE", "<vertical-tab>")); |
4b10dd6c UD |
3165 | else |
3166 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
f0a4b6b1 UD |
3167 | |
3168 | /* No need to search. */ | |
ce177a84 | 3169 | ELEM (ctype, class_collection, , L'\v') |= BITw (tok_space); |
19bc17a9 RM |
3170 | } |
3171 | ||
4b10dd6c | 3172 | if ((ctype->class_done & BITw (tok_xdigit)) == 0) |
19bc17a9 RM |
3173 | /* "If this keyword is not specified, the digits `0' to `9', the |
3174 | uppercase letters `A' through `F', and the lowercase letters `a' | |
3175 | through `f', ..., shell automatically belong to this class, with | |
3176 | implementation defined character values." [P1003.2, 2.5.2.1] */ | |
3177 | { | |
4b10dd6c UD |
3178 | set_default (BITPOS (tok_xdigit), '0', '9'); |
3179 | set_default (BITPOS (tok_xdigit), 'A', 'F'); | |
3180 | set_default (BITPOS (tok_xdigit), 'a', 'f'); | |
19bc17a9 RM |
3181 | } |
3182 | ||
4b10dd6c | 3183 | if ((ctype->class_done & BITw (tok_blank)) == 0) |
19bc17a9 RM |
3184 | /* "If this keyword [blank] is unspecified, the characters <space> and |
3185 | <tab> shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
3186 | { | |
4b10dd6c | 3187 | struct charseq *seq; |
19bc17a9 | 3188 | |
4b10dd6c | 3189 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3190 | if (seq == NULL) |
3191 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3192 | if (seq == NULL) |
3193 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 3194 | if (seq == NULL) |
880f421f UD |
3195 | { |
3196 | if (!be_quiet) | |
f2b98f97 | 3197 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3198 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3199 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3200 | } |
3201 | else if (seq->nbytes != 1) | |
f2b98f97 | 3202 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3203 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3204 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3205 | else |
3206 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
3207 | ||
f0a4b6b1 | 3208 | /* No need to search. */ |
ce177a84 | 3209 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_blank); |
4b10dd6c | 3210 | |
4b10dd6c UD |
3211 | |
3212 | seq = charmap_find_value (charmap, "tab", 3); | |
f0a4b6b1 UD |
3213 | if (seq == NULL) |
3214 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c UD |
3215 | if (seq == NULL) |
3216 | { | |
3217 | if (!be_quiet) | |
f2b98f97 | 3218 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3219 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3220 | "LC_CTYPE", "<tab>")); |
4b10dd6c UD |
3221 | } |
3222 | else if (seq->nbytes != 1) | |
f2b98f97 | 3223 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3224 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3225 | "LC_CTYPE", "<tab>")); |
4b10dd6c UD |
3226 | else |
3227 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
f0a4b6b1 UD |
3228 | |
3229 | /* No need to search. */ | |
ce177a84 | 3230 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_blank); |
19bc17a9 RM |
3231 | } |
3232 | ||
4b10dd6c | 3233 | if ((ctype->class_done & BITw (tok_graph)) == 0) |
19bc17a9 RM |
3234 | /* "If this keyword [graph] is not specified, characters specified for |
3235 | the keywords `upper', `lower', `alpha', `digit', `xdigit' and `punct', | |
3236 | shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
3237 | { | |
3238 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3239 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3240 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3241 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3242 | BITw (tok_punct); | |
19bc17a9 RM |
3243 | size_t cnt; |
3244 | ||
3245 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3246 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3247 | ctype->class_collection[cnt] |= BITw (tok_graph); | |
4b10dd6c UD |
3248 | |
3249 | for (cnt = 0; cnt < 256; ++cnt) | |
3250 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3251 | ctype->class256_collection[cnt] |= BIT (tok_graph); | |
19bc17a9 RM |
3252 | } |
3253 | ||
4b10dd6c | 3254 | if ((ctype->class_done & BITw (tok_print)) == 0) |
19bc17a9 RM |
3255 | /* "If this keyword [print] is not provided, characters specified for |
3256 | the keywords `upper', `lower', `alpha', `digit', `xdigit', `punct', | |
3257 | and the <space> character shall belong to this character class." | |
3258 | [P1003.2, 2.5.2.1] */ | |
3259 | { | |
3260 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3261 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3262 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3263 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3264 | BITw (tok_punct); | |
19bc17a9 | 3265 | size_t cnt; |
4b10dd6c | 3266 | struct charseq *seq; |
19bc17a9 RM |
3267 | |
3268 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3269 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3270 | ctype->class_collection[cnt] |= BITw (tok_print); | |
19bc17a9 | 3271 | |
4b10dd6c UD |
3272 | for (cnt = 0; cnt < 256; ++cnt) |
3273 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3274 | ctype->class256_collection[cnt] |= BIT (tok_print); | |
3275 | ||
3276 | ||
4b10dd6c | 3277 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3278 | if (seq == NULL) |
3279 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3280 | if (seq == NULL) |
3281 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c UD |
3282 | if (seq == NULL) |
3283 | { | |
3284 | if (!be_quiet) | |
f2b98f97 | 3285 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3286 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3287 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3288 | } |
3289 | else if (seq->nbytes != 1) | |
f2b98f97 | 3290 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3291 | %s: character `%s' in charmap not representable with one byte"), |
f2b98f97 | 3292 | "LC_CTYPE", "<space>")); |
4b10dd6c UD |
3293 | else |
3294 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_print); | |
f0a4b6b1 UD |
3295 | |
3296 | /* No need to search. */ | |
ce177a84 | 3297 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_print); |
19bc17a9 RM |
3298 | } |
3299 | ||
4b10dd6c | 3300 | if (ctype->tomap_done[0] == 0) |
6d52618b | 3301 | /* "If this keyword [toupper] is not specified, the lowercase letters |
19bc17a9 RM |
3302 | `a' through `z', and their corresponding uppercase letters `A' to |
3303 | `Z', ..., shall automatically be included, with implementation- | |
3304 | defined character values." [P1003.2, 2.5.2.1] */ | |
3305 | { | |
3306 | char tmp[4]; | |
3307 | int ch; | |
3308 | ||
3309 | strcpy (tmp, "<?>"); | |
3310 | ||
3311 | for (ch = 'a'; ch <= 'z'; ++ch) | |
3312 | { | |
4b10dd6c | 3313 | struct charseq *seq_from, *seq_to; |
19bc17a9 RM |
3314 | |
3315 | tmp[1] = (char) ch; | |
3316 | ||
4b10dd6c | 3317 | seq_from = charmap_find_value (charmap, &tmp[1], 1); |
69c69fe1 UD |
3318 | if (seq_from == NULL) |
3319 | { | |
3320 | char buf[10]; | |
3321 | sprintf (buf, "U%08X", ch); | |
3322 | seq_from = charmap_find_value (charmap, buf, 9); | |
3323 | } | |
4b10dd6c | 3324 | if (seq_from == NULL) |
19bc17a9 | 3325 | { |
880f421f | 3326 | if (!be_quiet) |
f2b98f97 | 3327 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3328 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3329 | "LC_CTYPE", tmp)); |
4b10dd6c UD |
3330 | } |
3331 | else if (seq_from->nbytes != 1) | |
3332 | { | |
3333 | if (!be_quiet) | |
f2b98f97 | 3334 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3335 | %s: character `%s' needed as default value not representable with one byte"), |
f2b98f97 | 3336 | "LC_CTYPE", tmp)); |
4b10dd6c UD |
3337 | } |
3338 | else | |
3339 | { | |
3340 | /* This conversion is implementation defined. */ | |
3341 | tmp[1] = (char) (ch + ('A' - 'a')); | |
3342 | seq_to = charmap_find_value (charmap, &tmp[1], 1); | |
69c69fe1 UD |
3343 | if (seq_to == NULL) |
3344 | { | |
3345 | char buf[10]; | |
3346 | sprintf (buf, "U%08X", ch + ('A' - 'a')); | |
3347 | seq_to = charmap_find_value (charmap, buf, 9); | |
3348 | } | |
4b10dd6c UD |
3349 | if (seq_to == NULL) |
3350 | { | |
3351 | if (!be_quiet) | |
f2b98f97 | 3352 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3353 | %s: character `%s' not defined while needed as default value"), |
f2b98f97 | 3354 | "LC_CTYPE", tmp)); |
4b10dd6c UD |
3355 | } |
3356 | else if (seq_to->nbytes != 1) | |
3357 | { | |
3358 | if (!be_quiet) | |
f2b98f97 | 3359 | WITH_CUR_LOCALE (error (0, 0, _("\ |
4b10dd6c | 3360 | %s: character `%s' needed as default value not representable with one byte"), |
f2b98f97 | 3361 | "LC_CTYPE", tmp)); |
4b10dd6c UD |
3362 | } |
3363 | else | |
3364 | /* The index [0] is determined by the order of the | |
3365 | `ctype_map_newP' calls in `ctype_startup'. */ | |
3366 | ctype->map256_collection[0][seq_from->bytes[0]] | |
3367 | = seq_to->bytes[0]; | |
19bc17a9 | 3368 | } |
f0a4b6b1 UD |
3369 | |
3370 | /* No need to search. */ | |
3371 | ELEM (ctype, map_collection, [0], ch) = ch + ('A' - 'a'); | |
19bc17a9 RM |
3372 | } |
3373 | } | |
3374 | ||
4b10dd6c | 3375 | if (ctype->tomap_done[1] == 0) |
19bc17a9 RM |
3376 | /* "If this keyword [tolower] is not specified, the mapping shall be |
3377 | the reverse mapping of the one specified to `toupper'." [P1003.2] */ | |
3378 | { | |
19bc17a9 RM |
3379 | for (cnt = 0; cnt < ctype->map_collection_act[0]; ++cnt) |
3380 | if (ctype->map_collection[0][cnt] != 0) | |
3381 | ELEM (ctype, map_collection, [1], | |
3382 | ctype->map_collection[0][cnt]) | |
3383 | = ctype->charnames[cnt]; | |
4b10dd6c UD |
3384 | |
3385 | for (cnt = 0; cnt < 256; ++cnt) | |
3386 | if (ctype->map256_collection[0][cnt] != 0) | |
85cb60ff | 3387 | ctype->map256_collection[1][ctype->map256_collection[0][cnt]] = cnt; |
4b10dd6c UD |
3388 | } |
3389 | ||
69c69fe1 | 3390 | if (ctype->outdigits_act != 10) |
4b10dd6c | 3391 | { |
69c69fe1 | 3392 | if (ctype->outdigits_act != 0) |
f2b98f97 UD |
3393 | WITH_CUR_LOCALE (error (0, 0, _("\ |
3394 | %s: field `%s' does not contain exactly ten entries"), | |
3395 | "LC_CTYPE", "outdigit")); | |
69c69fe1 UD |
3396 | |
3397 | for (cnt = ctype->outdigits_act; cnt < 10; ++cnt) | |
4b10dd6c UD |
3398 | { |
3399 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
701666b7 UD |
3400 | (char *) digits + cnt, |
3401 | 1); | |
4b10dd6c UD |
3402 | |
3403 | if (ctype->mboutdigits[cnt] == NULL) | |
1b97149d UD |
3404 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, |
3405 | longnames[cnt], | |
3406 | strlen (longnames[cnt])); | |
b9eb05d6 | 3407 | |
1b97149d UD |
3408 | if (ctype->mboutdigits[cnt] == NULL) |
3409 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
3410 | uninames[cnt], 9); | |
b9eb05d6 | 3411 | |
1b97149d | 3412 | if (ctype->mboutdigits[cnt] == NULL) |
b9eb05d6 | 3413 | { |
1b97149d | 3414 | /* Provide a replacement. */ |
f2b98f97 UD |
3415 | WITH_CUR_LOCALE (error (0, 0, _("\ |
3416 | no output digits defined and none of the standard names in the charmap"))); | |
b9eb05d6 | 3417 | |
47e8b443 | 3418 | ctype->mboutdigits[cnt] = obstack_alloc (&((struct charmap_t *) charmap)->mem_pool, |
1b97149d UD |
3419 | sizeof (struct charseq) |
3420 | + 1); | |
b9eb05d6 | 3421 | |
1b97149d UD |
3422 | /* This is better than nothing. */ |
3423 | ctype->mboutdigits[cnt]->bytes[0] = digits[cnt]; | |
3424 | ctype->mboutdigits[cnt]->nbytes = 1; | |
b9eb05d6 | 3425 | } |
1b97149d UD |
3426 | |
3427 | ctype->wcoutdigits[cnt] = L'0' + cnt; | |
4b10dd6c UD |
3428 | } |
3429 | ||
3430 | ctype->outdigits_act = 10; | |
19bc17a9 RM |
3431 | } |
3432 | } | |
3433 | ||
3434 | ||
ef446144 UD |
3435 | /* Initialize. Assumes t->p and t->q have already been set. */ |
3436 | static inline void | |
3437 | wctype_table_init (struct wctype_table *t) | |
3438 | { | |
d6040f17 | 3439 | t->level1 = NULL; |
ef446144 | 3440 | t->level1_alloc = t->level1_size = 0; |
d6040f17 | 3441 | t->level2 = NULL; |
ef446144 | 3442 | t->level2_alloc = t->level2_size = 0; |
d6040f17 | 3443 | t->level3 = NULL; |
ef446144 UD |
3444 | t->level3_alloc = t->level3_size = 0; |
3445 | } | |
3446 | ||
ec08818d UD |
3447 | /* Retrieve an entry. */ |
3448 | static inline int | |
3449 | wctype_table_get (struct wctype_table *t, uint32_t wc) | |
3450 | { | |
3451 | uint32_t index1 = wc >> (t->q + t->p + 5); | |
3452 | if (index1 < t->level1_size) | |
3453 | { | |
3454 | uint32_t lookup1 = t->level1[index1]; | |
bd75759f | 3455 | if (lookup1 != EMPTY) |
ec08818d UD |
3456 | { |
3457 | uint32_t index2 = ((wc >> (t->p + 5)) & ((1 << t->q) - 1)) | |
3458 | + (lookup1 << t->q); | |
3459 | uint32_t lookup2 = t->level2[index2]; | |
bd75759f | 3460 | if (lookup2 != EMPTY) |
ec08818d UD |
3461 | { |
3462 | uint32_t index3 = ((wc >> 5) & ((1 << t->p) - 1)) | |
3463 | + (lookup2 << t->p); | |
3464 | uint32_t lookup3 = t->level3[index3]; | |
3465 | uint32_t index4 = wc & 0x1f; | |
3466 | ||
3467 | return (lookup3 >> index4) & 1; | |
3468 | } | |
3469 | } | |
3470 | } | |
3471 | return 0; | |
3472 | } | |
3473 | ||
ef446144 UD |
3474 | /* Add one entry. */ |
3475 | static void | |
3476 | wctype_table_add (struct wctype_table *t, uint32_t wc) | |
3477 | { | |
3478 | uint32_t index1 = wc >> (t->q + t->p + 5); | |
3479 | uint32_t index2 = (wc >> (t->p + 5)) & ((1 << t->q) - 1); | |
3480 | uint32_t index3 = (wc >> 5) & ((1 << t->p) - 1); | |
3481 | uint32_t index4 = wc & 0x1f; | |
3482 | size_t i, i1, i2; | |
3483 | ||
3484 | if (index1 >= t->level1_size) | |
3485 | { | |
3486 | if (index1 >= t->level1_alloc) | |
3487 | { | |
3488 | size_t alloc = 2 * t->level1_alloc; | |
3489 | if (alloc <= index1) | |
3490 | alloc = index1 + 1; | |
d6040f17 UD |
3491 | t->level1 = (uint32_t *) xrealloc ((char *) t->level1, |
3492 | alloc * sizeof (uint32_t)); | |
ef446144 UD |
3493 | t->level1_alloc = alloc; |
3494 | } | |
3495 | while (index1 >= t->level1_size) | |
bd75759f | 3496 | t->level1[t->level1_size++] = EMPTY; |
ef446144 UD |
3497 | } |
3498 | ||
bd75759f | 3499 | if (t->level1[index1] == EMPTY) |
ef446144 UD |
3500 | { |
3501 | if (t->level2_size == t->level2_alloc) | |
3502 | { | |
3503 | size_t alloc = 2 * t->level2_alloc + 1; | |
d6040f17 UD |
3504 | t->level2 = (uint32_t *) xrealloc ((char *) t->level2, |
3505 | (alloc << t->q) * sizeof (uint32_t)); | |
ef446144 UD |
3506 | t->level2_alloc = alloc; |
3507 | } | |
3508 | i1 = t->level2_size << t->q; | |
3509 | i2 = (t->level2_size + 1) << t->q; | |
3510 | for (i = i1; i < i2; i++) | |
bd75759f | 3511 | t->level2[i] = EMPTY; |
ef446144 UD |
3512 | t->level1[index1] = t->level2_size++; |
3513 | } | |
3514 | ||
3515 | index2 += t->level1[index1] << t->q; | |
3516 | ||
bd75759f | 3517 | if (t->level2[index2] == EMPTY) |
ef446144 UD |
3518 | { |
3519 | if (t->level3_size == t->level3_alloc) | |
3520 | { | |
3521 | size_t alloc = 2 * t->level3_alloc + 1; | |
d6040f17 UD |
3522 | t->level3 = (uint32_t *) xrealloc ((char *) t->level3, |
3523 | (alloc << t->p) * sizeof (uint32_t)); | |
ef446144 UD |
3524 | t->level3_alloc = alloc; |
3525 | } | |
3526 | i1 = t->level3_size << t->p; | |
3527 | i2 = (t->level3_size + 1) << t->p; | |
3528 | for (i = i1; i < i2; i++) | |
3529 | t->level3[i] = 0; | |
3530 | t->level2[index2] = t->level3_size++; | |
3531 | } | |
3532 | ||
3533 | index3 += t->level2[index2] << t->p; | |
3534 | ||
3535 | t->level3[index3] |= (uint32_t)1 << index4; | |
3536 | } | |
3537 | ||
3538 | /* Finalize and shrink. */ | |
3539 | static void | |
1ecbb381 | 3540 | add_locale_wctype_table (struct locale_file *file, struct wctype_table *t) |
ef446144 UD |
3541 | { |
3542 | size_t i, j, k; | |
3543 | uint32_t reorder3[t->level3_size]; | |
3544 | uint32_t reorder2[t->level2_size]; | |
1ecbb381 | 3545 | uint32_t level2_offset, level3_offset; |
ef446144 UD |
3546 | |
3547 | /* Uniquify level3 blocks. */ | |
3548 | k = 0; | |
3549 | for (j = 0; j < t->level3_size; j++) | |
3550 | { | |
3551 | for (i = 0; i < k; i++) | |
3552 | if (memcmp (&t->level3[i << t->p], &t->level3[j << t->p], | |
3553 | (1 << t->p) * sizeof (uint32_t)) == 0) | |
3554 | break; | |
3555 | /* Relocate block j to block i. */ | |
3556 | reorder3[j] = i; | |
3557 | if (i == k) | |
3558 | { | |
3559 | if (i != j) | |
3560 | memcpy (&t->level3[i << t->p], &t->level3[j << t->p], | |
3561 | (1 << t->p) * sizeof (uint32_t)); | |
3562 | k++; | |
3563 | } | |
3564 | } | |
3565 | t->level3_size = k; | |
3566 | ||
3567 | for (i = 0; i < (t->level2_size << t->q); i++) | |
bd75759f | 3568 | if (t->level2[i] != EMPTY) |
ef446144 UD |
3569 | t->level2[i] = reorder3[t->level2[i]]; |
3570 | ||
3571 | /* Uniquify level2 blocks. */ | |
3572 | k = 0; | |
3573 | for (j = 0; j < t->level2_size; j++) | |
3574 | { | |
3575 | for (i = 0; i < k; i++) | |
3576 | if (memcmp (&t->level2[i << t->q], &t->level2[j << t->q], | |
3577 | (1 << t->q) * sizeof (uint32_t)) == 0) | |
3578 | break; | |
3579 | /* Relocate block j to block i. */ | |
3580 | reorder2[j] = i; | |
3581 | if (i == k) | |
3582 | { | |
3583 | if (i != j) | |
3584 | memcpy (&t->level2[i << t->q], &t->level2[j << t->q], | |
3585 | (1 << t->q) * sizeof (uint32_t)); | |
3586 | k++; | |
3587 | } | |
3588 | } | |
3589 | t->level2_size = k; | |
3590 | ||
3591 | for (i = 0; i < t->level1_size; i++) | |
bd75759f | 3592 | if (t->level1[i] != EMPTY) |
ef446144 UD |
3593 | t->level1[i] = reorder2[t->level1[i]]; |
3594 | ||
ef446144 UD |
3595 | t->result_size = |
3596 | 5 * sizeof (uint32_t) | |
3597 | + t->level1_size * sizeof (uint32_t) | |
3598 | + (t->level2_size << t->q) * sizeof (uint32_t) | |
3599 | + (t->level3_size << t->p) * sizeof (uint32_t); | |
ef446144 | 3600 | |
ef446144 UD |
3601 | level2_offset = |
3602 | 5 * sizeof (uint32_t) | |
3603 | + t->level1_size * sizeof (uint32_t); | |
3604 | level3_offset = | |
3605 | 5 * sizeof (uint32_t) | |
3606 | + t->level1_size * sizeof (uint32_t) | |
3607 | + (t->level2_size << t->q) * sizeof (uint32_t); | |
3608 | ||
1ecbb381 RS |
3609 | start_locale_structure (file); |
3610 | add_locale_uint32 (file, t->q + t->p + 5); | |
3611 | add_locale_uint32 (file, t->level1_size); | |
3612 | add_locale_uint32 (file, t->p + 5); | |
3613 | add_locale_uint32 (file, (1 << t->q) - 1); | |
3614 | add_locale_uint32 (file, (1 << t->p) - 1); | |
ef446144 UD |
3615 | |
3616 | for (i = 0; i < t->level1_size; i++) | |
1ecbb381 RS |
3617 | add_locale_uint32 |
3618 | (file, | |
3619 | t->level1[i] == EMPTY | |
ef446144 UD |
3620 | ? 0 |
3621 | : (t->level1[i] << t->q) * sizeof (uint32_t) + level2_offset); | |
3622 | ||
3623 | for (i = 0; i < (t->level2_size << t->q); i++) | |
1ecbb381 RS |
3624 | add_locale_uint32 |
3625 | (file, | |
3626 | t->level2[i] == EMPTY | |
ef446144 UD |
3627 | ? 0 |
3628 | : (t->level2[i] << t->p) * sizeof (uint32_t) + level3_offset); | |
3629 | ||
1ecbb381 RS |
3630 | add_locale_uint32_array (file, t->level3, t->level3_size << t->p); |
3631 | end_locale_structure (file); | |
ef446144 UD |
3632 | |
3633 | if (t->level1_alloc > 0) | |
3634 | free (t->level1); | |
3635 | if (t->level2_alloc > 0) | |
3636 | free (t->level2); | |
3637 | if (t->level3_alloc > 0) | |
3638 | free (t->level3); | |
3639 | } | |
3640 | ||
02fb3d17 UD |
3641 | /* Flattens the included transliterations into a translit list. |
3642 | Inserts them in the list at `cursor', and returns the new cursor. */ | |
3643 | static struct translit_t ** | |
47e8b443 UD |
3644 | translit_flatten (struct locale_ctype_t *ctype, |
3645 | const struct charmap_t *charmap, | |
02fb3d17 UD |
3646 | struct translit_t **cursor) |
3647 | { | |
3648 | while (ctype->translit_include != NULL) | |
3649 | { | |
3650 | const char *copy_locale = ctype->translit_include->copy_locale; | |
3651 | const char *copy_repertoire = ctype->translit_include->copy_repertoire; | |
3652 | struct localedef_t *other; | |
3653 | ||
3654 | /* Unchain the include statement. During the depth-first traversal | |
3655 | we don't want to visit any locale more than once. */ | |
3656 | ctype->translit_include = ctype->translit_include->next; | |
3657 | ||
3658 | other = find_locale (LC_CTYPE, copy_locale, copy_repertoire, charmap); | |
3659 | ||
6e310111 | 3660 | if (other == NULL || other->categories[LC_CTYPE].ctype == NULL) |
02fb3d17 | 3661 | { |
f2b98f97 | 3662 | WITH_CUR_LOCALE (error (0, 0, _("\ |
02fb3d17 | 3663 | %s: transliteration data from locale `%s' not available"), |
f2b98f97 | 3664 | "LC_CTYPE", copy_locale)); |
02fb3d17 UD |
3665 | } |
3666 | else | |
3667 | { | |
3668 | struct locale_ctype_t *other_ctype = | |
3669 | other->categories[LC_CTYPE].ctype; | |
3670 | ||
3671 | cursor = translit_flatten (other_ctype, charmap, cursor); | |
3672 | assert (other_ctype->translit_include == NULL); | |
3673 | ||
3674 | if (other_ctype->translit != NULL) | |
3675 | { | |
3676 | /* Insert the other_ctype->translit list at *cursor. */ | |
3677 | struct translit_t *endp = other_ctype->translit; | |
3678 | while (endp->next != NULL) | |
3679 | endp = endp->next; | |
3680 | ||
3681 | endp->next = *cursor; | |
3682 | *cursor = other_ctype->translit; | |
3683 | ||
3684 | /* Avoid any risk of circular lists. */ | |
3685 | other_ctype->translit = NULL; | |
3686 | ||
3687 | cursor = &endp->next; | |
3688 | } | |
3689 | ||
3690 | if (ctype->default_missing == NULL) | |
3691 | ctype->default_missing = other_ctype->default_missing; | |
3692 | } | |
3693 | } | |
3694 | ||
3695 | return cursor; | |
3696 | } | |
3697 | ||
19bc17a9 | 3698 | static void |
47e8b443 | 3699 | allocate_arrays (struct locale_ctype_t *ctype, const struct charmap_t *charmap, |
4b10dd6c | 3700 | struct repertoire_t *repertoire) |
19bc17a9 | 3701 | { |
4c7d276e | 3702 | size_t idx, nr; |
0e16ecfa UD |
3703 | const void *key; |
3704 | size_t len; | |
3705 | void *vdata; | |
3706 | void *curs; | |
5d431a3e | 3707 | |
19bc17a9 RM |
3708 | /* You wonder about this amount of memory? This is only because some |
3709 | users do not manage to address the array with unsigned values or | |
3710 | data types with range >= 256. '\200' would result in the array | |
3711 | index -128. To help these poor people we duplicate the entries for | |
3712 | 128 up to 255 below the entry for \0. */ | |
4c7d276e UD |
3713 | ctype->ctype_b = (char_class_t *) xcalloc (256 + 128, sizeof (char_class_t)); |
3714 | ctype->ctype32_b = (char_class32_t *) xcalloc (256, sizeof (char_class32_t)); | |
3715 | ctype->class_b = (uint32_t **) | |
3716 | xmalloc (ctype->nr_charclass * sizeof (uint32_t *)); | |
1ecbb381 RS |
3717 | ctype->class_3level = (struct wctype_table *) |
3718 | xmalloc (ctype->nr_charclass * sizeof (struct wctype_table)); | |
19bc17a9 | 3719 | |
4a33c2f5 | 3720 | /* This is the array accessed using the multibyte string elements. */ |
4b10dd6c | 3721 | for (idx = 0; idx < 256; ++idx) |
4a33c2f5 | 3722 | ctype->ctype_b[128 + idx] = ctype->class256_collection[idx]; |
19bc17a9 | 3723 | |
75cd5204 RM |
3724 | /* Mirror first 127 entries. We must take care that entry -1 is not |
3725 | mirrored because EOF == -1. */ | |
3726 | for (idx = 0; idx < 127; ++idx) | |
19bc17a9 RM |
3727 | ctype->ctype_b[idx] = ctype->ctype_b[256 + idx]; |
3728 | ||
4c7d276e UD |
3729 | /* The 32 bit array contains all characters < 0x100. */ |
3730 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
3731 | if (ctype->charnames[idx] < 0x100) | |
3732 | ctype->ctype32_b[ctype->charnames[idx]] = ctype->class_collection[idx]; | |
ef446144 | 3733 | |
4c7d276e | 3734 | for (nr = 0; nr < ctype->nr_charclass; nr++) |
ef446144 | 3735 | { |
4c7d276e | 3736 | ctype->class_b[nr] = (uint32_t *) xcalloc (256 / 32, sizeof (uint32_t)); |
ef446144 | 3737 | |
9a5c46e8 UD |
3738 | /* We only set CLASS_B for the bits in the ISO C classes, not |
3739 | the user defined classes. The number should not change but | |
3740 | who knows. */ | |
3741 | #define LAST_ISO_C_BIT 11 | |
3742 | if (nr <= LAST_ISO_C_BIT) | |
3743 | for (idx = 0; idx < 256; ++idx) | |
3744 | if (ctype->class256_collection[idx] & _ISbit (nr)) | |
3745 | ctype->class_b[nr][idx >> 5] |= (uint32_t) 1 << (idx & 0x1f); | |
4c7d276e | 3746 | } |
8fb81470 | 3747 | |
4c7d276e UD |
3748 | for (nr = 0; nr < ctype->nr_charclass; nr++) |
3749 | { | |
1ecbb381 | 3750 | struct wctype_table *t; |
ef446144 | 3751 | |
1ecbb381 RS |
3752 | t = &ctype->class_3level[nr]; |
3753 | t->p = 4; /* or: 5 */ | |
3754 | t->q = 7; /* or: 6 */ | |
3755 | wctype_table_init (t); | |
ef446144 | 3756 | |
4c7d276e UD |
3757 | for (idx = 0; idx < ctype->class_collection_act; ++idx) |
3758 | if (ctype->class_collection[idx] & _ISwbit (nr)) | |
1ecbb381 | 3759 | wctype_table_add (t, ctype->charnames[idx]); |
ef446144 | 3760 | |
4c7d276e | 3761 | if (verbose) |
f2b98f97 UD |
3762 | WITH_CUR_LOCALE (fprintf (stderr, _("\ |
3763 | %s: table for class \"%s\": %lu bytes\n"), | |
3764 | "LC_CTYPE", ctype->classnames[nr], | |
1ecbb381 | 3765 | (unsigned long int) t->result_size)); |
ef446144 | 3766 | } |
19bc17a9 RM |
3767 | |
3768 | /* Room for table of mappings. */ | |
4c7d276e UD |
3769 | ctype->map_b = (uint32_t **) xmalloc (2 * sizeof (uint32_t *)); |
3770 | ctype->map32_b = (uint32_t **) xmalloc (ctype->map_collection_nr | |
3771 | * sizeof (uint32_t *)); | |
1ecbb381 RS |
3772 | ctype->map_3level = (struct wctrans_table *) |
3773 | xmalloc (ctype->map_collection_nr * sizeof (struct wctrans_table)); | |
19bc17a9 RM |
3774 | |
3775 | /* Fill in all mappings. */ | |
49f2be5b | 3776 | for (idx = 0; idx < 2; ++idx) |
19bc17a9 RM |
3777 | { |
3778 | unsigned int idx2; | |
3779 | ||
3780 | /* Allocate table. */ | |
4c7d276e UD |
3781 | ctype->map_b[idx] = (uint32_t *) |
3782 | xmalloc ((256 + 128) * sizeof (uint32_t)); | |
19bc17a9 RM |
3783 | |
3784 | /* Copy values from collection. */ | |
4b10dd6c | 3785 | for (idx2 = 0; idx2 < 256; ++idx2) |
4c7d276e | 3786 | ctype->map_b[idx][128 + idx2] = ctype->map256_collection[idx][idx2]; |
19bc17a9 | 3787 | |
75cd5204 RM |
3788 | /* Mirror first 127 entries. We must take care not to map entry |
3789 | -1 because EOF == -1. */ | |
3790 | for (idx2 = 0; idx2 < 127; ++idx2) | |
4c7d276e | 3791 | ctype->map_b[idx][idx2] = ctype->map_b[idx][256 + idx2]; |
19bc17a9 | 3792 | |
75cd5204 | 3793 | /* EOF must map to EOF. */ |
4c7d276e | 3794 | ctype->map_b[idx][127] = EOF; |
49f2be5b | 3795 | } |
a9c27b3e | 3796 | |
49f2be5b UD |
3797 | for (idx = 0; idx < ctype->map_collection_nr; ++idx) |
3798 | { | |
3799 | unsigned int idx2; | |
3800 | ||
3801 | /* Allocate table. */ | |
4c7d276e | 3802 | ctype->map32_b[idx] = (uint32_t *) xmalloc (256 * sizeof (uint32_t)); |
49f2be5b | 3803 | |
4c7d276e | 3804 | /* Copy values from collection. Default is identity mapping. */ |
49f2be5b | 3805 | for (idx2 = 0; idx2 < 256; ++idx2) |
4c7d276e UD |
3806 | ctype->map32_b[idx][idx2] = |
3807 | (ctype->map_collection[idx][idx2] != 0 | |
3808 | ? ctype->map_collection[idx][idx2] | |
3809 | : idx2); | |
ef446144 UD |
3810 | } |
3811 | ||
4c7d276e | 3812 | for (nr = 0; nr < ctype->map_collection_nr; nr++) |
ef446144 | 3813 | { |
1ecbb381 | 3814 | struct wctrans_table *t; |
ef446144 | 3815 | |
1ecbb381 RS |
3816 | t = &ctype->map_3level[nr]; |
3817 | t->p = 7; | |
3818 | t->q = 9; | |
3819 | wctrans_table_init (t); | |
ef446144 | 3820 | |
4c7d276e UD |
3821 | for (idx = 0; idx < ctype->map_collection_act[nr]; ++idx) |
3822 | if (ctype->map_collection[nr][idx] != 0) | |
1ecbb381 | 3823 | wctrans_table_add (t, ctype->charnames[idx], |
4c7d276e | 3824 | ctype->map_collection[nr][idx]); |
ef446144 | 3825 | |
4c7d276e | 3826 | if (verbose) |
f2b98f97 UD |
3827 | WITH_CUR_LOCALE (fprintf (stderr, _("\ |
3828 | %s: table for map \"%s\": %lu bytes\n"), | |
3829 | "LC_CTYPE", ctype->mapnames[nr], | |
1ecbb381 | 3830 | (unsigned long int) t->result_size)); |
19bc17a9 RM |
3831 | } |
3832 | ||
3833 | /* Extra array for class and map names. */ | |
4b10dd6c UD |
3834 | ctype->class_name_ptr = (uint32_t *) xmalloc (ctype->nr_charclass |
3835 | * sizeof (uint32_t)); | |
3836 | ctype->map_name_ptr = (uint32_t *) xmalloc (ctype->map_collection_nr | |
3837 | * sizeof (uint32_t)); | |
75cd5204 | 3838 | |
4c7d276e UD |
3839 | ctype->class_offset = _NL_ITEM_INDEX (_NL_CTYPE_EXTRA_MAP_1); |
3840 | ctype->map_offset = ctype->class_offset + ctype->nr_charclass; | |
5866b131 | 3841 | |
4a9dcff1 UD |
3842 | /* Array for width information. Because the expected widths are very |
3843 | small (never larger than 2) we use only one single byte. This | |
3844 | saves space. | |
3845 | We put only printable characters in the table. wcwidth is specified | |
3846 | to return -1 for non-printable characters. Doing the check here | |
3847 | saves a run-time check. | |
3848 | But we put L'\0' in the table. This again saves a run-time check. */ | |
4c7d276e | 3849 | { |
1ecbb381 | 3850 | struct wcwidth_table *t; |
ef446144 | 3851 | |
1ecbb381 RS |
3852 | t = &ctype->width; |
3853 | t->p = 7; | |
3854 | t->q = 9; | |
3855 | wcwidth_table_init (t); | |
ef446144 | 3856 | |
4a9dcff1 UD |
3857 | /* First set all the printable characters of the character set to |
3858 | the default width. */ | |
4c7d276e UD |
3859 | curs = NULL; |
3860 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
3861 | { | |
3862 | struct charseq *data = (struct charseq *) vdata; | |
0e16ecfa | 3863 | |
4c7d276e UD |
3864 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) |
3865 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
3866 | data->name, len); | |
ef446144 | 3867 | |
4c7d276e | 3868 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) |
4a9dcff1 UD |
3869 | { |
3870 | uint32_t *class_bits = | |
3871 | find_idx (ctype, &ctype->class_collection, NULL, | |
3872 | &ctype->class_collection_act, data->ucs4); | |
3873 | ||
3874 | if (class_bits != NULL && (*class_bits & BITw (tok_print))) | |
1ecbb381 | 3875 | wcwidth_table_add (t, data->ucs4, charmap->width_default); |
4a9dcff1 | 3876 | } |
4c7d276e | 3877 | } |
ef446144 | 3878 | |
4c7d276e UD |
3879 | /* Now add the explicitly specified widths. */ |
3880 | if (charmap->width_rules != NULL) | |
3881 | { | |
3882 | size_t cnt; | |
ef446144 | 3883 | |
4c7d276e UD |
3884 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
3885 | { | |
3886 | unsigned char bytes[charmap->mb_cur_max]; | |
3887 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
3888 | ||
3889 | /* We have the range of character for which the width is | |
3890 | specified described using byte sequences of the multibyte | |
3891 | charset. We have to convert this to UCS4 now. And we | |
3892 | cannot simply convert the beginning and the end of the | |
3893 | sequence, we have to iterate over the byte sequence and | |
3894 | convert it for every single character. */ | |
3895 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
3896 | ||
3897 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
3898 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
3899 | nbytes) <= 0) | |
3900 | { | |
3901 | /* Find the UCS value for `bytes'. */ | |
3902 | int inner; | |
3903 | uint32_t wch; | |
3904 | struct charseq *seq = | |
701666b7 | 3905 | charmap_find_symbol (charmap, (char *) bytes, nbytes); |
0e16ecfa | 3906 | |
4c7d276e UD |
3907 | if (seq == NULL) |
3908 | wch = ILLEGAL_CHAR_VALUE; | |
3909 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
3910 | wch = seq->ucs4; | |
3911 | else | |
3912 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
3913 | strlen (seq->name)); | |
ef446144 | 3914 | |
4c7d276e | 3915 | if (wch != ILLEGAL_CHAR_VALUE) |
4a9dcff1 UD |
3916 | { |
3917 | /* Store the value. */ | |
3918 | uint32_t *class_bits = | |
3919 | find_idx (ctype, &ctype->class_collection, NULL, | |
3920 | &ctype->class_collection_act, wch); | |
3921 | ||
3922 | if (class_bits != NULL && (*class_bits & BITw (tok_print))) | |
1ecbb381 | 3923 | wcwidth_table_add (t, wch, |
4a9dcff1 UD |
3924 | charmap->width_rules[cnt].width); |
3925 | } | |
ef446144 | 3926 | |
4c7d276e UD |
3927 | /* "Increment" the bytes sequence. */ |
3928 | inner = nbytes - 1; | |
3929 | while (inner >= 0 && bytes[inner] == 0xff) | |
3930 | --inner; | |
0e16ecfa | 3931 | |
4c7d276e UD |
3932 | if (inner < 0) |
3933 | { | |
3934 | /* We have to extend the byte sequence. */ | |
3935 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
3936 | break; | |
ef446144 | 3937 | |
4c7d276e UD |
3938 | bytes[0] = 1; |
3939 | memset (&bytes[1], 0, nbytes); | |
3940 | ++nbytes; | |
3941 | } | |
3942 | else | |
3943 | { | |
3944 | ++bytes[inner]; | |
3945 | while (++inner < nbytes) | |
3946 | bytes[inner] = 0; | |
3947 | } | |
3948 | } | |
3949 | } | |
3950 | } | |
ef446144 | 3951 | |
4a9dcff1 | 3952 | /* Set the width of L'\0' to 0. */ |
1ecbb381 | 3953 | wcwidth_table_add (t, 0, 0); |
ef446144 | 3954 | |
4c7d276e | 3955 | if (verbose) |
f2b98f97 | 3956 | WITH_CUR_LOCALE (fprintf (stderr, _("%s: table for width: %lu bytes\n"), |
1ecbb381 | 3957 | "LC_CTYPE", (unsigned long int) t->result_size)); |
4c7d276e | 3958 | } |
0e16ecfa | 3959 | |
4b10dd6c UD |
3960 | /* Set MB_CUR_MAX. */ |
3961 | ctype->mb_cur_max = charmap->mb_cur_max; | |
6990326c | 3962 | |
4b10dd6c UD |
3963 | /* Now determine the table for the transliteration information. |
3964 | ||
3965 | XXX It is not yet clear to me whether it is worth implementing a | |
3966 | complicated algorithm which uses a hash table to locate the entries. | |
3967 | For now I'll use a simple array which can be searching using binary | |
3968 | search. */ | |
02fb3d17 UD |
3969 | if (ctype->translit_include != NULL) |
3970 | /* Traverse the locales mentioned in the `include' statements in a | |
3971 | depth-first way and fold in their transliteration information. */ | |
3972 | translit_flatten (ctype, charmap, &ctype->translit); | |
4b10dd6c UD |
3973 | |
3974 | if (ctype->translit != NULL) | |
3975 | { | |
3976 | /* First count how many entries we have. This is the upper limit | |
3977 | since some entries from the included files might be overwritten. */ | |
3978 | size_t number = 0; | |
3979 | size_t cnt; | |
3980 | struct translit_t *runp = ctype->translit; | |
3981 | struct translit_t **sorted; | |
3982 | size_t from_len, to_len; | |
3983 | ||
3984 | while (runp != NULL) | |
3985 | { | |
3986 | ++number; | |
3987 | runp = runp->next; | |
3988 | } | |
3989 | ||
3990 | /* Next we allocate an array large enough and fill in the values. */ | |
a9c27b3e UD |
3991 | sorted = (struct translit_t **) alloca (number |
3992 | * sizeof (struct translit_t **)); | |
4b10dd6c UD |
3993 | runp = ctype->translit; |
3994 | number = 0; | |
3995 | do | |
3996 | { | |
3997 | /* Search for the place where to insert this string. | |
3998 | XXX Better use a real sorting algorithm later. */ | |
3999 | size_t idx = 0; | |
4000 | int replace = 0; | |
4001 | ||
4002 | while (idx < number) | |
4003 | { | |
4004 | int res = wcscmp ((const wchar_t *) sorted[idx]->from, | |
4005 | (const wchar_t *) runp->from); | |
4006 | if (res == 0) | |
4007 | { | |
4008 | replace = 1; | |
4009 | break; | |
4010 | } | |
4011 | if (res > 0) | |
4012 | break; | |
4013 | ++idx; | |
4014 | } | |
4015 | ||
4016 | if (replace) | |
4017 | sorted[idx] = runp; | |
4018 | else | |
4019 | { | |
4020 | memmove (&sorted[idx + 1], &sorted[idx], | |
4021 | (number - idx) * sizeof (struct translit_t *)); | |
4022 | sorted[idx] = runp; | |
4023 | ++number; | |
4024 | } | |
4025 | ||
4026 | runp = runp->next; | |
4027 | } | |
4028 | while (runp != NULL); | |
4029 | ||
4030 | /* The next step is putting all the possible transliteration | |
4031 | strings in one memory block so that we can write it out. | |
4032 | We need several different blocks: | |
9ca23765 | 4033 | - index to the from-string array |
4b10dd6c UD |
4034 | - from-string array |
4035 | - index to the to-string array | |
4036 | - to-string array. | |
4b10dd6c UD |
4037 | */ |
4038 | from_len = to_len = 0; | |
4039 | for (cnt = 0; cnt < number; ++cnt) | |
4040 | { | |
4041 | struct translit_to_t *srunp; | |
4042 | from_len += wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4043 | srunp = sorted[cnt]->to; | |
4044 | while (srunp != NULL) | |
4045 | { | |
4046 | to_len += wcslen ((const wchar_t *) srunp->str) + 1; | |
4047 | srunp = srunp->next; | |
4048 | } | |
4049 | /* Plus one for the extra NUL character marking the end of | |
4050 | the list for the current entry. */ | |
4051 | ++to_len; | |
4052 | } | |
4053 | ||
4054 | /* We can allocate the arrays for the results. */ | |
4a33c2f5 UD |
4055 | ctype->translit_from_idx = xmalloc (number * sizeof (uint32_t)); |
4056 | ctype->translit_from_tbl = xmalloc (from_len * sizeof (uint32_t)); | |
4057 | ctype->translit_to_idx = xmalloc (number * sizeof (uint32_t)); | |
4058 | ctype->translit_to_tbl = xmalloc (to_len * sizeof (uint32_t)); | |
4b10dd6c UD |
4059 | |
4060 | from_len = 0; | |
4061 | to_len = 0; | |
4062 | for (cnt = 0; cnt < number; ++cnt) | |
4063 | { | |
4064 | size_t len; | |
4065 | struct translit_to_t *srunp; | |
4066 | ||
4a33c2f5 UD |
4067 | ctype->translit_from_idx[cnt] = from_len; |
4068 | ctype->translit_to_idx[cnt] = to_len; | |
4b10dd6c UD |
4069 | |
4070 | len = wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4a33c2f5 | 4071 | wmemcpy ((wchar_t *) &ctype->translit_from_tbl[from_len], |
4b10dd6c UD |
4072 | (const wchar_t *) sorted[cnt]->from, len); |
4073 | from_len += len; | |
4074 | ||
4a33c2f5 | 4075 | ctype->translit_to_idx[cnt] = to_len; |
4b10dd6c UD |
4076 | srunp = sorted[cnt]->to; |
4077 | while (srunp != NULL) | |
4078 | { | |
4079 | len = wcslen ((const wchar_t *) srunp->str) + 1; | |
4a33c2f5 | 4080 | wmemcpy ((wchar_t *) &ctype->translit_to_tbl[to_len], |
4b10dd6c UD |
4081 | (const wchar_t *) srunp->str, len); |
4082 | to_len += len; | |
4083 | srunp = srunp->next; | |
4084 | } | |
4a33c2f5 | 4085 | ctype->translit_to_tbl[to_len++] = L'\0'; |
4b10dd6c | 4086 | } |
4b10dd6c UD |
4087 | |
4088 | /* Store the information about the length. */ | |
04fbc779 | 4089 | ctype->translit_idx_size = number; |
4b10dd6c UD |
4090 | ctype->translit_from_tbl_size = from_len * sizeof (uint32_t); |
4091 | ctype->translit_to_tbl_size = to_len * sizeof (uint32_t); | |
4092 | } | |
4093 | else | |
4094 | { | |
4095 | /* Provide some dummy pointers since we have nothing to write out. */ | |
4096 | static uint32_t no_str = { 0 }; | |
4097 | ||
4a33c2f5 UD |
4098 | ctype->translit_from_idx = &no_str; |
4099 | ctype->translit_from_tbl = &no_str; | |
4100 | ctype->translit_to_tbl = &no_str; | |
4b10dd6c UD |
4101 | ctype->translit_idx_size = 0; |
4102 | ctype->translit_from_tbl_size = 0; | |
4103 | ctype->translit_to_tbl_size = 0; | |
4104 | } | |
19bc17a9 | 4105 | } |