]>
Commit | Line | Data |
---|---|---|
01ff9d0b | 1 | /* Copyright (C) 1995-1999, 2000 Free Software Foundation, Inc. |
c84142e8 | 2 | This file is part of the GNU C Library. |
4b10dd6c | 3 | Contributed by Ulrich Drepper <drepper@gnu.org>, 1995. |
19bc17a9 | 4 | |
c84142e8 UD |
5 | The GNU C Library is free software; you can redistribute it and/or |
6 | modify it under the terms of the GNU Library General Public License as | |
7 | published by the Free Software Foundation; either version 2 of the | |
8 | License, or (at your option) any later version. | |
19bc17a9 | 9 | |
c84142e8 UD |
10 | The GNU C Library is distributed in the hope that it will be useful, |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
13 | Library General Public License for more details. | |
19bc17a9 | 14 | |
c84142e8 UD |
15 | You should have received a copy of the GNU Library General Public |
16 | License along with the GNU C Library; see the file COPYING.LIB. If not, | |
17 | write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
18 | Boston, MA 02111-1307, USA. */ | |
19bc17a9 RM |
19 | |
20 | #ifdef HAVE_CONFIG_H | |
21 | # include <config.h> | |
22 | #endif | |
23 | ||
a68b0d31 | 24 | #include <alloca.h> |
4b10dd6c | 25 | #include <byteswap.h> |
19bc17a9 | 26 | #include <endian.h> |
4b10dd6c | 27 | #include <errno.h> |
19bc17a9 | 28 | #include <limits.h> |
4b10dd6c UD |
29 | #include <obstack.h> |
30 | #include <stdlib.h> | |
19bc17a9 | 31 | #include <string.h> |
4b10dd6c UD |
32 | #include <wchar.h> |
33 | #include <wctype.h> | |
34 | #include <sys/uio.h> | |
19bc17a9 | 35 | |
4b10dd6c | 36 | #include "charmap.h" |
19bc17a9 RM |
37 | #include "localeinfo.h" |
38 | #include "langinfo.h" | |
4b10dd6c | 39 | #include "linereader.h" |
19bc17a9 | 40 | #include "locfile-token.h" |
4b10dd6c UD |
41 | #include "locfile.h" |
42 | #include "localedef.h" | |
19bc17a9 | 43 | |
19bc17a9 RM |
44 | #include <assert.h> |
45 | ||
46 | ||
011ebfab | 47 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
48 | /* These are the extra bits not in wctype.h since these are not preallocated |
49 | classes. */ | |
011ebfab UD |
50 | # define _ISwspecial1 (1 << 29) |
51 | # define _ISwspecial2 (1 << 30) | |
52 | # define _ISwspecial3 (1 << 31) | |
53 | #endif | |
19bc17a9 RM |
54 | |
55 | ||
56 | /* The bit used for representing a special class. */ | |
57 | #define BITPOS(class) ((class) - tok_upper) | |
4b10dd6c UD |
58 | #define BIT(class) (_ISbit (BITPOS (class))) |
59 | #define BITw(class) (_ISwbit (BITPOS (class))) | |
19bc17a9 RM |
60 | |
61 | #define ELEM(ctype, collection, idx, value) \ | |
62 | *find_idx (ctype, &ctype->collection idx, &ctype->collection##_max idx, \ | |
63 | &ctype->collection##_act idx, value) | |
64 | ||
19bc17a9 RM |
65 | |
66 | /* To be compatible with former implementations we for now restrict | |
67 | the number of bits for character classes to 16. When compatibility | |
68 | is not necessary anymore increase the number to 32. */ | |
4b10dd6c | 69 | #define char_class_t uint16_t |
4b10dd6c | 70 | #define char_class32_t uint32_t |
4b10dd6c UD |
71 | |
72 | ||
73 | /* Type to describe a transliteration action. We have a possibly | |
74 | multiple character from-string and a set of multiple character | |
75 | to-strings. All are 32bit values since this is what is used in | |
76 | the gconv functions. */ | |
77 | struct translit_to_t | |
78 | { | |
79 | uint32_t *str; | |
80 | ||
81 | struct translit_to_t *next; | |
82 | }; | |
83 | ||
84 | struct translit_t | |
85 | { | |
86 | uint32_t *from; | |
87 | ||
a673fbcb UD |
88 | const char *fname; |
89 | size_t lineno; | |
90 | ||
4b10dd6c UD |
91 | struct translit_to_t *to; |
92 | ||
93 | struct translit_t *next; | |
94 | }; | |
19bc17a9 | 95 | |
a673fbcb UD |
96 | struct translit_ignore_t |
97 | { | |
98 | uint32_t from; | |
99 | uint32_t to; | |
a0dc5206 | 100 | uint32_t step; |
a673fbcb UD |
101 | |
102 | const char *fname; | |
103 | size_t lineno; | |
104 | ||
105 | struct translit_ignore_t *next; | |
106 | }; | |
107 | ||
19bc17a9 RM |
108 | |
109 | /* The real definition of the struct for the LC_CTYPE locale. */ | |
110 | struct locale_ctype_t | |
111 | { | |
4b10dd6c | 112 | uint32_t *charnames; |
19bc17a9 RM |
113 | size_t charnames_max; |
114 | size_t charnames_act; | |
115 | ||
4b10dd6c UD |
116 | struct repertoire_t *repertoire; |
117 | ||
118 | /* We will allow up to 8 * sizeof (uint32_t) character classes. */ | |
119 | #define MAX_NR_CHARCLASS (8 * sizeof (uint32_t)) | |
ba1ffaa1 | 120 | size_t nr_charclass; |
19bc17a9 | 121 | const char *classnames[MAX_NR_CHARCLASS]; |
4b10dd6c UD |
122 | uint32_t last_class_char; |
123 | uint32_t class256_collection[256]; | |
124 | uint32_t *class_collection; | |
19bc17a9 RM |
125 | size_t class_collection_max; |
126 | size_t class_collection_act; | |
4b10dd6c | 127 | uint32_t class_done; |
ef446144 | 128 | uint32_t class_offset; |
4b10dd6c UD |
129 | |
130 | struct charseq **mbdigits; | |
131 | size_t mbdigits_act; | |
132 | size_t mbdigits_max; | |
133 | uint32_t *wcdigits; | |
134 | size_t wcdigits_act; | |
135 | size_t wcdigits_max; | |
136 | ||
137 | struct charseq *mboutdigits[10]; | |
138 | uint32_t wcoutdigits[10]; | |
139 | size_t outdigits_act; | |
19bc17a9 RM |
140 | |
141 | /* If the following number ever turns out to be too small simply | |
142 | increase it. But I doubt it will. --drepper@gnu */ | |
143 | #define MAX_NR_CHARMAP 16 | |
144 | const char *mapnames[MAX_NR_CHARMAP]; | |
4b10dd6c UD |
145 | uint32_t *map_collection[MAX_NR_CHARMAP]; |
146 | uint32_t map256_collection[2][256]; | |
9a0a462c UD |
147 | size_t map_collection_max[MAX_NR_CHARMAP]; |
148 | size_t map_collection_act[MAX_NR_CHARMAP]; | |
19bc17a9 RM |
149 | size_t map_collection_nr; |
150 | size_t last_map_idx; | |
4b10dd6c | 151 | int tomap_done[MAX_NR_CHARMAP]; |
ef446144 | 152 | uint32_t map_offset; |
4b10dd6c UD |
153 | |
154 | /* Transliteration information. */ | |
155 | const char *translit_copy_locale; | |
156 | const char *translit_copy_repertoire; | |
157 | struct translit_t *translit; | |
a673fbcb | 158 | struct translit_ignore_t *translit_ignore; |
a8e4c924 | 159 | uint32_t ntranslit_ignore; |
a673fbcb UD |
160 | |
161 | uint32_t *default_missing; | |
162 | const char *default_missing_file; | |
163 | size_t default_missing_lineno; | |
19bc17a9 RM |
164 | |
165 | /* The arrays for the binary representation. */ | |
4b10dd6c UD |
166 | uint32_t plane_size; |
167 | uint32_t plane_cnt; | |
19bc17a9 RM |
168 | char_class_t *ctype_b; |
169 | char_class32_t *ctype32_b; | |
4a33c2f5 UD |
170 | uint32_t *names; |
171 | uint32_t **map; | |
49f2be5b | 172 | uint32_t **map32; |
ef446144 UD |
173 | struct iovec *class_3level; |
174 | struct iovec *map_3level; | |
4b10dd6c UD |
175 | uint32_t *class_name_ptr; |
176 | uint32_t *map_name_ptr; | |
75cd5204 | 177 | unsigned char *width; |
ef446144 | 178 | struct iovec width_3level; |
4b10dd6c | 179 | uint32_t mb_cur_max; |
6990326c | 180 | const char *codeset_name; |
4a33c2f5 UD |
181 | uint32_t *translit_from_idx; |
182 | uint32_t *translit_from_tbl; | |
183 | uint32_t *translit_to_idx; | |
184 | uint32_t *translit_to_tbl; | |
04fbc779 | 185 | uint32_t translit_idx_size; |
4b10dd6c UD |
186 | size_t translit_from_tbl_size; |
187 | size_t translit_to_tbl_size; | |
188 | ||
a673fbcb | 189 | struct obstack mempool; |
19bc17a9 RM |
190 | }; |
191 | ||
192 | ||
4b10dd6c UD |
193 | #define obstack_chunk_alloc xmalloc |
194 | #define obstack_chunk_free free | |
195 | ||
196 | ||
19bc17a9 | 197 | /* Prototypes for local functions. */ |
4b10dd6c UD |
198 | static void ctype_startup (struct linereader *lr, struct localedef_t *locale, |
199 | struct charmap_t *charmap, int ignore_content); | |
200 | static void ctype_class_new (struct linereader *lr, | |
201 | struct locale_ctype_t *ctype, const char *name); | |
202 | static void ctype_map_new (struct linereader *lr, | |
203 | struct locale_ctype_t *ctype, | |
204 | const char *name, struct charmap_t *charmap); | |
205 | static uint32_t *find_idx (struct locale_ctype_t *ctype, uint32_t **table, | |
206 | size_t *max, size_t *act, unsigned int idx); | |
19bc17a9 | 207 | static void set_class_defaults (struct locale_ctype_t *ctype, |
4b10dd6c UD |
208 | struct charmap_t *charmap, |
209 | struct repertoire_t *repertoire); | |
75cd5204 | 210 | static void allocate_arrays (struct locale_ctype_t *ctype, |
4b10dd6c UD |
211 | struct charmap_t *charmap, |
212 | struct repertoire_t *repertoire); | |
19bc17a9 RM |
213 | |
214 | ||
4b10dd6c UD |
215 | static const char *longnames[] = |
216 | { | |
217 | "zero", "one", "two", "three", "four", | |
218 | "five", "six", "seven", "eight", "nine" | |
219 | }; | |
1b97149d UD |
220 | static const char *uninames[] = |
221 | { | |
222 | "U00000030", "U00000031", "U00000032", "U00000033", "U00000034", | |
223 | "U00000035", "U00000036", "U00000037", "U00000038", "U00000039" | |
224 | }; | |
4b10dd6c UD |
225 | static const unsigned char digits[] = "0123456789"; |
226 | ||
227 | ||
228 | static void | |
19bc17a9 | 229 | ctype_startup (struct linereader *lr, struct localedef_t *locale, |
4b10dd6c | 230 | struct charmap_t *charmap, int ignore_content) |
19bc17a9 RM |
231 | { |
232 | unsigned int cnt; | |
233 | struct locale_ctype_t *ctype; | |
234 | ||
4b10dd6c | 235 | if (!ignore_content) |
19bc17a9 | 236 | { |
4b10dd6c UD |
237 | /* Allocate the needed room. */ |
238 | locale->categories[LC_CTYPE].ctype = ctype = | |
239 | (struct locale_ctype_t *) xcalloc (1, sizeof (struct locale_ctype_t)); | |
240 | ||
241 | /* We have seen no names yet. */ | |
242 | ctype->charnames_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
243 | ctype->charnames = | |
244 | (unsigned int *) xmalloc (ctype->charnames_max | |
245 | * sizeof (unsigned int)); | |
246 | for (cnt = 0; cnt < 256; ++cnt) | |
247 | ctype->charnames[cnt] = cnt; | |
248 | ctype->charnames_act = 256; | |
249 | ||
250 | /* Fill character class information. */ | |
251 | ctype->last_class_char = ILLEGAL_CHAR_VALUE; | |
252 | /* The order of the following instructions determines the bit | |
253 | positions! */ | |
254 | ctype_class_new (lr, ctype, "upper"); | |
255 | ctype_class_new (lr, ctype, "lower"); | |
256 | ctype_class_new (lr, ctype, "alpha"); | |
257 | ctype_class_new (lr, ctype, "digit"); | |
258 | ctype_class_new (lr, ctype, "xdigit"); | |
259 | ctype_class_new (lr, ctype, "space"); | |
260 | ctype_class_new (lr, ctype, "print"); | |
261 | ctype_class_new (lr, ctype, "graph"); | |
262 | ctype_class_new (lr, ctype, "blank"); | |
263 | ctype_class_new (lr, ctype, "cntrl"); | |
264 | ctype_class_new (lr, ctype, "punct"); | |
265 | ctype_class_new (lr, ctype, "alnum"); | |
011ebfab | 266 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
267 | /* The following are extensions from ISO 14652. */ |
268 | ctype_class_new (lr, ctype, "left_to_right"); | |
269 | ctype_class_new (lr, ctype, "right_to_left"); | |
270 | ctype_class_new (lr, ctype, "num_terminator"); | |
271 | ctype_class_new (lr, ctype, "num_separator"); | |
272 | ctype_class_new (lr, ctype, "segment_separator"); | |
273 | ctype_class_new (lr, ctype, "block_separator"); | |
274 | ctype_class_new (lr, ctype, "direction_control"); | |
275 | ctype_class_new (lr, ctype, "sym_swap_layout"); | |
276 | ctype_class_new (lr, ctype, "char_shape_selector"); | |
277 | ctype_class_new (lr, ctype, "num_shape_selector"); | |
278 | ctype_class_new (lr, ctype, "non_spacing"); | |
279 | ctype_class_new (lr, ctype, "non_spacing_level3"); | |
280 | ctype_class_new (lr, ctype, "normal_connect"); | |
281 | ctype_class_new (lr, ctype, "r_connect"); | |
282 | ctype_class_new (lr, ctype, "no_connect"); | |
283 | ctype_class_new (lr, ctype, "no_connect-space"); | |
284 | ctype_class_new (lr, ctype, "vowel_connect"); | |
011ebfab | 285 | #endif |
4b10dd6c UD |
286 | |
287 | ctype->class_collection_max = charmap->mb_cur_max == 1 ? 256 : 512; | |
288 | ctype->class_collection | |
289 | = (uint32_t *) xcalloc (sizeof (unsigned long int), | |
290 | ctype->class_collection_max); | |
291 | ctype->class_collection_act = 256; | |
292 | ||
293 | /* Fill character map information. */ | |
4b10dd6c UD |
294 | ctype->last_map_idx = MAX_NR_CHARMAP; |
295 | ctype_map_new (lr, ctype, "toupper", charmap); | |
296 | ctype_map_new (lr, ctype, "tolower", charmap); | |
011ebfab | 297 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 298 | ctype_map_new (lr, ctype, "tosymmetric", charmap); |
011ebfab | 299 | #endif |
4b10dd6c UD |
300 | |
301 | /* Fill first 256 entries in `toXXX' arrays. */ | |
302 | for (cnt = 0; cnt < 256; ++cnt) | |
303 | { | |
304 | ctype->map_collection[0][cnt] = cnt; | |
305 | ctype->map_collection[1][cnt] = cnt; | |
9e2b7438 | 306 | #ifdef PREDEFINED_CLASSES |
4b10dd6c | 307 | ctype->map_collection[2][cnt] = cnt; |
9e2b7438 | 308 | #endif |
4b10dd6c UD |
309 | ctype->map256_collection[0][cnt] = cnt; |
310 | ctype->map256_collection[1][cnt] = cnt; | |
311 | } | |
312 | ||
a673fbcb | 313 | obstack_init (&ctype->mempool); |
19bc17a9 RM |
314 | } |
315 | } | |
316 | ||
317 | ||
318 | void | |
4b10dd6c | 319 | ctype_finish (struct localedef_t *locale, struct charmap_t *charmap) |
19bc17a9 RM |
320 | { |
321 | /* See POSIX.2, table 2-6 for the meaning of the following table. */ | |
322 | #define NCLASS 12 | |
323 | static const struct | |
324 | { | |
325 | const char *name; | |
326 | const char allow[NCLASS]; | |
327 | } | |
328 | valid_table[NCLASS] = | |
329 | { | |
330 | /* The order is important. See token.h for more information. | |
331 | M = Always, D = Default, - = Permitted, X = Mutually exclusive */ | |
332 | { "upper", "--MX-XDDXXX-" }, | |
333 | { "lower", "--MX-XDDXXX-" }, | |
334 | { "alpha", "---X-XDDXXX-" }, | |
335 | { "digit", "XXX--XDDXXX-" }, | |
336 | { "xdigit", "-----XDDXXX-" }, | |
337 | { "space", "XXXXX------X" }, | |
338 | { "print", "---------X--" }, | |
339 | { "graph", "---------X--" }, | |
340 | { "blank", "XXXXXM-----X" }, | |
341 | { "cntrl", "XXXXX-XX--XX" }, | |
342 | { "punct", "XXXXX-DD-X-X" }, | |
343 | { "alnum", "-----XDDXXX-" } | |
344 | }; | |
345 | size_t cnt; | |
346 | int cls1, cls2; | |
4b10dd6c UD |
347 | uint32_t space_value; |
348 | struct charseq *space_seq; | |
19bc17a9 | 349 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
4b10dd6c | 350 | int warned; |
0e16ecfa UD |
351 | const void *key; |
352 | size_t len; | |
353 | void *vdata; | |
354 | void *curs; | |
19bc17a9 | 355 | |
b9eb05d6 UD |
356 | /* Now resolve copying and also handle completely missing definitions. */ |
357 | if (ctype == NULL) | |
358 | { | |
70e51ab9 UD |
359 | const char *repertoire_name; |
360 | ||
b9eb05d6 UD |
361 | /* First see whether we were supposed to copy. If yes, find the |
362 | actual definition. */ | |
363 | if (locale->copy_name[LC_CTYPE] != NULL) | |
364 | { | |
365 | /* Find the copying locale. This has to happen transitively since | |
366 | the locale we are copying from might also copying another one. */ | |
367 | struct localedef_t *from = locale; | |
368 | ||
369 | do | |
370 | from = find_locale (LC_CTYPE, from->copy_name[LC_CTYPE], | |
371 | from->repertoire_name, charmap); | |
372 | while (from->categories[LC_CTYPE].ctype == NULL | |
373 | && from->copy_name[LC_CTYPE] != NULL); | |
374 | ||
375 | ctype = locale->categories[LC_CTYPE].ctype | |
376 | = from->categories[LC_CTYPE].ctype; | |
377 | } | |
378 | ||
379 | /* If there is still no definition issue an warning and create an | |
380 | empty one. */ | |
381 | if (ctype == NULL) | |
382 | { | |
f6ada7ad UD |
383 | if (! be_quiet) |
384 | error (0, 0, _("No definition for %s category found"), "LC_CTYPE"); | |
b9eb05d6 UD |
385 | ctype_startup (NULL, locale, charmap, 0); |
386 | ctype = locale->categories[LC_CTYPE].ctype; | |
387 | } | |
70e51ab9 UD |
388 | |
389 | /* Get the repertoire we have to use. */ | |
390 | repertoire_name = locale->repertoire_name ?: repertoire_global; | |
391 | if (repertoire_name != NULL) | |
392 | ctype->repertoire = repertoire_read (repertoire_name); | |
b9eb05d6 UD |
393 | } |
394 | ||
db76d943 UD |
395 | /* We need the name of the currently used 8-bit character set to |
396 | make correct conversion between this 8-bit representation and the | |
397 | ISO 10646 character set used internally for wide characters. */ | |
398 | ctype->codeset_name = charmap->code_set_name; | |
399 | if (ctype->codeset_name == NULL) | |
400 | { | |
401 | if (! be_quiet) | |
402 | error (0, 0, "no character set name specified in charmap"); | |
403 | ctype->codeset_name = "//UNKNOWN//"; | |
404 | } | |
405 | ||
19bc17a9 | 406 | /* Set default value for classes not specified. */ |
4b10dd6c | 407 | set_class_defaults (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
408 | |
409 | /* Check according to table. */ | |
42d7c593 | 410 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) |
19bc17a9 | 411 | { |
4b10dd6c | 412 | uint32_t tmp = ctype->class_collection[cnt]; |
19bc17a9 | 413 | |
4b10dd6c UD |
414 | if (tmp != 0) |
415 | { | |
416 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
417 | if ((tmp & _ISwbit (cls1)) != 0) | |
418 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
419 | if (valid_table[cls1].allow[cls2] != '-') | |
19bc17a9 | 420 | { |
4b10dd6c UD |
421 | int eq = (tmp & _ISwbit (cls2)) != 0; |
422 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 423 | { |
4b10dd6c UD |
424 | case 'M': |
425 | if (!eq) | |
426 | { | |
427 | uint32_t value = ctype->charnames[cnt]; | |
428 | ||
429 | if (!be_quiet) | |
430 | error (0, 0, _("\ | |
431 | character L'\\u%0*x' in class `%s' must be in class `%s'"), | |
432 | value > 0xffff ? 8 : 4, value, | |
433 | valid_table[cls1].name, | |
434 | valid_table[cls2].name); | |
435 | } | |
436 | break; | |
437 | ||
438 | case 'X': | |
439 | if (eq) | |
440 | { | |
441 | uint32_t value = ctype->charnames[cnt]; | |
442 | ||
443 | if (!be_quiet) | |
444 | error (0, 0, _("\ | |
445 | character L'\\u%0*x' in class `%s' must not be in class `%s'"), | |
446 | value > 0xffff ? 8 : 4, value, | |
447 | valid_table[cls1].name, | |
448 | valid_table[cls2].name); | |
449 | } | |
450 | break; | |
451 | ||
452 | case 'D': | |
453 | ctype->class_collection[cnt] |= _ISwbit (cls2); | |
454 | break; | |
455 | ||
456 | default: | |
457 | error (5, 0, _("internal error in %s, line %u"), | |
458 | __FUNCTION__, __LINE__); | |
19bc17a9 | 459 | } |
4b10dd6c UD |
460 | } |
461 | } | |
462 | } | |
463 | ||
464 | for (cnt = 0; cnt < 256; ++cnt) | |
465 | { | |
466 | uint32_t tmp = ctype->class256_collection[cnt]; | |
19bc17a9 | 467 | |
4b10dd6c UD |
468 | if (tmp != 0) |
469 | { | |
470 | for (cls1 = 0; cls1 < NCLASS; ++cls1) | |
471 | if ((tmp & _ISbit (cls1)) != 0) | |
472 | for (cls2 = 0; cls2 < NCLASS; ++cls2) | |
473 | if (valid_table[cls1].allow[cls2] != '-') | |
474 | { | |
475 | int eq = (tmp & _ISbit (cls2)) != 0; | |
476 | switch (valid_table[cls1].allow[cls2]) | |
19bc17a9 | 477 | { |
4b10dd6c UD |
478 | case 'M': |
479 | if (!eq) | |
480 | { | |
481 | char buf[17]; | |
482 | ||
5d431a3e | 483 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
484 | |
485 | if (!be_quiet) | |
486 | error (0, 0, _("\ | |
487 | character '%s' in class `%s' must be in class `%s'"), | |
488 | buf, valid_table[cls1].name, | |
489 | valid_table[cls2].name); | |
490 | } | |
491 | break; | |
492 | ||
493 | case 'X': | |
494 | if (eq) | |
495 | { | |
496 | char buf[17]; | |
497 | ||
5d431a3e | 498 | snprintf (buf, sizeof buf, "\\%Zo", cnt); |
4b10dd6c UD |
499 | |
500 | if (!be_quiet) | |
501 | error (0, 0, _("\ | |
502 | character '%s' in class `%s' must not be in class `%s'"), | |
503 | buf, valid_table[cls1].name, | |
504 | valid_table[cls2].name); | |
505 | } | |
506 | break; | |
507 | ||
508 | case 'D': | |
509 | ctype->class256_collection[cnt] |= _ISbit (cls2); | |
510 | break; | |
511 | ||
512 | default: | |
513 | error (5, 0, _("internal error in %s, line %u"), | |
514 | __FUNCTION__, __LINE__); | |
19bc17a9 | 515 | } |
4b10dd6c UD |
516 | } |
517 | } | |
19bc17a9 RM |
518 | } |
519 | ||
520 | /* ... and now test <SP> as a special case. */ | |
a0dc5206 UD |
521 | space_value = 32; |
522 | if (((cnt = BITPOS (tok_space), | |
523 | (ELEM (ctype, class_collection, , space_value) | |
524 | & BITw (tok_space)) == 0) | |
525 | || (cnt = BITPOS (tok_blank), | |
526 | (ELEM (ctype, class_collection, , space_value) | |
527 | & BITw (tok_blank)) == 0))) | |
880f421f UD |
528 | { |
529 | if (!be_quiet) | |
530 | error (0, 0, _("<SP> character not in class `%s'"), | |
531 | valid_table[cnt].name); | |
532 | } | |
c84142e8 UD |
533 | else if (((cnt = BITPOS (tok_punct), |
534 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 535 | & BITw (tok_punct)) != 0) |
c84142e8 UD |
536 | || (cnt = BITPOS (tok_graph), |
537 | (ELEM (ctype, class_collection, , space_value) | |
4b10dd6c | 538 | & BITw (tok_graph)) |
880f421f UD |
539 | != 0))) |
540 | { | |
541 | if (!be_quiet) | |
542 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
543 | valid_table[cnt].name); | |
544 | } | |
19bc17a9 | 545 | else |
4b10dd6c UD |
546 | ELEM (ctype, class_collection, , space_value) |= BITw (tok_print); |
547 | ||
548 | space_seq = charmap_find_value (charmap, "SP", 2); | |
ce177a84 | 549 | if (space_seq == NULL) |
45c95239 UD |
550 | space_seq = charmap_find_value (charmap, "space", 5); |
551 | if (space_seq == NULL) | |
1b97149d | 552 | space_seq = charmap_find_value (charmap, "U00000020", 9); |
4b10dd6c UD |
553 | if (space_seq == NULL || space_seq->nbytes != 1) |
554 | { | |
555 | if (!be_quiet) | |
556 | error (0, 0, _("character <SP> not defined in character map")); | |
557 | } | |
558 | else if (((cnt = BITPOS (tok_space), | |
559 | (ctype->class256_collection[space_seq->bytes[0]] | |
560 | & BIT (tok_space)) == 0) | |
561 | || (cnt = BITPOS (tok_blank), | |
562 | (ctype->class256_collection[space_seq->bytes[0]] | |
563 | & BIT (tok_blank)) == 0))) | |
564 | { | |
565 | if (!be_quiet) | |
566 | error (0, 0, _("<SP> character not in class `%s'"), | |
567 | valid_table[cnt].name); | |
568 | } | |
569 | else if (((cnt = BITPOS (tok_punct), | |
570 | (ctype->class256_collection[space_seq->bytes[0]] | |
571 | & BIT (tok_punct)) != 0) | |
572 | || (cnt = BITPOS (tok_graph), | |
573 | (ctype->class256_collection[space_seq->bytes[0]] | |
574 | & BIT (tok_graph)) != 0))) | |
575 | { | |
576 | if (!be_quiet) | |
577 | error (0, 0, _("<SP> character must not be in class `%s'"), | |
578 | valid_table[cnt].name); | |
579 | } | |
580 | else | |
581 | ctype->class256_collection[space_seq->bytes[0]] |= BIT (tok_print); | |
75cd5204 RM |
582 | |
583 | /* Now that the tests are done make sure the name array contains all | |
584 | characters which are handled in the WIDTH section of the | |
585 | character set definition file. */ | |
4b10dd6c UD |
586 | if (charmap->width_rules != NULL) |
587 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) | |
75cd5204 | 588 | { |
827ff758 UD |
589 | unsigned char bytes[charmap->mb_cur_max]; |
590 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
591 | ||
592 | /* We have the range of character for which the width is | |
593 | specified described using byte sequences of the multibyte | |
594 | charset. We have to convert this to UCS4 now. And we | |
595 | cannot simply convert the beginning and the end of the | |
596 | sequence, we have to iterate over the byte sequence and | |
597 | convert it for every single character. */ | |
598 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
599 | ||
600 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
601 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
602 | nbytes) <= 0) | |
603 | { | |
604 | /* Find the UCS value for `bytes'. */ | |
827ff758 | 605 | int inner; |
76e680a8 UD |
606 | uint32_t wch; |
607 | struct charseq *seq = charmap_find_symbol (charmap, bytes, nbytes); | |
608 | ||
609 | if (seq == NULL) | |
610 | wch = ILLEGAL_CHAR_VALUE; | |
611 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
612 | wch = seq->ucs4; | |
613 | else | |
614 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
615 | strlen (seq->name)); | |
827ff758 UD |
616 | |
617 | if (wch != ILLEGAL_CHAR_VALUE) | |
618 | /* We are only interested in the side-effects of the | |
619 | `find_idx' call. It will add appropriate entries in | |
620 | the name array if this is necessary. */ | |
621 | (void) find_idx (ctype, NULL, NULL, NULL, wch); | |
622 | ||
623 | /* "Increment" the bytes sequence. */ | |
624 | inner = nbytes - 1; | |
625 | while (inner >= 0 && bytes[inner] == 0xff) | |
626 | --inner; | |
627 | ||
628 | if (inner < 0) | |
629 | { | |
630 | /* We have to extend the byte sequence. */ | |
631 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
632 | break; | |
633 | ||
634 | bytes[0] = 1; | |
635 | memset (&bytes[1], 0, nbytes); | |
636 | ++nbytes; | |
637 | } | |
638 | else | |
639 | { | |
640 | ++bytes[inner]; | |
641 | while (++inner < nbytes) | |
642 | bytes[inner] = 0; | |
643 | } | |
644 | } | |
4b10dd6c UD |
645 | } |
646 | ||
0e16ecfa UD |
647 | /* Now set all the other characters of the character set to the |
648 | default width. */ | |
649 | curs = NULL; | |
650 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
651 | { | |
652 | struct charseq *data = (struct charseq *) vdata; | |
653 | ||
654 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
655 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
656 | data->name, len); | |
657 | ||
658 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) | |
659 | (void) find_idx (ctype, NULL, NULL, NULL, data->ucs4); | |
660 | } | |
661 | ||
4b10dd6c UD |
662 | /* There must be a multiple of 10 digits. */ |
663 | if (ctype->mbdigits_act % 10 != 0) | |
664 | { | |
665 | assert (ctype->mbdigits_act == ctype->wcdigits_act); | |
666 | ctype->wcdigits_act -= ctype->mbdigits_act % 10; | |
667 | ctype->mbdigits_act -= ctype->mbdigits_act % 10; | |
668 | error (0, 0, _("`digit' category has not entries in groups of ten")); | |
669 | } | |
670 | ||
671 | /* Check the input digits. There must be a multiple of ten available. | |
42d7c593 | 672 | In each group it could be that one or the other character is missing. |
4b10dd6c UD |
673 | In this case the whole group must be removed. */ |
674 | cnt = 0; | |
675 | while (cnt < ctype->mbdigits_act) | |
676 | { | |
677 | size_t inner; | |
678 | for (inner = 0; inner < 10; ++inner) | |
679 | if (ctype->mbdigits[cnt + inner] == NULL) | |
680 | break; | |
681 | ||
682 | if (inner == 10) | |
683 | cnt += 10; | |
684 | else | |
685 | { | |
686 | /* Remove the group. */ | |
687 | memmove (&ctype->mbdigits[cnt], &ctype->mbdigits[cnt + 10], | |
688 | ((ctype->wcdigits_act - cnt - 10) | |
689 | * sizeof (ctype->mbdigits[0]))); | |
690 | ctype->mbdigits_act -= 10; | |
691 | } | |
692 | } | |
693 | ||
694 | /* If no input digits are given use the default. */ | |
695 | if (ctype->mbdigits_act == 0) | |
696 | { | |
697 | if (ctype->mbdigits_max == 0) | |
698 | { | |
699 | ctype->mbdigits = obstack_alloc (&charmap->mem_pool, | |
700 | 10 * sizeof (struct charseq *)); | |
701 | ctype->mbdigits_max = 10; | |
702 | } | |
703 | ||
704 | for (cnt = 0; cnt < 10; ++cnt) | |
705 | { | |
706 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
707 | digits + cnt, 1); | |
708 | if (ctype->mbdigits[cnt] == NULL) | |
709 | { | |
710 | ctype->mbdigits[cnt] = charmap_find_symbol (charmap, | |
711 | longnames[cnt], | |
712 | strlen (longnames[cnt])); | |
713 | if (ctype->mbdigits[cnt] == NULL) | |
714 | { | |
715 | /* Hum, this ain't good. */ | |
716 | error (0, 0, _("\ | |
717 | no input digits defined and none of the standard names in the charmap")); | |
718 | ||
719 | ctype->mbdigits[cnt] = obstack_alloc (&charmap->mem_pool, | |
720 | sizeof (struct charseq) + 1); | |
721 | ||
722 | /* This is better than nothing. */ | |
723 | ctype->mbdigits[cnt]->bytes[0] = digits[cnt]; | |
724 | ctype->mbdigits[cnt]->nbytes = 1; | |
725 | } | |
726 | } | |
727 | } | |
728 | ||
729 | ctype->mbdigits_act = 10; | |
730 | } | |
731 | ||
732 | /* Check the wide character input digits. There must be a multiple | |
42d7c593 | 733 | of ten available. In each group it could be that one or the other |
4b10dd6c UD |
734 | character is missing. In this case the whole group must be |
735 | removed. */ | |
736 | cnt = 0; | |
737 | while (cnt < ctype->wcdigits_act) | |
738 | { | |
739 | size_t inner; | |
740 | for (inner = 0; inner < 10; ++inner) | |
741 | if (ctype->wcdigits[cnt + inner] == ILLEGAL_CHAR_VALUE) | |
742 | break; | |
743 | ||
744 | if (inner == 10) | |
745 | cnt += 10; | |
746 | else | |
747 | { | |
748 | /* Remove the group. */ | |
749 | memmove (&ctype->wcdigits[cnt], &ctype->wcdigits[cnt + 10], | |
750 | ((ctype->wcdigits_act - cnt - 10) | |
751 | * sizeof (ctype->wcdigits[0]))); | |
752 | ctype->wcdigits_act -= 10; | |
753 | } | |
754 | } | |
755 | ||
756 | /* If no input digits are given use the default. */ | |
757 | if (ctype->wcdigits_act == 0) | |
758 | { | |
759 | if (ctype->wcdigits_max == 0) | |
760 | { | |
761 | ctype->wcdigits = obstack_alloc (&charmap->mem_pool, | |
762 | 10 * sizeof (uint32_t)); | |
763 | ctype->wcdigits_max = 10; | |
764 | } | |
765 | ||
766 | for (cnt = 0; cnt < 10; ++cnt) | |
767 | ctype->wcdigits[cnt] = L'0' + cnt; | |
768 | ||
769 | ctype->mbdigits_act = 10; | |
770 | } | |
771 | ||
772 | /* Check the outdigits. */ | |
773 | warned = 0; | |
774 | for (cnt = 0; cnt < 10; ++cnt) | |
775 | if (ctype->mboutdigits[cnt] == NULL) | |
776 | { | |
777 | static struct charseq replace[2]; | |
778 | ||
779 | if (!warned) | |
780 | { | |
781 | error (0, 0, _("\ | |
782 | not all characters used in `outdigit' are available in the charmap")); | |
783 | warned = 1; | |
784 | } | |
785 | ||
786 | replace[0].nbytes = 1; | |
787 | replace[0].bytes[0] = '?'; | |
788 | replace[0].bytes[1] = '\0'; | |
789 | ctype->mboutdigits[cnt] = &replace[0]; | |
790 | } | |
791 | ||
792 | warned = 0; | |
793 | for (cnt = 0; cnt < 10; ++cnt) | |
794 | if (ctype->wcoutdigits[cnt] == 0) | |
795 | { | |
796 | if (!warned) | |
797 | { | |
798 | error (0, 0, _("\ | |
799 | not all characters used in `outdigit' are available in the repertoire")); | |
800 | warned = 1; | |
801 | } | |
802 | ||
803 | ctype->wcoutdigits[cnt] = L'?'; | |
75cd5204 | 804 | } |
a8e4c924 UD |
805 | |
806 | /* Sort the entries in the translit_ignore list. */ | |
807 | if (ctype->translit_ignore != NULL) | |
808 | { | |
809 | struct translit_ignore_t *firstp = ctype->translit_ignore; | |
810 | struct translit_ignore_t *runp; | |
811 | ||
812 | ctype->ntranslit_ignore = 1; | |
813 | ||
814 | for (runp = firstp->next; runp != NULL; runp = runp->next) | |
815 | { | |
816 | struct translit_ignore_t *lastp = NULL; | |
817 | struct translit_ignore_t *cmpp; | |
818 | ||
819 | ++ctype->ntranslit_ignore; | |
820 | ||
821 | for (cmpp = firstp; cmpp != NULL; lastp = cmpp, cmpp = cmpp->next) | |
822 | if (runp->from < cmpp->from) | |
823 | break; | |
824 | ||
825 | runp->next = lastp; | |
826 | if (lastp == NULL) | |
827 | firstp = runp; | |
828 | } | |
829 | ||
830 | ctype->translit_ignore = firstp; | |
831 | } | |
19bc17a9 RM |
832 | } |
833 | ||
834 | ||
835 | void | |
4b10dd6c | 836 | ctype_output (struct localedef_t *locale, struct charmap_t *charmap, |
75cd5204 | 837 | const char *output_path) |
19bc17a9 | 838 | { |
e43e0dd6 | 839 | static const char nulbytes[4] = { 0, 0, 0, 0 }; |
19bc17a9 RM |
840 | struct locale_ctype_t *ctype = locale->categories[LC_CTYPE].ctype; |
841 | const size_t nelems = (_NL_ITEM_INDEX (_NL_NUM_LC_CTYPE) | |
ef446144 UD |
842 | + (oldstyle_tables |
843 | ? (ctype->map_collection_nr - 2) | |
844 | : (ctype->nr_charclass + ctype->map_collection_nr))); | |
75cd5204 | 845 | struct iovec iov[2 + nelems + ctype->nr_charclass |
f175216d | 846 | + ctype->map_collection_nr + 2]; |
19bc17a9 | 847 | struct locale_file data; |
4b10dd6c | 848 | uint32_t idx[nelems + 1]; |
1d96d74d | 849 | uint32_t default_missing_len; |
75cd5204 | 850 | size_t elem, cnt, offset, total; |
4b10dd6c | 851 | char *cp; |
19bc17a9 RM |
852 | |
853 | /* Now prepare the output: Find the sizes of the table we can use. */ | |
4b10dd6c | 854 | allocate_arrays (ctype, charmap, ctype->repertoire); |
19bc17a9 RM |
855 | |
856 | data.magic = LIMAGIC (LC_CTYPE); | |
857 | data.n = nelems; | |
858 | iov[0].iov_base = (void *) &data; | |
859 | iov[0].iov_len = sizeof (data); | |
860 | ||
861 | iov[1].iov_base = (void *) idx; | |
a0edd63e | 862 | iov[1].iov_len = nelems * sizeof (uint32_t); |
19bc17a9 RM |
863 | |
864 | idx[0] = iov[0].iov_len + iov[1].iov_len; | |
865 | offset = 0; | |
866 | ||
867 | for (elem = 0; elem < nelems; ++elem) | |
868 | { | |
869 | if (elem < _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) | |
870 | switch (elem) | |
871 | { | |
c6df09ad UD |
872 | #define CTYPE_EMPTY(name) \ |
873 | case name: \ | |
04fbc779 | 874 | iov[2 + elem + offset].iov_base = (void *) ""; \ |
c6df09ad UD |
875 | iov[2 + elem + offset].iov_len = 0; \ |
876 | idx[elem + 1] = idx[elem]; \ | |
877 | break | |
878 | ||
879 | CTYPE_EMPTY(_NL_CTYPE_GAP1); | |
880 | CTYPE_EMPTY(_NL_CTYPE_GAP2); | |
881 | CTYPE_EMPTY(_NL_CTYPE_GAP3); | |
882 | ||
19bc17a9 RM |
883 | #define CTYPE_DATA(name, base, len) \ |
884 | case _NL_ITEM_INDEX (name): \ | |
ce7a5ef4 RM |
885 | iov[2 + elem + offset].iov_base = (base); \ |
886 | iov[2 + elem + offset].iov_len = (len); \ | |
1d96d74d | 887 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; \ |
19bc17a9 RM |
888 | break |
889 | ||
890 | CTYPE_DATA (_NL_CTYPE_CLASS, | |
891 | ctype->ctype_b, | |
892 | (256 + 128) * sizeof (char_class_t)); | |
893 | ||
4a33c2f5 UD |
894 | CTYPE_DATA (_NL_CTYPE_TOUPPER, |
895 | ctype->map[0], | |
f1d8b804 | 896 | (256 + 128) * sizeof (uint32_t)); |
4a33c2f5 UD |
897 | CTYPE_DATA (_NL_CTYPE_TOLOWER, |
898 | ctype->map[1], | |
f1d8b804 | 899 | (256 + 128) * sizeof (uint32_t)); |
19bc17a9 | 900 | |
49f2be5b UD |
901 | CTYPE_DATA (_NL_CTYPE_TOUPPER32, |
902 | ctype->map32[0], | |
ef446144 | 903 | (oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 256) |
49f2be5b UD |
904 | * sizeof (uint32_t)); |
905 | CTYPE_DATA (_NL_CTYPE_TOLOWER32, | |
906 | ctype->map32[1], | |
ef446144 | 907 | (oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 256) |
49f2be5b UD |
908 | * sizeof (uint32_t)); |
909 | ||
19bc17a9 RM |
910 | CTYPE_DATA (_NL_CTYPE_CLASS32, |
911 | ctype->ctype32_b, | |
ef446144 UD |
912 | (oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 256) |
913 | * sizeof (char_class32_t)); | |
19bc17a9 | 914 | |
4a33c2f5 | 915 | CTYPE_DATA (_NL_CTYPE_NAMES, |
ef446144 UD |
916 | ctype->names, |
917 | (oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 0) | |
918 | * sizeof (uint32_t)); | |
919 | ||
920 | CTYPE_DATA (_NL_CTYPE_CLASS_OFFSET, | |
921 | &ctype->class_offset, sizeof (uint32_t)); | |
922 | ||
923 | CTYPE_DATA (_NL_CTYPE_MAP_OFFSET, | |
924 | &ctype->map_offset, sizeof (uint32_t)); | |
4a33c2f5 | 925 | |
04fbc779 UD |
926 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TAB_SIZE, |
927 | &ctype->translit_idx_size, sizeof (uint32_t)); | |
4a33c2f5 UD |
928 | |
929 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_IDX, | |
930 | ctype->translit_from_idx, | |
04fbc779 | 931 | ctype->translit_idx_size * sizeof (uint32_t)); |
4b10dd6c | 932 | |
4a33c2f5 UD |
933 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_FROM_TBL, |
934 | ctype->translit_from_tbl, | |
4b10dd6c UD |
935 | ctype->translit_from_tbl_size); |
936 | ||
4a33c2f5 UD |
937 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_IDX, |
938 | ctype->translit_to_idx, | |
04fbc779 | 939 | ctype->translit_idx_size * sizeof (uint32_t)); |
4b10dd6c | 940 | |
4a33c2f5 UD |
941 | CTYPE_DATA (_NL_CTYPE_TRANSLIT_TO_TBL, |
942 | ctype->translit_to_tbl, ctype->translit_to_tbl_size); | |
4b10dd6c | 943 | |
4a33c2f5 | 944 | CTYPE_DATA (_NL_CTYPE_HASH_SIZE, |
4b10dd6c | 945 | &ctype->plane_size, sizeof (uint32_t)); |
4a33c2f5 | 946 | CTYPE_DATA (_NL_CTYPE_HASH_LAYERS, |
4b10dd6c | 947 | &ctype->plane_cnt, sizeof (uint32_t)); |
19bc17a9 | 948 | |
75cd5204 RM |
949 | case _NL_ITEM_INDEX (_NL_CTYPE_CLASS_NAMES): |
950 | /* The class name array. */ | |
951 | total = 0; | |
952 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt, ++offset) | |
953 | { | |
954 | iov[2 + elem + offset].iov_base | |
955 | = (void *) ctype->classnames[cnt]; | |
956 | iov[2 + elem + offset].iov_len | |
957 | = strlen (ctype->classnames[cnt]) + 1; | |
958 | total += iov[2 + elem + offset].iov_len; | |
959 | } | |
e43e0dd6 | 960 | iov[2 + elem + offset].iov_base = (void *) nulbytes; |
ce7a5ef4 RM |
961 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); |
962 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 963 | |
4b10dd6c | 964 | idx[elem + 1] = idx[elem] + total; |
75cd5204 RM |
965 | break; |
966 | ||
967 | case _NL_ITEM_INDEX (_NL_CTYPE_MAP_NAMES): | |
968 | /* The class name array. */ | |
969 | total = 0; | |
970 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt, ++offset) | |
971 | { | |
972 | iov[2 + elem + offset].iov_base | |
973 | = (void *) ctype->mapnames[cnt]; | |
974 | iov[2 + elem + offset].iov_len | |
975 | = strlen (ctype->mapnames[cnt]) + 1; | |
976 | total += iov[2 + elem + offset].iov_len; | |
977 | } | |
e43e0dd6 | 978 | iov[2 + elem + offset].iov_base = (void *) nulbytes; |
ce7a5ef4 RM |
979 | iov[2 + elem + offset].iov_len = 1 + (4 - ((total + 1) % 4)); |
980 | total += 1 + (4 - ((total + 1) % 4)); | |
75cd5204 | 981 | |
4b10dd6c | 982 | idx[elem + 1] = idx[elem] + total; |
75cd5204 | 983 | break; |
19bc17a9 RM |
984 | |
985 | CTYPE_DATA (_NL_CTYPE_WIDTH, | |
ef446144 UD |
986 | (oldstyle_tables |
987 | ? ctype->width | |
988 | : ctype->width_3level.iov_base), | |
989 | (oldstyle_tables | |
990 | ? (ctype->plane_size * ctype->plane_cnt + 3) & ~3ul | |
991 | : ctype->width_3level.iov_len)); | |
19bc17a9 | 992 | |
0200214b | 993 | CTYPE_DATA (_NL_CTYPE_MB_CUR_MAX, |
4b10dd6c | 994 | &ctype->mb_cur_max, sizeof (uint32_t)); |
0200214b | 995 | |
ce7a5ef4 RM |
996 | case _NL_ITEM_INDEX (_NL_CTYPE_CODESET_NAME): |
997 | total = strlen (ctype->codeset_name) + 1; | |
998 | if (total % 4 == 0) | |
999 | iov[2 + elem + offset].iov_base = (char *) ctype->codeset_name; | |
1000 | else | |
1001 | { | |
1002 | iov[2 + elem + offset].iov_base = alloca ((total + 3) & ~3); | |
9756dfe1 UD |
1003 | memset (mempcpy (iov[2 + elem + offset].iov_base, |
1004 | ctype->codeset_name, total), | |
1005 | '\0', 4 - (total & 3)); | |
ce7a5ef4 RM |
1006 | total = (total + 3) & ~3; |
1007 | } | |
1008 | iov[2 + elem + offset].iov_len = total; | |
4b10dd6c UD |
1009 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
1010 | break; | |
1011 | ||
4a33c2f5 | 1012 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_MB_LEN): |
4b10dd6c UD |
1013 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
1014 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
1015 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
1016 | ctype->mbdigits_act / 10; | |
a9c27b3e | 1017 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
1018 | break; |
1019 | ||
4a33c2f5 | 1020 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS_WC_LEN): |
e43e0dd6 UD |
1021 | /* Align entries. */ |
1022 | iov[2 + elem + offset].iov_base = (void *) nulbytes; | |
1023 | iov[2 + elem + offset].iov_len = (4 - idx[elem] % 4) % 4; | |
1024 | idx[elem] += iov[2 + elem + offset].iov_len; | |
1025 | ++offset; | |
1026 | ||
4b10dd6c UD |
1027 | iov[2 + elem + offset].iov_base = alloca (sizeof (uint32_t)); |
1028 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
4a33c2f5 UD |
1029 | *(uint32_t *) iov[2 + elem + offset].iov_base = |
1030 | ctype->wcdigits_act / 10; | |
a9c27b3e | 1031 | idx[elem + 1] = idx[elem] + sizeof (uint32_t); |
4b10dd6c UD |
1032 | break; |
1033 | ||
e43e0dd6 | 1034 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_MB): |
4b10dd6c UD |
1035 | /* Compute the length of all possible characters. For INDIGITS |
1036 | there might be more than one. We simply concatenate all of | |
1037 | them with a NUL byte following. The NUL byte wouldn't be | |
1038 | necessary but it makes it easier for the user. */ | |
1039 | total = 0; | |
f175216d | 1040 | |
498b733e | 1041 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB); |
4b10dd6c UD |
1042 | cnt < ctype->mbdigits_act; cnt += 10) |
1043 | total += ctype->mbdigits[cnt]->nbytes + 1; | |
1044 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
1045 | iov[2 + elem + offset].iov_len = total; | |
1046 | ||
1047 | cp = iov[2 + elem + offset].iov_base; | |
498b733e | 1048 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_MB); |
4b10dd6c UD |
1049 | cnt < ctype->mbdigits_act; cnt += 10) |
1050 | { | |
1051 | cp = mempcpy (cp, ctype->mbdigits[cnt]->bytes, | |
1052 | ctype->mbdigits[cnt]->nbytes); | |
1053 | *cp++ = '\0'; | |
1054 | } | |
a9c27b3e | 1055 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1056 | break; |
1057 | ||
1058 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_MB): | |
1059 | /* Compute the length of all possible characters. For INDIGITS | |
1060 | there might be more than one. We simply concatenate all of | |
1061 | them with a NUL byte following. The NUL byte wouldn't be | |
1062 | necessary but it makes it easier for the user. */ | |
498b733e | 1063 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_MB); |
4b10dd6c UD |
1064 | total = ctype->mboutdigits[cnt]->nbytes + 1; |
1065 | iov[2 + elem + offset].iov_base = (char *) alloca (total); | |
1066 | iov[2 + elem + offset].iov_len = total; | |
1067 | ||
1068 | *(char *) mempcpy (iov[2 + elem + offset].iov_base, | |
498b733e UD |
1069 | ctype->mboutdigits[cnt]->bytes, |
1070 | ctype->mboutdigits[cnt]->nbytes) = '\0'; | |
a9c27b3e | 1071 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1072 | break; |
1073 | ||
e43e0dd6 | 1074 | case _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS9_WC): |
4b10dd6c UD |
1075 | total = ctype->wcdigits_act / 10; |
1076 | ||
1077 | iov[2 + elem + offset].iov_base = | |
1078 | (uint32_t *) alloca (total * sizeof (uint32_t)); | |
1079 | iov[2 + elem + offset].iov_len = total * sizeof (uint32_t); | |
1080 | ||
498b733e | 1081 | for (cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_INDIGITS0_WC); |
4b10dd6c UD |
1082 | cnt < ctype->wcdigits_act; cnt += 10) |
1083 | ((uint32_t *) iov[2 + elem + offset].iov_base)[cnt / 10] | |
4a33c2f5 | 1084 | = ctype->wcdigits[cnt]; |
a9c27b3e | 1085 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1086 | break; |
1087 | ||
e43e0dd6 UD |
1088 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC): |
1089 | /* Align entries. */ | |
1090 | iov[2 + elem + offset].iov_base = (void *) nulbytes; | |
1091 | iov[2 + elem + offset].iov_len = (4 - idx[elem] % 4) % 4; | |
1092 | idx[elem] += iov[2 + elem + offset].iov_len; | |
1093 | ++offset; | |
1094 | /* FALLTRHOUGH */ | |
1095 | ||
1096 | case _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT1_WC) ... _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT9_WC): | |
498b733e | 1097 | cnt = elem - _NL_ITEM_INDEX (_NL_CTYPE_OUTDIGIT0_WC); |
4b10dd6c UD |
1098 | iov[2 + elem + offset].iov_base = &ctype->wcoutdigits[cnt]; |
1099 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
a9c27b3e | 1100 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
4b10dd6c UD |
1101 | break; |
1102 | ||
a8e4c924 UD |
1103 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING_LEN): |
1104 | default_missing_len = (ctype->default_missing | |
1105 | ? wcslen ((wchar_t *)ctype->default_missing) | |
7f455351 | 1106 | : 0); |
a8e4c924 UD |
1107 | iov[2 + elem + offset].iov_base = &default_missing_len; |
1108 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); | |
1109 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1110 | break; | |
1111 | ||
1d96d74d UD |
1112 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_DEFAULT_MISSING): |
1113 | iov[2 + elem + offset].iov_base = | |
1114 | ctype->default_missing ?: (uint32_t *) L""; | |
1115 | iov[2 + elem + offset].iov_len = | |
1116 | wcslen (iov[2 + elem + offset].iov_base); | |
1117 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1118 | break; | |
1119 | ||
a8e4c924 UD |
1120 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE_LEN): |
1121 | iov[2 + elem + offset].iov_base = &ctype->ntranslit_ignore; | |
1d96d74d | 1122 | iov[2 + elem + offset].iov_len = sizeof (uint32_t); |
a8e4c924 UD |
1123 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
1124 | break; | |
1125 | ||
1126 | case _NL_ITEM_INDEX(_NL_CTYPE_TRANSLIT_IGNORE): | |
1127 | { | |
1128 | uint32_t *ranges = (uint32_t *) alloca (ctype->ntranslit_ignore | |
1129 | * 3 * sizeof (uint32_t)); | |
1130 | struct translit_ignore_t *runp; | |
1131 | ||
1132 | iov[2 + elem + offset].iov_base = ranges; | |
1133 | iov[2 + elem + offset].iov_len = (ctype->ntranslit_ignore | |
1134 | * 3 * sizeof (uint32_t)); | |
1135 | ||
1136 | for (runp = ctype->translit_ignore; runp != NULL; | |
1137 | runp = runp->next) | |
1138 | { | |
1139 | *ranges++ = runp->from; | |
1140 | *ranges++ = runp->to; | |
1141 | *ranges++ = runp->step; | |
1142 | } | |
1143 | } | |
1d96d74d UD |
1144 | /* Remove the following line in case a new entry is added |
1145 | after _NL_CTYPE_TRANSLIT_DEFAULT_MISSING_LEN. */ | |
1146 | if (elem < nelems) | |
1147 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1148 | break; | |
1149 | ||
19bc17a9 RM |
1150 | default: |
1151 | assert (! "unknown CTYPE element"); | |
1152 | } | |
1153 | else | |
1154 | { | |
1155 | /* Handle extra maps. */ | |
ef446144 UD |
1156 | if (oldstyle_tables) |
1157 | { | |
1158 | size_t nr = (elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE)) + 2; | |
19bc17a9 | 1159 | |
ef446144 UD |
1160 | iov[2 + elem + offset].iov_base = ctype->map32[nr]; |
1161 | iov[2 + elem + offset].iov_len = ((ctype->plane_size | |
1162 | * ctype->plane_cnt) | |
1163 | * sizeof (uint32_t)); | |
19bc17a9 | 1164 | |
ef446144 UD |
1165 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; |
1166 | } | |
1167 | else | |
1168 | { | |
1169 | size_t nr = elem - _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE); | |
1170 | if (nr < ctype->nr_charclass) | |
1171 | { | |
1172 | iov[2 + elem + offset] = ctype->class_3level[nr]; | |
1173 | } | |
1174 | else | |
1175 | { | |
1176 | nr -= ctype->nr_charclass; | |
1177 | assert (nr < ctype->map_collection_nr); | |
1178 | iov[2 + elem + offset] = ctype->map_3level[nr]; | |
1179 | } | |
1180 | idx[elem + 1] = idx[elem] + iov[2 + elem + offset].iov_len; | |
1181 | } | |
75cd5204 | 1182 | } |
19bc17a9 | 1183 | } |
19bc17a9 | 1184 | |
75cd5204 | 1185 | assert (2 + elem + offset == (nelems + ctype->nr_charclass |
f175216d | 1186 | + ctype->map_collection_nr + 2 + 2)); |
19bc17a9 | 1187 | |
83b1b6d8 | 1188 | write_locale_data (output_path, "LC_CTYPE", 2 + elem + offset, iov); |
19bc17a9 RM |
1189 | } |
1190 | ||
1191 | ||
4b10dd6c UD |
1192 | /* Local functions. */ |
1193 | static void | |
1194 | ctype_class_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1195 | const char *name) | |
19bc17a9 | 1196 | { |
4b10dd6c | 1197 | size_t cnt; |
19bc17a9 | 1198 | |
4b10dd6c UD |
1199 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
1200 | if (strcmp (ctype->classnames[cnt], name) == 0) | |
1201 | break; | |
19bc17a9 | 1202 | |
4b10dd6c UD |
1203 | if (cnt < ctype->nr_charclass) |
1204 | { | |
1205 | lr_error (lr, _("character class `%s' already defined"), name); | |
1206 | return; | |
1207 | } | |
19bc17a9 | 1208 | |
4b10dd6c UD |
1209 | if (ctype->nr_charclass == MAX_NR_CHARCLASS) |
1210 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1211 | error (2, 0, _("\ | |
5d431a3e | 1212 | implementation limit: no more than %Zd character classes allowed"), |
4b10dd6c | 1213 | MAX_NR_CHARCLASS); |
19bc17a9 | 1214 | |
4b10dd6c | 1215 | ctype->classnames[ctype->nr_charclass++] = name; |
19bc17a9 RM |
1216 | } |
1217 | ||
1218 | ||
4b10dd6c UD |
1219 | static void |
1220 | ctype_map_new (struct linereader *lr, struct locale_ctype_t *ctype, | |
1221 | const char *name, struct charmap_t *charmap) | |
19bc17a9 | 1222 | { |
4b10dd6c | 1223 | size_t max_chars = 0; |
ba1ffaa1 | 1224 | size_t cnt; |
19bc17a9 | 1225 | |
4b10dd6c | 1226 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
19bc17a9 | 1227 | { |
4b10dd6c UD |
1228 | if (strcmp (ctype->mapnames[cnt], name) == 0) |
1229 | break; | |
1230 | ||
1231 | if (max_chars < ctype->map_collection_max[cnt]) | |
1232 | max_chars = ctype->map_collection_max[cnt]; | |
19bc17a9 RM |
1233 | } |
1234 | ||
4b10dd6c UD |
1235 | if (cnt < ctype->map_collection_nr) |
1236 | { | |
1237 | lr_error (lr, _("character map `%s' already defined"), name); | |
1238 | return; | |
1239 | } | |
19bc17a9 | 1240 | |
4b10dd6c UD |
1241 | if (ctype->map_collection_nr == MAX_NR_CHARMAP) |
1242 | /* Exit code 2 is prescribed in P1003.2b. */ | |
1243 | error (2, 0, _("\ | |
1244 | implementation limit: no more than %d character maps allowed"), | |
1245 | MAX_NR_CHARMAP); | |
19bc17a9 | 1246 | |
4b10dd6c UD |
1247 | ctype->mapnames[cnt] = name; |
1248 | ||
1249 | if (max_chars == 0) | |
1250 | ctype->map_collection_max[cnt] = charmap->mb_cur_max == 1 ? 256 : 512; | |
1251 | else | |
1252 | ctype->map_collection_max[cnt] = max_chars; | |
1253 | ||
1254 | ctype->map_collection[cnt] = (uint32_t *) | |
5866b131 | 1255 | xcalloc (sizeof (uint32_t), ctype->map_collection_max[cnt]); |
4b10dd6c | 1256 | ctype->map_collection_act[cnt] = 256; |
19bc17a9 | 1257 | |
4b10dd6c | 1258 | ++ctype->map_collection_nr; |
19bc17a9 RM |
1259 | } |
1260 | ||
1261 | ||
4b10dd6c | 1262 | /* We have to be prepared that TABLE, MAX, and ACT can be NULL. This |
42d7c593 | 1263 | is possible if we only want to extend the name array. */ |
4b10dd6c UD |
1264 | static uint32_t * |
1265 | find_idx (struct locale_ctype_t *ctype, uint32_t **table, size_t *max, | |
1266 | size_t *act, uint32_t idx) | |
19bc17a9 | 1267 | { |
4b10dd6c | 1268 | size_t cnt; |
19bc17a9 | 1269 | |
4b10dd6c UD |
1270 | if (idx < 256) |
1271 | return table == NULL ? NULL : &(*table)[idx]; | |
19bc17a9 | 1272 | |
4b10dd6c UD |
1273 | for (cnt = 256; cnt < ctype->charnames_act; ++cnt) |
1274 | if (ctype->charnames[cnt] == idx) | |
1275 | break; | |
19bc17a9 | 1276 | |
4b10dd6c UD |
1277 | /* We have to distinguish two cases: the name is found or not. */ |
1278 | if (cnt == ctype->charnames_act) | |
1279 | { | |
1280 | /* Extend the name array. */ | |
1281 | if (ctype->charnames_act == ctype->charnames_max) | |
1282 | { | |
1283 | ctype->charnames_max *= 2; | |
5866b131 | 1284 | ctype->charnames = (uint32_t *) |
4b10dd6c | 1285 | xrealloc (ctype->charnames, |
5866b131 | 1286 | sizeof (uint32_t) * ctype->charnames_max); |
4b10dd6c UD |
1287 | } |
1288 | ctype->charnames[ctype->charnames_act++] = idx; | |
1289 | } | |
19bc17a9 | 1290 | |
4b10dd6c UD |
1291 | if (table == NULL) |
1292 | /* We have done everything we are asked to do. */ | |
1293 | return NULL; | |
19bc17a9 | 1294 | |
4b10dd6c UD |
1295 | if (cnt >= *act) |
1296 | { | |
1297 | if (cnt >= *max) | |
1298 | { | |
1299 | size_t old_max = *max; | |
1300 | do | |
1301 | *max *= 2; | |
1302 | while (*max <= cnt); | |
19bc17a9 | 1303 | |
4b10dd6c | 1304 | *table = |
5866b131 | 1305 | (uint32_t *) xrealloc (*table, *max * sizeof (uint32_t)); |
4b10dd6c UD |
1306 | memset (&(*table)[old_max], '\0', |
1307 | (*max - old_max) * sizeof (uint32_t)); | |
1308 | } | |
19bc17a9 | 1309 | |
76e680a8 | 1310 | *act = cnt + 1; |
4b10dd6c | 1311 | } |
19bc17a9 | 1312 | |
4b10dd6c | 1313 | return &(*table)[cnt]; |
19bc17a9 RM |
1314 | } |
1315 | ||
1316 | ||
4b10dd6c UD |
1317 | static int |
1318 | get_character (struct token *now, struct charmap_t *charmap, | |
1319 | struct repertoire_t *repertoire, | |
1320 | struct charseq **seqp, uint32_t *wchp) | |
19bc17a9 | 1321 | { |
4b10dd6c UD |
1322 | if (now->tok == tok_bsymbol) |
1323 | { | |
1324 | /* This will hopefully be the normal case. */ | |
1325 | *wchp = repertoire_find_value (repertoire, now->val.str.startmb, | |
1326 | now->val.str.lenmb); | |
1327 | *seqp = charmap_find_value (charmap, now->val.str.startmb, | |
1328 | now->val.str.lenmb); | |
1329 | } | |
1330 | else if (now->tok == tok_ucs4) | |
1331 | { | |
f0a4b6b1 UD |
1332 | char utmp[10]; |
1333 | ||
1334 | snprintf (utmp, sizeof (utmp), "U%08X", now->val.ucs4); | |
1335 | *seqp = charmap_find_value (charmap, utmp, 9); | |
1336 | ||
1337 | if (*seqp == NULL) | |
1338 | *seqp = repertoire_find_seq (repertoire, now->val.ucs4); | |
19bc17a9 | 1339 | |
4b10dd6c UD |
1340 | if (*seqp == NULL) |
1341 | { | |
1342 | /* Compute the value in the charmap from the UCS value. */ | |
1343 | const char *symbol = repertoire_find_symbol (repertoire, | |
1344 | now->val.ucs4); | |
19bc17a9 | 1345 | |
4b10dd6c UD |
1346 | if (symbol == NULL) |
1347 | *seqp = NULL; | |
1348 | else | |
1349 | *seqp = charmap_find_value (charmap, symbol, strlen (symbol)); | |
19bc17a9 | 1350 | |
4b10dd6c UD |
1351 | if (*seqp == NULL) |
1352 | { | |
723faa38 UD |
1353 | if (repertoire != NULL) |
1354 | { | |
1355 | /* Insert a negative entry. */ | |
1356 | static const struct charseq negative | |
1357 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
1358 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, | |
1359 | sizeof (uint32_t)); | |
1360 | *newp = now->val.ucs4; | |
1361 | ||
1362 | insert_entry (&repertoire->seq_table, newp, | |
1363 | sizeof (uint32_t), (void *) &negative); | |
1364 | } | |
4b10dd6c UD |
1365 | } |
1366 | else | |
1367 | (*seqp)->ucs4 = now->val.ucs4; | |
1368 | } | |
1369 | else if ((*seqp)->ucs4 != now->val.ucs4) | |
1370 | *seqp = NULL; | |
19bc17a9 | 1371 | |
4b10dd6c UD |
1372 | *wchp = now->val.ucs4; |
1373 | } | |
1374 | else if (now->tok == tok_charcode) | |
1375 | { | |
1376 | /* We must map from the byte code to UCS4. */ | |
1377 | *seqp = charmap_find_symbol (charmap, now->val.str.startmb, | |
1378 | now->val.str.lenmb); | |
19bc17a9 | 1379 | |
4b10dd6c UD |
1380 | if (*seqp == NULL) |
1381 | *wchp = ILLEGAL_CHAR_VALUE; | |
1382 | else | |
1383 | { | |
1384 | if ((*seqp)->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1385 | (*seqp)->ucs4 = repertoire_find_value (repertoire, (*seqp)->name, | |
1386 | strlen ((*seqp)->name)); | |
1387 | *wchp = (*seqp)->ucs4; | |
1388 | } | |
1389 | } | |
1390 | else | |
1391 | return 1; | |
19bc17a9 RM |
1392 | |
1393 | return 0; | |
1394 | } | |
1395 | ||
1396 | ||
a0dc5206 UD |
1397 | /* Ellipsis like in `<foo123>..<foo12a>' or `<j1234>....<j1245>' and |
1398 | the .(2). counterparts. */ | |
4b10dd6c UD |
1399 | static void |
1400 | charclass_symbolic_ellipsis (struct linereader *ldfile, | |
1401 | struct locale_ctype_t *ctype, | |
1402 | struct charmap_t *charmap, | |
1403 | struct repertoire_t *repertoire, | |
1404 | struct token *now, | |
1405 | const char *last_str, | |
1406 | unsigned long int class256_bit, | |
1407 | unsigned long int class_bit, int base, | |
a0dc5206 | 1408 | int ignore_content, int handle_digits, int step) |
19bc17a9 | 1409 | { |
4b10dd6c UD |
1410 | const char *nowstr = now->val.str.startmb; |
1411 | char tmp[now->val.str.lenmb + 1]; | |
1412 | const char *cp; | |
1413 | char *endp; | |
1414 | unsigned long int from; | |
1415 | unsigned long int to; | |
19bc17a9 | 1416 | |
4b10dd6c UD |
1417 | /* We have to compute the ellipsis values using the symbolic names. */ |
1418 | assert (last_str != NULL); | |
1419 | ||
1420 | if (strlen (last_str) != now->val.str.lenmb) | |
19bc17a9 | 1421 | { |
4b10dd6c UD |
1422 | invalid_range: |
1423 | lr_error (ldfile, | |
549b3c3a | 1424 | _("`%s' and `%.*s' are no valid names for symbolic range"), |
f6ada7ad | 1425 | last_str, (int) now->val.str.lenmb, nowstr); |
4b10dd6c | 1426 | return; |
19bc17a9 RM |
1427 | } |
1428 | ||
4b10dd6c UD |
1429 | if (memcmp (last_str, nowstr, now->val.str.lenmb) == 0) |
1430 | /* Nothing to do, the names are the same. */ | |
1431 | return; | |
19bc17a9 | 1432 | |
4b10dd6c UD |
1433 | for (cp = last_str; *cp == *(nowstr + (cp - last_str)); ++cp) |
1434 | ; | |
19bc17a9 | 1435 | |
4b10dd6c UD |
1436 | errno = 0; |
1437 | from = strtoul (cp, &endp, base); | |
1438 | if ((from == UINT_MAX && errno == ERANGE) || *endp != '\0') | |
1439 | goto invalid_range; | |
19bc17a9 | 1440 | |
4b10dd6c | 1441 | to = strtoul (nowstr + (cp - last_str), &endp, base); |
549b3c3a UD |
1442 | if ((to == UINT_MAX && errno == ERANGE) |
1443 | || (endp - nowstr) != now->val.str.lenmb || from >= to) | |
4b10dd6c | 1444 | goto invalid_range; |
19bc17a9 | 1445 | |
4b10dd6c UD |
1446 | /* OK, we have a range FROM - TO. Now we can create the symbolic names. */ |
1447 | if (!ignore_content) | |
1448 | { | |
1449 | now->val.str.startmb = tmp; | |
a0dc5206 | 1450 | while ((from += step) <= to) |
4b10dd6c UD |
1451 | { |
1452 | struct charseq *seq; | |
1453 | uint32_t wch; | |
19bc17a9 | 1454 | |
4b10dd6c UD |
1455 | sprintf (tmp, (base == 10 ? "%.*s%0*d" : "%.*s%0*X"), cp - last_str, |
1456 | last_str, now->val.str.lenmb - (cp - last_str), from); | |
19bc17a9 | 1457 | |
4b10dd6c UD |
1458 | get_character (now, charmap, repertoire, &seq, &wch); |
1459 | ||
1460 | if (seq != NULL && seq->nbytes == 1) | |
1461 | /* Yep, we can store information about this byte sequence. */ | |
1462 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
19bc17a9 | 1463 | |
4b10dd6c UD |
1464 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) |
1465 | /* We have the UCS4 position. */ | |
1466 | *find_idx (ctype, &ctype->class_collection, | |
1467 | &ctype->class_collection_max, | |
1468 | &ctype->class_collection_act, wch) |= class_bit; | |
19bc17a9 | 1469 | |
4b10dd6c UD |
1470 | if (handle_digits == 1) |
1471 | { | |
1472 | /* We must store the digit values. */ | |
1473 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1474 | { | |
1475 | ctype->mbdigits_max *= 2; | |
1476 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1477 | (ctype->mbdigits_max | |
1478 | * sizeof (char *))); | |
1479 | ctype->wcdigits_max *= 2; | |
1480 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1481 | (ctype->wcdigits_max | |
1482 | * sizeof (uint32_t))); | |
1483 | } | |
1484 | ||
1485 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1486 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1487 | } | |
1488 | else if (handle_digits == 2) | |
1489 | { | |
1490 | /* We must store the digit values. */ | |
1491 | if (ctype->outdigits_act >= 10) | |
1492 | { | |
1493 | lr_error (ldfile, _("\ | |
1494 | %s: field `%s' does not contain exactly ten entries"), | |
1495 | "LC_CTYPE", "outdigit"); | |
1496 | return; | |
1497 | } | |
1498 | ||
1499 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1500 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1501 | ++ctype->outdigits_act; | |
1502 | } | |
1503 | } | |
1504 | } | |
19bc17a9 RM |
1505 | } |
1506 | ||
1507 | ||
a0dc5206 | 1508 | /* Ellipsis like in `<U1234>..<U2345>' or `<U1234>..(2)..<U2345>'. */ |
4b10dd6c UD |
1509 | static void |
1510 | charclass_ucs4_ellipsis (struct linereader *ldfile, | |
1511 | struct locale_ctype_t *ctype, | |
1512 | struct charmap_t *charmap, | |
1513 | struct repertoire_t *repertoire, | |
1514 | struct token *now, uint32_t last_wch, | |
1515 | unsigned long int class256_bit, | |
1516 | unsigned long int class_bit, int ignore_content, | |
a0dc5206 | 1517 | int handle_digits, int step) |
19bc17a9 | 1518 | { |
4b10dd6c | 1519 | if (last_wch > now->val.ucs4) |
19bc17a9 | 1520 | { |
4b10dd6c UD |
1521 | lr_error (ldfile, _("\ |
1522 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1523 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, now->val.ucs4, | |
1524 | (now->val.ucs4 | last_wch) < 65536 ? 4 : 8, last_wch); | |
19bc17a9 RM |
1525 | return; |
1526 | } | |
1527 | ||
4b10dd6c | 1528 | if (!ignore_content) |
a0dc5206 | 1529 | while ((last_wch += step) <= now->val.ucs4) |
4b10dd6c UD |
1530 | { |
1531 | /* We have to find out whether there is a byte sequence corresponding | |
1532 | to this UCS4 value. */ | |
f0a4b6b1 UD |
1533 | struct charseq *seq; |
1534 | char utmp[10]; | |
1535 | ||
1536 | snprintf (utmp, sizeof (utmp), "U%08X", last_wch); | |
1537 | seq = charmap_find_value (charmap, utmp, 9); | |
a0dc5206 UD |
1538 | if (seq == NULL) |
1539 | { | |
1540 | snprintf (utmp, sizeof (utmp), "U%04X", last_wch); | |
1541 | seq = charmap_find_value (charmap, utmp, 5); | |
1542 | } | |
f0a4b6b1 UD |
1543 | |
1544 | if (seq == NULL) | |
1545 | /* Try looking in the repertoire map. */ | |
1546 | seq = repertoire_find_seq (repertoire, last_wch); | |
19bc17a9 | 1547 | |
4b10dd6c UD |
1548 | /* If this is the first time we look for this sequence create a new |
1549 | entry. */ | |
1550 | if (seq == NULL) | |
1551 | { | |
f0a4b6b1 UD |
1552 | static const struct charseq negative |
1553 | = { .ucs4 = ILLEGAL_CHAR_VALUE }; | |
19bc17a9 | 1554 | |
f0a4b6b1 UD |
1555 | /* Find the symbolic name for this UCS4 value. */ |
1556 | if (repertoire != NULL) | |
4b10dd6c | 1557 | { |
f0a4b6b1 UD |
1558 | const char *symbol = repertoire_find_symbol (repertoire, |
1559 | last_wch); | |
5866b131 UD |
1560 | uint32_t *newp = obstack_alloc (&repertoire->mem_pool, |
1561 | sizeof (uint32_t)); | |
f0a4b6b1 UD |
1562 | *newp = last_wch; |
1563 | ||
1564 | if (symbol != NULL) | |
1565 | /* We have a name, now search the multibyte value. */ | |
1566 | seq = charmap_find_value (charmap, symbol, strlen (symbol)); | |
1567 | ||
1568 | if (seq == NULL) | |
1569 | /* We have to create a fake entry. */ | |
1570 | seq = (struct charseq *) &negative; | |
1571 | else | |
1572 | seq->ucs4 = last_wch; | |
1573 | ||
5866b131 UD |
1574 | insert_entry (&repertoire->seq_table, newp, sizeof (uint32_t), |
1575 | seq); | |
4b10dd6c UD |
1576 | } |
1577 | else | |
f0a4b6b1 UD |
1578 | /* We have to create a fake entry. */ |
1579 | seq = (struct charseq *) &negative; | |
4b10dd6c UD |
1580 | } |
1581 | ||
1582 | /* We have a name, now search the multibyte value. */ | |
1583 | if (seq->ucs4 == last_wch && seq->nbytes == 1) | |
1584 | /* Yep, we can store information about this byte sequence. */ | |
1585 | ctype->class256_collection[(size_t) seq->bytes[0]] | |
1586 | |= class256_bit; | |
1587 | ||
1588 | /* And of course we have the UCS4 position. */ | |
5866b131 | 1589 | if (class_bit != 0) |
4b10dd6c UD |
1590 | *find_idx (ctype, &ctype->class_collection, |
1591 | &ctype->class_collection_max, | |
1592 | &ctype->class_collection_act, last_wch) |= class_bit; | |
1593 | ||
1594 | if (handle_digits == 1) | |
1595 | { | |
1596 | /* We must store the digit values. */ | |
1597 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1598 | { | |
1599 | ctype->mbdigits_max *= 2; | |
1600 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1601 | (ctype->mbdigits_max | |
1602 | * sizeof (char *))); | |
1603 | ctype->wcdigits_max *= 2; | |
1604 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1605 | (ctype->wcdigits_max | |
1606 | * sizeof (uint32_t))); | |
1607 | } | |
1608 | ||
1609 | ctype->mbdigits[ctype->mbdigits_act++] = (seq->ucs4 == last_wch | |
1610 | ? seq : NULL); | |
1611 | ctype->wcdigits[ctype->wcdigits_act++] = last_wch; | |
1612 | } | |
1613 | else if (handle_digits == 2) | |
1614 | { | |
1615 | /* We must store the digit values. */ | |
1616 | if (ctype->outdigits_act >= 10) | |
1617 | { | |
1618 | lr_error (ldfile, _("\ | |
1619 | %s: field `%s' does not contain exactly ten entries"), | |
1620 | "LC_CTYPE", "outdigit"); | |
1621 | return; | |
1622 | } | |
19bc17a9 | 1623 | |
4b10dd6c UD |
1624 | ctype->mboutdigits[ctype->outdigits_act] = (seq->ucs4 == last_wch |
1625 | ? seq : NULL); | |
1626 | ctype->wcoutdigits[ctype->outdigits_act] = last_wch; | |
1627 | ++ctype->outdigits_act; | |
1628 | } | |
1629 | } | |
19bc17a9 RM |
1630 | } |
1631 | ||
1632 | ||
4b10dd6c | 1633 | /* Ellipsis as in `/xea/x12.../xea/x34'. */ |
19bc17a9 | 1634 | static void |
4b10dd6c UD |
1635 | charclass_charcode_ellipsis (struct linereader *ldfile, |
1636 | struct locale_ctype_t *ctype, | |
1637 | struct charmap_t *charmap, | |
1638 | struct repertoire_t *repertoire, | |
1639 | struct token *now, char *last_charcode, | |
1640 | uint32_t last_charcode_len, | |
1641 | unsigned long int class256_bit, | |
1642 | unsigned long int class_bit, int ignore_content, | |
1643 | int handle_digits) | |
19bc17a9 | 1644 | { |
4b10dd6c UD |
1645 | /* First check whether the to-value is larger. */ |
1646 | if (now->val.charcode.nbytes != last_charcode_len) | |
1647 | { | |
1648 | lr_error (ldfile, _("\ | |
1649 | start end end character sequence of range must have the same length")); | |
1650 | return; | |
1651 | } | |
19bc17a9 | 1652 | |
4b10dd6c | 1653 | if (memcmp (last_charcode, now->val.charcode.bytes, last_charcode_len) > 0) |
19bc17a9 | 1654 | { |
4b10dd6c UD |
1655 | lr_error (ldfile, _("\ |
1656 | to-value character sequence is smaller than from-value sequence")); | |
19bc17a9 RM |
1657 | return; |
1658 | } | |
1659 | ||
4b10dd6c UD |
1660 | if (!ignore_content) |
1661 | { | |
1662 | do | |
1663 | { | |
1664 | /* Increment the byte sequence value. */ | |
1665 | struct charseq *seq; | |
1666 | uint32_t wch; | |
1667 | int i; | |
1668 | ||
1669 | for (i = last_charcode_len - 1; i >= 0; --i) | |
1670 | if (++last_charcode[i] != 0) | |
1671 | break; | |
1672 | ||
1673 | if (last_charcode_len == 1) | |
1674 | /* Of course we have the charcode value. */ | |
1675 | ctype->class256_collection[(size_t) last_charcode[0]] | |
1676 | |= class256_bit; | |
1677 | ||
1678 | /* Find the symbolic name. */ | |
1679 | seq = charmap_find_symbol (charmap, last_charcode, | |
1680 | last_charcode_len); | |
1681 | if (seq != NULL) | |
1682 | { | |
1683 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1684 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1685 | strlen (seq->name)); | |
f0a4b6b1 | 1686 | wch = seq == NULL ? ILLEGAL_CHAR_VALUE : seq->ucs4; |
4b10dd6c UD |
1687 | |
1688 | if (wch != ILLEGAL_CHAR_VALUE && class_bit != 0) | |
1689 | *find_idx (ctype, &ctype->class_collection, | |
1690 | &ctype->class_collection_max, | |
1691 | &ctype->class_collection_act, wch) |= class_bit; | |
1692 | } | |
1693 | else | |
1694 | wch = ILLEGAL_CHAR_VALUE; | |
19bc17a9 | 1695 | |
4b10dd6c UD |
1696 | if (handle_digits == 1) |
1697 | { | |
1698 | /* We must store the digit values. */ | |
1699 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
1700 | { | |
1701 | ctype->mbdigits_max *= 2; | |
1702 | ctype->mbdigits = xrealloc (ctype->mbdigits, | |
1703 | (ctype->mbdigits_max | |
1704 | * sizeof (char *))); | |
1705 | ctype->wcdigits_max *= 2; | |
1706 | ctype->wcdigits = xrealloc (ctype->wcdigits, | |
1707 | (ctype->wcdigits_max | |
1708 | * sizeof (uint32_t))); | |
1709 | } | |
1710 | ||
1711 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1712 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1713 | seq->nbytes = last_charcode_len; | |
1714 | ||
1715 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
1716 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
1717 | } | |
1718 | else if (handle_digits == 2) | |
1719 | { | |
1720 | struct charseq *seq; | |
1721 | /* We must store the digit values. */ | |
1722 | if (ctype->outdigits_act >= 10) | |
1723 | { | |
1724 | lr_error (ldfile, _("\ | |
1725 | %s: field `%s' does not contain exactly ten entries"), | |
1726 | "LC_CTYPE", "outdigit"); | |
1727 | return; | |
1728 | } | |
1729 | ||
1730 | seq = xmalloc (sizeof (struct charseq) + last_charcode_len); | |
1731 | memcpy ((char *) (seq + 1), last_charcode, last_charcode_len); | |
1732 | seq->nbytes = last_charcode_len; | |
1733 | ||
1734 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
1735 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
1736 | ++ctype->outdigits_act; | |
1737 | } | |
1738 | } | |
1739 | while (memcmp (last_charcode, now->val.charcode.bytes, | |
1740 | last_charcode_len) != 0); | |
1741 | } | |
19bc17a9 RM |
1742 | } |
1743 | ||
1744 | ||
4b10dd6c UD |
1745 | /* Read one transliteration entry. */ |
1746 | static uint32_t * | |
1747 | read_widestring (struct linereader *ldfile, struct token *now, | |
1748 | struct charmap_t *charmap, struct repertoire_t *repertoire) | |
19bc17a9 | 1749 | { |
4b10dd6c | 1750 | uint32_t *wstr; |
19bc17a9 | 1751 | |
4b10dd6c UD |
1752 | if (now->tok == tok_default_missing) |
1753 | /* The special name "" will denote this case. */ | |
5866b131 | 1754 | wstr = ((uint32_t *) { 0 }); |
4b10dd6c | 1755 | else if (now->tok == tok_bsymbol) |
19bc17a9 | 1756 | { |
4b10dd6c | 1757 | /* Get the value from the repertoire. */ |
a673fbcb | 1758 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1759 | wstr[0] = repertoire_find_value (repertoire, now->val.str.startmb, |
1760 | now->val.str.lenmb); | |
1761 | if (wstr[0] == ILLEGAL_CHAR_VALUE) | |
f0a4b6b1 UD |
1762 | { |
1763 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1764 | free (wstr); | |
1765 | return NULL; | |
1766 | } | |
4b10dd6c UD |
1767 | |
1768 | wstr[1] = 0; | |
19bc17a9 | 1769 | } |
4b10dd6c | 1770 | else if (now->tok == tok_ucs4) |
19bc17a9 | 1771 | { |
a673fbcb | 1772 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1773 | wstr[0] = now->val.ucs4; |
1774 | wstr[1] = 0; | |
1775 | } | |
1776 | else if (now->tok == tok_charcode) | |
1777 | { | |
1778 | /* Argh, we have to convert to the symbol name first and then to the | |
1779 | UCS4 value. */ | |
1780 | struct charseq *seq = charmap_find_symbol (charmap, | |
1781 | now->val.str.startmb, | |
1782 | now->val.str.lenmb); | |
1783 | if (seq == NULL) | |
1784 | /* Cannot find the UCS4 value. */ | |
1785 | return NULL; | |
1786 | ||
1787 | if (seq->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
1788 | seq->ucs4 = repertoire_find_value (repertoire, seq->name, | |
1789 | strlen (seq->name)); | |
1790 | if (seq->ucs4 == ILLEGAL_CHAR_VALUE) | |
1791 | /* We cannot proceed, we don't know the UCS4 value. */ | |
1792 | return NULL; | |
1793 | ||
a673fbcb | 1794 | wstr = (uint32_t *) xmalloc (2 * sizeof (uint32_t)); |
4b10dd6c UD |
1795 | wstr[0] = seq->ucs4; |
1796 | wstr[1] = 0; | |
1797 | } | |
1798 | else if (now->tok == tok_string) | |
1799 | { | |
1800 | wstr = now->val.str.startwc; | |
a673fbcb | 1801 | if (wstr == NULL || wstr[0] == 0) |
4b10dd6c UD |
1802 | return NULL; |
1803 | } | |
1804 | else | |
1805 | { | |
1806 | if (now->tok != tok_eol && now->tok != tok_eof) | |
1807 | lr_ignore_rest (ldfile, 0); | |
1808 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
1809 | return (uint32_t *) -1l; | |
19bc17a9 RM |
1810 | } |
1811 | ||
4b10dd6c UD |
1812 | return wstr; |
1813 | } | |
19bc17a9 | 1814 | |
19bc17a9 | 1815 | |
4b10dd6c UD |
1816 | static void |
1817 | read_translit_entry (struct linereader *ldfile, struct locale_ctype_t *ctype, | |
1818 | struct token *now, struct charmap_t *charmap, | |
1819 | struct repertoire_t *repertoire) | |
1820 | { | |
1821 | uint32_t *from_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1822 | struct translit_t *result; | |
1823 | struct translit_to_t **top; | |
a673fbcb | 1824 | struct obstack *ob = &ctype->mempool; |
4b10dd6c UD |
1825 | int first; |
1826 | int ignore; | |
1827 | ||
1828 | if (from_wstr == NULL) | |
1829 | /* There is no valid from string. */ | |
1830 | return; | |
19bc17a9 | 1831 | |
4b10dd6c UD |
1832 | result = (struct translit_t *) obstack_alloc (ob, |
1833 | sizeof (struct translit_t)); | |
1834 | result->from = from_wstr; | |
a673fbcb UD |
1835 | result->fname = ldfile->fname; |
1836 | result->lineno = ldfile->lineno; | |
4b10dd6c UD |
1837 | result->next = NULL; |
1838 | result->to = NULL; | |
1839 | top = &result->to; | |
1840 | first = 1; | |
1841 | ignore = 0; | |
1842 | ||
1843 | while (1) | |
1844 | { | |
1845 | uint32_t *to_wstr; | |
1846 | ||
1847 | /* Next we have one or more transliterations. They are | |
1848 | separated by semicolons. */ | |
1849 | now = lr_token (ldfile, charmap, repertoire); | |
1850 | ||
1851 | if (!first && (now->tok == tok_semicolon || now->tok == tok_eol)) | |
1852 | { | |
1853 | /* One string read. */ | |
1854 | const uint32_t zero = 0; | |
1855 | ||
1856 | if (!ignore) | |
1857 | { | |
1858 | obstack_grow (ob, &zero, 4); | |
1859 | to_wstr = obstack_finish (ob); | |
1860 | ||
1861 | *top = obstack_alloc (ob, sizeof (struct translit_to_t)); | |
1862 | (*top)->str = to_wstr; | |
1863 | (*top)->next = NULL; | |
1864 | } | |
1865 | ||
1866 | if (now->tok == tok_eol) | |
1867 | { | |
1868 | result->next = ctype->translit; | |
1869 | ctype->translit = result; | |
1870 | return; | |
1871 | } | |
1872 | ||
1873 | if (!ignore) | |
1874 | top = &(*top)->next; | |
1875 | ignore = 0; | |
1876 | } | |
1877 | else | |
1878 | { | |
1879 | to_wstr = read_widestring (ldfile, now, charmap, repertoire); | |
1880 | if (to_wstr == (uint32_t *) -1l) | |
1881 | { | |
1882 | /* An error occurred. */ | |
1883 | obstack_free (ob, result); | |
1884 | return; | |
1885 | } | |
1886 | ||
1887 | if (to_wstr == NULL) | |
1888 | ignore = 1; | |
1889 | else | |
1890 | /* This value is usable. */ | |
1891 | obstack_grow (ob, to_wstr, wcslen ((wchar_t *) to_wstr) * 4); | |
19bc17a9 | 1892 | |
4b10dd6c UD |
1893 | first = 0; |
1894 | } | |
1895 | } | |
19bc17a9 RM |
1896 | } |
1897 | ||
1898 | ||
a673fbcb UD |
1899 | static void |
1900 | read_translit_ignore_entry (struct linereader *ldfile, | |
1901 | struct locale_ctype_t *ctype, | |
1902 | struct charmap_t *charmap, | |
1903 | struct repertoire_t *repertoire) | |
1904 | { | |
1905 | /* We expect a semicolon-separated list of characters we ignore. We are | |
1906 | only interested in the wide character definitions. These must be | |
1907 | single characters, possibly defining a range when an ellipsis is used. */ | |
1908 | while (1) | |
1909 | { | |
1910 | struct token *now = lr_token (ldfile, charmap, repertoire); | |
1911 | struct translit_ignore_t *newp; | |
1912 | uint32_t from; | |
1913 | ||
1914 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1915 | { | |
1916 | lr_error (ldfile, | |
1917 | _("premature end of `translit_ignore' definition")); | |
1918 | return; | |
1919 | } | |
1920 | ||
1921 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
1922 | { | |
1923 | lr_error (ldfile, _("syntax error")); | |
1924 | lr_ignore_rest (ldfile, 0); | |
1925 | return; | |
1926 | } | |
1927 | ||
1928 | if (now->tok == tok_ucs4) | |
1929 | from = now->val.ucs4; | |
1930 | else | |
f0a4b6b1 UD |
1931 | /* Try to get the value. */ |
1932 | from = repertoire_find_value (repertoire, now->val.str.startmb, | |
1933 | now->val.str.lenmb); | |
a673fbcb UD |
1934 | |
1935 | if (from == ILLEGAL_CHAR_VALUE) | |
1936 | { | |
1937 | lr_error (ldfile, "invalid character name"); | |
1938 | newp = NULL; | |
1939 | } | |
1940 | else | |
1941 | { | |
1942 | newp = (struct translit_ignore_t *) | |
1943 | obstack_alloc (&ctype->mempool, sizeof (struct translit_ignore_t)); | |
1944 | newp->from = from; | |
1945 | newp->to = from; | |
a0dc5206 | 1946 | newp->step = 1; |
a673fbcb UD |
1947 | |
1948 | newp->next = ctype->translit_ignore; | |
1949 | ctype->translit_ignore = newp; | |
1950 | } | |
1951 | ||
1952 | /* Now we expect either a semicolon, an ellipsis, or the end of the | |
1953 | line. */ | |
1954 | now = lr_token (ldfile, charmap, repertoire); | |
1955 | ||
a0dc5206 | 1956 | if (now->tok == tok_ellipsis2 || now->tok == tok_ellipsis2_2) |
a673fbcb UD |
1957 | { |
1958 | /* XXX Should we bother implementing `....'? `...' certainly | |
1959 | will not be implemented. */ | |
1960 | uint32_t to; | |
a0dc5206 | 1961 | int step = now->tok == tok_ellipsis2_2 ? 2 : 1; |
a673fbcb UD |
1962 | |
1963 | now = lr_token (ldfile, charmap, repertoire); | |
1964 | ||
1965 | if (now->tok == tok_eol || now->tok == tok_eof) | |
1966 | { | |
1967 | lr_error (ldfile, | |
1968 | _("premature end of `translit_ignore' definition")); | |
1969 | return; | |
1970 | } | |
1971 | ||
1972 | if (now->tok != tok_bsymbol && now->tok != tok_ucs4) | |
1973 | { | |
1974 | lr_error (ldfile, _("syntax error")); | |
1975 | lr_ignore_rest (ldfile, 0); | |
1976 | return; | |
1977 | } | |
1978 | ||
1979 | if (now->tok == tok_ucs4) | |
1980 | to = now->val.ucs4; | |
1981 | else | |
f0a4b6b1 UD |
1982 | /* Try to get the value. */ |
1983 | to = repertoire_find_value (repertoire, now->val.str.startmb, | |
1984 | now->val.str.lenmb); | |
a673fbcb UD |
1985 | |
1986 | if (to == ILLEGAL_CHAR_VALUE) | |
1987 | lr_error (ldfile, "invalid character name"); | |
1988 | else | |
1989 | { | |
1990 | /* Make sure the `to'-value is larger. */ | |
1991 | if (to >= from) | |
a0dc5206 UD |
1992 | { |
1993 | newp->to = to; | |
1994 | newp->step = step; | |
1995 | } | |
a673fbcb UD |
1996 | else |
1997 | lr_error (ldfile, _("\ | |
1998 | to-value <U%0*X> of range is smaller than from-value <U%0*X>"), | |
1999 | (to | from) < 65536 ? 4 : 8, to, | |
2000 | (to | from) < 65536 ? 4 : 8, from); | |
2001 | } | |
2002 | ||
2003 | /* And the next token. */ | |
2004 | now = lr_token (ldfile, charmap, repertoire); | |
2005 | } | |
2006 | ||
2007 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2008 | /* We are done. */ | |
2009 | return; | |
2010 | ||
2011 | if (now->tok == tok_semicolon) | |
2012 | /* Next round. */ | |
2013 | continue; | |
2014 | ||
2015 | /* If we come here something is wrong. */ | |
2016 | lr_error (ldfile, _("syntax error")); | |
2017 | lr_ignore_rest (ldfile, 0); | |
2018 | return; | |
2019 | } | |
2020 | } | |
2021 | ||
2022 | ||
4b10dd6c UD |
2023 | /* The parser for the LC_CTYPE section of the locale definition. */ |
2024 | void | |
2025 | ctype_read (struct linereader *ldfile, struct localedef_t *result, | |
2026 | struct charmap_t *charmap, const char *repertoire_name, | |
2027 | int ignore_content) | |
19bc17a9 | 2028 | { |
4b10dd6c UD |
2029 | struct repertoire_t *repertoire = NULL; |
2030 | struct locale_ctype_t *ctype; | |
2031 | struct token *now; | |
2032 | enum token_t nowtok; | |
19bc17a9 | 2033 | size_t cnt; |
4b10dd6c UD |
2034 | struct charseq *last_seq; |
2035 | uint32_t last_wch = 0; | |
2036 | enum token_t last_token; | |
2037 | enum token_t ellipsis_token; | |
a0dc5206 | 2038 | int step; |
4b10dd6c UD |
2039 | char last_charcode[16]; |
2040 | size_t last_charcode_len = 0; | |
2041 | const char *last_str = NULL; | |
2042 | int mapidx; | |
19bc17a9 | 2043 | |
4b10dd6c UD |
2044 | /* Get the repertoire we have to use. */ |
2045 | if (repertoire_name != NULL) | |
2046 | repertoire = repertoire_read (repertoire_name); | |
19bc17a9 | 2047 | |
4b10dd6c UD |
2048 | /* The rest of the line containing `LC_CTYPE' must be free. */ |
2049 | lr_ignore_rest (ldfile, 1); | |
19bc17a9 | 2050 | |
4b10dd6c UD |
2051 | |
2052 | do | |
19bc17a9 | 2053 | { |
4b10dd6c UD |
2054 | now = lr_token (ldfile, charmap, NULL); |
2055 | nowtok = now->tok; | |
19bc17a9 | 2056 | } |
4b10dd6c | 2057 | while (nowtok == tok_eol); |
19bc17a9 | 2058 | |
4b10dd6c UD |
2059 | /* If we see `copy' now we are almost done. */ |
2060 | if (nowtok == tok_copy) | |
2061 | { | |
01ff9d0b UD |
2062 | handle_copy (ldfile, charmap, repertoire_name, result, tok_lc_ctype, |
2063 | LC_CTYPE, "LC_CTYPE", ignore_content); | |
4b10dd6c UD |
2064 | return; |
2065 | } | |
75cd5204 | 2066 | |
4b10dd6c UD |
2067 | /* Prepare the data structures. */ |
2068 | ctype_startup (ldfile, result, charmap, ignore_content); | |
2069 | ctype = result->categories[LC_CTYPE].ctype; | |
2070 | ||
2071 | /* Remember the repertoire we use. */ | |
2072 | if (!ignore_content) | |
2073 | ctype->repertoire = repertoire; | |
2074 | ||
2075 | while (1) | |
19bc17a9 | 2076 | { |
4b10dd6c UD |
2077 | unsigned long int class_bit = 0; |
2078 | unsigned long int class256_bit = 0; | |
2079 | int handle_digits = 0; | |
2080 | ||
2081 | /* Of course we don't proceed beyond the end of file. */ | |
2082 | if (nowtok == tok_eof) | |
2083 | break; | |
2084 | ||
2085 | /* Ingore empty lines. */ | |
2086 | if (nowtok == tok_eol) | |
19bc17a9 | 2087 | { |
4b10dd6c UD |
2088 | now = lr_token (ldfile, charmap, NULL); |
2089 | nowtok = now->tok; | |
2090 | continue; | |
2091 | } | |
19bc17a9 | 2092 | |
4b10dd6c UD |
2093 | switch (nowtok) |
2094 | { | |
5491da0d UD |
2095 | case tok_charclass: |
2096 | now = lr_token (ldfile, charmap, NULL); | |
2097 | while (now->tok == tok_ident || now->tok == tok_string) | |
2098 | { | |
2099 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
2100 | now = lr_token (ldfile, charmap, NULL); | |
2101 | if (now->tok != tok_semicolon) | |
2102 | break; | |
2103 | now = lr_token (ldfile, charmap, NULL); | |
2104 | } | |
2105 | if (now->tok != tok_eol) | |
2106 | SYNTAX_ERROR (_("\ | |
2107 | %s: syntax error in definition of new character class"), "LC_CTYPE"); | |
2108 | break; | |
2109 | ||
2110 | case tok_charconv: | |
2111 | now = lr_token (ldfile, charmap, NULL); | |
2112 | while (now->tok == tok_ident || now->tok == tok_string) | |
2113 | { | |
2114 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2115 | now = lr_token (ldfile, charmap, NULL); | |
2116 | if (now->tok != tok_semicolon) | |
2117 | break; | |
2118 | now = lr_token (ldfile, charmap, NULL); | |
2119 | } | |
2120 | if (now->tok != tok_eol) | |
2121 | SYNTAX_ERROR (_("\ | |
2122 | %s: syntax error in definition of new character map"), "LC_CTYPE"); | |
2123 | break; | |
2124 | ||
4b10dd6c | 2125 | case tok_class: |
b9eb05d6 UD |
2126 | /* Ignore the rest of the line if we don't need the input of |
2127 | this line. */ | |
2128 | if (ignore_content) | |
2129 | { | |
2130 | lr_ignore_rest (ldfile, 0); | |
2131 | break; | |
2132 | } | |
2133 | ||
4b10dd6c UD |
2134 | /* We simply forget the `class' keyword and use the following |
2135 | operand to determine the bit. */ | |
2136 | now = lr_token (ldfile, charmap, NULL); | |
2137 | if (now->tok == tok_ident || now->tok == tok_string) | |
2138 | { | |
87372aa9 | 2139 | /* Must can be one of the predefined class names. */ |
4b10dd6c UD |
2140 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) |
2141 | if (strcmp (ctype->classnames[cnt], now->val.str.startmb) == 0) | |
2142 | break; | |
2143 | if (cnt >= ctype->nr_charclass) | |
2144 | { | |
011ebfab | 2145 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2146 | if (now->val.str.lenmb == 8 |
2147 | && memcmp ("special1", now->val.str.startmb, 8) == 0) | |
2148 | class_bit = _ISwspecial1; | |
2149 | else if (now->val.str.lenmb == 8 | |
2150 | && memcmp ("special2", now->val.str.startmb, 8) == 0) | |
2151 | class_bit = _ISwspecial2; | |
2152 | else if (now->val.str.lenmb == 8 | |
2153 | && memcmp ("special3", now->val.str.startmb, 8) == 0) | |
2154 | class_bit = _ISwspecial3; | |
2155 | else | |
011ebfab | 2156 | #endif |
4b10dd6c | 2157 | { |
87372aa9 UD |
2158 | /* OK, it's a new class. */ |
2159 | ctype_class_new (ldfile, ctype, now->val.str.startmb); | |
4b10dd6c | 2160 | |
87372aa9 | 2161 | class_bit = _ISwbit (ctype->nr_charclass - 1); |
4b10dd6c UD |
2162 | } |
2163 | } | |
2164 | else | |
7f653277 UD |
2165 | { |
2166 | class_bit = _ISwbit (cnt); | |
4b10dd6c | 2167 | |
7f653277 UD |
2168 | free (now->val.str.startmb); |
2169 | } | |
4b10dd6c UD |
2170 | } |
2171 | else if (now->tok == tok_digit) | |
2172 | goto handle_tok_digit; | |
2173 | else if (now->tok < tok_upper || now->tok > tok_blank) | |
2174 | goto err_label; | |
2175 | else | |
2176 | { | |
2177 | class_bit = BITw (now->tok); | |
2178 | class256_bit = BIT (now->tok); | |
2179 | } | |
2180 | ||
2181 | /* The next character must be a semicolon. */ | |
2182 | now = lr_token (ldfile, charmap, NULL); | |
2183 | if (now->tok != tok_semicolon) | |
2184 | goto err_label; | |
2185 | goto read_charclass; | |
2186 | ||
2187 | case tok_upper: | |
2188 | case tok_lower: | |
2189 | case tok_alpha: | |
2190 | case tok_alnum: | |
2191 | case tok_space: | |
2192 | case tok_cntrl: | |
2193 | case tok_punct: | |
2194 | case tok_graph: | |
2195 | case tok_print: | |
2196 | case tok_xdigit: | |
2197 | case tok_blank: | |
b9eb05d6 UD |
2198 | /* Ignore the rest of the line if we don't need the input of |
2199 | this line. */ | |
2200 | if (ignore_content) | |
2201 | { | |
2202 | lr_ignore_rest (ldfile, 0); | |
2203 | break; | |
2204 | } | |
2205 | ||
4b10dd6c UD |
2206 | class_bit = BITw (now->tok); |
2207 | class256_bit = BIT (now->tok); | |
2208 | handle_digits = 0; | |
2209 | read_charclass: | |
2210 | ctype->class_done |= class_bit; | |
2211 | last_token = tok_none; | |
2212 | ellipsis_token = tok_none; | |
a0dc5206 | 2213 | step = 1; |
4b10dd6c UD |
2214 | now = lr_token (ldfile, charmap, NULL); |
2215 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2216 | { | |
2217 | uint32_t wch; | |
2218 | struct charseq *seq; | |
2219 | ||
2220 | if (ellipsis_token == tok_none) | |
2221 | { | |
2222 | if (get_character (now, charmap, repertoire, &seq, &wch)) | |
2223 | goto err_label; | |
2224 | ||
2225 | if (!ignore_content && seq != NULL && seq->nbytes == 1) | |
2226 | /* Yep, we can store information about this byte | |
2227 | sequence. */ | |
2228 | ctype->class256_collection[seq->bytes[0]] |= class256_bit; | |
2229 | ||
2230 | if (!ignore_content && wch != ILLEGAL_CHAR_VALUE | |
2231 | && class_bit != 0) | |
2232 | /* We have the UCS4 position. */ | |
2233 | *find_idx (ctype, &ctype->class_collection, | |
2234 | &ctype->class_collection_max, | |
2235 | &ctype->class_collection_act, wch) |= class_bit; | |
2236 | ||
2237 | last_token = now->tok; | |
549b3c3a | 2238 | /* Terminate the string. */ |
9e2b7438 UD |
2239 | if (last_token == tok_bsymbol) |
2240 | { | |
2241 | now->val.str.startmb[now->val.str.lenmb] = '\0'; | |
2242 | last_str = now->val.str.startmb; | |
2243 | } | |
2244 | else | |
2245 | last_str = NULL; | |
4b10dd6c UD |
2246 | last_seq = seq; |
2247 | last_wch = wch; | |
2248 | memcpy (last_charcode, now->val.charcode.bytes, 16); | |
2249 | last_charcode_len = now->val.charcode.nbytes; | |
2250 | ||
2251 | if (!ignore_content && handle_digits == 1) | |
2252 | { | |
2253 | /* We must store the digit values. */ | |
2254 | if (ctype->mbdigits_act == ctype->mbdigits_max) | |
2255 | { | |
b9eb05d6 | 2256 | ctype->mbdigits_max += 10; |
4b10dd6c UD |
2257 | ctype->mbdigits = xrealloc (ctype->mbdigits, |
2258 | (ctype->mbdigits_max | |
2259 | * sizeof (char *))); | |
b9eb05d6 | 2260 | ctype->wcdigits_max += 10; |
4b10dd6c UD |
2261 | ctype->wcdigits = xrealloc (ctype->wcdigits, |
2262 | (ctype->wcdigits_max | |
2263 | * sizeof (uint32_t))); | |
2264 | } | |
2265 | ||
2266 | ctype->mbdigits[ctype->mbdigits_act++] = seq; | |
2267 | ctype->wcdigits[ctype->wcdigits_act++] = wch; | |
2268 | } | |
2269 | else if (!ignore_content && handle_digits == 2) | |
2270 | { | |
2271 | /* We must store the digit values. */ | |
2272 | if (ctype->outdigits_act >= 10) | |
2273 | { | |
2274 | lr_error (ldfile, _("\ | |
2275 | %s: field `%s' does not contain exactly ten entries"), | |
2276 | "LC_CTYPE", "outdigit"); | |
2277 | goto err_label; | |
2278 | } | |
2279 | ||
2280 | ctype->mboutdigits[ctype->outdigits_act] = seq; | |
2281 | ctype->wcoutdigits[ctype->outdigits_act] = wch; | |
2282 | ++ctype->outdigits_act; | |
2283 | } | |
2284 | } | |
2285 | else | |
2286 | { | |
2287 | /* Now it gets complicated. We have to resolve the | |
2288 | ellipsis problem. First we must distinguish between | |
2289 | the different kind of ellipsis and this must match the | |
2290 | tokens we have seen. */ | |
2291 | assert (last_token != tok_none); | |
2292 | ||
2293 | if (last_token != now->tok) | |
2294 | { | |
2295 | lr_error (ldfile, _("\ | |
2296 | ellipsis range must be marked by two operands of same type")); | |
2297 | lr_ignore_rest (ldfile, 0); | |
2298 | break; | |
2299 | } | |
2300 | ||
2301 | if (last_token == tok_bsymbol) | |
2302 | { | |
2303 | if (ellipsis_token == tok_ellipsis3) | |
2304 | lr_error (ldfile, _("with symbolic name range values \ | |
2305 | the absolute ellipsis `...' must not be used")); | |
2306 | ||
2307 | charclass_symbolic_ellipsis (ldfile, ctype, charmap, | |
2308 | repertoire, now, last_str, | |
2309 | class256_bit, class_bit, | |
2310 | (ellipsis_token | |
2311 | == tok_ellipsis4 | |
2312 | ? 10 : 16), | |
2313 | ignore_content, | |
a0dc5206 | 2314 | handle_digits, step); |
4b10dd6c UD |
2315 | } |
2316 | else if (last_token == tok_ucs4) | |
2317 | { | |
2318 | if (ellipsis_token != tok_ellipsis2) | |
2319 | lr_error (ldfile, _("\ | |
2320 | with UCS range values one must use the hexadecimal symbolic ellipsis `..'")); | |
2321 | ||
2322 | charclass_ucs4_ellipsis (ldfile, ctype, charmap, | |
2323 | repertoire, now, last_wch, | |
2324 | class256_bit, class_bit, | |
a0dc5206 UD |
2325 | ignore_content, handle_digits, |
2326 | step); | |
4b10dd6c UD |
2327 | } |
2328 | else | |
2329 | { | |
2330 | assert (last_token == tok_charcode); | |
2331 | ||
2332 | if (ellipsis_token != tok_ellipsis3) | |
2333 | lr_error (ldfile, _("\ | |
2334 | with character code range values one must use the absolute ellipsis `...'")); | |
2335 | ||
2336 | charclass_charcode_ellipsis (ldfile, ctype, charmap, | |
2337 | repertoire, now, | |
2338 | last_charcode, | |
2339 | last_charcode_len, | |
2340 | class256_bit, class_bit, | |
2341 | ignore_content, | |
2342 | handle_digits); | |
2343 | } | |
2344 | ||
2345 | /* Now we have used the last value. */ | |
2346 | last_token = tok_none; | |
2347 | } | |
2348 | ||
2349 | /* Next we expect a semicolon or the end of the line. */ | |
2350 | now = lr_token (ldfile, charmap, NULL); | |
2351 | if (now->tok == tok_eol || now->tok == tok_eof) | |
2352 | break; | |
2353 | ||
2354 | if (last_token != tok_none | |
a0dc5206 | 2355 | && now->tok >= tok_ellipsis2 && now->tok <= tok_ellipsis4_2) |
4b10dd6c | 2356 | { |
a0dc5206 UD |
2357 | if (now->tok == tok_ellipsis2_2) |
2358 | { | |
2359 | now->tok = tok_ellipsis2; | |
2360 | step = 2; | |
2361 | } | |
2362 | else if (now->tok == tok_ellipsis4_2) | |
2363 | { | |
2364 | now->tok = tok_ellipsis4; | |
2365 | step = 2; | |
2366 | } | |
2367 | ||
4b10dd6c | 2368 | ellipsis_token = now->tok; |
a0dc5206 | 2369 | |
4b10dd6c UD |
2370 | now = lr_token (ldfile, charmap, NULL); |
2371 | continue; | |
2372 | } | |
2373 | ||
2374 | if (now->tok != tok_semicolon) | |
2375 | goto err_label; | |
2376 | ||
2377 | /* And get the next character. */ | |
2378 | now = lr_token (ldfile, charmap, NULL); | |
2379 | ||
2380 | ellipsis_token = tok_none; | |
a0dc5206 | 2381 | step = 1; |
4b10dd6c UD |
2382 | } |
2383 | break; | |
2384 | ||
2385 | case tok_digit: | |
b9eb05d6 UD |
2386 | /* Ignore the rest of the line if we don't need the input of |
2387 | this line. */ | |
2388 | if (ignore_content) | |
42d7c593 UD |
2389 | { |
2390 | lr_ignore_rest (ldfile, 0); | |
2391 | break; | |
2392 | } | |
b9eb05d6 | 2393 | |
4b10dd6c UD |
2394 | handle_tok_digit: |
2395 | class_bit = _ISwdigit; | |
2396 | class256_bit = _ISdigit; | |
2397 | handle_digits = 1; | |
2398 | goto read_charclass; | |
2399 | ||
2400 | case tok_outdigit: | |
b9eb05d6 UD |
2401 | /* Ignore the rest of the line if we don't need the input of |
2402 | this line. */ | |
2403 | if (ignore_content) | |
2404 | { | |
2405 | lr_ignore_rest (ldfile, 0); | |
2406 | break; | |
2407 | } | |
2408 | ||
4b10dd6c UD |
2409 | if (ctype->outdigits_act != 0) |
2410 | lr_error (ldfile, _("\ | |
2411 | %s: field `%s' declared more than once"), | |
2412 | "LC_CTYPE", "outdigit"); | |
2413 | class_bit = 0; | |
2414 | class256_bit = 0; | |
2415 | handle_digits = 2; | |
2416 | goto read_charclass; | |
2417 | ||
2418 | case tok_toupper: | |
b9eb05d6 UD |
2419 | /* Ignore the rest of the line if we don't need the input of |
2420 | this line. */ | |
2421 | if (ignore_content) | |
2422 | { | |
2423 | lr_ignore_rest (ldfile, 0); | |
2424 | break; | |
2425 | } | |
2426 | ||
4b10dd6c UD |
2427 | mapidx = 0; |
2428 | goto read_mapping; | |
2429 | ||
2430 | case tok_tolower: | |
b9eb05d6 UD |
2431 | /* Ignore the rest of the line if we don't need the input of |
2432 | this line. */ | |
2433 | if (ignore_content) | |
2434 | { | |
2435 | lr_ignore_rest (ldfile, 0); | |
2436 | break; | |
2437 | } | |
2438 | ||
4b10dd6c UD |
2439 | mapidx = 1; |
2440 | goto read_mapping; | |
2441 | ||
2442 | case tok_map: | |
b9eb05d6 UD |
2443 | /* Ignore the rest of the line if we don't need the input of |
2444 | this line. */ | |
2445 | if (ignore_content) | |
2446 | { | |
2447 | lr_ignore_rest (ldfile, 0); | |
2448 | break; | |
2449 | } | |
2450 | ||
4b10dd6c UD |
2451 | /* We simply forget the `map' keyword and use the following |
2452 | operand to determine the mapping. */ | |
2453 | now = lr_token (ldfile, charmap, NULL); | |
2454 | if (now->tok == tok_ident || now->tok == tok_string) | |
2455 | { | |
2456 | size_t cnt; | |
2457 | ||
2458 | for (cnt = 2; cnt < ctype->map_collection_nr; ++cnt) | |
2459 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2460 | break; | |
2461 | ||
7f653277 UD |
2462 | if (cnt < ctype->map_collection_nr) |
2463 | free (now->val.str.startmb); | |
2464 | else | |
87372aa9 UD |
2465 | /* OK, it's a new map. */ |
2466 | ctype_map_new (ldfile, ctype, now->val.str.startmb, charmap); | |
2467 | ||
2468 | mapidx = cnt; | |
4b10dd6c UD |
2469 | } |
2470 | else if (now->tok < tok_toupper || now->tok > tok_tolower) | |
2471 | goto err_label; | |
2472 | else | |
2473 | mapidx = now->tok - tok_toupper; | |
2474 | ||
2475 | now = lr_token (ldfile, charmap, NULL); | |
2476 | /* This better should be a semicolon. */ | |
2477 | if (now->tok != tok_semicolon) | |
2478 | goto err_label; | |
2479 | ||
2480 | read_mapping: | |
2481 | /* Test whether this mapping was already defined. */ | |
2482 | if (ctype->tomap_done[mapidx]) | |
2483 | { | |
2484 | lr_error (ldfile, _("duplicated definition for mapping `%s'"), | |
2485 | ctype->mapnames[mapidx]); | |
2486 | lr_ignore_rest (ldfile, 0); | |
2487 | break; | |
2488 | } | |
2489 | ctype->tomap_done[mapidx] = 1; | |
2490 | ||
2491 | now = lr_token (ldfile, charmap, NULL); | |
2492 | while (now->tok != tok_eol && now->tok != tok_eof) | |
2493 | { | |
2494 | struct charseq *from_seq; | |
2495 | uint32_t from_wch; | |
2496 | struct charseq *to_seq; | |
2497 | uint32_t to_wch; | |
2498 | ||
2499 | /* Every pair starts with an opening brace. */ | |
2500 | if (now->tok != tok_open_brace) | |
2501 | goto err_label; | |
2502 | ||
2503 | /* Next comes the from-value. */ | |
2504 | now = lr_token (ldfile, charmap, NULL); | |
2505 | if (get_character (now, charmap, repertoire, &from_seq, | |
2506 | &from_wch) != 0) | |
2507 | goto err_label; | |
2508 | ||
2509 | /* The next is a comma. */ | |
2510 | now = lr_token (ldfile, charmap, NULL); | |
2511 | if (now->tok != tok_comma) | |
2512 | goto err_label; | |
2513 | ||
2514 | /* And the other value. */ | |
2515 | now = lr_token (ldfile, charmap, NULL); | |
2516 | if (get_character (now, charmap, repertoire, &to_seq, | |
2517 | &to_wch) != 0) | |
2518 | goto err_label; | |
2519 | ||
2520 | /* And the last thing is the closing brace. */ | |
2521 | now = lr_token (ldfile, charmap, NULL); | |
2522 | if (now->tok != tok_close_brace) | |
2523 | goto err_label; | |
2524 | ||
2525 | if (!ignore_content) | |
2526 | { | |
2527 | if (mapidx < 2 && from_seq != NULL && to_seq != NULL | |
2528 | && from_seq->nbytes == 1 && to_seq->nbytes == 1) | |
2529 | /* We can use this value. */ | |
2530 | ctype->map256_collection[mapidx][from_seq->bytes[0]] | |
2531 | = to_seq->bytes[0]; | |
2532 | ||
2533 | if (from_wch != ILLEGAL_CHAR_VALUE | |
2534 | && to_wch != ILLEGAL_CHAR_VALUE) | |
2535 | /* Both correct values. */ | |
2536 | *find_idx (ctype, &ctype->map_collection[mapidx], | |
2537 | &ctype->map_collection_max[mapidx], | |
2538 | &ctype->map_collection_act[mapidx], | |
2539 | from_wch) = to_wch; | |
2540 | } | |
2541 | ||
2542 | /* Now comes a semicolon or the end of the line/file. */ | |
2543 | now = lr_token (ldfile, charmap, NULL); | |
2544 | if (now->tok == tok_semicolon) | |
2545 | now = lr_token (ldfile, charmap, NULL); | |
2546 | } | |
2547 | break; | |
2548 | ||
2549 | case tok_translit_start: | |
b9eb05d6 UD |
2550 | /* Ignore the rest of the line if we don't need the input of |
2551 | this line. */ | |
2552 | if (ignore_content) | |
2553 | { | |
2554 | lr_ignore_rest (ldfile, 0); | |
2555 | break; | |
2556 | } | |
2557 | ||
4b10dd6c UD |
2558 | /* The rest of the line better should be empty. */ |
2559 | lr_ignore_rest (ldfile, 1); | |
2560 | ||
2561 | /* We count here the number of allocated entries in the `translit' | |
2562 | array. */ | |
2563 | cnt = 0; | |
2564 | ||
2565 | /* We proceed until we see the `translit_end' token. */ | |
2566 | while (now = lr_token (ldfile, charmap, repertoire), | |
2567 | now->tok != tok_translit_end && now->tok != tok_eof) | |
2568 | { | |
2569 | if (now->tok == tok_eol) | |
2570 | /* Ignore empty lines. */ | |
2571 | continue; | |
2572 | ||
2573 | if (now->tok == tok_translit_end) | |
2574 | { | |
2575 | lr_ignore_rest (ldfile, 0); | |
2576 | break; | |
2577 | } | |
2578 | ||
2579 | if (now->tok == tok_include) | |
2580 | { | |
2581 | /* We have to include locale. */ | |
2582 | const char *locale_name; | |
2583 | const char *repertoire_name; | |
2584 | ||
2585 | now = lr_token (ldfile, charmap, NULL); | |
2586 | /* This should be a string or an identifier. In any | |
2587 | case something to name a locale. */ | |
2588 | if (now->tok != tok_string && now->tok != tok_ident) | |
2589 | { | |
2590 | translit_syntax: | |
2591 | lr_error (ldfile, _("%s: syntax error"), "LC_CTYPE"); | |
2592 | lr_ignore_rest (ldfile, 0); | |
2593 | continue; | |
2594 | } | |
2595 | locale_name = now->val.str.startmb; | |
2596 | ||
2597 | /* Next should be a semicolon. */ | |
2598 | now = lr_token (ldfile, charmap, NULL); | |
2599 | if (now->tok != tok_semicolon) | |
2600 | goto translit_syntax; | |
2601 | ||
2602 | /* Now the repertoire name. */ | |
2603 | now = lr_token (ldfile, charmap, NULL); | |
2604 | if ((now->tok != tok_string && now->tok != tok_ident) | |
2605 | || now->val.str.startmb == NULL) | |
2606 | goto translit_syntax; | |
2607 | repertoire_name = now->val.str.startmb; | |
2608 | ||
2609 | /* We must not have more than one `include'. */ | |
2610 | if (ctype->translit_copy_locale != NULL) | |
2611 | { | |
2612 | lr_error (ldfile, _("\ | |
2613 | %s: only one `include' instruction allowed"), "LC_CTYPE"); | |
2614 | lr_ignore_rest (ldfile, 0); | |
2615 | continue; | |
2616 | } | |
2617 | ||
2618 | ctype->translit_copy_locale = locale_name; | |
2619 | ctype->translit_copy_repertoire = repertoire_name; | |
2620 | ||
2621 | /* The rest of the line must be empty. */ | |
2622 | lr_ignore_rest (ldfile, 1); | |
a673fbcb UD |
2623 | |
2624 | /* Make sure the locale is read. */ | |
2625 | add_to_readlist (LC_CTYPE, ctype->translit_copy_locale, | |
07dab0c3 | 2626 | repertoire_name, 1, NULL); |
a673fbcb UD |
2627 | continue; |
2628 | } | |
2629 | else if (now->tok == tok_default_missing) | |
2630 | { | |
2631 | uint32_t *wstr; | |
2632 | ||
2633 | /* We expect a single character or string as the | |
2634 | argument. */ | |
2635 | now = lr_token (ldfile, charmap, NULL); | |
2636 | wstr = read_widestring (ldfile, now, charmap, repertoire); | |
2637 | ||
2638 | if (wstr != NULL) | |
2639 | { | |
2640 | if (ctype->default_missing != NULL) | |
2641 | { | |
2642 | lr_error (ldfile, _("\ | |
2643 | %s: duplicate `default_missing' definition"), "LC_CTYPE"); | |
2644 | error_at_line (0, 0, ctype->default_missing_file, | |
2645 | ctype->default_missing_lineno, | |
2646 | _("previous definition was here")); | |
2647 | } | |
2648 | else | |
2649 | { | |
2650 | ctype->default_missing = wstr; | |
2651 | ctype->default_missing_file = ldfile->fname; | |
2652 | ctype->default_missing_lineno = ldfile->lineno; | |
2653 | } | |
2654 | } | |
2655 | lr_ignore_rest (ldfile, 1); | |
2656 | continue; | |
2657 | } | |
2658 | else if (now->tok == tok_translit_ignore) | |
2659 | { | |
2660 | read_translit_ignore_entry (ldfile, ctype, charmap, | |
2661 | repertoire); | |
4b10dd6c UD |
2662 | continue; |
2663 | } | |
2664 | ||
2665 | read_translit_entry (ldfile, ctype, now, charmap, repertoire); | |
2666 | } | |
2667 | break; | |
2668 | ||
2669 | case tok_ident: | |
b9eb05d6 UD |
2670 | /* Ignore the rest of the line if we don't need the input of |
2671 | this line. */ | |
2672 | if (ignore_content) | |
2673 | { | |
2674 | lr_ignore_rest (ldfile, 0); | |
2675 | break; | |
2676 | } | |
2677 | ||
4b10dd6c UD |
2678 | /* This could mean one of several things. First test whether |
2679 | it's a character class name. */ | |
2680 | for (cnt = 0; cnt < ctype->nr_charclass; ++cnt) | |
2681 | if (strcmp (now->val.str.startmb, ctype->classnames[cnt]) == 0) | |
2682 | break; | |
2683 | if (cnt < ctype->nr_charclass) | |
2684 | { | |
2685 | class_bit = _ISwbit (cnt); | |
2686 | class256_bit = cnt <= 11 ? _ISbit (cnt) : 0; | |
2687 | free (now->val.str.startmb); | |
2688 | goto read_charclass; | |
2689 | } | |
5491da0d UD |
2690 | for (cnt = 0; cnt < ctype->map_collection_nr; ++cnt) |
2691 | if (strcmp (now->val.str.startmb, ctype->mapnames[cnt]) == 0) | |
2692 | break; | |
2693 | if (cnt < ctype->map_collection_nr) | |
2694 | { | |
2695 | mapidx = cnt; | |
2696 | free (now->val.str.startmb); | |
2697 | goto read_mapping; | |
2698 | } | |
011ebfab | 2699 | #ifdef PREDEFINED_CLASSES |
4b10dd6c UD |
2700 | if (strcmp (now->val.str.startmb, "special1") == 0) |
2701 | { | |
2702 | class_bit = _ISwspecial1; | |
2703 | free (now->val.str.startmb); | |
2704 | goto read_charclass; | |
2705 | } | |
2706 | if (strcmp (now->val.str.startmb, "special2") == 0) | |
2707 | { | |
2708 | class_bit = _ISwspecial2; | |
2709 | free (now->val.str.startmb); | |
2710 | goto read_charclass; | |
2711 | } | |
2712 | if (strcmp (now->val.str.startmb, "special3") == 0) | |
2713 | { | |
2714 | class_bit = _ISwspecial3; | |
2715 | free (now->val.str.startmb); | |
2716 | goto read_charclass; | |
2717 | } | |
2718 | if (strcmp (now->val.str.startmb, "tosymmetric") == 0) | |
2719 | { | |
2720 | mapidx = 2; | |
2721 | goto read_mapping; | |
2722 | } | |
011ebfab | 2723 | #endif |
4b10dd6c UD |
2724 | break; |
2725 | ||
2726 | case tok_end: | |
2727 | /* Next we assume `LC_CTYPE'. */ | |
2728 | now = lr_token (ldfile, charmap, NULL); | |
2729 | if (now->tok == tok_eof) | |
2730 | break; | |
2731 | if (now->tok == tok_eol) | |
2732 | lr_error (ldfile, _("%s: incomplete `END' line"), | |
2733 | "LC_CTYPE"); | |
2734 | else if (now->tok != tok_lc_ctype) | |
2735 | lr_error (ldfile, _("\ | |
2736 | %1$s: definition does not end with `END %1$s'"), "LC_CTYPE"); | |
2737 | lr_ignore_rest (ldfile, now->tok == tok_lc_ctype); | |
2738 | return; | |
2739 | ||
2740 | default: | |
2741 | err_label: | |
2742 | if (now->tok != tok_eof) | |
2743 | SYNTAX_ERROR (_("%s: syntax error"), "LC_CTYPE"); | |
19bc17a9 RM |
2744 | } |
2745 | ||
4b10dd6c UD |
2746 | /* Prepare for the next round. */ |
2747 | now = lr_token (ldfile, charmap, NULL); | |
2748 | nowtok = now->tok; | |
19bc17a9 RM |
2749 | } |
2750 | ||
4b10dd6c UD |
2751 | /* When we come here we reached the end of the file. */ |
2752 | lr_error (ldfile, _("%s: premature end of file"), "LC_CTYPE"); | |
19bc17a9 RM |
2753 | } |
2754 | ||
2755 | ||
2756 | static void | |
4b10dd6c UD |
2757 | set_class_defaults (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
2758 | struct repertoire_t *repertoire) | |
19bc17a9 | 2759 | { |
4b10dd6c UD |
2760 | size_t cnt; |
2761 | ||
19bc17a9 RM |
2762 | /* These function defines the default values for the classes and conversions |
2763 | according to POSIX.2 2.5.2.1. | |
2764 | It may seem that the order of these if-blocks is arbitrary but it is NOT. | |
2765 | Don't move them unless you know what you do! */ | |
2766 | ||
4b10dd6c | 2767 | void set_default (int bitpos, int from, int to) |
19bc17a9 RM |
2768 | { |
2769 | char tmp[2]; | |
2770 | int ch; | |
4b10dd6c UD |
2771 | int bit = _ISbit (bitpos); |
2772 | int bitw = _ISwbit (bitpos); | |
19bc17a9 RM |
2773 | /* Define string. */ |
2774 | strcpy (tmp, "?"); | |
2775 | ||
2776 | for (ch = from; ch <= to; ++ch) | |
2777 | { | |
4b10dd6c | 2778 | struct charseq *seq; |
19bc17a9 RM |
2779 | tmp[0] = ch; |
2780 | ||
4b10dd6c UD |
2781 | seq = charmap_find_value (charmap, tmp, 1); |
2782 | if (seq == NULL) | |
2783 | { | |
2784 | if (!be_quiet) | |
2785 | error (0, 0, _("\ | |
2786 | %s: character `%s' not defined in charmap while needed as default value"), | |
2787 | "LC_CTYPE", tmp); | |
19bc17a9 | 2788 | } |
4b10dd6c UD |
2789 | else if (seq->nbytes != 1) |
2790 | error (0, 0, _("\ | |
2791 | %s: character `%s' in charmap not representable with one byte"), | |
2792 | "LC_CTYPE", tmp); | |
19bc17a9 | 2793 | else |
4b10dd6c | 2794 | ctype->class256_collection[seq->bytes[0]] |= bit; |
f0a4b6b1 UD |
2795 | |
2796 | /* No need to search here, the ASCII value is also the Unicode | |
2797 | value. */ | |
2798 | ELEM (ctype, class_collection, , ch) |= bitw; | |
19bc17a9 RM |
2799 | } |
2800 | } | |
2801 | ||
2802 | /* Set default values if keyword was not present. */ | |
4b10dd6c | 2803 | if ((ctype->class_done & BITw (tok_upper)) == 0) |
19bc17a9 RM |
2804 | /* "If this keyword [lower] is not specified, the lowercase letters |
2805 | `A' through `Z', ..., shall automatically belong to this class, | |
2806 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2807 | set_default (BITPOS (tok_upper), 'A', 'Z'); |
19bc17a9 | 2808 | |
4b10dd6c | 2809 | if ((ctype->class_done & BITw (tok_lower)) == 0) |
19bc17a9 RM |
2810 | /* "If this keyword [lower] is not specified, the lowercase letters |
2811 | `a' through `z', ..., shall automatically belong to this class, | |
2812 | with implementation defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2813 | set_default (BITPOS (tok_lower), 'a', 'z'); |
19bc17a9 | 2814 | |
4b10dd6c | 2815 | if ((ctype->class_done & BITw (tok_alpha)) == 0) |
19bc17a9 RM |
2816 | { |
2817 | /* Table 2-6 in P1003.2 says that characters in class `upper' or | |
2818 | class `lower' *must* be in class `alpha'. */ | |
2819 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower); | |
96f0d1f5 UD |
2820 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower); |
2821 | ||
2822 | for (cnt = 0; cnt < 256; ++cnt) | |
2823 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2824 | ctype->class256_collection[cnt] |= BIT (tok_alpha); | |
19bc17a9 RM |
2825 | |
2826 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2827 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2828 | ctype->class_collection[cnt] |= BITw (tok_alpha); | |
19bc17a9 RM |
2829 | } |
2830 | ||
4b10dd6c | 2831 | if ((ctype->class_done & BITw (tok_digit)) == 0) |
19bc17a9 RM |
2832 | /* "If this keyword [digit] is not specified, the digits `0' through |
2833 | `9', ..., shall automatically belong to this class, with | |
2834 | implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
4b10dd6c | 2835 | set_default (BITPOS (tok_digit), '0', '9'); |
19bc17a9 RM |
2836 | |
2837 | /* "Only characters specified for the `alpha' and `digit' keyword | |
2838 | shall be specified. Characters specified for the keyword `alpha' | |
2839 | and `digit' are automatically included in this class. */ | |
2840 | { | |
2841 | unsigned long int mask = BIT (tok_alpha) | BIT (tok_digit); | |
96f0d1f5 UD |
2842 | unsigned long int maskw = BITw (tok_alpha) | BITw (tok_digit); |
2843 | ||
2844 | for (cnt = 0; cnt < 256; ++cnt) | |
2845 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
2846 | ctype->class256_collection[cnt] |= BIT (tok_alnum); | |
19bc17a9 RM |
2847 | |
2848 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
96f0d1f5 UD |
2849 | if ((ctype->class_collection[cnt] & maskw) != 0) |
2850 | ctype->class_collection[cnt] |= BITw (tok_alnum); | |
19bc17a9 RM |
2851 | } |
2852 | ||
4b10dd6c | 2853 | if ((ctype->class_done & BITw (tok_space)) == 0) |
19bc17a9 RM |
2854 | /* "If this keyword [space] is not specified, the characters <space>, |
2855 | <form-feed>, <newline>, <carriage-return>, <tab>, and | |
2856 | <vertical-tab>, ..., shall automatically belong to this class, | |
2857 | with implementation-defined character values." [P1003.2, 2.5.2.1] */ | |
2858 | { | |
4b10dd6c | 2859 | struct charseq *seq; |
19bc17a9 | 2860 | |
4b10dd6c | 2861 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
2862 | if (seq == NULL) |
2863 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
2864 | if (seq == NULL) |
2865 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 2866 | if (seq == NULL) |
880f421f UD |
2867 | { |
2868 | if (!be_quiet) | |
2869 | error (0, 0, _("\ | |
4b10dd6c UD |
2870 | %s: character `%s' not defined while needed as default value"), |
2871 | "LC_CTYPE", "<space>"); | |
2872 | } | |
2873 | else if (seq->nbytes != 1) | |
2874 | error (0, 0, _("\ | |
2875 | %s: character `%s' in charmap not representable with one byte"), | |
2876 | "LC_CTYPE", "<space>"); | |
2877 | else | |
2878 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2879 | ||
f0a4b6b1 | 2880 | /* No need to search. */ |
ce177a84 | 2881 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_space); |
19bc17a9 | 2882 | |
4b10dd6c | 2883 | seq = charmap_find_value (charmap, "form-feed", 9); |
f0a4b6b1 UD |
2884 | if (seq == NULL) |
2885 | seq = charmap_find_value (charmap, "U0000000C", 9); | |
4b10dd6c | 2886 | if (seq == NULL) |
880f421f UD |
2887 | { |
2888 | if (!be_quiet) | |
2889 | error (0, 0, _("\ | |
4b10dd6c UD |
2890 | %s: character `%s' not defined while needed as default value"), |
2891 | "LC_CTYPE", "<form-feed>"); | |
2892 | } | |
2893 | else if (seq->nbytes != 1) | |
2894 | error (0, 0, _("\ | |
2895 | %s: character `%s' in charmap not representable with one byte"), | |
2896 | "LC_CTYPE", "<form-feed>"); | |
2897 | else | |
2898 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2899 | ||
f0a4b6b1 | 2900 | /* No need to search. */ |
ce177a84 | 2901 | ELEM (ctype, class_collection, , L'\f') |= BITw (tok_space); |
4b10dd6c | 2902 | |
19bc17a9 | 2903 | |
4b10dd6c | 2904 | seq = charmap_find_value (charmap, "newline", 7); |
f0a4b6b1 UD |
2905 | if (seq == NULL) |
2906 | seq = charmap_find_value (charmap, "U0000000A", 9); | |
4b10dd6c | 2907 | if (seq == NULL) |
880f421f UD |
2908 | { |
2909 | if (!be_quiet) | |
2910 | error (0, 0, _("\ | |
19bc17a9 | 2911 | character `%s' not defined while needed as default value"), |
4b10dd6c UD |
2912 | "<newline>"); |
2913 | } | |
2914 | else if (seq->nbytes != 1) | |
2915 | error (0, 0, _("\ | |
2916 | %s: character `%s' in charmap not representable with one byte"), | |
2917 | "LC_CTYPE", "<newline>"); | |
2918 | else | |
2919 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2920 | ||
f0a4b6b1 | 2921 | /* No need to search. */ |
ce177a84 | 2922 | ELEM (ctype, class_collection, , L'\n') |= BITw (tok_space); |
4b10dd6c | 2923 | |
19bc17a9 | 2924 | |
4b10dd6c | 2925 | seq = charmap_find_value (charmap, "carriage-return", 15); |
f0a4b6b1 UD |
2926 | if (seq == NULL) |
2927 | seq = charmap_find_value (charmap, "U0000000D", 9); | |
4b10dd6c | 2928 | if (seq == NULL) |
880f421f UD |
2929 | { |
2930 | if (!be_quiet) | |
2931 | error (0, 0, _("\ | |
4b10dd6c UD |
2932 | %s: character `%s' not defined while needed as default value"), |
2933 | "LC_CTYPE", "<carriage-return>"); | |
2934 | } | |
2935 | else if (seq->nbytes != 1) | |
2936 | error (0, 0, _("\ | |
2937 | %s: character `%s' in charmap not representable with one byte"), | |
2938 | "LC_CTYPE", "<carriage-return>"); | |
2939 | else | |
2940 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2941 | ||
f0a4b6b1 | 2942 | /* No need to search. */ |
ce177a84 | 2943 | ELEM (ctype, class_collection, , L'\r') |= BITw (tok_space); |
4b10dd6c | 2944 | |
19bc17a9 | 2945 | |
4b10dd6c | 2946 | seq = charmap_find_value (charmap, "tab", 3); |
f0a4b6b1 UD |
2947 | if (seq == NULL) |
2948 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c | 2949 | if (seq == NULL) |
880f421f UD |
2950 | { |
2951 | if (!be_quiet) | |
2952 | error (0, 0, _("\ | |
4b10dd6c UD |
2953 | %s: character `%s' not defined while needed as default value"), |
2954 | "LC_CTYPE", "<tab>"); | |
2955 | } | |
2956 | else if (seq->nbytes != 1) | |
2957 | error (0, 0, _("\ | |
2958 | %s: character `%s' in charmap not representable with one byte"), | |
2959 | "LC_CTYPE", "<tab>"); | |
2960 | else | |
2961 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
2962 | ||
f0a4b6b1 | 2963 | /* No need to search. */ |
ce177a84 | 2964 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_space); |
4b10dd6c | 2965 | |
4b10dd6c UD |
2966 | |
2967 | seq = charmap_find_value (charmap, "vertical-tab", 12); | |
f0a4b6b1 UD |
2968 | if (seq == NULL) |
2969 | seq = charmap_find_value (charmap, "U0000000B", 9); | |
4b10dd6c UD |
2970 | if (seq == NULL) |
2971 | { | |
2972 | if (!be_quiet) | |
2973 | error (0, 0, _("\ | |
2974 | %s: character `%s' not defined while needed as default value"), | |
2975 | "LC_CTYPE", "<vertical-tab>"); | |
2976 | } | |
2977 | else if (seq->nbytes != 1) | |
2978 | error (0, 0, _("\ | |
2979 | %s: character `%s' in charmap not representable with one byte"), | |
2980 | "LC_CTYPE", "<vertical-tab>"); | |
2981 | else | |
2982 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_space); | |
f0a4b6b1 UD |
2983 | |
2984 | /* No need to search. */ | |
ce177a84 | 2985 | ELEM (ctype, class_collection, , L'\v') |= BITw (tok_space); |
19bc17a9 RM |
2986 | } |
2987 | ||
4b10dd6c | 2988 | if ((ctype->class_done & BITw (tok_xdigit)) == 0) |
19bc17a9 RM |
2989 | /* "If this keyword is not specified, the digits `0' to `9', the |
2990 | uppercase letters `A' through `F', and the lowercase letters `a' | |
2991 | through `f', ..., shell automatically belong to this class, with | |
2992 | implementation defined character values." [P1003.2, 2.5.2.1] */ | |
2993 | { | |
4b10dd6c UD |
2994 | set_default (BITPOS (tok_xdigit), '0', '9'); |
2995 | set_default (BITPOS (tok_xdigit), 'A', 'F'); | |
2996 | set_default (BITPOS (tok_xdigit), 'a', 'f'); | |
19bc17a9 RM |
2997 | } |
2998 | ||
4b10dd6c | 2999 | if ((ctype->class_done & BITw (tok_blank)) == 0) |
19bc17a9 RM |
3000 | /* "If this keyword [blank] is unspecified, the characters <space> and |
3001 | <tab> shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
3002 | { | |
4b10dd6c | 3003 | struct charseq *seq; |
19bc17a9 | 3004 | |
4b10dd6c | 3005 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3006 | if (seq == NULL) |
3007 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3008 | if (seq == NULL) |
3009 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c | 3010 | if (seq == NULL) |
880f421f UD |
3011 | { |
3012 | if (!be_quiet) | |
3013 | error (0, 0, _("\ | |
4b10dd6c UD |
3014 | %s: character `%s' not defined while needed as default value"), |
3015 | "LC_CTYPE", "<space>"); | |
3016 | } | |
3017 | else if (seq->nbytes != 1) | |
3018 | error (0, 0, _("\ | |
3019 | %s: character `%s' in charmap not representable with one byte"), | |
3020 | "LC_CTYPE", "<space>"); | |
3021 | else | |
3022 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
3023 | ||
f0a4b6b1 | 3024 | /* No need to search. */ |
ce177a84 | 3025 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_blank); |
4b10dd6c | 3026 | |
4b10dd6c UD |
3027 | |
3028 | seq = charmap_find_value (charmap, "tab", 3); | |
f0a4b6b1 UD |
3029 | if (seq == NULL) |
3030 | seq = charmap_find_value (charmap, "U00000009", 9); | |
4b10dd6c UD |
3031 | if (seq == NULL) |
3032 | { | |
3033 | if (!be_quiet) | |
3034 | error (0, 0, _("\ | |
3035 | %s: character `%s' not defined while needed as default value"), | |
3036 | "LC_CTYPE", "<tab>"); | |
3037 | } | |
3038 | else if (seq->nbytes != 1) | |
3039 | error (0, 0, _("\ | |
3040 | %s: character `%s' in charmap not representable with one byte"), | |
3041 | "LC_CTYPE", "<tab>"); | |
3042 | else | |
3043 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_blank); | |
f0a4b6b1 UD |
3044 | |
3045 | /* No need to search. */ | |
ce177a84 | 3046 | ELEM (ctype, class_collection, , L'\t') |= BITw (tok_blank); |
19bc17a9 RM |
3047 | } |
3048 | ||
4b10dd6c | 3049 | if ((ctype->class_done & BITw (tok_graph)) == 0) |
19bc17a9 RM |
3050 | /* "If this keyword [graph] is not specified, characters specified for |
3051 | the keywords `upper', `lower', `alpha', `digit', `xdigit' and `punct', | |
3052 | shall belong to this character class." [P1003.2, 2.5.2.1] */ | |
3053 | { | |
3054 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3055 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3056 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3057 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3058 | BITw (tok_punct); | |
19bc17a9 RM |
3059 | size_t cnt; |
3060 | ||
3061 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3062 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3063 | ctype->class_collection[cnt] |= BITw (tok_graph); | |
4b10dd6c UD |
3064 | |
3065 | for (cnt = 0; cnt < 256; ++cnt) | |
3066 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3067 | ctype->class256_collection[cnt] |= BIT (tok_graph); | |
19bc17a9 RM |
3068 | } |
3069 | ||
4b10dd6c | 3070 | if ((ctype->class_done & BITw (tok_print)) == 0) |
19bc17a9 RM |
3071 | /* "If this keyword [print] is not provided, characters specified for |
3072 | the keywords `upper', `lower', `alpha', `digit', `xdigit', `punct', | |
3073 | and the <space> character shall belong to this character class." | |
3074 | [P1003.2, 2.5.2.1] */ | |
3075 | { | |
3076 | unsigned long int mask = BIT (tok_upper) | BIT (tok_lower) | | |
3077 | BIT (tok_alpha) | BIT (tok_digit) | BIT (tok_xdigit) | BIT (tok_punct); | |
ce177a84 UD |
3078 | unsigned long int maskw = BITw (tok_upper) | BITw (tok_lower) | |
3079 | BITw (tok_alpha) | BITw (tok_digit) | BITw (tok_xdigit) | | |
3080 | BITw (tok_punct); | |
19bc17a9 | 3081 | size_t cnt; |
4b10dd6c | 3082 | struct charseq *seq; |
19bc17a9 RM |
3083 | |
3084 | for (cnt = 0; cnt < ctype->class_collection_act; ++cnt) | |
ce177a84 UD |
3085 | if ((ctype->class_collection[cnt] & maskw) != 0) |
3086 | ctype->class_collection[cnt] |= BITw (tok_print); | |
19bc17a9 | 3087 | |
4b10dd6c UD |
3088 | for (cnt = 0; cnt < 256; ++cnt) |
3089 | if ((ctype->class256_collection[cnt] & mask) != 0) | |
3090 | ctype->class256_collection[cnt] |= BIT (tok_print); | |
3091 | ||
3092 | ||
4b10dd6c | 3093 | seq = charmap_find_value (charmap, "space", 5); |
45c95239 UD |
3094 | if (seq == NULL) |
3095 | seq = charmap_find_value (charmap, "SP", 2); | |
f0a4b6b1 UD |
3096 | if (seq == NULL) |
3097 | seq = charmap_find_value (charmap, "U00000020", 9); | |
4b10dd6c UD |
3098 | if (seq == NULL) |
3099 | { | |
3100 | if (!be_quiet) | |
3101 | error (0, 0, _("\ | |
3102 | %s: character `%s' not defined while needed as default value"), | |
3103 | "LC_CTYPE", "<space>"); | |
3104 | } | |
3105 | else if (seq->nbytes != 1) | |
3106 | error (0, 0, _("\ | |
3107 | %s: character `%s' in charmap not representable with one byte"), | |
3108 | "LC_CTYPE", "<space>"); | |
3109 | else | |
3110 | ctype->class256_collection[seq->bytes[0]] |= BIT (tok_print); | |
f0a4b6b1 UD |
3111 | |
3112 | /* No need to search. */ | |
ce177a84 | 3113 | ELEM (ctype, class_collection, , L' ') |= BITw (tok_print); |
19bc17a9 RM |
3114 | } |
3115 | ||
4b10dd6c | 3116 | if (ctype->tomap_done[0] == 0) |
6d52618b | 3117 | /* "If this keyword [toupper] is not specified, the lowercase letters |
19bc17a9 RM |
3118 | `a' through `z', and their corresponding uppercase letters `A' to |
3119 | `Z', ..., shall automatically be included, with implementation- | |
3120 | defined character values." [P1003.2, 2.5.2.1] */ | |
3121 | { | |
3122 | char tmp[4]; | |
3123 | int ch; | |
3124 | ||
3125 | strcpy (tmp, "<?>"); | |
3126 | ||
3127 | for (ch = 'a'; ch <= 'z'; ++ch) | |
3128 | { | |
4b10dd6c | 3129 | struct charseq *seq_from, *seq_to; |
19bc17a9 RM |
3130 | |
3131 | tmp[1] = (char) ch; | |
3132 | ||
4b10dd6c UD |
3133 | seq_from = charmap_find_value (charmap, &tmp[1], 1); |
3134 | if (seq_from == NULL) | |
19bc17a9 | 3135 | { |
880f421f UD |
3136 | if (!be_quiet) |
3137 | error (0, 0, _("\ | |
4b10dd6c UD |
3138 | %s: character `%s' not defined while needed as default value"), |
3139 | "LC_CTYPE", tmp); | |
3140 | } | |
3141 | else if (seq_from->nbytes != 1) | |
3142 | { | |
3143 | if (!be_quiet) | |
3144 | error (0, 0, _("\ | |
3145 | %s: character `%s' needed as default value not representable with one byte"), | |
3146 | "LC_CTYPE", tmp); | |
3147 | } | |
3148 | else | |
3149 | { | |
3150 | /* This conversion is implementation defined. */ | |
3151 | tmp[1] = (char) (ch + ('A' - 'a')); | |
3152 | seq_to = charmap_find_value (charmap, &tmp[1], 1); | |
3153 | if (seq_to == NULL) | |
3154 | { | |
3155 | if (!be_quiet) | |
3156 | error (0, 0, _("\ | |
3157 | %s: character `%s' not defined while needed as default value"), | |
3158 | "LC_CTYPE", tmp); | |
3159 | } | |
3160 | else if (seq_to->nbytes != 1) | |
3161 | { | |
3162 | if (!be_quiet) | |
3163 | error (0, 0, _("\ | |
3164 | %s: character `%s' needed as default value not representable with one byte"), | |
3165 | "LC_CTYPE", tmp); | |
3166 | } | |
3167 | else | |
3168 | /* The index [0] is determined by the order of the | |
3169 | `ctype_map_newP' calls in `ctype_startup'. */ | |
3170 | ctype->map256_collection[0][seq_from->bytes[0]] | |
3171 | = seq_to->bytes[0]; | |
19bc17a9 | 3172 | } |
f0a4b6b1 UD |
3173 | |
3174 | /* No need to search. */ | |
3175 | ELEM (ctype, map_collection, [0], ch) = ch + ('A' - 'a'); | |
19bc17a9 RM |
3176 | } |
3177 | } | |
3178 | ||
4b10dd6c | 3179 | if (ctype->tomap_done[1] == 0) |
19bc17a9 RM |
3180 | /* "If this keyword [tolower] is not specified, the mapping shall be |
3181 | the reverse mapping of the one specified to `toupper'." [P1003.2] */ | |
3182 | { | |
19bc17a9 RM |
3183 | for (cnt = 0; cnt < ctype->map_collection_act[0]; ++cnt) |
3184 | if (ctype->map_collection[0][cnt] != 0) | |
3185 | ELEM (ctype, map_collection, [1], | |
3186 | ctype->map_collection[0][cnt]) | |
3187 | = ctype->charnames[cnt]; | |
4b10dd6c UD |
3188 | |
3189 | for (cnt = 0; cnt < 256; ++cnt) | |
3190 | if (ctype->map256_collection[0][cnt] != 0) | |
85cb60ff | 3191 | ctype->map256_collection[1][ctype->map256_collection[0][cnt]] = cnt; |
4b10dd6c UD |
3192 | } |
3193 | ||
3194 | if (ctype->outdigits_act == 0) | |
3195 | { | |
3196 | for (cnt = 0; cnt < 10; ++cnt) | |
3197 | { | |
3198 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
3199 | digits + cnt, 1); | |
3200 | ||
3201 | if (ctype->mboutdigits[cnt] == NULL) | |
1b97149d UD |
3202 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, |
3203 | longnames[cnt], | |
3204 | strlen (longnames[cnt])); | |
b9eb05d6 | 3205 | |
1b97149d UD |
3206 | if (ctype->mboutdigits[cnt] == NULL) |
3207 | ctype->mboutdigits[cnt] = charmap_find_symbol (charmap, | |
3208 | uninames[cnt], 9); | |
b9eb05d6 | 3209 | |
1b97149d | 3210 | if (ctype->mboutdigits[cnt] == NULL) |
b9eb05d6 | 3211 | { |
1b97149d UD |
3212 | /* Provide a replacement. */ |
3213 | error (0, 0, _("\ | |
3214 | no output digits defined and none of the standard names in the charmap")); | |
b9eb05d6 | 3215 | |
1b97149d UD |
3216 | ctype->mboutdigits[cnt] = obstack_alloc (&charmap->mem_pool, |
3217 | sizeof (struct charseq) | |
3218 | + 1); | |
b9eb05d6 | 3219 | |
1b97149d UD |
3220 | /* This is better than nothing. */ |
3221 | ctype->mboutdigits[cnt]->bytes[0] = digits[cnt]; | |
3222 | ctype->mboutdigits[cnt]->nbytes = 1; | |
b9eb05d6 | 3223 | } |
1b97149d UD |
3224 | |
3225 | ctype->wcoutdigits[cnt] = L'0' + cnt; | |
4b10dd6c UD |
3226 | } |
3227 | ||
3228 | ctype->outdigits_act = 10; | |
19bc17a9 RM |
3229 | } |
3230 | } | |
3231 | ||
3232 | ||
ef446144 UD |
3233 | /* Construction of sparse 3-level tables. |
3234 | See wchar-lookup.h for their structure and the meaning of p and q. */ | |
3235 | ||
3236 | struct wctype_table | |
3237 | { | |
3238 | /* Parameters. */ | |
3239 | unsigned int p; | |
3240 | unsigned int q; | |
3241 | /* Working representation. */ | |
3242 | size_t level1_alloc; | |
3243 | size_t level1_size; | |
3244 | uint32_t *level1; | |
3245 | size_t level2_alloc; | |
3246 | size_t level2_size; | |
3247 | uint32_t *level2; | |
3248 | size_t level3_alloc; | |
3249 | size_t level3_size; | |
3250 | uint32_t *level3; | |
3251 | /* Compressed representation. */ | |
3252 | size_t result_size; | |
3253 | char *result; | |
3254 | }; | |
3255 | ||
3256 | /* Initialize. Assumes t->p and t->q have already been set. */ | |
3257 | static inline void | |
3258 | wctype_table_init (struct wctype_table *t) | |
3259 | { | |
3260 | t->level1_alloc = t->level1_size = 0; | |
3261 | t->level2_alloc = t->level2_size = 0; | |
3262 | t->level3_alloc = t->level3_size = 0; | |
3263 | } | |
3264 | ||
ec08818d UD |
3265 | /* Retrieve an entry. */ |
3266 | static inline int | |
3267 | wctype_table_get (struct wctype_table *t, uint32_t wc) | |
3268 | { | |
3269 | uint32_t index1 = wc >> (t->q + t->p + 5); | |
3270 | if (index1 < t->level1_size) | |
3271 | { | |
3272 | uint32_t lookup1 = t->level1[index1]; | |
3273 | if (lookup1 != ~((uint32_t) 0)) | |
3274 | { | |
3275 | uint32_t index2 = ((wc >> (t->p + 5)) & ((1 << t->q) - 1)) | |
3276 | + (lookup1 << t->q); | |
3277 | uint32_t lookup2 = t->level2[index2]; | |
3278 | if (lookup2 != ~((uint32_t) 0)) | |
3279 | { | |
3280 | uint32_t index3 = ((wc >> 5) & ((1 << t->p) - 1)) | |
3281 | + (lookup2 << t->p); | |
3282 | uint32_t lookup3 = t->level3[index3]; | |
3283 | uint32_t index4 = wc & 0x1f; | |
3284 | ||
3285 | return (lookup3 >> index4) & 1; | |
3286 | } | |
3287 | } | |
3288 | } | |
3289 | return 0; | |
3290 | } | |
3291 | ||
ef446144 UD |
3292 | /* Add one entry. */ |
3293 | static void | |
3294 | wctype_table_add (struct wctype_table *t, uint32_t wc) | |
3295 | { | |
3296 | uint32_t index1 = wc >> (t->q + t->p + 5); | |
3297 | uint32_t index2 = (wc >> (t->p + 5)) & ((1 << t->q) - 1); | |
3298 | uint32_t index3 = (wc >> 5) & ((1 << t->p) - 1); | |
3299 | uint32_t index4 = wc & 0x1f; | |
3300 | size_t i, i1, i2; | |
3301 | ||
3302 | if (index1 >= t->level1_size) | |
3303 | { | |
3304 | if (index1 >= t->level1_alloc) | |
3305 | { | |
3306 | size_t alloc = 2 * t->level1_alloc; | |
3307 | if (alloc <= index1) | |
3308 | alloc = index1 + 1; | |
3309 | t->level1 = (t->level1_alloc > 0 | |
3310 | ? (uint32_t *) xrealloc ((char *) t->level1, | |
3311 | alloc * sizeof (uint32_t)) | |
3312 | : (uint32_t *) xmalloc (alloc * sizeof (uint32_t))); | |
3313 | t->level1_alloc = alloc; | |
3314 | } | |
3315 | while (index1 >= t->level1_size) | |
3316 | t->level1[t->level1_size++] = ~((uint32_t) 0); | |
3317 | } | |
3318 | ||
3319 | if (t->level1[index1] == ~((uint32_t) 0)) | |
3320 | { | |
3321 | if (t->level2_size == t->level2_alloc) | |
3322 | { | |
3323 | size_t alloc = 2 * t->level2_alloc + 1; | |
3324 | t->level2 = (t->level2_alloc > 0 | |
3325 | ? (uint32_t *) xrealloc ((char *) t->level2, | |
3326 | (alloc << t->q) * sizeof (uint32_t)) | |
3327 | : (uint32_t *) xmalloc ((alloc << t->q) * sizeof (uint32_t))); | |
3328 | t->level2_alloc = alloc; | |
3329 | } | |
3330 | i1 = t->level2_size << t->q; | |
3331 | i2 = (t->level2_size + 1) << t->q; | |
3332 | for (i = i1; i < i2; i++) | |
3333 | t->level2[i] = ~((uint32_t) 0); | |
3334 | t->level1[index1] = t->level2_size++; | |
3335 | } | |
3336 | ||
3337 | index2 += t->level1[index1] << t->q; | |
3338 | ||
3339 | if (t->level2[index2] == ~((uint32_t) 0)) | |
3340 | { | |
3341 | if (t->level3_size == t->level3_alloc) | |
3342 | { | |
3343 | size_t alloc = 2 * t->level3_alloc + 1; | |
3344 | t->level3 = (t->level3_alloc > 0 | |
3345 | ? (uint32_t *) xrealloc ((char *) t->level3, | |
3346 | (alloc << t->p) * sizeof (uint32_t)) | |
3347 | : (uint32_t *) xmalloc ((alloc << t->p) * sizeof (uint32_t))); | |
3348 | t->level3_alloc = alloc; | |
3349 | } | |
3350 | i1 = t->level3_size << t->p; | |
3351 | i2 = (t->level3_size + 1) << t->p; | |
3352 | for (i = i1; i < i2; i++) | |
3353 | t->level3[i] = 0; | |
3354 | t->level2[index2] = t->level3_size++; | |
3355 | } | |
3356 | ||
3357 | index3 += t->level2[index2] << t->p; | |
3358 | ||
3359 | t->level3[index3] |= (uint32_t)1 << index4; | |
3360 | } | |
3361 | ||
3362 | /* Finalize and shrink. */ | |
3363 | static void | |
3364 | wctype_table_finalize (struct wctype_table *t) | |
3365 | { | |
3366 | size_t i, j, k; | |
3367 | uint32_t reorder3[t->level3_size]; | |
3368 | uint32_t reorder2[t->level2_size]; | |
3369 | uint32_t level1_offset, level2_offset, level3_offset; | |
3370 | ||
3371 | /* Uniquify level3 blocks. */ | |
3372 | k = 0; | |
3373 | for (j = 0; j < t->level3_size; j++) | |
3374 | { | |
3375 | for (i = 0; i < k; i++) | |
3376 | if (memcmp (&t->level3[i << t->p], &t->level3[j << t->p], | |
3377 | (1 << t->p) * sizeof (uint32_t)) == 0) | |
3378 | break; | |
3379 | /* Relocate block j to block i. */ | |
3380 | reorder3[j] = i; | |
3381 | if (i == k) | |
3382 | { | |
3383 | if (i != j) | |
3384 | memcpy (&t->level3[i << t->p], &t->level3[j << t->p], | |
3385 | (1 << t->p) * sizeof (uint32_t)); | |
3386 | k++; | |
3387 | } | |
3388 | } | |
3389 | t->level3_size = k; | |
3390 | ||
3391 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3392 | if (t->level2[i] != ~((uint32_t) 0)) | |
3393 | t->level2[i] = reorder3[t->level2[i]]; | |
3394 | ||
3395 | /* Uniquify level2 blocks. */ | |
3396 | k = 0; | |
3397 | for (j = 0; j < t->level2_size; j++) | |
3398 | { | |
3399 | for (i = 0; i < k; i++) | |
3400 | if (memcmp (&t->level2[i << t->q], &t->level2[j << t->q], | |
3401 | (1 << t->q) * sizeof (uint32_t)) == 0) | |
3402 | break; | |
3403 | /* Relocate block j to block i. */ | |
3404 | reorder2[j] = i; | |
3405 | if (i == k) | |
3406 | { | |
3407 | if (i != j) | |
3408 | memcpy (&t->level2[i << t->q], &t->level2[j << t->q], | |
3409 | (1 << t->q) * sizeof (uint32_t)); | |
3410 | k++; | |
3411 | } | |
3412 | } | |
3413 | t->level2_size = k; | |
3414 | ||
3415 | for (i = 0; i < t->level1_size; i++) | |
3416 | if (t->level1[i] != ~((uint32_t) 0)) | |
3417 | t->level1[i] = reorder2[t->level1[i]]; | |
3418 | ||
3419 | /* Create and fill the resulting compressed representation. */ | |
3420 | t->result_size = | |
3421 | 5 * sizeof (uint32_t) | |
3422 | + t->level1_size * sizeof (uint32_t) | |
3423 | + (t->level2_size << t->q) * sizeof (uint32_t) | |
3424 | + (t->level3_size << t->p) * sizeof (uint32_t); | |
3425 | t->result = (char *) xmalloc (t->result_size); | |
3426 | ||
3427 | level1_offset = | |
3428 | 5 * sizeof (uint32_t); | |
3429 | level2_offset = | |
3430 | 5 * sizeof (uint32_t) | |
3431 | + t->level1_size * sizeof (uint32_t); | |
3432 | level3_offset = | |
3433 | 5 * sizeof (uint32_t) | |
3434 | + t->level1_size * sizeof (uint32_t) | |
3435 | + (t->level2_size << t->q) * sizeof (uint32_t); | |
3436 | ||
3437 | ((uint32_t *) t->result)[0] = t->q + t->p + 5; | |
3438 | ((uint32_t *) t->result)[1] = t->level1_size; | |
3439 | ((uint32_t *) t->result)[2] = t->p + 5; | |
3440 | ((uint32_t *) t->result)[3] = (1 << t->q) - 1; | |
3441 | ((uint32_t *) t->result)[4] = (1 << t->p) - 1; | |
3442 | ||
3443 | for (i = 0; i < t->level1_size; i++) | |
3444 | ((uint32_t *) (t->result + level1_offset))[i] = | |
3445 | (t->level1[i] == ~((uint32_t) 0) | |
3446 | ? 0 | |
3447 | : (t->level1[i] << t->q) * sizeof (uint32_t) + level2_offset); | |
3448 | ||
3449 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3450 | ((uint32_t *) (t->result + level2_offset))[i] = | |
3451 | (t->level2[i] == ~((uint32_t) 0) | |
3452 | ? 0 | |
3453 | : (t->level2[i] << t->p) * sizeof (uint32_t) + level3_offset); | |
3454 | ||
3455 | for (i = 0; i < (t->level3_size << t->p); i++) | |
3456 | ((uint32_t *) (t->result + level3_offset))[i] = t->level3[i]; | |
3457 | ||
3458 | if (t->level1_alloc > 0) | |
3459 | free (t->level1); | |
3460 | if (t->level2_alloc > 0) | |
3461 | free (t->level2); | |
3462 | if (t->level3_alloc > 0) | |
3463 | free (t->level3); | |
3464 | } | |
3465 | ||
3466 | struct wcwidth_table | |
3467 | { | |
3468 | /* Parameters. */ | |
3469 | unsigned int p; | |
3470 | unsigned int q; | |
3471 | /* Working representation. */ | |
3472 | size_t level1_alloc; | |
3473 | size_t level1_size; | |
3474 | uint32_t *level1; | |
3475 | size_t level2_alloc; | |
3476 | size_t level2_size; | |
3477 | uint32_t *level2; | |
3478 | size_t level3_alloc; | |
3479 | size_t level3_size; | |
3480 | uint8_t *level3; | |
3481 | /* Compressed representation. */ | |
3482 | size_t result_size; | |
3483 | char *result; | |
3484 | }; | |
3485 | ||
3486 | /* Initialize. Assumes t->p and t->q have already been set. */ | |
3487 | static inline void | |
3488 | wcwidth_table_init (struct wcwidth_table *t) | |
3489 | { | |
3490 | t->level1_alloc = t->level1_size = 0; | |
3491 | t->level2_alloc = t->level2_size = 0; | |
3492 | t->level3_alloc = t->level3_size = 0; | |
3493 | } | |
3494 | ||
ec08818d UD |
3495 | /* Retrieve an entry. */ |
3496 | static inline uint8_t | |
3497 | wcwidth_table_get (struct wcwidth_table *t, uint32_t wc) | |
3498 | { | |
3499 | uint32_t index1 = wc >> (t->q + t->p); | |
3500 | if (index1 < t->level1_size) | |
3501 | { | |
3502 | uint32_t lookup1 = t->level1[index1]; | |
3503 | if (lookup1 != ~((uint32_t) 0)) | |
3504 | { | |
3505 | uint32_t index2 = ((wc >> t->p) & ((1 << t->q) - 1)) | |
3506 | + (lookup1 << t->q); | |
3507 | uint32_t lookup2 = t->level2[index2]; | |
3508 | if (lookup2 != ~((uint32_t) 0)) | |
3509 | { | |
3510 | uint32_t index3 = (wc & ((1 << t->p) - 1)) | |
3511 | + (lookup2 << t->p); | |
3512 | uint8_t lookup3 = t->level3[index3]; | |
3513 | ||
3514 | return lookup3; | |
3515 | } | |
3516 | } | |
3517 | } | |
3518 | return 0xff; | |
3519 | } | |
3520 | ||
ef446144 UD |
3521 | /* Add one entry. */ |
3522 | static void | |
3523 | wcwidth_table_add (struct wcwidth_table *t, uint32_t wc, uint8_t width) | |
3524 | { | |
3525 | uint32_t index1 = wc >> (t->q + t->p); | |
3526 | uint32_t index2 = (wc >> t->p) & ((1 << t->q) - 1); | |
3527 | uint32_t index3 = wc & ((1 << t->p) - 1); | |
3528 | size_t i, i1, i2; | |
3529 | ||
ec08818d | 3530 | if (width == wcwidth_table_get (t, wc)) |
ef446144 UD |
3531 | return; |
3532 | ||
3533 | if (index1 >= t->level1_size) | |
3534 | { | |
3535 | if (index1 >= t->level1_alloc) | |
3536 | { | |
3537 | size_t alloc = 2 * t->level1_alloc; | |
3538 | if (alloc <= index1) | |
3539 | alloc = index1 + 1; | |
3540 | t->level1 = (t->level1_alloc > 0 | |
3541 | ? (uint32_t *) xrealloc ((char *) t->level1, | |
3542 | alloc * sizeof (uint32_t)) | |
3543 | : (uint32_t *) xmalloc (alloc * sizeof (uint32_t))); | |
3544 | t->level1_alloc = alloc; | |
3545 | } | |
3546 | while (index1 >= t->level1_size) | |
3547 | t->level1[t->level1_size++] = ~((uint32_t) 0); | |
3548 | } | |
3549 | ||
3550 | if (t->level1[index1] == ~((uint32_t) 0)) | |
3551 | { | |
3552 | if (t->level2_size == t->level2_alloc) | |
3553 | { | |
3554 | size_t alloc = 2 * t->level2_alloc + 1; | |
3555 | t->level2 = (t->level2_alloc > 0 | |
3556 | ? (uint32_t *) xrealloc ((char *) t->level2, | |
3557 | (alloc << t->q) * sizeof (uint32_t)) | |
3558 | : (uint32_t *) xmalloc ((alloc << t->q) * sizeof (uint32_t))); | |
3559 | t->level2_alloc = alloc; | |
3560 | } | |
3561 | i1 = t->level2_size << t->q; | |
3562 | i2 = (t->level2_size + 1) << t->q; | |
3563 | for (i = i1; i < i2; i++) | |
3564 | t->level2[i] = ~((uint32_t) 0); | |
3565 | t->level1[index1] = t->level2_size++; | |
3566 | } | |
3567 | ||
3568 | index2 += t->level1[index1] << t->q; | |
3569 | ||
3570 | if (t->level2[index2] == ~((uint32_t) 0)) | |
3571 | { | |
3572 | if (t->level3_size == t->level3_alloc) | |
3573 | { | |
3574 | size_t alloc = 2 * t->level3_alloc + 1; | |
3575 | t->level3 = (t->level3_alloc > 0 | |
3576 | ? (uint8_t *) xrealloc ((char *) t->level3, | |
3577 | (alloc << t->p) * sizeof (uint8_t)) | |
3578 | : (uint8_t *) xmalloc ((alloc << t->p) * sizeof (uint8_t))); | |
3579 | t->level3_alloc = alloc; | |
3580 | } | |
3581 | i1 = t->level3_size << t->p; | |
3582 | i2 = (t->level3_size + 1) << t->p; | |
3583 | for (i = i1; i < i2; i++) | |
3584 | t->level3[i] = 0xff; | |
3585 | t->level2[index2] = t->level3_size++; | |
3586 | } | |
3587 | ||
3588 | index3 += t->level2[index2] << t->p; | |
3589 | ||
3590 | t->level3[index3] = width; | |
3591 | } | |
3592 | ||
3593 | /* Finalize and shrink. */ | |
3594 | static void | |
3595 | wcwidth_table_finalize (struct wcwidth_table *t) | |
3596 | { | |
3597 | size_t i, j, k; | |
3598 | uint32_t reorder3[t->level3_size]; | |
3599 | uint32_t reorder2[t->level2_size]; | |
3600 | uint32_t level1_offset, level2_offset, level3_offset, last_offset; | |
3601 | ||
3602 | /* Uniquify level3 blocks. */ | |
3603 | k = 0; | |
3604 | for (j = 0; j < t->level3_size; j++) | |
3605 | { | |
3606 | for (i = 0; i < k; i++) | |
3607 | if (memcmp (&t->level3[i << t->p], &t->level3[j << t->p], | |
3608 | (1 << t->p) * sizeof (uint8_t)) == 0) | |
3609 | break; | |
3610 | /* Relocate block j to block i. */ | |
3611 | reorder3[j] = i; | |
3612 | if (i == k) | |
3613 | { | |
3614 | if (i != j) | |
3615 | memcpy (&t->level3[i << t->p], &t->level3[j << t->p], | |
3616 | (1 << t->p) * sizeof (uint8_t)); | |
3617 | k++; | |
3618 | } | |
3619 | } | |
3620 | t->level3_size = k; | |
3621 | ||
3622 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3623 | if (t->level2[i] != ~((uint32_t) 0)) | |
3624 | t->level2[i] = reorder3[t->level2[i]]; | |
3625 | ||
3626 | /* Uniquify level2 blocks. */ | |
3627 | k = 0; | |
3628 | for (j = 0; j < t->level2_size; j++) | |
3629 | { | |
3630 | for (i = 0; i < k; i++) | |
3631 | if (memcmp (&t->level2[i << t->q], &t->level2[j << t->q], | |
3632 | (1 << t->q) * sizeof (uint32_t)) == 0) | |
3633 | break; | |
3634 | /* Relocate block j to block i. */ | |
3635 | reorder2[j] = i; | |
3636 | if (i == k) | |
3637 | { | |
3638 | if (i != j) | |
3639 | memcpy (&t->level2[i << t->q], &t->level2[j << t->q], | |
3640 | (1 << t->q) * sizeof (uint32_t)); | |
3641 | k++; | |
3642 | } | |
3643 | } | |
3644 | t->level2_size = k; | |
3645 | ||
3646 | for (i = 0; i < t->level1_size; i++) | |
3647 | if (t->level1[i] != ~((uint32_t) 0)) | |
3648 | t->level1[i] = reorder2[t->level1[i]]; | |
3649 | ||
3650 | /* Create and fill the resulting compressed representation. */ | |
3651 | last_offset = | |
3652 | 5 * sizeof (uint32_t) | |
3653 | + t->level1_size * sizeof (uint32_t) | |
3654 | + (t->level2_size << t->q) * sizeof (uint32_t) | |
3655 | + (t->level3_size << t->p) * sizeof (uint8_t); | |
3656 | t->result_size = (last_offset + 3) & ~3ul; | |
3657 | t->result = (char *) xmalloc (t->result_size); | |
3658 | ||
3659 | level1_offset = | |
3660 | 5 * sizeof (uint32_t); | |
3661 | level2_offset = | |
3662 | 5 * sizeof (uint32_t) | |
3663 | + t->level1_size * sizeof (uint32_t); | |
3664 | level3_offset = | |
3665 | 5 * sizeof (uint32_t) | |
3666 | + t->level1_size * sizeof (uint32_t) | |
3667 | + (t->level2_size << t->q) * sizeof (uint32_t); | |
3668 | ||
3669 | ((uint32_t *) t->result)[0] = t->q + t->p; | |
3670 | ((uint32_t *) t->result)[1] = t->level1_size; | |
3671 | ((uint32_t *) t->result)[2] = t->p; | |
3672 | ((uint32_t *) t->result)[3] = (1 << t->q) - 1; | |
3673 | ((uint32_t *) t->result)[4] = (1 << t->p) - 1; | |
3674 | ||
3675 | for (i = 0; i < t->level1_size; i++) | |
3676 | ((uint32_t *) (t->result + level1_offset))[i] = | |
3677 | (t->level1[i] == ~((uint32_t) 0) | |
3678 | ? 0 | |
3679 | : (t->level1[i] << t->q) * sizeof (uint32_t) + level2_offset); | |
3680 | ||
3681 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3682 | ((uint32_t *) (t->result + level2_offset))[i] = | |
3683 | (t->level2[i] == ~((uint32_t) 0) | |
3684 | ? 0 | |
3685 | : (t->level2[i] << t->p) * sizeof (uint8_t) + level3_offset); | |
3686 | ||
3687 | for (i = 0; i < (t->level3_size << t->p); i++) | |
3688 | ((uint8_t *) (t->result + level3_offset))[i] = t->level3[i]; | |
3689 | ||
3690 | if (last_offset < t->result_size) | |
3691 | memset (t->result + last_offset, 0, t->result_size - last_offset); | |
3692 | ||
3693 | if (t->level1_alloc > 0) | |
3694 | free (t->level1); | |
3695 | if (t->level2_alloc > 0) | |
3696 | free (t->level2); | |
3697 | if (t->level3_alloc > 0) | |
3698 | free (t->level3); | |
3699 | } | |
3700 | ||
3701 | struct wctrans_table | |
3702 | { | |
3703 | /* Parameters. */ | |
3704 | unsigned int p; | |
3705 | unsigned int q; | |
3706 | /* Working representation. */ | |
3707 | size_t level1_alloc; | |
3708 | size_t level1_size; | |
3709 | uint32_t *level1; | |
3710 | size_t level2_alloc; | |
3711 | size_t level2_size; | |
3712 | uint32_t *level2; | |
3713 | size_t level3_alloc; | |
3714 | size_t level3_size; | |
3715 | int32_t *level3; | |
3716 | /* Compressed representation. */ | |
3717 | size_t result_size; | |
3718 | char *result; | |
3719 | }; | |
3720 | ||
3721 | /* Initialize. Assumes t->p and t->q have already been set. */ | |
3722 | static inline void | |
3723 | wctrans_table_init (struct wctrans_table *t) | |
3724 | { | |
3725 | t->level1_alloc = t->level1_size = 0; | |
3726 | t->level2_alloc = t->level2_size = 0; | |
3727 | t->level3_alloc = t->level3_size = 0; | |
3728 | } | |
3729 | ||
ec08818d UD |
3730 | /* Retrieve an entry. */ |
3731 | static inline uint32_t | |
3732 | wctrans_table_get (struct wctrans_table *t, uint32_t wc) | |
3733 | { | |
3734 | uint32_t index1 = wc >> (t->q + t->p); | |
3735 | if (index1 < t->level1_size) | |
3736 | { | |
3737 | uint32_t lookup1 = t->level1[index1]; | |
3738 | if (lookup1 != ~((uint32_t) 0)) | |
3739 | { | |
3740 | uint32_t index2 = ((wc >> t->p) & ((1 << t->q) - 1)) | |
3741 | + (lookup1 << t->q); | |
3742 | uint32_t lookup2 = t->level2[index2]; | |
3743 | if (lookup2 != ~((uint32_t) 0)) | |
3744 | { | |
3745 | uint32_t index3 = (wc & ((1 << t->p) - 1)) | |
3746 | + (lookup2 << t->p); | |
3747 | int32_t lookup3 = t->level3[index3]; | |
3748 | ||
3749 | return wc + lookup3; | |
3750 | } | |
3751 | } | |
3752 | } | |
3753 | return wc; | |
3754 | } | |
3755 | ||
ef446144 UD |
3756 | /* Add one entry. */ |
3757 | static void | |
3758 | wctrans_table_add (struct wctrans_table *t, uint32_t wc, uint32_t mapped_wc) | |
3759 | { | |
3760 | uint32_t index1 = wc >> (t->q + t->p); | |
3761 | uint32_t index2 = (wc >> t->p) & ((1 << t->q) - 1); | |
3762 | uint32_t index3 = wc & ((1 << t->p) - 1); | |
ec08818d | 3763 | int32_t value; |
ef446144 UD |
3764 | size_t i, i1, i2; |
3765 | ||
ec08818d | 3766 | if (mapped_wc == wctrans_table_get (t, wc)) |
ef446144 UD |
3767 | return; |
3768 | ||
ec08818d UD |
3769 | value = (int32_t) mapped_wc - (int32_t) wc; |
3770 | ||
ef446144 UD |
3771 | if (index1 >= t->level1_size) |
3772 | { | |
3773 | if (index1 >= t->level1_alloc) | |
3774 | { | |
3775 | size_t alloc = 2 * t->level1_alloc; | |
3776 | if (alloc <= index1) | |
3777 | alloc = index1 + 1; | |
3778 | t->level1 = (t->level1_alloc > 0 | |
3779 | ? (uint32_t *) xrealloc ((char *) t->level1, | |
3780 | alloc * sizeof (uint32_t)) | |
3781 | : (uint32_t *) xmalloc (alloc * sizeof (uint32_t))); | |
3782 | t->level1_alloc = alloc; | |
3783 | } | |
3784 | while (index1 >= t->level1_size) | |
3785 | t->level1[t->level1_size++] = ~((uint32_t) 0); | |
3786 | } | |
3787 | ||
3788 | if (t->level1[index1] == ~((uint32_t) 0)) | |
3789 | { | |
3790 | if (t->level2_size == t->level2_alloc) | |
3791 | { | |
3792 | size_t alloc = 2 * t->level2_alloc + 1; | |
3793 | t->level2 = (t->level2_alloc > 0 | |
3794 | ? (uint32_t *) xrealloc ((char *) t->level2, | |
3795 | (alloc << t->q) * sizeof (uint32_t)) | |
3796 | : (uint32_t *) xmalloc ((alloc << t->q) * sizeof (uint32_t))); | |
3797 | t->level2_alloc = alloc; | |
3798 | } | |
3799 | i1 = t->level2_size << t->q; | |
3800 | i2 = (t->level2_size + 1) << t->q; | |
3801 | for (i = i1; i < i2; i++) | |
3802 | t->level2[i] = ~((uint32_t) 0); | |
3803 | t->level1[index1] = t->level2_size++; | |
3804 | } | |
3805 | ||
3806 | index2 += t->level1[index1] << t->q; | |
3807 | ||
3808 | if (t->level2[index2] == ~((uint32_t) 0)) | |
3809 | { | |
3810 | if (t->level3_size == t->level3_alloc) | |
3811 | { | |
3812 | size_t alloc = 2 * t->level3_alloc + 1; | |
3813 | t->level3 = (t->level3_alloc > 0 | |
3814 | ? (int32_t *) xrealloc ((char *) t->level3, | |
3815 | (alloc << t->p) * sizeof (int32_t)) | |
3816 | : (int32_t *) xmalloc ((alloc << t->p) * sizeof (int32_t))); | |
3817 | t->level3_alloc = alloc; | |
3818 | } | |
3819 | i1 = t->level3_size << t->p; | |
3820 | i2 = (t->level3_size + 1) << t->p; | |
3821 | for (i = i1; i < i2; i++) | |
3822 | t->level3[i] = 0; | |
3823 | t->level2[index2] = t->level3_size++; | |
3824 | } | |
3825 | ||
3826 | index3 += t->level2[index2] << t->p; | |
3827 | ||
3828 | t->level3[index3] = value; | |
3829 | } | |
3830 | ||
3831 | /* Finalize and shrink. */ | |
3832 | static void | |
3833 | wctrans_table_finalize (struct wctrans_table *t) | |
3834 | { | |
3835 | size_t i, j, k; | |
3836 | uint32_t reorder3[t->level3_size]; | |
3837 | uint32_t reorder2[t->level2_size]; | |
3838 | uint32_t level1_offset, level2_offset, level3_offset; | |
3839 | ||
3840 | /* Uniquify level3 blocks. */ | |
3841 | k = 0; | |
3842 | for (j = 0; j < t->level3_size; j++) | |
3843 | { | |
3844 | for (i = 0; i < k; i++) | |
3845 | if (memcmp (&t->level3[i << t->p], &t->level3[j << t->p], | |
3846 | (1 << t->p) * sizeof (int32_t)) == 0) | |
3847 | break; | |
3848 | /* Relocate block j to block i. */ | |
3849 | reorder3[j] = i; | |
3850 | if (i == k) | |
3851 | { | |
3852 | if (i != j) | |
3853 | memcpy (&t->level3[i << t->p], &t->level3[j << t->p], | |
3854 | (1 << t->p) * sizeof (int32_t)); | |
3855 | k++; | |
3856 | } | |
3857 | } | |
3858 | t->level3_size = k; | |
3859 | ||
3860 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3861 | if (t->level2[i] != ~((uint32_t) 0)) | |
3862 | t->level2[i] = reorder3[t->level2[i]]; | |
3863 | ||
3864 | /* Uniquify level2 blocks. */ | |
3865 | k = 0; | |
3866 | for (j = 0; j < t->level2_size; j++) | |
3867 | { | |
3868 | for (i = 0; i < k; i++) | |
3869 | if (memcmp (&t->level2[i << t->q], &t->level2[j << t->q], | |
3870 | (1 << t->q) * sizeof (uint32_t)) == 0) | |
3871 | break; | |
3872 | /* Relocate block j to block i. */ | |
3873 | reorder2[j] = i; | |
3874 | if (i == k) | |
3875 | { | |
3876 | if (i != j) | |
3877 | memcpy (&t->level2[i << t->q], &t->level2[j << t->q], | |
3878 | (1 << t->q) * sizeof (uint32_t)); | |
3879 | k++; | |
3880 | } | |
3881 | } | |
3882 | t->level2_size = k; | |
3883 | ||
3884 | for (i = 0; i < t->level1_size; i++) | |
3885 | if (t->level1[i] != ~((uint32_t) 0)) | |
3886 | t->level1[i] = reorder2[t->level1[i]]; | |
3887 | ||
3888 | /* Create and fill the resulting compressed representation. */ | |
3889 | t->result_size = | |
3890 | 5 * sizeof (uint32_t) | |
3891 | + t->level1_size * sizeof (uint32_t) | |
3892 | + (t->level2_size << t->q) * sizeof (uint32_t) | |
3893 | + (t->level3_size << t->p) * sizeof (int32_t); | |
3894 | t->result = (char *) xmalloc (t->result_size); | |
3895 | ||
3896 | level1_offset = | |
3897 | 5 * sizeof (uint32_t); | |
3898 | level2_offset = | |
3899 | 5 * sizeof (uint32_t) | |
3900 | + t->level1_size * sizeof (uint32_t); | |
3901 | level3_offset = | |
3902 | 5 * sizeof (uint32_t) | |
3903 | + t->level1_size * sizeof (uint32_t) | |
3904 | + (t->level2_size << t->q) * sizeof (uint32_t); | |
3905 | ||
3906 | ((uint32_t *) t->result)[0] = t->q + t->p; | |
3907 | ((uint32_t *) t->result)[1] = t->level1_size; | |
3908 | ((uint32_t *) t->result)[2] = t->p; | |
3909 | ((uint32_t *) t->result)[3] = (1 << t->q) - 1; | |
3910 | ((uint32_t *) t->result)[4] = (1 << t->p) - 1; | |
3911 | ||
3912 | for (i = 0; i < t->level1_size; i++) | |
3913 | ((uint32_t *) (t->result + level1_offset))[i] = | |
3914 | (t->level1[i] == ~((uint32_t) 0) | |
3915 | ? 0 | |
3916 | : (t->level1[i] << t->q) * sizeof (uint32_t) + level2_offset); | |
3917 | ||
3918 | for (i = 0; i < (t->level2_size << t->q); i++) | |
3919 | ((uint32_t *) (t->result + level2_offset))[i] = | |
3920 | (t->level2[i] == ~((uint32_t) 0) | |
3921 | ? 0 | |
3922 | : (t->level2[i] << t->p) * sizeof (int32_t) + level3_offset); | |
3923 | ||
3924 | for (i = 0; i < (t->level3_size << t->p); i++) | |
3925 | ((int32_t *) (t->result + level3_offset))[i] = t->level3[i]; | |
3926 | ||
3927 | if (t->level1_alloc > 0) | |
3928 | free (t->level1); | |
3929 | if (t->level2_alloc > 0) | |
3930 | free (t->level2); | |
3931 | if (t->level3_alloc > 0) | |
3932 | free (t->level3); | |
3933 | } | |
3934 | ||
3935 | ||
19bc17a9 | 3936 | static void |
4b10dd6c UD |
3937 | allocate_arrays (struct locale_ctype_t *ctype, struct charmap_t *charmap, |
3938 | struct repertoire_t *repertoire) | |
19bc17a9 RM |
3939 | { |
3940 | size_t idx; | |
a53e3292 | 3941 | size_t width_table_size; |
0e16ecfa UD |
3942 | const void *key; |
3943 | size_t len; | |
3944 | void *vdata; | |
3945 | void *curs; | |
5d431a3e | 3946 | |
6d52618b UD |
3947 | /* First we have to decide how we organize the arrays. It is easy |
3948 | for a one-byte character set. But multi-byte character set | |
3949 | cannot be stored flat because the chars might be sparsely used. | |
3950 | So we determine an optimal hashing function for the used | |
3951 | characters. | |
3952 | ||
3953 | We use a very trivial hashing function to store the sparse | |
3954 | table. CH % TABSIZE is used as an index. To solve multiple hits | |
3955 | we have N planes. This guarantees a fixed search time for a | |
42d7c593 | 3956 | character [N / 2]. In the following code we determine the minimum |
66ac0abe UD |
3957 | value for TABSIZE * N, where TABSIZE >= 256. |
3958 | ||
3959 | Some people complained that this algorithm takes too long. Well, | |
3960 | go on, improve it. But changing the step size is *not* an | |
3961 | option. Some people changed this to use only sizes of prime | |
3962 | numbers. Think again, do some math. We are looking for the | |
3963 | optimal solution, not something which works in general. Unless | |
3964 | somebody can provide a dynamic programming solution I think this | |
3965 | implementation is as good as it can get. */ | |
19bc17a9 RM |
3966 | size_t min_total = UINT_MAX; |
3967 | size_t act_size = 256; | |
3968 | ||
ef446144 UD |
3969 | if (oldstyle_tables) |
3970 | { | |
3971 | if (!be_quiet && ctype->charnames_act > 512) | |
3972 | fputs (_("\ | |
19bc17a9 | 3973 | Computing table size for character classes might take a while..."), |
ef446144 | 3974 | stderr); |
19bc17a9 | 3975 | |
ef446144 UD |
3976 | /* While we want to have a small total size we are willing to use a |
3977 | little bit larger table if this reduces the number of layers. | |
3978 | Therefore we add a little penalty to the number of planes. | |
3979 | Maybe this constant has to be adjusted a bit. */ | |
66ac0abe | 3980 | #define PENALTY 128 |
ef446144 UD |
3981 | do |
3982 | { | |
3983 | size_t cnt[act_size]; | |
3984 | size_t act_planes = 1; | |
19bc17a9 | 3985 | |
ef446144 | 3986 | memset (cnt, '\0', sizeof cnt); |
19bc17a9 | 3987 | |
ef446144 UD |
3988 | for (idx = 0; idx < 256; ++idx) |
3989 | cnt[idx] = 1; | |
19bc17a9 | 3990 | |
ef446144 UD |
3991 | for (idx = 0; idx < ctype->charnames_act; ++idx) |
3992 | if (ctype->charnames[idx] >= 256) | |
19bc17a9 | 3993 | { |
ef446144 UD |
3994 | size_t nr = ctype->charnames[idx] % act_size; |
3995 | ||
3996 | if (++cnt[nr] > act_planes) | |
3997 | { | |
3998 | act_planes = cnt[nr]; | |
3999 | if ((act_size + PENALTY) * act_planes >= min_total) | |
4000 | break; | |
4001 | } | |
19bc17a9 | 4002 | } |
19bc17a9 | 4003 | |
ef446144 UD |
4004 | if ((act_size + PENALTY) * act_planes < min_total) |
4005 | { | |
4006 | min_total = (act_size + PENALTY) * act_planes; | |
4007 | ctype->plane_size = act_size; | |
4008 | ctype->plane_cnt = act_planes; | |
4009 | } | |
19bc17a9 | 4010 | |
ef446144 UD |
4011 | ++act_size; |
4012 | } | |
4013 | while (act_size < min_total); | |
19bc17a9 | 4014 | |
ef446144 UD |
4015 | if (!be_quiet && ctype->charnames_act > 512) |
4016 | fputs (_(" done\n"), stderr); | |
19bc17a9 | 4017 | |
75cd5204 | 4018 | |
ef446144 UD |
4019 | ctype->names = (uint32_t *) xcalloc (ctype->plane_size |
4020 | * ctype->plane_cnt, | |
4021 | sizeof (uint32_t)); | |
19bc17a9 | 4022 | |
ef446144 UD |
4023 | for (idx = 1; idx < 256; ++idx) |
4024 | ctype->names[idx] = idx; | |
19bc17a9 | 4025 | |
ef446144 UD |
4026 | /* Trick: change the 0th entry's name to 1 to mark the cell occupied. */ |
4027 | ctype->names[0] = 1; | |
19bc17a9 | 4028 | |
ef446144 UD |
4029 | for (idx = 256; idx < ctype->charnames_act; ++idx) |
4030 | { | |
4031 | size_t nr = (ctype->charnames[idx] % ctype->plane_size); | |
4032 | size_t depth = 0; | |
19bc17a9 | 4033 | |
ef446144 UD |
4034 | while (ctype->names[nr + depth * ctype->plane_size]) |
4035 | ++depth; | |
4036 | assert (depth < ctype->plane_cnt); | |
19bc17a9 | 4037 | |
ef446144 | 4038 | ctype->names[nr + depth * ctype->plane_size] = ctype->charnames[idx]; |
19bc17a9 | 4039 | |
ef446144 UD |
4040 | /* Now for faster access remember the index in the NAMES_B array. */ |
4041 | ctype->charnames[idx] = nr + depth * ctype->plane_size; | |
4042 | } | |
4043 | ctype->names[0] = 0; | |
4044 | } | |
4045 | else | |
4046 | { | |
4047 | ctype->plane_size = 0; | |
4048 | ctype->plane_cnt = 0; | |
4049 | ctype->names = NULL; | |
19bc17a9 | 4050 | } |
19bc17a9 RM |
4051 | |
4052 | /* You wonder about this amount of memory? This is only because some | |
4053 | users do not manage to address the array with unsigned values or | |
4054 | data types with range >= 256. '\200' would result in the array | |
4055 | index -128. To help these poor people we duplicate the entries for | |
4056 | 128 up to 255 below the entry for \0. */ | |
4057 | ctype->ctype_b = (char_class_t *) xcalloc (256 + 128, | |
4058 | sizeof (char_class_t)); | |
ef446144 UD |
4059 | ctype->ctype32_b = (char_class32_t *) |
4060 | xcalloc ((oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 256), | |
4061 | sizeof (char_class32_t)); | |
4062 | if (!oldstyle_tables) | |
4063 | ctype->class_3level = (struct iovec *) | |
4064 | xmalloc (ctype->nr_charclass * sizeof (struct iovec)); | |
19bc17a9 | 4065 | |
4a33c2f5 | 4066 | /* This is the array accessed using the multibyte string elements. */ |
4b10dd6c | 4067 | for (idx = 0; idx < 256; ++idx) |
4a33c2f5 | 4068 | ctype->ctype_b[128 + idx] = ctype->class256_collection[idx]; |
19bc17a9 | 4069 | |
75cd5204 RM |
4070 | /* Mirror first 127 entries. We must take care that entry -1 is not |
4071 | mirrored because EOF == -1. */ | |
4072 | for (idx = 0; idx < 127; ++idx) | |
19bc17a9 RM |
4073 | ctype->ctype_b[idx] = ctype->ctype_b[256 + idx]; |
4074 | ||
ef446144 UD |
4075 | if (oldstyle_tables) |
4076 | { | |
4077 | /* The 32 bit array contains all characters. */ | |
4078 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
4079 | ctype->ctype32_b[ctype->charnames[idx]] = ctype->class_collection[idx]; | |
4080 | } | |
4081 | else | |
4082 | { | |
4083 | /* The 32 bit array contains all characters < 0x100. */ | |
4084 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
4085 | if (ctype->charnames[idx] < 0x100) | |
4086 | ctype->ctype32_b[ctype->charnames[idx]] = ctype->class_collection[idx]; | |
4087 | } | |
4088 | ||
4089 | if (!oldstyle_tables) | |
4090 | { | |
4091 | size_t nr; | |
4092 | ||
4093 | for (nr = 0; nr < ctype->nr_charclass; nr++) | |
4094 | { | |
4095 | struct wctype_table t; | |
4096 | ||
4097 | t.p = 4; /* or: 5 */ | |
4098 | t.q = 7; /* or: 6 */ | |
4099 | wctype_table_init (&t); | |
4100 | ||
4101 | for (idx = 0; idx < ctype->class_collection_act; ++idx) | |
4102 | if (ctype->class_collection[idx] & _ISwbit (nr)) | |
4103 | wctype_table_add (&t, ctype->charnames[idx]); | |
4104 | ||
4105 | wctype_table_finalize (&t); | |
4106 | ||
4107 | if (verbose) | |
4108 | fprintf (stderr, _("%s: table for class \"%s\": %lu bytes\n"), | |
4109 | "LC_CTYPE", ctype->classnames[nr], | |
4110 | (unsigned long int) t.result_size); | |
4111 | ||
4112 | ctype->class_3level[nr].iov_base = t.result; | |
4113 | ctype->class_3level[nr].iov_len = t.result_size; | |
4114 | } | |
4115 | } | |
19bc17a9 RM |
4116 | |
4117 | /* Room for table of mappings. */ | |
49f2be5b UD |
4118 | ctype->map = (uint32_t **) xmalloc (2 * sizeof (uint32_t *)); |
4119 | ctype->map32 = (uint32_t **) xmalloc (ctype->map_collection_nr | |
ef446144 UD |
4120 | * sizeof (uint32_t *)); |
4121 | if (!oldstyle_tables) | |
4122 | ctype->map_3level = (struct iovec *) | |
4123 | xmalloc (ctype->map_collection_nr * sizeof (struct iovec)); | |
19bc17a9 RM |
4124 | |
4125 | /* Fill in all mappings. */ | |
49f2be5b | 4126 | for (idx = 0; idx < 2; ++idx) |
19bc17a9 RM |
4127 | { |
4128 | unsigned int idx2; | |
4129 | ||
4130 | /* Allocate table. */ | |
49f2be5b | 4131 | ctype->map[idx] = (uint32_t *) xmalloc ((256 + 128) * sizeof (uint32_t)); |
19bc17a9 RM |
4132 | |
4133 | /* Copy values from collection. */ | |
4b10dd6c | 4134 | for (idx2 = 0; idx2 < 256; ++idx2) |
4a33c2f5 | 4135 | ctype->map[idx][128 + idx2] = ctype->map256_collection[idx][idx2]; |
19bc17a9 | 4136 | |
75cd5204 RM |
4137 | /* Mirror first 127 entries. We must take care not to map entry |
4138 | -1 because EOF == -1. */ | |
4139 | for (idx2 = 0; idx2 < 127; ++idx2) | |
4a33c2f5 | 4140 | ctype->map[idx][idx2] = ctype->map[idx][256 + idx2]; |
19bc17a9 | 4141 | |
75cd5204 | 4142 | /* EOF must map to EOF. */ |
4a33c2f5 | 4143 | ctype->map[idx][127] = EOF; |
49f2be5b | 4144 | } |
a9c27b3e | 4145 | |
49f2be5b UD |
4146 | for (idx = 0; idx < ctype->map_collection_nr; ++idx) |
4147 | { | |
4148 | unsigned int idx2; | |
4149 | ||
4150 | /* Allocate table. */ | |
ef446144 UD |
4151 | ctype->map32[idx] = (uint32_t *) |
4152 | xmalloc ((oldstyle_tables ? ctype->plane_size * ctype->plane_cnt : 256) | |
4153 | * sizeof (uint32_t)); | |
49f2be5b UD |
4154 | |
4155 | /* Copy default value (identity mapping). */ | |
ef446144 UD |
4156 | if (oldstyle_tables) |
4157 | memcpy (ctype->map32[idx], ctype->names, | |
4158 | ctype->plane_size * ctype->plane_cnt * sizeof (uint32_t)); | |
4159 | else | |
4160 | for (idx2 = 0; idx2 < 256; ++idx2) | |
4161 | ctype->map32[idx][idx2] = idx2; | |
49f2be5b UD |
4162 | |
4163 | /* Copy values from collection. */ | |
4164 | for (idx2 = 0; idx2 < 256; ++idx2) | |
a9c27b3e | 4165 | if (ctype->map_collection[idx][idx2] != 0) |
f1d8b804 UD |
4166 | ctype->map32[idx][idx2] = ctype->map_collection[idx][idx2]; |
4167 | ||
ef446144 UD |
4168 | if (oldstyle_tables) |
4169 | while (idx2 < ctype->map_collection_act[idx]) | |
4170 | { | |
4171 | if (ctype->map_collection[idx][idx2] != 0) | |
4172 | ctype->map32[idx][ctype->charnames[idx2]] = | |
4173 | ctype->map_collection[idx][idx2]; | |
4174 | ++idx2; | |
4175 | } | |
4176 | } | |
4177 | ||
4178 | if (!oldstyle_tables) | |
4179 | { | |
4180 | size_t nr; | |
4181 | ||
4182 | for (nr = 0; nr < ctype->map_collection_nr; nr++) | |
b06c53e7 | 4183 | { |
ef446144 UD |
4184 | struct wctrans_table t; |
4185 | ||
4186 | t.p = 7; | |
4187 | t.q = 9; | |
4188 | wctrans_table_init (&t); | |
4189 | ||
4190 | for (idx = 0; idx < ctype->map_collection_act[nr]; ++idx) | |
4191 | if (ctype->map_collection[nr][idx] != 0) | |
4192 | wctrans_table_add (&t, ctype->charnames[idx], | |
4193 | ctype->map_collection[nr][idx]); | |
4194 | ||
4195 | wctrans_table_finalize (&t); | |
4196 | ||
4197 | if (verbose) | |
4198 | fprintf (stderr, _("%s: table for map \"%s\": %lu bytes\n"), | |
4199 | "LC_CTYPE", ctype->mapnames[nr], | |
4200 | (unsigned long int) t.result_size); | |
4201 | ||
4202 | ctype->map_3level[nr].iov_base = t.result; | |
4203 | ctype->map_3level[nr].iov_len = t.result_size; | |
b06c53e7 | 4204 | } |
19bc17a9 RM |
4205 | } |
4206 | ||
4207 | /* Extra array for class and map names. */ | |
4b10dd6c UD |
4208 | ctype->class_name_ptr = (uint32_t *) xmalloc (ctype->nr_charclass |
4209 | * sizeof (uint32_t)); | |
4210 | ctype->map_name_ptr = (uint32_t *) xmalloc (ctype->map_collection_nr | |
4211 | * sizeof (uint32_t)); | |
75cd5204 | 4212 | |
ef446144 UD |
4213 | if (oldstyle_tables) |
4214 | { | |
4215 | ctype->class_offset = 0; /* not really used */ | |
4216 | ctype->map_offset = 0; /* not really used */ | |
4217 | } | |
4218 | else | |
4219 | { | |
4220 | ctype->class_offset = _NL_ITEM_INDEX (_NL_NUM_LC_CTYPE); | |
4221 | ctype->map_offset = ctype->class_offset + ctype->nr_charclass; | |
4222 | } | |
5866b131 | 4223 | |
ef446144 UD |
4224 | /* Array for width information. Because the expected width are very |
4225 | small we use only one single byte. This saves space. */ | |
4226 | if (oldstyle_tables) | |
75cd5204 | 4227 | { |
ef446144 UD |
4228 | width_table_size = (ctype->plane_size * ctype->plane_cnt + 3) & ~3ul; |
4229 | ctype->width = (unsigned char *) xmalloc (width_table_size); | |
75cd5204 | 4230 | |
ef446144 UD |
4231 | /* Initialize with -1. */ |
4232 | memset (ctype->width, '\xff', width_table_size); | |
4233 | if (charmap->width_rules != NULL) | |
827ff758 | 4234 | { |
ef446144 | 4235 | size_t cnt; |
827ff758 | 4236 | |
ef446144 | 4237 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
75cd5204 | 4238 | { |
ef446144 UD |
4239 | unsigned char bytes[charmap->mb_cur_max]; |
4240 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
4241 | ||
4242 | /* We have the range of character for which the width is | |
4243 | specified described using byte sequences of the multibyte | |
4244 | charset. We have to convert this to UCS4 now. And we | |
4245 | cannot simply convert the beginning and the end of the | |
4246 | sequence, we have to iterate over the byte sequence and | |
4247 | convert it for every single character. */ | |
4248 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
4249 | ||
4250 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
4251 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
4252 | nbytes) <= 0) | |
827ff758 | 4253 | { |
ef446144 UD |
4254 | /* Find the UCS value for `bytes'. */ |
4255 | int inner; | |
4256 | uint32_t wch; | |
4257 | struct charseq *seq = | |
4258 | charmap_find_symbol (charmap, bytes, nbytes); | |
4259 | ||
4260 | if (seq == NULL) | |
4261 | wch = ILLEGAL_CHAR_VALUE; | |
4262 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
4263 | wch = seq->ucs4; | |
4264 | else | |
4265 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
4266 | strlen (seq->name)); | |
827ff758 | 4267 | |
ef446144 | 4268 | if (wch != ILLEGAL_CHAR_VALUE) |
0e16ecfa | 4269 | { |
ef446144 UD |
4270 | /* Store the value. */ |
4271 | size_t nr = wch % ctype->plane_size; | |
4272 | size_t depth = 0; | |
4273 | ||
4274 | while (ctype->names[nr + depth * ctype->plane_size] != wch) | |
4275 | { | |
4276 | ++depth; | |
4277 | assert (depth < ctype->plane_cnt); | |
4278 | } | |
4279 | ||
4280 | ctype->width[nr + depth * ctype->plane_size] | |
4281 | = charmap->width_rules[cnt].width; | |
0e16ecfa | 4282 | } |
827ff758 | 4283 | |
ef446144 UD |
4284 | /* "Increment" the bytes sequence. */ |
4285 | inner = nbytes - 1; | |
4286 | while (inner >= 0 && bytes[inner] == 0xff) | |
4287 | --inner; | |
4288 | ||
4289 | if (inner < 0) | |
4290 | { | |
4291 | /* We have to extend the byte sequence. */ | |
4292 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
4293 | break; | |
4294 | ||
4295 | bytes[0] = 1; | |
4296 | memset (&bytes[1], 0, nbytes); | |
4297 | ++nbytes; | |
4298 | } | |
4299 | else | |
4300 | { | |
4301 | ++bytes[inner]; | |
4302 | while (++inner < nbytes) | |
4303 | bytes[inner] = 0; | |
4304 | } | |
827ff758 | 4305 | } |
ef446144 UD |
4306 | } |
4307 | } | |
827ff758 | 4308 | |
ef446144 UD |
4309 | /* Now set all the other characters of the character set to the |
4310 | default width. */ | |
4311 | curs = NULL; | |
4312 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
4313 | { | |
4314 | struct charseq *data = (struct charseq *) vdata; | |
4315 | size_t nr; | |
4316 | size_t depth; | |
75cd5204 | 4317 | |
ef446144 UD |
4318 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) |
4319 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
4320 | data->name, len); | |
75cd5204 | 4321 | |
ef446144 UD |
4322 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) |
4323 | { | |
4324 | nr = data->ucs4 % ctype->plane_size; | |
4325 | depth = 0; | |
4326 | ||
4327 | while (ctype->names[nr + depth * ctype->plane_size] != data->ucs4) | |
827ff758 | 4328 | { |
ef446144 UD |
4329 | ++depth; |
4330 | assert (depth < ctype->plane_cnt); | |
827ff758 | 4331 | } |
ef446144 UD |
4332 | |
4333 | if (ctype->width[nr + depth * ctype->plane_size] | |
4334 | == (unsigned char) '\xff') | |
4335 | ctype->width[nr + depth * ctype->plane_size] = | |
4336 | charmap->width_default; | |
75cd5204 | 4337 | } |
827ff758 | 4338 | } |
75cd5204 | 4339 | } |
ef446144 | 4340 | else |
0e16ecfa | 4341 | { |
ef446144 | 4342 | struct wcwidth_table t; |
0e16ecfa | 4343 | |
ef446144 UD |
4344 | t.p = 7; |
4345 | t.q = 9; | |
4346 | wcwidth_table_init (&t); | |
0e16ecfa | 4347 | |
ef446144 UD |
4348 | /* First set all the characters of the character set to the default width. */ |
4349 | curs = NULL; | |
4350 | while (iterate_table (&charmap->char_table, &curs, &key, &len, &vdata) == 0) | |
4351 | { | |
4352 | struct charseq *data = (struct charseq *) vdata; | |
4353 | ||
4354 | if (data->ucs4 == UNINITIALIZED_CHAR_VALUE) | |
4355 | data->ucs4 = repertoire_find_value (ctype->repertoire, | |
4356 | data->name, len); | |
4357 | ||
4358 | if (data->ucs4 != ILLEGAL_CHAR_VALUE) | |
4359 | wcwidth_table_add (&t, data->ucs4, charmap->width_default); | |
4360 | } | |
4361 | ||
4362 | /* Now add the explicitly specified widths. */ | |
4363 | if (charmap->width_rules != NULL) | |
0e16ecfa | 4364 | { |
ef446144 | 4365 | size_t cnt; |
0e16ecfa | 4366 | |
ef446144 | 4367 | for (cnt = 0; cnt < charmap->nwidth_rules; ++cnt) |
0e16ecfa | 4368 | { |
ef446144 UD |
4369 | unsigned char bytes[charmap->mb_cur_max]; |
4370 | int nbytes = charmap->width_rules[cnt].from->nbytes; | |
4371 | ||
4372 | /* We have the range of character for which the width is | |
4373 | specified described using byte sequences of the multibyte | |
4374 | charset. We have to convert this to UCS4 now. And we | |
4375 | cannot simply convert the beginning and the end of the | |
4376 | sequence, we have to iterate over the byte sequence and | |
4377 | convert it for every single character. */ | |
4378 | memcpy (bytes, charmap->width_rules[cnt].from->bytes, nbytes); | |
4379 | ||
4380 | while (nbytes < charmap->width_rules[cnt].to->nbytes | |
4381 | || memcmp (bytes, charmap->width_rules[cnt].to->bytes, | |
4382 | nbytes) <= 0) | |
4383 | { | |
4384 | /* Find the UCS value for `bytes'. */ | |
4385 | int inner; | |
4386 | uint32_t wch; | |
4387 | struct charseq *seq = | |
4388 | charmap_find_symbol (charmap, bytes, nbytes); | |
4389 | ||
4390 | if (seq == NULL) | |
4391 | wch = ILLEGAL_CHAR_VALUE; | |
4392 | else if (seq->ucs4 != UNINITIALIZED_CHAR_VALUE) | |
4393 | wch = seq->ucs4; | |
4394 | else | |
4395 | wch = repertoire_find_value (ctype->repertoire, seq->name, | |
4396 | strlen (seq->name)); | |
4397 | ||
4398 | if (wch != ILLEGAL_CHAR_VALUE) | |
4399 | /* Store the value. */ | |
4400 | wcwidth_table_add (&t, wch, charmap->width_rules[cnt].width); | |
4401 | ||
4402 | /* "Increment" the bytes sequence. */ | |
4403 | inner = nbytes - 1; | |
4404 | while (inner >= 0 && bytes[inner] == 0xff) | |
4405 | --inner; | |
0e16ecfa | 4406 | |
ef446144 UD |
4407 | if (inner < 0) |
4408 | { | |
4409 | /* We have to extend the byte sequence. */ | |
4410 | if (nbytes >= charmap->width_rules[cnt].to->nbytes) | |
4411 | break; | |
4412 | ||
4413 | bytes[0] = 1; | |
4414 | memset (&bytes[1], 0, nbytes); | |
4415 | ++nbytes; | |
4416 | } | |
4417 | else | |
4418 | { | |
4419 | ++bytes[inner]; | |
4420 | while (++inner < nbytes) | |
4421 | bytes[inner] = 0; | |
4422 | } | |
4423 | } | |
4424 | } | |
0e16ecfa | 4425 | } |
ef446144 UD |
4426 | |
4427 | wcwidth_table_finalize (&t); | |
4428 | ||
4429 | if (verbose) | |
4430 | fprintf (stderr, _("%s: table for width: %lu bytes\n"), | |
4431 | "LC_CTYPE", (unsigned long int) t.result_size); | |
4432 | ||
4433 | ctype->width_3level.iov_base = t.result; | |
4434 | ctype->width_3level.iov_len = t.result_size; | |
0e16ecfa UD |
4435 | } |
4436 | ||
4b10dd6c UD |
4437 | /* Set MB_CUR_MAX. */ |
4438 | ctype->mb_cur_max = charmap->mb_cur_max; | |
6990326c | 4439 | |
4b10dd6c UD |
4440 | /* Now determine the table for the transliteration information. |
4441 | ||
4442 | XXX It is not yet clear to me whether it is worth implementing a | |
4443 | complicated algorithm which uses a hash table to locate the entries. | |
4444 | For now I'll use a simple array which can be searching using binary | |
4445 | search. */ | |
4446 | if (ctype->translit_copy_locale != NULL) | |
4447 | { | |
4448 | /* Fold in the transliteration information from the locale mentioned | |
4449 | in the `include' statement. */ | |
4450 | struct locale_ctype_t *here = ctype; | |
4451 | ||
4452 | do | |
4453 | { | |
4454 | struct localedef_t *other = find_locale (LC_CTYPE, | |
4455 | here->translit_copy_locale, | |
4456 | repertoire->name, charmap); | |
4457 | ||
4458 | if (other == NULL) | |
4459 | { | |
4460 | error (0, 0, _("\ | |
4461 | %s: transliteration data from locale `%s' not available"), | |
4462 | "LC_CTYPE", here->translit_copy_locale); | |
4463 | break; | |
4464 | } | |
4465 | ||
4466 | here = other->categories[LC_CTYPE].ctype; | |
4467 | ||
4468 | /* Enqueue the information if necessary. */ | |
4469 | if (here->translit != NULL) | |
4470 | { | |
4471 | struct translit_t *endp = here->translit; | |
4472 | while (endp->next != NULL) | |
4473 | endp = endp->next; | |
4474 | ||
4475 | endp->next = ctype->translit; | |
4476 | ctype->translit = here->translit; | |
4477 | } | |
4478 | } | |
4479 | while (here->translit_copy_locale != NULL); | |
4480 | } | |
4481 | ||
4482 | if (ctype->translit != NULL) | |
4483 | { | |
4484 | /* First count how many entries we have. This is the upper limit | |
4485 | since some entries from the included files might be overwritten. */ | |
4486 | size_t number = 0; | |
4487 | size_t cnt; | |
4488 | struct translit_t *runp = ctype->translit; | |
4489 | struct translit_t **sorted; | |
4490 | size_t from_len, to_len; | |
4491 | ||
4492 | while (runp != NULL) | |
4493 | { | |
4494 | ++number; | |
4495 | runp = runp->next; | |
4496 | } | |
4497 | ||
4498 | /* Next we allocate an array large enough and fill in the values. */ | |
a9c27b3e UD |
4499 | sorted = (struct translit_t **) alloca (number |
4500 | * sizeof (struct translit_t **)); | |
4b10dd6c UD |
4501 | runp = ctype->translit; |
4502 | number = 0; | |
4503 | do | |
4504 | { | |
4505 | /* Search for the place where to insert this string. | |
4506 | XXX Better use a real sorting algorithm later. */ | |
4507 | size_t idx = 0; | |
4508 | int replace = 0; | |
4509 | ||
4510 | while (idx < number) | |
4511 | { | |
4512 | int res = wcscmp ((const wchar_t *) sorted[idx]->from, | |
4513 | (const wchar_t *) runp->from); | |
4514 | if (res == 0) | |
4515 | { | |
4516 | replace = 1; | |
4517 | break; | |
4518 | } | |
4519 | if (res > 0) | |
4520 | break; | |
4521 | ++idx; | |
4522 | } | |
4523 | ||
4524 | if (replace) | |
4525 | sorted[idx] = runp; | |
4526 | else | |
4527 | { | |
4528 | memmove (&sorted[idx + 1], &sorted[idx], | |
4529 | (number - idx) * sizeof (struct translit_t *)); | |
4530 | sorted[idx] = runp; | |
4531 | ++number; | |
4532 | } | |
4533 | ||
4534 | runp = runp->next; | |
4535 | } | |
4536 | while (runp != NULL); | |
4537 | ||
4538 | /* The next step is putting all the possible transliteration | |
4539 | strings in one memory block so that we can write it out. | |
4540 | We need several different blocks: | |
9ca23765 | 4541 | - index to the from-string array |
4b10dd6c UD |
4542 | - from-string array |
4543 | - index to the to-string array | |
4544 | - to-string array. | |
4b10dd6c UD |
4545 | */ |
4546 | from_len = to_len = 0; | |
4547 | for (cnt = 0; cnt < number; ++cnt) | |
4548 | { | |
4549 | struct translit_to_t *srunp; | |
4550 | from_len += wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4551 | srunp = sorted[cnt]->to; | |
4552 | while (srunp != NULL) | |
4553 | { | |
4554 | to_len += wcslen ((const wchar_t *) srunp->str) + 1; | |
4555 | srunp = srunp->next; | |
4556 | } | |
4557 | /* Plus one for the extra NUL character marking the end of | |
4558 | the list for the current entry. */ | |
4559 | ++to_len; | |
4560 | } | |
4561 | ||
4562 | /* We can allocate the arrays for the results. */ | |
4a33c2f5 UD |
4563 | ctype->translit_from_idx = xmalloc (number * sizeof (uint32_t)); |
4564 | ctype->translit_from_tbl = xmalloc (from_len * sizeof (uint32_t)); | |
4565 | ctype->translit_to_idx = xmalloc (number * sizeof (uint32_t)); | |
4566 | ctype->translit_to_tbl = xmalloc (to_len * sizeof (uint32_t)); | |
4b10dd6c UD |
4567 | |
4568 | from_len = 0; | |
4569 | to_len = 0; | |
4570 | for (cnt = 0; cnt < number; ++cnt) | |
4571 | { | |
4572 | size_t len; | |
4573 | struct translit_to_t *srunp; | |
4574 | ||
4a33c2f5 UD |
4575 | ctype->translit_from_idx[cnt] = from_len; |
4576 | ctype->translit_to_idx[cnt] = to_len; | |
4b10dd6c UD |
4577 | |
4578 | len = wcslen ((const wchar_t *) sorted[cnt]->from) + 1; | |
4a33c2f5 | 4579 | wmemcpy ((wchar_t *) &ctype->translit_from_tbl[from_len], |
4b10dd6c UD |
4580 | (const wchar_t *) sorted[cnt]->from, len); |
4581 | from_len += len; | |
4582 | ||
4a33c2f5 | 4583 | ctype->translit_to_idx[cnt] = to_len; |
4b10dd6c UD |
4584 | srunp = sorted[cnt]->to; |
4585 | while (srunp != NULL) | |
4586 | { | |
4587 | len = wcslen ((const wchar_t *) srunp->str) + 1; | |
4a33c2f5 | 4588 | wmemcpy ((wchar_t *) &ctype->translit_to_tbl[to_len], |
4b10dd6c UD |
4589 | (const wchar_t *) srunp->str, len); |
4590 | to_len += len; | |
4591 | srunp = srunp->next; | |
4592 | } | |
4a33c2f5 | 4593 | ctype->translit_to_tbl[to_len++] = L'\0'; |
4b10dd6c | 4594 | } |
4b10dd6c UD |
4595 | |
4596 | /* Store the information about the length. */ | |
04fbc779 | 4597 | ctype->translit_idx_size = number; |
4b10dd6c UD |
4598 | ctype->translit_from_tbl_size = from_len * sizeof (uint32_t); |
4599 | ctype->translit_to_tbl_size = to_len * sizeof (uint32_t); | |
4600 | } | |
4601 | else | |
4602 | { | |
4603 | /* Provide some dummy pointers since we have nothing to write out. */ | |
4604 | static uint32_t no_str = { 0 }; | |
4605 | ||
4a33c2f5 UD |
4606 | ctype->translit_from_idx = &no_str; |
4607 | ctype->translit_from_tbl = &no_str; | |
4608 | ctype->translit_to_tbl = &no_str; | |
4b10dd6c UD |
4609 | ctype->translit_idx_size = 0; |
4610 | ctype->translit_from_tbl_size = 0; | |
4611 | ctype->translit_to_tbl_size = 0; | |
4612 | } | |
19bc17a9 | 4613 | } |