Commit | Line | Data |
---|---|---|
3263d5a2 | 1 | /* Basic character set support. |
73b0cd50 | 2 | Copyright (C) 2001-2011 Free Software Foundation, Inc. |
7976eda0 | 3 | Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, |
5df4f04c | 4 | 2005, 2006, 2007, 2008, 2009, 2010, 2011 |
ce03bf76 KH |
5 | National Institute of Advanced Industrial Science and Technology (AIST) |
6 | Registration Number H14PRO021 | |
4ed46869 | 7 | |
327719ee | 8 | Copyright (C) 2003, 2004 |
3263d5a2 KH |
9 | National Institute of Advanced Industrial Science and Technology (AIST) |
10 | Registration Number H13PRO009 | |
4ed46869 | 11 | |
369314dc KH |
12 | This file is part of GNU Emacs. |
13 | ||
9ec0b715 | 14 | GNU Emacs is free software: you can redistribute it and/or modify |
369314dc | 15 | it under the terms of the GNU General Public License as published by |
9ec0b715 GM |
16 | the Free Software Foundation, either version 3 of the License, or |
17 | (at your option) any later version. | |
4ed46869 | 18 | |
369314dc KH |
19 | GNU Emacs is distributed in the hope that it will be useful, |
20 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
21 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
22 | GNU General Public License for more details. | |
4ed46869 | 23 | |
369314dc | 24 | You should have received a copy of the GNU General Public License |
9ec0b715 | 25 | along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>. */ |
4ed46869 | 26 | |
68c45bf0 | 27 | #include <config.h> |
68c45bf0 | 28 | |
4ed46869 | 29 | #include <stdio.h> |
3263d5a2 KH |
30 | #include <unistd.h> |
31 | #include <ctype.h> | |
726929c4 | 32 | #include <limits.h> |
4ed46869 | 33 | #include <sys/types.h> |
d7306fe6 | 34 | #include <setjmp.h> |
4ed46869 | 35 | #include "lisp.h" |
3263d5a2 | 36 | #include "character.h" |
4ed46869 KH |
37 | #include "charset.h" |
38 | #include "coding.h" | |
fc6b09bf | 39 | #include "disptab.h" |
3263d5a2 | 40 | #include "buffer.h" |
4ed46869 | 41 | |
04c2f2c5 | 42 | /*** GENERAL NOTES on CODED CHARACTER SETS (CHARSETS) *** |
4ed46869 | 43 | |
3263d5a2 | 44 | A coded character set ("charset" hereafter) is a meaningful |
04c2f2c5 | 45 | collection (i.e. language, culture, functionality, etc.) of |
3263d5a2 | 46 | characters. Emacs handles multiple charsets at once. In Emacs Lisp |
04c2f2c5 DL |
47 | code, a charset is represented by a symbol. In C code, a charset is |
48 | represented by its ID number or by a pointer to a struct charset. | |
4ed46869 | 49 | |
3263d5a2 KH |
50 | The actual information about each charset is stored in two places. |
51 | Lispy information is stored in the hash table Vcharset_hash_table as | |
52 | a vector (charset attributes). The other information is stored in | |
04c2f2c5 | 53 | charset_table as a struct charset. |
4ed46869 | 54 | |
3263d5a2 | 55 | */ |
4ed46869 | 56 | |
3263d5a2 KH |
57 | /* Hash table that contains attributes of each charset. Keys are |
58 | charset symbols, and values are vectors of charset attributes. */ | |
59 | Lisp_Object Vcharset_hash_table; | |
4ed46869 | 60 | |
3263d5a2 KH |
61 | /* Table of struct charset. */ |
62 | struct charset *charset_table; | |
4ed46869 | 63 | |
3263d5a2 | 64 | static int charset_table_size; |
5af5dd92 | 65 | static int charset_table_used; |
4ed46869 | 66 | |
3263d5a2 | 67 | Lisp_Object Qcharsetp; |
4ed46869 | 68 | |
3263d5a2 KH |
69 | /* Special charset symbols. */ |
70 | Lisp_Object Qascii; | |
955cbe7b PE |
71 | static Lisp_Object Qeight_bit; |
72 | static Lisp_Object Qiso_8859_1; | |
73 | static Lisp_Object Qunicode; | |
74 | static Lisp_Object Qemacs; | |
4ed46869 | 75 | |
3263d5a2 KH |
76 | /* The corresponding charsets. */ |
77 | int charset_ascii; | |
2fe1edd1 | 78 | int charset_eight_bit; |
a2cb4e63 | 79 | static int charset_iso_8859_1; |
3263d5a2 | 80 | int charset_unicode; |
a2cb4e63 | 81 | static int charset_emacs; |
b0e3cf2b | 82 | |
7c7dceee KH |
83 | /* The other special charsets. */ |
84 | int charset_jisx0201_roman; | |
85 | int charset_jisx0208_1978; | |
86 | int charset_jisx0208; | |
d32320c4 | 87 | int charset_ksc5601; |
c1a08b4c | 88 | |
3263d5a2 | 89 | /* Value of charset attribute `charset-iso-plane'. */ |
955cbe7b | 90 | static Lisp_Object Qgl, Qgr; |
c1a08b4c | 91 | |
d1a04588 KH |
92 | /* Charset of unibyte characters. */ |
93 | int charset_unibyte; | |
4ed46869 | 94 | |
3263d5a2 KH |
95 | /* List of charsets ordered by the priority. */ |
96 | Lisp_Object Vcharset_ordered_list; | |
4ed46869 | 97 | |
6a9c90ec KH |
98 | /* Sub-list of Vcharset_ordered_list that contains all non-preferred |
99 | charsets. */ | |
100 | Lisp_Object Vcharset_non_preferred_head; | |
101 | ||
dbbb237d | 102 | /* Incremented everytime we change Vcharset_ordered_list. This is |
64165ae2 | 103 | unsigned short so that it fits in Lisp_Int and never matches |
dbbb237d KH |
104 | -1. */ |
105 | unsigned short charset_ordered_list_tick; | |
4ed46869 | 106 | |
3263d5a2 KH |
107 | /* List of iso-2022 charsets. */ |
108 | Lisp_Object Viso_2022_charset_list; | |
35e623fb | 109 | |
3263d5a2 KH |
110 | /* List of emacs-mule charsets. */ |
111 | Lisp_Object Vemacs_mule_charset_list; | |
112 | ||
b84ae584 | 113 | int emacs_mule_charset[256]; |
4ed46869 KH |
114 | |
115 | /* Mapping table from ISO2022's charset (specified by DIMENSION, | |
116 | CHARS, and FINAL-CHAR) to Emacs' charset. */ | |
3263d5a2 KH |
117 | int iso_charset_table[ISO_MAX_DIMENSION][ISO_MAX_CHARS][ISO_MAX_FINAL]; |
118 | ||
69f8de5b KH |
119 | #define CODE_POINT_TO_INDEX(charset, code) \ |
120 | ((charset)->code_linear_p \ | |
121 | ? (code) - (charset)->min_code \ | |
122 | : (((charset)->code_space_mask[(code) >> 24] & 0x8) \ | |
123 | && ((charset)->code_space_mask[((code) >> 16) & 0xFF] & 0x4) \ | |
124 | && ((charset)->code_space_mask[((code) >> 8) & 0xFF] & 0x2) \ | |
125 | && ((charset)->code_space_mask[(code) & 0xFF] & 0x1)) \ | |
126 | ? (((((code) >> 24) - (charset)->code_space[12]) \ | |
127 | * (charset)->code_space[11]) \ | |
128 | + (((((code) >> 16) & 0xFF) - (charset)->code_space[8]) \ | |
129 | * (charset)->code_space[7]) \ | |
130 | + (((((code) >> 8) & 0xFF) - (charset)->code_space[4]) \ | |
131 | * (charset)->code_space[3]) \ | |
820ee249 KH |
132 | + (((code) & 0xFF) - (charset)->code_space[0]) \ |
133 | - ((charset)->char_index_offset)) \ | |
3263d5a2 KH |
134 | : -1) |
135 | ||
136 | ||
137 | /* Convert the character index IDX to code-point CODE for CHARSET. | |
138 | It is assumed that IDX is in a valid range. */ | |
139 | ||
820ee249 KH |
140 | #define INDEX_TO_CODE_POINT(charset, idx) \ |
141 | ((charset)->code_linear_p \ | |
142 | ? (idx) + (charset)->min_code \ | |
143 | : (idx += (charset)->char_index_offset, \ | |
144 | (((charset)->code_space[0] + (idx) % (charset)->code_space[2]) \ | |
145 | | (((charset)->code_space[4] \ | |
146 | + ((idx) / (charset)->code_space[3] % (charset)->code_space[6])) \ | |
147 | << 8) \ | |
148 | | (((charset)->code_space[8] \ | |
149 | + ((idx) / (charset)->code_space[7] % (charset)->code_space[10])) \ | |
150 | << 16) \ | |
151 | | (((charset)->code_space[12] + ((idx) / (charset)->code_space[11])) \ | |
152 | << 24)))) | |
4cf9710d | 153 | |
ecca2aad KH |
154 | /* Structure to hold mapping tables for a charset. Used by temacs |
155 | invoked for dumping. */ | |
8a73a704 | 156 | |
ecca2aad KH |
157 | static struct |
158 | { | |
159 | /* The current charset for which the following tables are setup. */ | |
160 | struct charset *current; | |
161 | ||
162 | /* 1 iff the following table is used for encoder. */ | |
163 | short for_encoder; | |
164 | ||
165 | /* When the following table is used for encoding, mininum and | |
166 | maxinum character of the current charset. */ | |
167 | int min_char, max_char; | |
168 | ||
169 | /* A Unicode character correspoinding to the code indice 0 (i.e. the | |
170 | minimum code-point) of the current charset, or -1 if the code | |
171 | indice 0 is not a Unicode character. This is checked when | |
172 | table.encoder[CHAR] is zero. */ | |
173 | int zero_index_char; | |
174 | ||
175 | union { | |
176 | /* Table mapping code-indices (not code-points) of the current | |
177 | charset to Unicode characters. If decoder[CHAR] is -1, CHAR | |
178 | doesn't belong to the current charset. */ | |
179 | int decoder[0x10000]; | |
180 | /* Table mapping Unicode characters to code-indices of the current | |
181 | charset. The first 0x10000 elements are for BMP (0..0xFFFF), | |
182 | and the last 0x10000 are for SMP (0x10000..0x1FFFF) or SIP | |
183 | (0x20000..0x2FFFF). Note that there is no charset map that | |
184 | uses both SMP and SIP. */ | |
185 | unsigned short encoder[0x20000]; | |
186 | } table; | |
187 | } *temp_charset_work; | |
188 | ||
189 | #define SET_TEMP_CHARSET_WORK_ENCODER(C, CODE) \ | |
190 | do { \ | |
191 | if ((CODE) == 0) \ | |
192 | temp_charset_work->zero_index_char = (C); \ | |
193 | else if ((C) < 0x20000) \ | |
194 | temp_charset_work->table.encoder[(C)] = (CODE); \ | |
195 | else \ | |
196 | temp_charset_work->table.encoder[(C) - 0x10000] = (CODE); \ | |
197 | } while (0) | |
198 | ||
199 | #define GET_TEMP_CHARSET_WORK_ENCODER(C) \ | |
200 | ((C) == temp_charset_work->zero_index_char ? 0 \ | |
201 | : (C) < 0x20000 ? (temp_charset_work->table.encoder[(C)] \ | |
202 | ? (int) temp_charset_work->table.encoder[(C)] : -1) \ | |
203 | : temp_charset_work->table.encoder[(C) - 0x10000] \ | |
204 | ? temp_charset_work->table.encoder[(C) - 0x10000] : -1) | |
205 | ||
206 | #define SET_TEMP_CHARSET_WORK_DECODER(C, CODE) \ | |
207 | (temp_charset_work->table.decoder[(CODE)] = (C)) | |
208 | ||
209 | #define GET_TEMP_CHARSET_WORK_DECODER(CODE) \ | |
210 | (temp_charset_work->table.decoder[(CODE)]) | |
046b1f03 | 211 | \f |
93bcb785 | 212 | |
e9ce014c KH |
213 | /* Set to 1 to warn that a charset map is loaded and thus a buffer |
214 | text and a string data may be relocated. */ | |
3263d5a2 | 215 | int charset_map_loaded; |
4ed46869 | 216 | |
e9ce014c | 217 | struct charset_map_entries |
4ed46869 | 218 | { |
e9ce014c KH |
219 | struct { |
220 | unsigned from, to; | |
221 | int c; | |
222 | } entry[0x10000]; | |
223 | struct charset_map_entries *next; | |
224 | }; | |
225 | ||
ecca2aad KH |
226 | /* Load the mapping information of CHARSET from ENTRIES for |
227 | initializing (CONTROL_FLAG == 0), decoding (CONTROL_FLAG == 1), and | |
228 | encoding (CONTROL_FLAG == 2). | |
229 | ||
230 | If CONTROL_FLAG is 0, setup CHARSET->min_char, CHARSET->max_char, | |
231 | and CHARSET->fast_map. | |
232 | ||
233 | If CONTROL_FLAG is 1, setup the following tables according to | |
234 | CHARSET->method and inhibit_load_charset_map. | |
4cf9710d | 235 | |
ecca2aad KH |
236 | CHARSET->method | inhibit_lcm == 0 | inhibit_lcm == 1 |
237 | ----------------------+--------------------+--------------------------- | |
238 | CHARSET_METHOD_MAP | CHARSET->decoder | temp_charset_work->decoder | |
239 | ----------------------+--------------------+--------------------------- | |
240 | CHARSET_METHOD_OFFSET | Vchar_unify_table | temp_charset_work->decoder | |
8a73a704 | 241 | |
ecca2aad | 242 | If CONTROL_FLAG is 2, setup the following tables. |
93bcb785 | 243 | |
ecca2aad KH |
244 | CHARSET->method | inhibit_lcm == 0 | inhibit_lcm == 1 |
245 | ----------------------+--------------------+--------------------------- | |
246 | CHARSET_METHOD_MAP | CHARSET->encoder | temp_charset_work->encoder | |
247 | ----------------------+--------------------+-------------------------- | |
248 | CHARSET_METHOD_OFFSET | CHARSET->deunifier | temp_charset_work->encoder | |
249 | */ | |
4ed46869 | 250 | |
3263d5a2 | 251 | static void |
971de7fb | 252 | load_charset_map (struct charset *charset, struct charset_map_entries *entries, int n_entries, int control_flag) |
4ed46869 | 253 | { |
89ef49df | 254 | Lisp_Object vec IF_LINT (= Qnil), table IF_LINT (= Qnil); |
3263d5a2 KH |
255 | unsigned max_code = CHARSET_MAX_CODE (charset); |
256 | int ascii_compatible_p = charset->ascii_compatible_p; | |
257 | int min_char, max_char, nonascii_min_char; | |
3263d5a2 | 258 | int i; |
3263d5a2 | 259 | unsigned char *fast_map = charset->fast_map; |
99529c2c | 260 | |
e9ce014c KH |
261 | if (n_entries <= 0) |
262 | return; | |
263 | ||
ecca2aad | 264 | if (control_flag) |
8ac5a9cc | 265 | { |
ecca2aad KH |
266 | if (! inhibit_load_charset_map) |
267 | { | |
268 | if (control_flag == 1) | |
269 | { | |
270 | if (charset->method == CHARSET_METHOD_MAP) | |
271 | { | |
272 | int n = CODE_POINT_TO_INDEX (charset, max_code) + 1; | |
6662e69b | 273 | |
ecca2aad KH |
274 | vec = CHARSET_DECODER (charset) |
275 | = Fmake_vector (make_number (n), make_number (-1)); | |
276 | } | |
277 | else | |
278 | { | |
279 | char_table_set_range (Vchar_unify_table, | |
280 | charset->min_char, charset->max_char, | |
281 | Qnil); | |
282 | } | |
283 | } | |
284 | else | |
285 | { | |
286 | table = Fmake_char_table (Qnil, Qnil); | |
287 | if (charset->method == CHARSET_METHOD_MAP) | |
288 | CHARSET_ENCODER (charset) = table; | |
289 | else | |
290 | CHARSET_DEUNIFIER (charset) = table; | |
291 | } | |
292 | } | |
293 | else | |
294 | { | |
295 | if (! temp_charset_work) | |
546961a9 | 296 | temp_charset_work = xmalloc (sizeof (*temp_charset_work)); |
ecca2aad KH |
297 | if (control_flag == 1) |
298 | { | |
299 | memset (temp_charset_work->table.decoder, -1, | |
300 | sizeof (int) * 0x10000); | |
ecca2aad KH |
301 | } |
302 | else | |
303 | { | |
304 | memset (temp_charset_work->table.encoder, 0, | |
305 | sizeof (unsigned short) * 0x20000); | |
306 | temp_charset_work->zero_index_char = -1; | |
307 | } | |
308 | temp_charset_work->current = charset; | |
309 | temp_charset_work->for_encoder = (control_flag == 2); | |
310 | control_flag += 2; | |
311 | } | |
3263d5a2 | 312 | charset_map_loaded = 1; |
2e344af3 | 313 | } |
6662e69b | 314 | |
e9ce014c | 315 | min_char = max_char = entries->entry[0].c; |
3263d5a2 | 316 | nonascii_min_char = MAX_CHAR; |
e9ce014c | 317 | for (i = 0; i < n_entries; i++) |
2e344af3 | 318 | { |
e9ce014c | 319 | unsigned from, to; |
3befa583 | 320 | int from_index, to_index, lim_index; |
3b4f4446 | 321 | int from_c, to_c; |
e9ce014c | 322 | int idx = i % 0x10000; |
3263d5a2 | 323 | |
e9ce014c KH |
324 | if (i > 0 && idx == 0) |
325 | entries = entries->next; | |
326 | from = entries->entry[idx].from; | |
327 | to = entries->entry[idx].to; | |
3b4f4446 KH |
328 | from_c = entries->entry[idx].c; |
329 | from_index = CODE_POINT_TO_INDEX (charset, from); | |
330 | if (from == to) | |
6662e69b | 331 | { |
3b4f4446 KH |
332 | to_index = from_index; |
333 | to_c = from_c; | |
6662e69b | 334 | } |
3b4f4446 | 335 | else |
6662e69b | 336 | { |
3b4f4446 KH |
337 | to_index = CODE_POINT_TO_INDEX (charset, to); |
338 | to_c = from_c + (to_index - from_index); | |
6662e69b | 339 | } |
3b4f4446 KH |
340 | if (from_index < 0 || to_index < 0) |
341 | continue; | |
3befa583 | 342 | lim_index = to_index + 1; |
3263d5a2 | 343 | |
ecca2aad KH |
344 | if (to_c > max_char) |
345 | max_char = to_c; | |
346 | else if (from_c < min_char) | |
347 | min_char = from_c; | |
6662e69b | 348 | |
ecca2aad KH |
349 | if (control_flag == 1) |
350 | { | |
351 | if (charset->method == CHARSET_METHOD_MAP) | |
3befa583 | 352 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
353 | ASET (vec, from_index, make_number (from_c)); |
354 | else | |
3befa583 | 355 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
356 | CHAR_TABLE_SET (Vchar_unify_table, |
357 | CHARSET_CODE_OFFSET (charset) + from_index, | |
358 | make_number (from_c)); | |
359 | } | |
360 | else if (control_flag == 2) | |
361 | { | |
362 | if (charset->method == CHARSET_METHOD_MAP | |
363 | && CHARSET_COMPACT_CODES_P (charset)) | |
3befa583 | 364 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
365 | { |
366 | unsigned code = INDEX_TO_CODE_POINT (charset, from_index); | |
367 | ||
368 | if (NILP (CHAR_TABLE_REF (table, from_c))) | |
369 | CHAR_TABLE_SET (table, from_c, make_number (code)); | |
370 | } | |
371 | else | |
3befa583 | 372 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
373 | { |
374 | if (NILP (CHAR_TABLE_REF (table, from_c))) | |
375 | CHAR_TABLE_SET (table, from_c, make_number (from_index)); | |
376 | } | |
377 | } | |
378 | else if (control_flag == 3) | |
3befa583 | 379 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
380 | SET_TEMP_CHARSET_WORK_DECODER (from_c, from_index); |
381 | else if (control_flag == 4) | |
3befa583 | 382 | for (; from_index < lim_index; from_index++, from_c++) |
ecca2aad KH |
383 | SET_TEMP_CHARSET_WORK_ENCODER (from_c, from_index); |
384 | else /* control_flag == 0 */ | |
385 | { | |
3b4f4446 KH |
386 | if (ascii_compatible_p) |
387 | { | |
388 | if (! ASCII_BYTE_P (from_c)) | |
389 | { | |
390 | if (from_c < nonascii_min_char) | |
391 | nonascii_min_char = from_c; | |
392 | } | |
393 | else if (! ASCII_BYTE_P (to_c)) | |
394 | { | |
395 | nonascii_min_char = 0x80; | |
396 | } | |
397 | } | |
177c0ea7 | 398 | |
ecca2aad KH |
399 | for (; from_c <= to_c; from_c++) |
400 | CHARSET_FAST_MAP_SET (from_c, fast_map); | |
2e344af3 | 401 | } |
8ac5a9cc | 402 | } |
3263d5a2 | 403 | |
ecca2aad | 404 | if (control_flag == 0) |
4ed46869 | 405 | { |
3263d5a2 KH |
406 | CHARSET_MIN_CHAR (charset) = (ascii_compatible_p |
407 | ? nonascii_min_char : min_char); | |
408 | CHARSET_MAX_CHAR (charset) = max_char; | |
4ed46869 | 409 | } |
ecca2aad KH |
410 | else if (control_flag == 4) |
411 | { | |
412 | temp_charset_work->min_char = min_char; | |
413 | temp_charset_work->max_char = max_char; | |
414 | } | |
4ed46869 KH |
415 | } |
416 | ||
12bcae05 | 417 | |
3263d5a2 | 418 | /* Read a hexadecimal number (preceded by "0x") from the file FP while |
3b59c351 | 419 | paying attention to comment character '#'. */ |
12bcae05 | 420 | |
55d4c1b2 | 421 | static inline unsigned |
971de7fb | 422 | read_hex (FILE *fp, int *eof) |
12bcae05 | 423 | { |
3263d5a2 KH |
424 | int c; |
425 | unsigned n; | |
12bcae05 | 426 | |
3263d5a2 KH |
427 | while ((c = getc (fp)) != EOF) |
428 | { | |
69f8de5b | 429 | if (c == '#') |
3263d5a2 KH |
430 | { |
431 | while ((c = getc (fp)) != EOF && c != '\n'); | |
432 | } | |
433 | else if (c == '0') | |
434 | { | |
435 | if ((c = getc (fp)) == EOF || c == 'x') | |
436 | break; | |
437 | } | |
8f924df7 | 438 | } |
3263d5a2 KH |
439 | if (c == EOF) |
440 | { | |
441 | *eof = 1; | |
442 | return 0; | |
443 | } | |
444 | *eof = 0; | |
445 | n = 0; | |
446 | if (c == 'x') | |
447 | while ((c = getc (fp)) != EOF && isxdigit (c)) | |
448 | n = ((n << 4) | |
449 | | (c <= '9' ? c - '0' : c <= 'F' ? c - 'A' + 10 : c - 'a' + 10)); | |
450 | else | |
451 | while ((c = getc (fp)) != EOF && isdigit (c)) | |
452 | n = (n * 10) + c - '0'; | |
e9ce014c KH |
453 | if (c != EOF) |
454 | ungetc (c, fp); | |
3263d5a2 KH |
455 | return n; |
456 | } | |
12bcae05 | 457 | |
3263d5a2 | 458 | /* Return a mapping vector for CHARSET loaded from MAPFILE. |
e9ce014c KH |
459 | Each line of MAPFILE has this form |
460 | 0xAAAA 0xCCCC | |
461 | where 0xAAAA is a code-point and 0xCCCC is the corresponding | |
462 | character code, or this form | |
463 | 0xAAAA-0xBBBB 0xCCCC | |
464 | where 0xAAAA and 0xBBBB are code-points specifying a range, and | |
465 | 0xCCCC is the first character code of the range. | |
4ed46869 | 466 | |
3263d5a2 KH |
467 | The returned vector has this form: |
468 | [ CODE1 CHAR1 CODE2 CHAR2 .... ] | |
e9ce014c | 469 | where CODE1 is a code-point or a cons of code-points specifying a |
dde2559c KH |
470 | range. |
471 | ||
18a10a21 JB |
472 | Note that this function uses `openp' to open MAPFILE but ignores |
473 | `file-name-handler-alist' to avoid running any Lisp code. */ | |
4ed46869 | 474 | |
e9ce014c | 475 | static void |
971de7fb | 476 | load_charset_map_from_file (struct charset *charset, Lisp_Object mapfile, int control_flag) |
4ed46869 | 477 | { |
e9ce014c KH |
478 | unsigned min_code = CHARSET_MIN_CODE (charset); |
479 | unsigned max_code = CHARSET_MAX_CODE (charset); | |
3263d5a2 KH |
480 | int fd; |
481 | FILE *fp; | |
3263d5a2 KH |
482 | int eof; |
483 | Lisp_Object suffixes; | |
e9ce014c | 484 | struct charset_map_entries *head, *entries; |
a2f3eb19 CY |
485 | int n_entries, count; |
486 | USE_SAFE_ALLOCA; | |
4ed46869 | 487 | |
3263d5a2 KH |
488 | suffixes = Fcons (build_string (".map"), |
489 | Fcons (build_string (".TXT"), Qnil)); | |
4ed46869 | 490 | |
a2f3eb19 | 491 | count = SPECPDL_INDEX (); |
dde2559c | 492 | specbind (Qfile_name_handler_alist, Qnil); |
4beef065 | 493 | fd = openp (Vcharset_map_path, mapfile, suffixes, NULL, Qnil); |
dde2559c | 494 | unbind_to (count, Qnil); |
3263d5a2 KH |
495 | if (fd < 0 |
496 | || ! (fp = fdopen (fd, "r"))) | |
5fdb398c | 497 | error ("Failure in loading charset map: %s", SDATA (mapfile)); |
4ed46869 | 498 | |
a2f3eb19 CY |
499 | /* Use SAFE_ALLOCA instead of alloca, as `charset_map_entries' is |
500 | large (larger than MAX_ALLOCA). */ | |
501 | SAFE_ALLOCA (head, struct charset_map_entries *, | |
502 | sizeof (struct charset_map_entries)); | |
503 | entries = head; | |
72af86bd | 504 | memset (entries, 0, sizeof (struct charset_map_entries)); |
a2f3eb19 | 505 | |
e9ce014c | 506 | n_entries = 0; |
3263d5a2 KH |
507 | eof = 0; |
508 | while (1) | |
509 | { | |
e9ce014c KH |
510 | unsigned from, to; |
511 | int c; | |
512 | int idx; | |
4ed46869 | 513 | |
e9ce014c | 514 | from = read_hex (fp, &eof); |
3263d5a2 KH |
515 | if (eof) |
516 | break; | |
e9ce014c KH |
517 | if (getc (fp) == '-') |
518 | to = read_hex (fp, &eof); | |
519 | else | |
520 | to = from; | |
521 | c = (int) read_hex (fp, &eof); | |
ac4137cc | 522 | |
e9ce014c KH |
523 | if (from < min_code || to > max_code || from > to || c > MAX_CHAR) |
524 | continue; | |
ac4137cc | 525 | |
e9ce014c | 526 | if (n_entries > 0 && (n_entries % 0x10000) == 0) |
3263d5a2 | 527 | { |
a2f3eb19 CY |
528 | SAFE_ALLOCA (entries->next, struct charset_map_entries *, |
529 | sizeof (struct charset_map_entries)); | |
e9ce014c | 530 | entries = entries->next; |
72af86bd | 531 | memset (entries, 0, sizeof (struct charset_map_entries)); |
3263d5a2 | 532 | } |
e9ce014c KH |
533 | idx = n_entries % 0x10000; |
534 | entries->entry[idx].from = from; | |
535 | entries->entry[idx].to = to; | |
536 | entries->entry[idx].c = c; | |
537 | n_entries++; | |
3263d5a2 KH |
538 | } |
539 | fclose (fp); | |
177c0ea7 | 540 | |
e9ce014c | 541 | load_charset_map (charset, head, n_entries, control_flag); |
a2f3eb19 | 542 | SAFE_FREE (); |
4ed46869 KH |
543 | } |
544 | ||
e9ce014c | 545 | static void |
971de7fb | 546 | load_charset_map_from_vector (struct charset *charset, Lisp_Object vec, int control_flag) |
23d2a7f1 | 547 | { |
e9ce014c KH |
548 | unsigned min_code = CHARSET_MIN_CODE (charset); |
549 | unsigned max_code = CHARSET_MAX_CODE (charset); | |
550 | struct charset_map_entries *head, *entries; | |
551 | int n_entries; | |
552 | int len = ASIZE (vec); | |
553 | int i; | |
a2f3eb19 | 554 | USE_SAFE_ALLOCA; |
23d2a7f1 | 555 | |
e9ce014c | 556 | if (len % 2 == 1) |
3263d5a2 | 557 | { |
e9ce014c KH |
558 | add_to_log ("Failure in loading charset map: %V", vec, Qnil); |
559 | return; | |
3263d5a2 | 560 | } |
35e623fb | 561 | |
a2f3eb19 CY |
562 | /* Use SAFE_ALLOCA instead of alloca, as `charset_map_entries' is |
563 | large (larger than MAX_ALLOCA). */ | |
564 | SAFE_ALLOCA (head, struct charset_map_entries *, | |
565 | sizeof (struct charset_map_entries)); | |
566 | entries = head; | |
72af86bd | 567 | memset (entries, 0, sizeof (struct charset_map_entries)); |
a2f3eb19 | 568 | |
e9ce014c KH |
569 | n_entries = 0; |
570 | for (i = 0; i < len; i += 2) | |
35e623fb | 571 | { |
e9ce014c KH |
572 | Lisp_Object val, val2; |
573 | unsigned from, to; | |
574 | int c; | |
575 | int idx; | |
d2665018 | 576 | |
e9ce014c KH |
577 | val = AREF (vec, i); |
578 | if (CONSP (val)) | |
bbf12bb3 | 579 | { |
e9ce014c KH |
580 | val2 = XCDR (val); |
581 | val = XCAR (val); | |
582 | CHECK_NATNUM (val); | |
583 | CHECK_NATNUM (val2); | |
584 | from = XFASTINT (val); | |
585 | to = XFASTINT (val2); | |
bbf12bb3 | 586 | } |
e9ce014c | 587 | else |
bbf12bb3 | 588 | { |
e9ce014c KH |
589 | CHECK_NATNUM (val); |
590 | from = to = XFASTINT (val); | |
bbf12bb3 | 591 | } |
e9ce014c KH |
592 | val = AREF (vec, i + 1); |
593 | CHECK_NATNUM (val); | |
594 | c = XFASTINT (val); | |
76d7b829 | 595 | |
e9ce014c KH |
596 | if (from < min_code || to > max_code || from > to || c > MAX_CHAR) |
597 | continue; | |
76d7b829 | 598 | |
dcc694d7 | 599 | if (n_entries > 0 && (n_entries % 0x10000) == 0) |
e9ce014c | 600 | { |
a2f3eb19 CY |
601 | SAFE_ALLOCA (entries->next, struct charset_map_entries *, |
602 | sizeof (struct charset_map_entries)); | |
e9ce014c | 603 | entries = entries->next; |
72af86bd | 604 | memset (entries, 0, sizeof (struct charset_map_entries)); |
e9ce014c KH |
605 | } |
606 | idx = n_entries % 0x10000; | |
607 | entries->entry[idx].from = from; | |
608 | entries->entry[idx].to = to; | |
609 | entries->entry[idx].c = c; | |
610 | n_entries++; | |
611 | } | |
76d7b829 | 612 | |
e9ce014c | 613 | load_charset_map (charset, head, n_entries, control_flag); |
a2f3eb19 | 614 | SAFE_FREE (); |
ac4137cc KH |
615 | } |
616 | ||
ecca2aad KH |
617 | |
618 | /* Load a mapping table for CHARSET. CONTROL-FLAG tells what kind of | |
619 | map it is (see the comment of load_charset_map for the detail). */ | |
620 | ||
3263d5a2 | 621 | static void |
971de7fb | 622 | load_charset (struct charset *charset, int control_flag) |
76d7b829 | 623 | { |
ecca2aad | 624 | Lisp_Object map; |
76d7b829 | 625 | |
ecca2aad KH |
626 | if (inhibit_load_charset_map |
627 | && temp_charset_work | |
628 | && charset == temp_charset_work->current | |
56f00ed2 | 629 | && ((control_flag == 2) == temp_charset_work->for_encoder)) |
ecca2aad KH |
630 | return; |
631 | ||
632 | if (CHARSET_METHOD (charset) == CHARSET_METHOD_MAP) | |
633 | map = CHARSET_MAP (charset); | |
0ac2c299 PE |
634 | else |
635 | { | |
636 | if (! CHARSET_UNIFIED_P (charset)) | |
637 | abort (); | |
638 | map = CHARSET_UNIFY_MAP (charset); | |
639 | } | |
ecca2aad KH |
640 | if (STRINGP (map)) |
641 | load_charset_map_from_file (charset, map, control_flag); | |
642 | else | |
643 | load_charset_map_from_vector (charset, map, control_flag); | |
4ed46869 | 644 | } |
76d7b829 | 645 | |
3263d5a2 KH |
646 | |
647 | DEFUN ("charsetp", Fcharsetp, Scharsetp, 1, 1, 0, | |
648 | doc: /* Return non-nil if and only if OBJECT is a charset.*/) | |
5842a27b | 649 | (Lisp_Object object) |
23d2a7f1 | 650 | { |
3263d5a2 | 651 | return (CHARSETP (object) ? Qt : Qnil); |
76d7b829 KH |
652 | } |
653 | ||
4ed46869 | 654 | |
a2cb4e63 PE |
655 | static void |
656 | map_charset_for_dump (void (*c_function) (Lisp_Object, Lisp_Object), | |
657 | Lisp_Object function, Lisp_Object arg, | |
658 | unsigned int from, unsigned int to) | |
ecca2aad KH |
659 | { |
660 | int from_idx = CODE_POINT_TO_INDEX (temp_charset_work->current, from); | |
661 | int to_idx = CODE_POINT_TO_INDEX (temp_charset_work->current, to); | |
662 | Lisp_Object range; | |
663 | int c, stop; | |
664 | struct gcpro gcpro1; | |
665 | ||
666 | range = Fcons (Qnil, Qnil); | |
667 | GCPRO1 (range); | |
668 | ||
669 | c = temp_charset_work->min_char; | |
670 | stop = (temp_charset_work->max_char < 0x20000 | |
671 | ? temp_charset_work->max_char : 0xFFFF); | |
51b59d79 | 672 | |
ecca2aad KH |
673 | while (1) |
674 | { | |
f6095868 | 675 | int idx = GET_TEMP_CHARSET_WORK_ENCODER (c); |
ecca2aad | 676 | |
f6095868 | 677 | if (idx >= from_idx && idx <= to_idx) |
ecca2aad KH |
678 | { |
679 | if (NILP (XCAR (range))) | |
680 | XSETCAR (range, make_number (c)); | |
681 | } | |
682 | else if (! NILP (XCAR (range))) | |
683 | { | |
684 | XSETCDR (range, make_number (c - 1)); | |
685 | if (c_function) | |
686 | (*c_function) (arg, range); | |
687 | else | |
688 | call2 (function, range, arg); | |
689 | XSETCAR (range, Qnil); | |
690 | } | |
691 | if (c == stop) | |
692 | { | |
693 | if (c == temp_charset_work->max_char) | |
694 | { | |
695 | if (! NILP (XCAR (range))) | |
696 | { | |
697 | XSETCDR (range, make_number (c)); | |
698 | if (c_function) | |
699 | (*c_function) (arg, range); | |
700 | else | |
701 | call2 (function, range, arg); | |
702 | } | |
703 | break; | |
704 | } | |
705 | c = 0x1FFFF; | |
706 | stop = temp_charset_work->max_char; | |
707 | } | |
708 | c++; | |
709 | } | |
c542407d | 710 | UNGCPRO; |
ecca2aad KH |
711 | } |
712 | ||
4ed46869 | 713 | void |
6f704c76 DN |
714 | map_charset_chars (void (*c_function)(Lisp_Object, Lisp_Object), Lisp_Object function, |
715 | Lisp_Object arg, struct charset *charset, unsigned from, unsigned to) | |
4ed46869 | 716 | { |
3263d5a2 | 717 | Lisp_Object range; |
374c5cfd | 718 | int partial; |
3263d5a2 | 719 | |
374c5cfd KH |
720 | partial = (from > CHARSET_MIN_CODE (charset) |
721 | || to < CHARSET_MAX_CODE (charset)); | |
722 | ||
3263d5a2 | 723 | if (CHARSET_METHOD (charset) == CHARSET_METHOD_OFFSET) |
4ed46869 | 724 | { |
374c5cfd KH |
725 | int from_idx = CODE_POINT_TO_INDEX (charset, from); |
726 | int to_idx = CODE_POINT_TO_INDEX (charset, to); | |
727 | int from_c = from_idx + CHARSET_CODE_OFFSET (charset); | |
728 | int to_c = to_idx + CHARSET_CODE_OFFSET (charset); | |
729 | ||
ecca2aad KH |
730 | if (CHARSET_UNIFIED_P (charset)) |
731 | { | |
732 | if (! CHAR_TABLE_P (CHARSET_DEUNIFIER (charset))) | |
733 | load_charset (charset, 2); | |
734 | if (CHAR_TABLE_P (CHARSET_DEUNIFIER (charset))) | |
735 | map_char_table_for_charset (c_function, function, | |
736 | CHARSET_DEUNIFIER (charset), arg, | |
737 | partial ? charset : NULL, from, to); | |
738 | else | |
739 | map_charset_for_dump (c_function, function, arg, from, to); | |
740 | } | |
741 | ||
374c5cfd | 742 | range = Fcons (make_number (from_c), make_number (to_c)); |
3263d5a2 | 743 | if (NILP (function)) |
5af5dd92 | 744 | (*c_function) (arg, range); |
3263d5a2 KH |
745 | else |
746 | call2 (function, range, arg); | |
c83ef371 | 747 | } |
3263d5a2 KH |
748 | else if (CHARSET_METHOD (charset) == CHARSET_METHOD_MAP) |
749 | { | |
750 | if (! CHAR_TABLE_P (CHARSET_ENCODER (charset))) | |
ecca2aad KH |
751 | load_charset (charset, 2); |
752 | if (CHAR_TABLE_P (CHARSET_ENCODER (charset))) | |
753 | map_char_table_for_charset (c_function, function, | |
754 | CHARSET_ENCODER (charset), arg, | |
755 | partial ? charset : NULL, from, to); | |
756 | else | |
757 | map_charset_for_dump (c_function, function, arg, from, to); | |
3263d5a2 | 758 | } |
374c5cfd | 759 | else if (CHARSET_METHOD (charset) == CHARSET_METHOD_SUBSET) |
4ed46869 | 760 | { |
374c5cfd KH |
761 | Lisp_Object subset_info; |
762 | int offset; | |
763 | ||
764 | subset_info = CHARSET_SUBSET (charset); | |
765 | charset = CHARSET_FROM_ID (XFASTINT (AREF (subset_info, 0))); | |
766 | offset = XINT (AREF (subset_info, 3)); | |
767 | from -= offset; | |
768 | if (from < XFASTINT (AREF (subset_info, 1))) | |
769 | from = XFASTINT (AREF (subset_info, 1)); | |
770 | to -= offset; | |
771 | if (to > XFASTINT (AREF (subset_info, 2))) | |
772 | to = XFASTINT (AREF (subset_info, 2)); | |
773 | map_charset_chars (c_function, function, arg, charset, from, to); | |
4ed46869 | 774 | } |
374c5cfd KH |
775 | else /* i.e. CHARSET_METHOD_SUPERSET */ |
776 | { | |
777 | Lisp_Object parents; | |
4ed46869 | 778 | |
374c5cfd KH |
779 | for (parents = CHARSET_SUPERSET (charset); CONSP (parents); |
780 | parents = XCDR (parents)) | |
bbf12bb3 | 781 | { |
374c5cfd KH |
782 | int offset; |
783 | unsigned this_from, this_to; | |
784 | ||
785 | charset = CHARSET_FROM_ID (XFASTINT (XCAR (XCAR (parents)))); | |
786 | offset = XINT (XCDR (XCAR (parents))); | |
12b55765 KH |
787 | this_from = from > offset ? from - offset : 0; |
788 | this_to = to > offset ? to - offset : 0; | |
374c5cfd KH |
789 | if (this_from < CHARSET_MIN_CODE (charset)) |
790 | this_from = CHARSET_MIN_CODE (charset); | |
791 | if (this_to > CHARSET_MAX_CODE (charset)) | |
792 | this_to = CHARSET_MAX_CODE (charset); | |
111daccf KH |
793 | map_charset_chars (c_function, function, arg, charset, |
794 | this_from, this_to); | |
bbf12bb3 | 795 | } |
35e623fb | 796 | } |
4ed46869 KH |
797 | } |
798 | ||
374c5cfd | 799 | DEFUN ("map-charset-chars", Fmap_charset_chars, Smap_charset_chars, 2, 5, 0, |
04c2f2c5 | 800 | doc: /* Call FUNCTION for all characters in CHARSET. |
374c5cfd | 801 | FUNCTION is called with an argument RANGE and the optional 3rd |
3263d5a2 | 802 | argument ARG. |
4ed46869 | 803 | |
374c5cfd KH |
804 | RANGE is a cons (FROM . TO), where FROM and TO indicate a range of |
805 | characters contained in CHARSET. | |
4ed46869 | 806 | |
374c5cfd | 807 | The optional 4th and 5th arguments FROM-CODE and TO-CODE specify the |
72d51285 | 808 | range of code points (in CHARSET) of target characters. */) |
5842a27b | 809 | (Lisp_Object function, Lisp_Object charset, Lisp_Object arg, Lisp_Object from_code, Lisp_Object to_code) |
4ed46869 | 810 | { |
374c5cfd | 811 | struct charset *cs; |
16fed1fc | 812 | unsigned from, to; |
4ed46869 | 813 | |
374c5cfd KH |
814 | CHECK_CHARSET_GET_CHARSET (charset, cs); |
815 | if (NILP (from_code)) | |
16fed1fc | 816 | from = CHARSET_MIN_CODE (cs); |
970b7474 | 817 | else |
4ed46869 | 818 | { |
970b7474 KH |
819 | CHECK_NATNUM (from_code); |
820 | from = XINT (from_code); | |
821 | if (from < CHARSET_MIN_CODE (cs)) | |
822 | from = CHARSET_MIN_CODE (cs); | |
4ed46869 | 823 | } |
374c5cfd | 824 | if (NILP (to_code)) |
970b7474 | 825 | to = CHARSET_MAX_CODE (cs); |
4ed46869 KH |
826 | else |
827 | { | |
970b7474 KH |
828 | CHECK_NATNUM (to_code); |
829 | to = XINT (to_code); | |
830 | if (to > CHARSET_MAX_CODE (cs)) | |
831 | to = CHARSET_MAX_CODE (cs); | |
4ed46869 | 832 | } |
16fed1fc | 833 | map_charset_chars (NULL, function, arg, cs, from, to); |
3263d5a2 | 834 | return Qnil; |
35e623fb | 835 | } |
4ed46869 | 836 | |
4ed46869 | 837 | |
3263d5a2 KH |
838 | /* Define a charset according to the arguments. The Nth argument is |
839 | the Nth attribute of the charset (the last attribute `charset-id' | |
840 | is not included). See the docstring of `define-charset' for the | |
841 | detail. */ | |
4ed46869 | 842 | |
3263d5a2 KH |
843 | DEFUN ("define-charset-internal", Fdefine_charset_internal, |
844 | Sdefine_charset_internal, charset_arg_max, MANY, 0, | |
04c2f2c5 DL |
845 | doc: /* For internal use only. |
846 | usage: (define-charset-internal ...) */) | |
c5101a77 | 847 | (size_t nargs, Lisp_Object *args) |
4ed46869 | 848 | { |
3263d5a2 KH |
849 | /* Charset attr vector. */ |
850 | Lisp_Object attrs; | |
851 | Lisp_Object val; | |
0de4bb68 | 852 | EMACS_UINT hash_code; |
3263d5a2 | 853 | struct Lisp_Hash_Table *hash_table = XHASH_TABLE (Vcharset_hash_table); |
69f8de5b | 854 | int i, j; |
3263d5a2 KH |
855 | struct charset charset; |
856 | int id; | |
857 | int dimension; | |
858 | int new_definition_p; | |
859 | int nchars; | |
860 | ||
861 | if (nargs != charset_arg_max) | |
862 | return Fsignal (Qwrong_number_of_arguments, | |
863 | Fcons (intern ("define-charset-internal"), | |
864 | make_number (nargs))); | |
865 | ||
866 | attrs = Fmake_vector (make_number (charset_attr_max), Qnil); | |
867 | ||
868 | CHECK_SYMBOL (args[charset_arg_name]); | |
869 | ASET (attrs, charset_name, args[charset_arg_name]); | |
870 | ||
871 | val = args[charset_arg_code_space]; | |
c032b5f8 | 872 | for (i = 0, dimension = 0, nchars = 1; ; i++) |
76d7b829 | 873 | { |
3263d5a2 KH |
874 | int min_byte, max_byte; |
875 | ||
876 | min_byte = XINT (Faref (val, make_number (i * 2))); | |
877 | max_byte = XINT (Faref (val, make_number (i * 2 + 1))); | |
878 | if (min_byte < 0 || min_byte > max_byte || max_byte >= 256) | |
879 | error ("Invalid :code-space value"); | |
880 | charset.code_space[i * 4] = min_byte; | |
881 | charset.code_space[i * 4 + 1] = max_byte; | |
882 | charset.code_space[i * 4 + 2] = max_byte - min_byte + 1; | |
3263d5a2 KH |
883 | if (max_byte > 0) |
884 | dimension = i + 1; | |
c032b5f8 PE |
885 | if (i == 3) |
886 | break; | |
887 | nchars *= charset.code_space[i * 4 + 2]; | |
888 | charset.code_space[i * 4 + 3] = nchars; | |
3263d5a2 | 889 | } |
4ed46869 | 890 | |
3263d5a2 KH |
891 | val = args[charset_arg_dimension]; |
892 | if (NILP (val)) | |
893 | charset.dimension = dimension; | |
894 | else | |
4ed46869 | 895 | { |
3263d5a2 KH |
896 | CHECK_NATNUM (val); |
897 | charset.dimension = XINT (val); | |
898 | if (charset.dimension < 1 || charset.dimension > 4) | |
899 | args_out_of_range_3 (val, make_number (1), make_number (4)); | |
4ed46869 KH |
900 | } |
901 | ||
3263d5a2 KH |
902 | charset.code_linear_p |
903 | = (charset.dimension == 1 | |
904 | || (charset.code_space[2] == 256 | |
905 | && (charset.dimension == 2 | |
906 | || (charset.code_space[6] == 256 | |
907 | && (charset.dimension == 3 | |
908 | || charset.code_space[10] == 256))))); | |
909 | ||
69f8de5b | 910 | if (! charset.code_linear_p) |
4ed46869 | 911 | { |
69f8de5b | 912 | charset.code_space_mask = (unsigned char *) xmalloc (256); |
72af86bd | 913 | memset (charset.code_space_mask, 0, 256); |
69f8de5b KH |
914 | for (i = 0; i < 4; i++) |
915 | for (j = charset.code_space[i * 4]; j <= charset.code_space[i * 4 + 1]; | |
916 | j++) | |
917 | charset.code_space_mask[j] |= (1 << i); | |
4ed46869 KH |
918 | } |
919 | ||
3263d5a2 | 920 | charset.iso_chars_96 = charset.code_space[2] == 96; |
4ed46869 | 921 | |
3263d5a2 KH |
922 | charset.min_code = (charset.code_space[0] |
923 | | (charset.code_space[4] << 8) | |
924 | | (charset.code_space[8] << 16) | |
925 | | (charset.code_space[12] << 24)); | |
926 | charset.max_code = (charset.code_space[1] | |
927 | | (charset.code_space[5] << 8) | |
928 | | (charset.code_space[9] << 16) | |
929 | | (charset.code_space[13] << 24)); | |
820ee249 | 930 | charset.char_index_offset = 0; |
8a73a704 | 931 | |
820ee249 KH |
932 | val = args[charset_arg_min_code]; |
933 | if (! NILP (val)) | |
934 | { | |
be44ca6c | 935 | unsigned code = cons_to_unsigned (val, UINT_MAX); |
fdb82f93 | 936 | |
820ee249 KH |
937 | if (code < charset.min_code |
938 | || code > charset.max_code) | |
939 | args_out_of_range_3 (make_number (charset.min_code), | |
940 | make_number (charset.max_code), val); | |
941 | charset.char_index_offset = CODE_POINT_TO_INDEX (&charset, code); | |
942 | charset.min_code = code; | |
943 | } | |
3fac5a51 | 944 | |
820ee249 KH |
945 | val = args[charset_arg_max_code]; |
946 | if (! NILP (val)) | |
3fac5a51 | 947 | { |
be44ca6c | 948 | unsigned code = cons_to_unsigned (val, UINT_MAX); |
820ee249 | 949 | |
820ee249 KH |
950 | if (code < charset.min_code |
951 | || code > charset.max_code) | |
952 | args_out_of_range_3 (make_number (charset.min_code), | |
953 | make_number (charset.max_code), val); | |
954 | charset.max_code = code; | |
3fac5a51 | 955 | } |
3fac5a51 | 956 | |
ecca2aad | 957 | charset.compact_codes_p = charset.max_code < 0x10000; |
4ed46869 | 958 | |
3263d5a2 KH |
959 | val = args[charset_arg_invalid_code]; |
960 | if (NILP (val)) | |
961 | { | |
962 | if (charset.min_code > 0) | |
963 | charset.invalid_code = 0; | |
bbf12bb3 KH |
964 | else |
965 | { | |
3263d5a2 KH |
966 | XSETINT (val, charset.max_code + 1); |
967 | if (XINT (val) == charset.max_code + 1) | |
968 | charset.invalid_code = charset.max_code + 1; | |
969 | else | |
970 | error ("Attribute :invalid-code must be specified"); | |
76d7b829 | 971 | } |
76d7b829 | 972 | } |
3263d5a2 KH |
973 | else |
974 | { | |
975 | CHECK_NATNUM (val); | |
976 | charset.invalid_code = XFASTINT (val); | |
977 | } | |
4ed46869 | 978 | |
3263d5a2 KH |
979 | val = args[charset_arg_iso_final]; |
980 | if (NILP (val)) | |
981 | charset.iso_final = -1; | |
982 | else | |
983 | { | |
984 | CHECK_NUMBER (val); | |
985 | if (XINT (val) < '0' || XINT (val) > 127) | |
c2982e87 | 986 | error ("Invalid iso-final-char: %"pI"d", XINT (val)); |
3263d5a2 KH |
987 | charset.iso_final = XINT (val); |
988 | } | |
4ed46869 | 989 | |
3263d5a2 KH |
990 | val = args[charset_arg_iso_revision]; |
991 | if (NILP (val)) | |
992 | charset.iso_revision = -1; | |
993 | else | |
4ed46869 | 994 | { |
3263d5a2 KH |
995 | CHECK_NUMBER (val); |
996 | if (XINT (val) > 63) | |
997 | args_out_of_range (make_number (63), val); | |
998 | charset.iso_revision = XINT (val); | |
4ed46869 | 999 | } |
4ed46869 | 1000 | |
3263d5a2 KH |
1001 | val = args[charset_arg_emacs_mule_id]; |
1002 | if (NILP (val)) | |
1003 | charset.emacs_mule_id = -1; | |
4ed46869 KH |
1004 | else |
1005 | { | |
3263d5a2 KH |
1006 | CHECK_NATNUM (val); |
1007 | if ((XINT (val) > 0 && XINT (val) <= 128) || XINT (val) >= 256) | |
c2982e87 | 1008 | error ("Invalid emacs-mule-id: %"pI"d", XINT (val)); |
3263d5a2 | 1009 | charset.emacs_mule_id = XINT (val); |
c83ef371 | 1010 | } |
f6302ac9 | 1011 | |
3263d5a2 | 1012 | charset.ascii_compatible_p = ! NILP (args[charset_arg_ascii_compatible_p]); |
1d67c29b | 1013 | |
3263d5a2 | 1014 | charset.supplementary_p = ! NILP (args[charset_arg_supplementary_p]); |
4ed46869 | 1015 | |
3263d5a2 KH |
1016 | charset.unified_p = 0; |
1017 | ||
72af86bd | 1018 | memset (charset.fast_map, 0, sizeof (charset.fast_map)); |
3263d5a2 KH |
1019 | |
1020 | if (! NILP (args[charset_arg_code_offset])) | |
1021 | { | |
1022 | val = args[charset_arg_code_offset]; | |
1023 | CHECK_NUMBER (val); | |
1024 | ||
1025 | charset.method = CHARSET_METHOD_OFFSET; | |
1026 | charset.code_offset = XINT (val); | |
1027 | ||
1028 | i = CODE_POINT_TO_INDEX (&charset, charset.min_code); | |
1029 | charset.min_char = i + charset.code_offset; | |
1030 | i = CODE_POINT_TO_INDEX (&charset, charset.max_code); | |
1031 | charset.max_char = i + charset.code_offset; | |
1032 | if (charset.max_char > MAX_CHAR) | |
1033 | error ("Unsupported max char: %d", charset.max_char); | |
1034 | ||
f148205f KH |
1035 | i = (charset.min_char >> 7) << 7; |
1036 | for (; i < 0x10000 && i <= charset.max_char; i += 128) | |
3263d5a2 | 1037 | CHARSET_FAST_MAP_SET (i, charset.fast_map); |
f148205f | 1038 | i = (i >> 12) << 12; |
3263d5a2 KH |
1039 | for (; i <= charset.max_char; i += 0x1000) |
1040 | CHARSET_FAST_MAP_SET (i, charset.fast_map); | |
3620330b KH |
1041 | if (charset.code_offset == 0 && charset.max_char >= 0x80) |
1042 | charset.ascii_compatible_p = 1; | |
3263d5a2 KH |
1043 | } |
1044 | else if (! NILP (args[charset_arg_map])) | |
1045 | { | |
1046 | val = args[charset_arg_map]; | |
1047 | ASET (attrs, charset_map, val); | |
ecca2aad | 1048 | charset.method = CHARSET_METHOD_MAP; |
3263d5a2 | 1049 | } |
374c5cfd | 1050 | else if (! NILP (args[charset_arg_subset])) |
3263d5a2 | 1051 | { |
374c5cfd KH |
1052 | Lisp_Object parent; |
1053 | Lisp_Object parent_min_code, parent_max_code, parent_code_offset; | |
1054 | struct charset *parent_charset; | |
1055 | ||
1056 | val = args[charset_arg_subset]; | |
1057 | parent = Fcar (val); | |
1058 | CHECK_CHARSET_GET_CHARSET (parent, parent_charset); | |
1059 | parent_min_code = Fnth (make_number (1), val); | |
1060 | CHECK_NATNUM (parent_min_code); | |
1061 | parent_max_code = Fnth (make_number (2), val); | |
1062 | CHECK_NATNUM (parent_max_code); | |
1063 | parent_code_offset = Fnth (make_number (3), val); | |
1064 | CHECK_NUMBER (parent_code_offset); | |
1065 | val = Fmake_vector (make_number (4), Qnil); | |
1066 | ASET (val, 0, make_number (parent_charset->id)); | |
1067 | ASET (val, 1, parent_min_code); | |
1068 | ASET (val, 2, parent_max_code); | |
1069 | ASET (val, 3, parent_code_offset); | |
1070 | ASET (attrs, charset_subset, val); | |
1071 | ||
1072 | charset.method = CHARSET_METHOD_SUBSET; | |
1073 | /* Here, we just copy the parent's fast_map. It's not accurate, | |
1074 | but at least it works for quickly detecting which character | |
1075 | DOESN'T belong to this charset. */ | |
1076 | for (i = 0; i < 190; i++) | |
1077 | charset.fast_map[i] = parent_charset->fast_map[i]; | |
1078 | ||
1079 | /* We also copy these for parents. */ | |
1080 | charset.min_char = parent_charset->min_char; | |
1081 | charset.max_char = parent_charset->max_char; | |
1082 | } | |
1083 | else if (! NILP (args[charset_arg_superset])) | |
0282eb69 | 1084 | { |
374c5cfd KH |
1085 | val = args[charset_arg_superset]; |
1086 | charset.method = CHARSET_METHOD_SUPERSET; | |
3263d5a2 | 1087 | val = Fcopy_sequence (val); |
374c5cfd | 1088 | ASET (attrs, charset_superset, val); |
3263d5a2 KH |
1089 | |
1090 | charset.min_char = MAX_CHAR; | |
1091 | charset.max_char = 0; | |
1092 | for (; ! NILP (val); val = Fcdr (val)) | |
0282eb69 | 1093 | { |
3263d5a2 KH |
1094 | Lisp_Object elt, car_part, cdr_part; |
1095 | int this_id, offset; | |
1096 | struct charset *this_charset; | |
2e344af3 | 1097 | |
3263d5a2 KH |
1098 | elt = Fcar (val); |
1099 | if (CONSP (elt)) | |
2e344af3 | 1100 | { |
3263d5a2 KH |
1101 | car_part = XCAR (elt); |
1102 | cdr_part = XCDR (elt); | |
1103 | CHECK_CHARSET_GET_ID (car_part, this_id); | |
1104 | CHECK_NUMBER (cdr_part); | |
1105 | offset = XINT (cdr_part); | |
177c0ea7 | 1106 | } |
3263d5a2 | 1107 | else |
4ed46869 | 1108 | { |
3263d5a2 KH |
1109 | CHECK_CHARSET_GET_ID (elt, this_id); |
1110 | offset = 0; | |
4ed46869 | 1111 | } |
3263d5a2 KH |
1112 | XSETCAR (val, Fcons (make_number (this_id), make_number (offset))); |
1113 | ||
1114 | this_charset = CHARSET_FROM_ID (this_id); | |
1115 | if (charset.min_char > this_charset->min_char) | |
1116 | charset.min_char = this_charset->min_char; | |
1117 | if (charset.max_char < this_charset->max_char) | |
1118 | charset.max_char = this_charset->max_char; | |
1119 | for (i = 0; i < 190; i++) | |
1120 | charset.fast_map[i] |= this_charset->fast_map[i]; | |
0282eb69 | 1121 | } |
0282eb69 | 1122 | } |
2e344af3 | 1123 | else |
3263d5a2 | 1124 | error ("None of :code-offset, :map, :parents are specified"); |
05505664 | 1125 | |
3263d5a2 KH |
1126 | val = args[charset_arg_unify_map]; |
1127 | if (! NILP (val) && !STRINGP (val)) | |
1128 | CHECK_VECTOR (val); | |
1129 | ASET (attrs, charset_unify_map, val); | |
05505664 | 1130 | |
3263d5a2 KH |
1131 | CHECK_LIST (args[charset_arg_plist]); |
1132 | ASET (attrs, charset_plist, args[charset_arg_plist]); | |
4ed46869 | 1133 | |
3263d5a2 KH |
1134 | charset.hash_index = hash_lookup (hash_table, args[charset_arg_name], |
1135 | &hash_code); | |
1136 | if (charset.hash_index >= 0) | |
1137 | { | |
1138 | new_definition_p = 0; | |
4f65af01 | 1139 | id = XFASTINT (CHARSET_SYMBOL_ID (args[charset_arg_name])); |
3263d5a2 KH |
1140 | HASH_VALUE (hash_table, charset.hash_index) = attrs; |
1141 | } | |
1a45ff10 | 1142 | else |
3263d5a2 KH |
1143 | { |
1144 | charset.hash_index = hash_put (hash_table, args[charset_arg_name], attrs, | |
1145 | hash_code); | |
1146 | if (charset_table_used == charset_table_size) | |
1147 | { | |
2fe1edd1 KH |
1148 | struct charset *new_table |
1149 | = (struct charset *) xmalloc (sizeof (struct charset) | |
1150 | * (charset_table_size + 16)); | |
72af86bd AS |
1151 | memcpy (new_table, charset_table, |
1152 | sizeof (struct charset) * charset_table_size); | |
2fe1edd1 KH |
1153 | charset_table_size += 16; |
1154 | charset_table = new_table; | |
3263d5a2 KH |
1155 | } |
1156 | id = charset_table_used++; | |
3263d5a2 KH |
1157 | new_definition_p = 1; |
1158 | } | |
2e344af3 | 1159 | |
4f65af01 | 1160 | ASET (attrs, charset_id, make_number (id)); |
3263d5a2 KH |
1161 | charset.id = id; |
1162 | charset_table[id] = charset; | |
2e344af3 | 1163 | |
ecca2aad | 1164 | if (charset.method == CHARSET_METHOD_MAP) |
b8ebe9dd KH |
1165 | { |
1166 | load_charset (&charset, 0); | |
1167 | charset_table[id] = charset; | |
1168 | } | |
ecca2aad | 1169 | |
3263d5a2 | 1170 | if (charset.iso_final >= 0) |
4ed46869 | 1171 | { |
3263d5a2 KH |
1172 | ISO_CHARSET_TABLE (charset.dimension, charset.iso_chars_96, |
1173 | charset.iso_final) = id; | |
1174 | if (new_definition_p) | |
1175 | Viso_2022_charset_list = nconc2 (Viso_2022_charset_list, | |
1176 | Fcons (make_number (id), Qnil)); | |
7c7dceee KH |
1177 | if (ISO_CHARSET_TABLE (1, 0, 'J') == id) |
1178 | charset_jisx0201_roman = id; | |
1179 | else if (ISO_CHARSET_TABLE (2, 0, '@') == id) | |
1180 | charset_jisx0208_1978 = id; | |
1181 | else if (ISO_CHARSET_TABLE (2, 0, 'B') == id) | |
1182 | charset_jisx0208 = id; | |
d32320c4 KH |
1183 | else if (ISO_CHARSET_TABLE (2, 0, 'C') == id) |
1184 | charset_ksc5601 = id; | |
4ed46869 | 1185 | } |
d0cf2d48 | 1186 | |
3263d5a2 | 1187 | if (charset.emacs_mule_id >= 0) |
4ed46869 | 1188 | { |
b84ae584 | 1189 | emacs_mule_charset[charset.emacs_mule_id] = id; |
4f65af01 KH |
1190 | if (charset.emacs_mule_id < 0xA0) |
1191 | emacs_mule_bytes[charset.emacs_mule_id] = charset.dimension + 1; | |
3b1ae89b KH |
1192 | else |
1193 | emacs_mule_bytes[charset.emacs_mule_id] = charset.dimension + 2; | |
3263d5a2 KH |
1194 | if (new_definition_p) |
1195 | Vemacs_mule_charset_list = nconc2 (Vemacs_mule_charset_list, | |
1196 | Fcons (make_number (id), Qnil)); | |
4ed46869 KH |
1197 | } |
1198 | ||
3263d5a2 KH |
1199 | if (new_definition_p) |
1200 | { | |
1201 | Vcharset_list = Fcons (args[charset_arg_name], Vcharset_list); | |
8055c66a KH |
1202 | if (charset.supplementary_p) |
1203 | Vcharset_ordered_list = nconc2 (Vcharset_ordered_list, | |
1204 | Fcons (make_number (id), Qnil)); | |
1205 | else | |
880820fe KH |
1206 | { |
1207 | Lisp_Object tail; | |
1208 | ||
1209 | for (tail = Vcharset_ordered_list; CONSP (tail); tail = XCDR (tail)) | |
1210 | { | |
1211 | struct charset *cs = CHARSET_FROM_ID (XINT (XCAR (tail))); | |
1212 | ||
1213 | if (cs->supplementary_p) | |
1214 | break; | |
1215 | } | |
1216 | if (EQ (tail, Vcharset_ordered_list)) | |
1217 | Vcharset_ordered_list = Fcons (make_number (id), | |
1218 | Vcharset_ordered_list); | |
1219 | else if (NILP (tail)) | |
1220 | Vcharset_ordered_list = nconc2 (Vcharset_ordered_list, | |
1221 | Fcons (make_number (id), Qnil)); | |
1222 | else | |
1223 | { | |
1224 | val = Fcons (XCAR (tail), XCDR (tail)); | |
1225 | XSETCDR (tail, val); | |
1226 | XSETCAR (tail, make_number (id)); | |
1227 | } | |
1228 | } | |
dbbb237d | 1229 | charset_ordered_list_tick++; |
3263d5a2 | 1230 | } |
4ed46869 | 1231 | |
3263d5a2 | 1232 | return Qnil; |
4ed46869 KH |
1233 | } |
1234 | ||
2fe1edd1 KH |
1235 | |
1236 | /* Same as Fdefine_charset_internal but arguments are more convenient | |
1237 | to call from C (typically in syms_of_charset). This can define a | |
1238 | charset of `offset' method only. Return the ID of the new | |
1239 | charset. */ | |
1240 | ||
1241 | static int | |
dd4c5104 DN |
1242 | define_charset_internal (Lisp_Object name, |
1243 | int dimension, | |
dfb6afda | 1244 | const char *code_space_chars, |
dd4c5104 DN |
1245 | unsigned min_code, unsigned max_code, |
1246 | int iso_final, int iso_revision, int emacs_mule_id, | |
1247 | int ascii_compatible, int supplementary, | |
1248 | int code_offset) | |
2fe1edd1 | 1249 | { |
dfb6afda | 1250 | const unsigned char *code_space = (const unsigned char *) code_space_chars; |
2fe1edd1 KH |
1251 | Lisp_Object args[charset_arg_max]; |
1252 | Lisp_Object plist[14]; | |
1253 | Lisp_Object val; | |
1254 | int i; | |
1255 | ||
1256 | args[charset_arg_name] = name; | |
1257 | args[charset_arg_dimension] = make_number (dimension); | |
1258 | val = Fmake_vector (make_number (8), make_number (0)); | |
1259 | for (i = 0; i < 8; i++) | |
1260 | ASET (val, i, make_number (code_space[i])); | |
1261 | args[charset_arg_code_space] = val; | |
1262 | args[charset_arg_min_code] = make_number (min_code); | |
1263 | args[charset_arg_max_code] = make_number (max_code); | |
1264 | args[charset_arg_iso_final] | |
1265 | = (iso_final < 0 ? Qnil : make_number (iso_final)); | |
1266 | args[charset_arg_iso_revision] = make_number (iso_revision); | |
1267 | args[charset_arg_emacs_mule_id] | |
1268 | = (emacs_mule_id < 0 ? Qnil : make_number (emacs_mule_id)); | |
1269 | args[charset_arg_ascii_compatible_p] = ascii_compatible ? Qt : Qnil; | |
7acf89e6 | 1270 | args[charset_arg_supplementary_p] = supplementary ? Qt : Qnil; |
2fe1edd1 KH |
1271 | args[charset_arg_invalid_code] = Qnil; |
1272 | args[charset_arg_code_offset] = make_number (code_offset); | |
1273 | args[charset_arg_map] = Qnil; | |
1274 | args[charset_arg_subset] = Qnil; | |
1275 | args[charset_arg_superset] = Qnil; | |
1276 | args[charset_arg_unify_map] = Qnil; | |
1277 | ||
d67b4f80 | 1278 | plist[0] = intern_c_string (":name"); |
2fe1edd1 | 1279 | plist[1] = args[charset_arg_name]; |
d67b4f80 | 1280 | plist[2] = intern_c_string (":dimension"); |
2fe1edd1 | 1281 | plist[3] = args[charset_arg_dimension]; |
d67b4f80 | 1282 | plist[4] = intern_c_string (":code-space"); |
2fe1edd1 | 1283 | plist[5] = args[charset_arg_code_space]; |
d67b4f80 | 1284 | plist[6] = intern_c_string (":iso-final-char"); |
2fe1edd1 | 1285 | plist[7] = args[charset_arg_iso_final]; |
d67b4f80 | 1286 | plist[8] = intern_c_string (":emacs-mule-id"); |
2fe1edd1 | 1287 | plist[9] = args[charset_arg_emacs_mule_id]; |
d67b4f80 | 1288 | plist[10] = intern_c_string (":ascii-compatible-p"); |
2fe1edd1 | 1289 | plist[11] = args[charset_arg_ascii_compatible_p]; |
d67b4f80 | 1290 | plist[12] = intern_c_string (":code-offset"); |
2fe1edd1 KH |
1291 | plist[13] = args[charset_arg_code_offset]; |
1292 | ||
1293 | args[charset_arg_plist] = Flist (14, plist); | |
1294 | Fdefine_charset_internal (charset_arg_max, args); | |
1295 | ||
1296 | return XINT (CHARSET_SYMBOL_ID (name)); | |
1297 | } | |
1298 | ||
1299 | ||
3263d5a2 KH |
1300 | DEFUN ("define-charset-alias", Fdefine_charset_alias, |
1301 | Sdefine_charset_alias, 2, 2, 0, | |
1302 | doc: /* Define ALIAS as an alias for charset CHARSET. */) | |
5842a27b | 1303 | (Lisp_Object alias, Lisp_Object charset) |
4ed46869 | 1304 | { |
3263d5a2 KH |
1305 | Lisp_Object attr; |
1306 | ||
1307 | CHECK_CHARSET_GET_ATTR (charset, attr); | |
1308 | Fputhash (alias, attr, Vcharset_hash_table); | |
528623a0 | 1309 | Vcharset_list = Fcons (alias, Vcharset_list); |
3263d5a2 KH |
1310 | return Qnil; |
1311 | } | |
4ed46869 | 1312 | |
4ed46869 | 1313 | |
3263d5a2 | 1314 | DEFUN ("charset-plist", Fcharset_plist, Scharset_plist, 1, 1, 0, |
56a46d1d | 1315 | doc: /* Return the property list of CHARSET. */) |
5842a27b | 1316 | (Lisp_Object charset) |
3263d5a2 KH |
1317 | { |
1318 | Lisp_Object attrs; | |
1319 | ||
1320 | CHECK_CHARSET_GET_ATTR (charset, attrs); | |
1321 | return CHARSET_ATTR_PLIST (attrs); | |
1322 | } | |
1323 | ||
1324 | ||
1325 | DEFUN ("set-charset-plist", Fset_charset_plist, Sset_charset_plist, 2, 2, 0, | |
1326 | doc: /* Set CHARSET's property list to PLIST. */) | |
5842a27b | 1327 | (Lisp_Object charset, Lisp_Object plist) |
3263d5a2 KH |
1328 | { |
1329 | Lisp_Object attrs; | |
1330 | ||
1331 | CHECK_CHARSET_GET_ATTR (charset, attrs); | |
1332 | CHARSET_ATTR_PLIST (attrs) = plist; | |
1333 | return plist; | |
1334 | } | |
1335 | ||
1336 | ||
dbbb237d | 1337 | DEFUN ("unify-charset", Funify_charset, Sunify_charset, 1, 3, 0, |
56a46d1d DL |
1338 | doc: /* Unify characters of CHARSET with Unicode. |
1339 | This means reading the relevant file and installing the table defined | |
dbbb237d KH |
1340 | by CHARSET's `:unify-map' property. |
1341 | ||
64165ae2 DL |
1342 | Optional second arg UNIFY-MAP is a file name string or a vector. It has |
1343 | the same meaning as the `:unify-map' attribute in the function | |
dbbb237d KH |
1344 | `define-charset' (which see). |
1345 | ||
1346 | Optional third argument DEUNIFY, if non-nil, means to de-unify CHARSET. */) | |
5842a27b | 1347 | (Lisp_Object charset, Lisp_Object unify_map, Lisp_Object deunify) |
8a73a704 | 1348 | { |
3263d5a2 KH |
1349 | int id; |
1350 | struct charset *cs; | |
8f924df7 | 1351 | |
3263d5a2 KH |
1352 | CHECK_CHARSET_GET_ID (charset, id); |
1353 | cs = CHARSET_FROM_ID (id); | |
dbbb237d KH |
1354 | if (NILP (deunify) |
1355 | ? CHARSET_UNIFIED_P (cs) && ! NILP (CHARSET_DEUNIFIER (cs)) | |
1356 | : ! CHARSET_UNIFIED_P (cs)) | |
3263d5a2 | 1357 | return Qnil; |
dbbb237d | 1358 | |
3263d5a2 | 1359 | CHARSET_UNIFIED_P (cs) = 0; |
dbbb237d KH |
1360 | if (NILP (deunify)) |
1361 | { | |
ecca2aad KH |
1362 | if (CHARSET_METHOD (cs) != CHARSET_METHOD_OFFSET |
1363 | || CHARSET_CODE_OFFSET (cs) < 0x110000) | |
8f924df7 | 1364 | error ("Can't unify charset: %s", SDATA (SYMBOL_NAME (charset))); |
dbbb237d KH |
1365 | if (NILP (unify_map)) |
1366 | unify_map = CHARSET_UNIFY_MAP (cs); | |
dbbb237d | 1367 | else |
ecca2aad KH |
1368 | { |
1369 | if (! STRINGP (unify_map) && ! VECTORP (unify_map)) | |
1370 | signal_error ("Bad unify-map", unify_map); | |
1371 | CHARSET_UNIFY_MAP (cs) = unify_map; | |
1372 | } | |
1373 | if (NILP (Vchar_unify_table)) | |
1374 | Vchar_unify_table = Fmake_char_table (Qnil, Qnil); | |
1375 | char_table_set_range (Vchar_unify_table, | |
1376 | cs->min_char, cs->max_char, charset); | |
dbbb237d KH |
1377 | CHARSET_UNIFIED_P (cs) = 1; |
1378 | } | |
1379 | else if (CHAR_TABLE_P (Vchar_unify_table)) | |
1380 | { | |
1381 | int min_code = CHARSET_MIN_CODE (cs); | |
1382 | int max_code = CHARSET_MAX_CODE (cs); | |
1383 | int min_char = DECODE_CHAR (cs, min_code); | |
1384 | int max_char = DECODE_CHAR (cs, max_code); | |
8f924df7 | 1385 | |
dbbb237d KH |
1386 | char_table_set_range (Vchar_unify_table, min_char, max_char, Qnil); |
1387 | } | |
8f924df7 | 1388 | |
3263d5a2 | 1389 | return Qnil; |
8a73a704 KH |
1390 | } |
1391 | ||
3fac5a51 KH |
1392 | DEFUN ("get-unused-iso-final-char", Fget_unused_iso_final_char, |
1393 | Sget_unused_iso_final_char, 2, 2, 0, | |
3263d5a2 | 1394 | doc: /* |
d0cf2d48 | 1395 | Return an unused ISO final char for a charset of DIMENSION and CHARS. |
fdb82f93 PJ |
1396 | DIMENSION is the number of bytes to represent a character: 1 or 2. |
1397 | CHARS is the number of characters in a dimension: 94 or 96. | |
1398 | ||
1399 | This final char is for private use, thus the range is `0' (48) .. `?' (63). | |
1721b6af | 1400 | If there's no unused final char for the specified kind of charset, |
fdb82f93 | 1401 | return nil. */) |
5842a27b | 1402 | (Lisp_Object dimension, Lisp_Object chars) |
3fac5a51 KH |
1403 | { |
1404 | int final_char; | |
1405 | ||
b7826503 PJ |
1406 | CHECK_NUMBER (dimension); |
1407 | CHECK_NUMBER (chars); | |
3263d5a2 KH |
1408 | if (XINT (dimension) != 1 && XINT (dimension) != 2 && XINT (dimension) != 3) |
1409 | args_out_of_range_3 (dimension, make_number (1), make_number (3)); | |
3fac5a51 | 1410 | if (XINT (chars) != 94 && XINT (chars) != 96) |
3263d5a2 | 1411 | args_out_of_range_3 (chars, make_number (94), make_number (96)); |
3fac5a51 | 1412 | for (final_char = '0'; final_char <= '?'; final_char++) |
3263d5a2 KH |
1413 | if (ISO_CHARSET_TABLE (XINT (dimension), XINT (chars), final_char) < 0) |
1414 | break; | |
3fac5a51 KH |
1415 | return (final_char <= '?' ? make_number (final_char) : Qnil); |
1416 | } | |
1417 | ||
3263d5a2 | 1418 | static void |
971de7fb | 1419 | check_iso_charset_parameter (Lisp_Object dimension, Lisp_Object chars, Lisp_Object final_char) |
4ed46869 | 1420 | { |
3263d5a2 KH |
1421 | CHECK_NATNUM (dimension); |
1422 | CHECK_NATNUM (chars); | |
e6c3da20 | 1423 | CHECK_CHARACTER (final_char); |
4ed46869 | 1424 | |
3263d5a2 | 1425 | if (XINT (dimension) > 3) |
c2982e87 | 1426 | error ("Invalid DIMENSION %"pI"d, it should be 1, 2, or 3", |
5fdb398c | 1427 | XINT (dimension)); |
4ed46869 | 1428 | if (XINT (chars) != 94 && XINT (chars) != 96) |
c2982e87 | 1429 | error ("Invalid CHARS %"pI"d, it should be 94 or 96", XINT (chars)); |
3263d5a2 | 1430 | if (XINT (final_char) < '0' || XINT (final_char) > '~') |
e6c3da20 EZ |
1431 | error ("Invalid FINAL-CHAR %c, it should be `0'..`~'", |
1432 | (int)XINT (final_char)); | |
3263d5a2 KH |
1433 | } |
1434 | ||
1435 | ||
1436 | DEFUN ("declare-equiv-charset", Fdeclare_equiv_charset, Sdeclare_equiv_charset, | |
1437 | 4, 4, 0, | |
cefd8c4f KH |
1438 | doc: /* Declare an equivalent charset for ISO-2022 decoding. |
1439 | ||
1440 | On decoding by an ISO-2022 base coding system, when a charset | |
1441 | specified by DIMENSION, CHARS, and FINAL-CHAR is designated, behave as | |
1442 | if CHARSET is designated instead. */) | |
5842a27b | 1443 | (Lisp_Object dimension, Lisp_Object chars, Lisp_Object final_char, Lisp_Object charset) |
3263d5a2 KH |
1444 | { |
1445 | int id; | |
82215ce9 | 1446 | int chars_flag; |
4ed46869 | 1447 | |
3263d5a2 KH |
1448 | CHECK_CHARSET_GET_ID (charset, id); |
1449 | check_iso_charset_parameter (dimension, chars, final_char); | |
82215ce9 KH |
1450 | chars_flag = XINT (chars) == 96; |
1451 | ISO_CHARSET_TABLE (XINT (dimension), chars_flag, XINT (final_char)) = id; | |
4ed46869 KH |
1452 | return Qnil; |
1453 | } | |
1454 | ||
3263d5a2 | 1455 | |
2e344af3 KH |
1456 | /* Return information about charsets in the text at PTR of NBYTES |
1457 | bytes, which are NCHARS characters. The value is: | |
f6302ac9 | 1458 | |
cfe34140 | 1459 | 0: Each character is represented by one byte. This is always |
3263d5a2 KH |
1460 | true for a unibyte string. For a multibyte string, true if |
1461 | it contains only ASCII characters. | |
1462 | ||
28c026cd DL |
1463 | 1: No charsets other than ascii, control-1, and latin-1 are |
1464 | found. | |
1d67c29b | 1465 | |
3263d5a2 KH |
1466 | 2: Otherwise. |
1467 | */ | |
4ed46869 KH |
1468 | |
1469 | int | |
971de7fb | 1470 | string_xstring_p (Lisp_Object string) |
4ed46869 | 1471 | { |
8f924df7 KH |
1472 | const unsigned char *p = SDATA (string); |
1473 | const unsigned char *endp = p + SBYTES (string); | |
3263d5a2 | 1474 | |
8f924df7 | 1475 | if (SCHARS (string) == SBYTES (string)) |
3263d5a2 KH |
1476 | return 0; |
1477 | ||
3263d5a2 | 1478 | while (p < endp) |
0282eb69 | 1479 | { |
3263d5a2 | 1480 | int c = STRING_CHAR_ADVANCE (p); |
2e344af3 | 1481 | |
3cc67a4d | 1482 | if (c >= 0x100) |
3263d5a2 | 1483 | return 2; |
0282eb69 | 1484 | } |
3263d5a2 KH |
1485 | return 1; |
1486 | } | |
05505664 | 1487 | |
05505664 | 1488 | |
3263d5a2 | 1489 | /* Find charsets in the string at PTR of NCHARS and NBYTES. |
4ed46869 | 1490 | |
3cc67a4d KH |
1491 | CHARSETS is a vector. If Nth element is non-nil, it means the |
1492 | charset whose id is N is already found. | |
2e344af3 | 1493 | |
3263d5a2 | 1494 | It may lookup a translation table TABLE if supplied. */ |
2e344af3 | 1495 | |
3263d5a2 | 1496 | static void |
971de7fb | 1497 | find_charsets_in_text (const unsigned char *ptr, EMACS_INT nchars, EMACS_INT nbytes, Lisp_Object charsets, Lisp_Object table, int multibyte) |
3263d5a2 | 1498 | { |
dbbb237d | 1499 | const unsigned char *pend = ptr + nbytes; |
3263d5a2 KH |
1500 | |
1501 | if (nchars == nbytes) | |
3263d5a2 | 1502 | { |
3cc67a4d KH |
1503 | if (multibyte) |
1504 | ASET (charsets, charset_ascii, Qt); | |
1505 | else | |
1506 | while (ptr < pend) | |
1507 | { | |
1508 | int c = *ptr++; | |
1509 | ||
1510 | if (!NILP (table)) | |
1511 | c = translate_char (table, c); | |
1512 | if (ASCII_BYTE_P (c)) | |
1513 | ASET (charsets, charset_ascii, Qt); | |
1514 | else | |
1515 | ASET (charsets, charset_eight_bit, Qt); | |
1516 | } | |
1517 | } | |
1518 | else | |
1519 | { | |
1520 | while (ptr < pend) | |
3263d5a2 | 1521 | { |
3cc67a4d KH |
1522 | int c = STRING_CHAR_ADVANCE (ptr); |
1523 | struct charset *charset; | |
3263d5a2 | 1524 | |
3cc67a4d KH |
1525 | if (!NILP (table)) |
1526 | c = translate_char (table, c); | |
1527 | charset = CHAR_CHARSET (c); | |
1528 | ASET (charsets, CHARSET_ID (charset), Qt); | |
4ed46869 | 1529 | } |
4ed46869 | 1530 | } |
4ed46869 KH |
1531 | } |
1532 | ||
1533 | DEFUN ("find-charset-region", Ffind_charset_region, Sfind_charset_region, | |
23d2a7f1 | 1534 | 2, 3, 0, |
fdb82f93 PJ |
1535 | doc: /* Return a list of charsets in the region between BEG and END. |
1536 | BEG and END are buffer positions. | |
1537 | Optional arg TABLE if non-nil is a translation table to look up. | |
1538 | ||
fdb82f93 PJ |
1539 | If the current buffer is unibyte, the returned list may contain |
1540 | only `ascii', `eight-bit-control', and `eight-bit-graphic'. */) | |
5842a27b | 1541 | (Lisp_Object beg, Lisp_Object end, Lisp_Object table) |
4ed46869 | 1542 | { |
3263d5a2 | 1543 | Lisp_Object charsets; |
42ca828e DL |
1544 | EMACS_INT from, from_byte, to, stop, stop_byte; |
1545 | int i; | |
4ed46869 | 1546 | Lisp_Object val; |
4b4deea2 | 1547 | int multibyte = ! NILP (BVAR (current_buffer, enable_multibyte_characters)); |
4ed46869 KH |
1548 | |
1549 | validate_region (&beg, &end); | |
1550 | from = XFASTINT (beg); | |
1551 | stop = to = XFASTINT (end); | |
6ae1f27e | 1552 | |
4ed46869 | 1553 | if (from < GPT && GPT < to) |
6ae1f27e RS |
1554 | { |
1555 | stop = GPT; | |
1556 | stop_byte = GPT_BYTE; | |
1557 | } | |
1558 | else | |
1559 | stop_byte = CHAR_TO_BYTE (stop); | |
1560 | ||
1561 | from_byte = CHAR_TO_BYTE (from); | |
1562 | ||
3263d5a2 | 1563 | charsets = Fmake_vector (make_number (charset_table_used), Qnil); |
4ed46869 KH |
1564 | while (1) |
1565 | { | |
3263d5a2 | 1566 | find_charsets_in_text (BYTE_POS_ADDR (from_byte), stop - from, |
3cc67a4d KH |
1567 | stop_byte - from_byte, charsets, table, |
1568 | multibyte); | |
4ed46869 | 1569 | if (stop < to) |
6ae1f27e RS |
1570 | { |
1571 | from = stop, from_byte = stop_byte; | |
1572 | stop = to, stop_byte = CHAR_TO_BYTE (stop); | |
1573 | } | |
4ed46869 KH |
1574 | else |
1575 | break; | |
1576 | } | |
6ae1f27e | 1577 | |
4ed46869 | 1578 | val = Qnil; |
3263d5a2 | 1579 | for (i = charset_table_used - 1; i >= 0; i--) |
3cc67a4d | 1580 | if (!NILP (AREF (charsets, i))) |
3263d5a2 | 1581 | val = Fcons (CHARSET_NAME (charset_table + i), val); |
4ed46869 KH |
1582 | return val; |
1583 | } | |
1584 | ||
1585 | DEFUN ("find-charset-string", Ffind_charset_string, Sfind_charset_string, | |
23d2a7f1 | 1586 | 1, 2, 0, |
fdb82f93 PJ |
1587 | doc: /* Return a list of charsets in STR. |
1588 | Optional arg TABLE if non-nil is a translation table to look up. | |
1589 | ||
fdb82f93 | 1590 | If STR is unibyte, the returned list may contain |
3263d5a2 | 1591 | only `ascii', `eight-bit-control', and `eight-bit-graphic'. */) |
5842a27b | 1592 | (Lisp_Object str, Lisp_Object table) |
4ed46869 | 1593 | { |
3263d5a2 | 1594 | Lisp_Object charsets; |
4ed46869 KH |
1595 | int i; |
1596 | Lisp_Object val; | |
1597 | ||
b7826503 | 1598 | CHECK_STRING (str); |
87b089ad | 1599 | |
3263d5a2 | 1600 | charsets = Fmake_vector (make_number (charset_table_used), Qnil); |
8f924df7 | 1601 | find_charsets_in_text (SDATA (str), SCHARS (str), SBYTES (str), |
3cc67a4d KH |
1602 | charsets, table, |
1603 | STRING_MULTIBYTE (str)); | |
4ed46869 | 1604 | val = Qnil; |
3263d5a2 | 1605 | for (i = charset_table_used - 1; i >= 0; i--) |
3cc67a4d | 1606 | if (!NILP (AREF (charsets, i))) |
3263d5a2 | 1607 | val = Fcons (CHARSET_NAME (charset_table + i), val); |
4ed46869 KH |
1608 | return val; |
1609 | } | |
2e344af3 | 1610 | |
4ed46869 | 1611 | \f |
3263d5a2 | 1612 | |
ecca2aad KH |
1613 | /* Return a unified character code for C (>= 0x110000). VAL is a |
1614 | value of Vchar_unify_table for C; i.e. it is nil, an integer, or a | |
1615 | charset symbol. */ | |
1616 | int | |
971de7fb | 1617 | maybe_unify_char (int c, Lisp_Object val) |
ecca2aad KH |
1618 | { |
1619 | struct charset *charset; | |
1620 | ||
1621 | if (INTEGERP (val)) | |
ccd9a01a | 1622 | return XFASTINT (val); |
ecca2aad KH |
1623 | if (NILP (val)) |
1624 | return c; | |
1625 | ||
1626 | CHECK_CHARSET_GET_CHARSET (val, charset); | |
1627 | load_charset (charset, 1); | |
1628 | if (! inhibit_load_charset_map) | |
1629 | { | |
1630 | val = CHAR_TABLE_REF (Vchar_unify_table, c); | |
1631 | if (! NILP (val)) | |
ccd9a01a | 1632 | c = XFASTINT (val); |
ecca2aad KH |
1633 | } |
1634 | else | |
1635 | { | |
1636 | int code_index = c - CHARSET_CODE_OFFSET (charset); | |
1637 | int unified = GET_TEMP_CHARSET_WORK_DECODER (code_index); | |
1638 | ||
1639 | if (unified > 0) | |
1640 | c = unified; | |
1641 | } | |
1642 | return c; | |
1643 | } | |
1644 | ||
1645 | ||
3263d5a2 KH |
1646 | /* Return a character correponding to the code-point CODE of |
1647 | CHARSET. */ | |
1648 | ||
1649 | int | |
971de7fb | 1650 | decode_char (struct charset *charset, unsigned int code) |
4ed46869 | 1651 | { |
3263d5a2 KH |
1652 | int c, char_index; |
1653 | enum charset_method method = CHARSET_METHOD (charset); | |
ac4137cc | 1654 | |
3263d5a2 KH |
1655 | if (code < CHARSET_MIN_CODE (charset) || code > CHARSET_MAX_CODE (charset)) |
1656 | return -1; | |
4ed46869 | 1657 | |
374c5cfd | 1658 | if (method == CHARSET_METHOD_SUBSET) |
2e344af3 | 1659 | { |
374c5cfd KH |
1660 | Lisp_Object subset_info; |
1661 | ||
1662 | subset_info = CHARSET_SUBSET (charset); | |
1663 | charset = CHARSET_FROM_ID (XFASTINT (AREF (subset_info, 0))); | |
1664 | code -= XINT (AREF (subset_info, 3)); | |
1665 | if (code < XFASTINT (AREF (subset_info, 1)) | |
1666 | || code > XFASTINT (AREF (subset_info, 2))) | |
1667 | c = -1; | |
1668 | else | |
1669 | c = DECODE_CHAR (charset, code); | |
2e344af3 | 1670 | } |
374c5cfd | 1671 | else if (method == CHARSET_METHOD_SUPERSET) |
2e344af3 | 1672 | { |
3263d5a2 | 1673 | Lisp_Object parents; |
4ed46869 | 1674 | |
374c5cfd | 1675 | parents = CHARSET_SUPERSET (charset); |
3263d5a2 KH |
1676 | c = -1; |
1677 | for (; CONSP (parents); parents = XCDR (parents)) | |
1678 | { | |
1679 | int id = XINT (XCAR (XCAR (parents))); | |
1680 | int code_offset = XINT (XCDR (XCAR (parents))); | |
374c5cfd | 1681 | unsigned this_code = code - code_offset; |
4ed46869 | 1682 | |
3263d5a2 KH |
1683 | charset = CHARSET_FROM_ID (id); |
1684 | if ((c = DECODE_CHAR (charset, this_code)) >= 0) | |
1685 | break; | |
1686 | } | |
1687 | } | |
1688 | else | |
ac4137cc | 1689 | { |
3263d5a2 | 1690 | char_index = CODE_POINT_TO_INDEX (charset, code); |
69f8de5b KH |
1691 | if (char_index < 0) |
1692 | return -1; | |
4ed46869 | 1693 | |
3263d5a2 | 1694 | if (method == CHARSET_METHOD_MAP) |
ac4137cc | 1695 | { |
3263d5a2 | 1696 | Lisp_Object decoder; |
4ed46869 | 1697 | |
3263d5a2 KH |
1698 | decoder = CHARSET_DECODER (charset); |
1699 | if (! VECTORP (decoder)) | |
ecca2aad KH |
1700 | { |
1701 | load_charset (charset, 1); | |
1702 | decoder = CHARSET_DECODER (charset); | |
1703 | } | |
1704 | if (VECTORP (decoder)) | |
1705 | c = XINT (AREF (decoder, char_index)); | |
1706 | else | |
1707 | c = GET_TEMP_CHARSET_WORK_DECODER (char_index); | |
ac4137cc | 1708 | } |
ecca2aad | 1709 | else /* method == CHARSET_METHOD_OFFSET */ |
ac4137cc | 1710 | { |
3263d5a2 | 1711 | c = char_index + CHARSET_CODE_OFFSET (charset); |
ecca2aad KH |
1712 | if (CHARSET_UNIFIED_P (charset) |
1713 | && c > MAX_UNICODE_CHAR) | |
1714 | MAYBE_UNIFY_CHAR (c); | |
ac4137cc KH |
1715 | } |
1716 | } | |
4ed46869 | 1717 | |
3263d5a2 | 1718 | return c; |
90d7b74e KH |
1719 | } |
1720 | ||
374c5cfd KH |
1721 | /* Variable used temporarily by the macro ENCODE_CHAR. */ |
1722 | Lisp_Object charset_work; | |
4ed46869 | 1723 | |
3263d5a2 | 1724 | /* Return a code-point of CHAR in CHARSET. If CHAR doesn't belong to |
28c026cd DL |
1725 | CHARSET, return CHARSET_INVALID_CODE (CHARSET). If STRICT is true, |
1726 | use CHARSET's strict_max_char instead of max_char. */ | |
4ed46869 | 1727 | |
3263d5a2 | 1728 | unsigned |
971de7fb | 1729 | encode_char (struct charset *charset, int c) |
9d3d8cba | 1730 | { |
3263d5a2 KH |
1731 | unsigned code; |
1732 | enum charset_method method = CHARSET_METHOD (charset); | |
9d3d8cba | 1733 | |
3263d5a2 | 1734 | if (CHARSET_UNIFIED_P (charset)) |
ac4137cc | 1735 | { |
6809ca75 | 1736 | Lisp_Object deunifier; |
ecca2aad | 1737 | int code_index = -1; |
4ed46869 | 1738 | |
3263d5a2 KH |
1739 | deunifier = CHARSET_DEUNIFIER (charset); |
1740 | if (! CHAR_TABLE_P (deunifier)) | |
ac4137cc | 1741 | { |
ecca2aad | 1742 | load_charset (charset, 2); |
3263d5a2 | 1743 | deunifier = CHARSET_DEUNIFIER (charset); |
ac4137cc | 1744 | } |
ecca2aad KH |
1745 | if (CHAR_TABLE_P (deunifier)) |
1746 | { | |
1747 | Lisp_Object deunified = CHAR_TABLE_REF (deunifier, c); | |
1748 | ||
1749 | if (INTEGERP (deunified)) | |
1750 | code_index = XINT (deunified); | |
1751 | } | |
1752 | else | |
1753 | { | |
1754 | code_index = GET_TEMP_CHARSET_WORK_ENCODER (c); | |
1755 | } | |
1756 | if (code_index >= 0) | |
1757 | c = CHARSET_CODE_OFFSET (charset) + code_index; | |
ac4137cc | 1758 | } |
9d3d8cba | 1759 | |
374c5cfd KH |
1760 | if (method == CHARSET_METHOD_SUBSET) |
1761 | { | |
1762 | Lisp_Object subset_info; | |
1763 | struct charset *this_charset; | |
1764 | ||
1765 | subset_info = CHARSET_SUBSET (charset); | |
1766 | this_charset = CHARSET_FROM_ID (XFASTINT (AREF (subset_info, 0))); | |
1767 | code = ENCODE_CHAR (this_charset, c); | |
1768 | if (code == CHARSET_INVALID_CODE (this_charset) | |
1769 | || code < XFASTINT (AREF (subset_info, 1)) | |
1770 | || code > XFASTINT (AREF (subset_info, 2))) | |
1771 | return CHARSET_INVALID_CODE (charset); | |
1772 | code += XINT (AREF (subset_info, 3)); | |
1773 | return code; | |
1774 | } | |
9d3d8cba | 1775 | |
374c5cfd | 1776 | if (method == CHARSET_METHOD_SUPERSET) |
859f2b3c | 1777 | { |
3263d5a2 | 1778 | Lisp_Object parents; |
d2665018 | 1779 | |
374c5cfd | 1780 | parents = CHARSET_SUPERSET (charset); |
3263d5a2 | 1781 | for (; CONSP (parents); parents = XCDR (parents)) |
beeedaad | 1782 | { |
3263d5a2 KH |
1783 | int id = XINT (XCAR (XCAR (parents))); |
1784 | int code_offset = XINT (XCDR (XCAR (parents))); | |
1785 | struct charset *this_charset = CHARSET_FROM_ID (id); | |
d2665018 | 1786 | |
3263d5a2 | 1787 | code = ENCODE_CHAR (this_charset, c); |
dbbb237d KH |
1788 | if (code != CHARSET_INVALID_CODE (this_charset)) |
1789 | return code + code_offset; | |
beeedaad | 1790 | } |
3263d5a2 KH |
1791 | return CHARSET_INVALID_CODE (charset); |
1792 | } | |
1bcc1567 | 1793 | |
15c85a88 KH |
1794 | if (! CHARSET_FAST_MAP_REF ((c), charset->fast_map) |
1795 | || c < CHARSET_MIN_CHAR (charset) || c > CHARSET_MAX_CHAR (charset)) | |
1796 | return CHARSET_INVALID_CODE (charset); | |
1bcc1567 | 1797 | |
3263d5a2 | 1798 | if (method == CHARSET_METHOD_MAP) |
3f62427c | 1799 | { |
3263d5a2 | 1800 | Lisp_Object encoder; |
beeedaad | 1801 | Lisp_Object val; |
9b6a601f | 1802 | |
3263d5a2 KH |
1803 | encoder = CHARSET_ENCODER (charset); |
1804 | if (! CHAR_TABLE_P (CHARSET_ENCODER (charset))) | |
b8ebe9dd KH |
1805 | { |
1806 | load_charset (charset, 2); | |
1807 | encoder = CHARSET_ENCODER (charset); | |
1808 | } | |
1809 | if (CHAR_TABLE_P (encoder)) | |
ecca2aad KH |
1810 | { |
1811 | val = CHAR_TABLE_REF (encoder, c); | |
1812 | if (NILP (val)) | |
1813 | return CHARSET_INVALID_CODE (charset); | |
1814 | code = XINT (val); | |
1815 | if (! CHARSET_COMPACT_CODES_P (charset)) | |
1816 | code = INDEX_TO_CODE_POINT (charset, code); | |
1817 | } | |
1818 | else | |
1819 | { | |
1820 | code = GET_TEMP_CHARSET_WORK_ENCODER (c); | |
51b59d79 | 1821 | code = INDEX_TO_CODE_POINT (charset, code); |
ecca2aad | 1822 | } |
3263d5a2 | 1823 | } |
820ee249 | 1824 | else /* method == CHARSET_METHOD_OFFSET */ |
beeedaad | 1825 | { |
ecca2aad KH |
1826 | int code_index = c - CHARSET_CODE_OFFSET (charset); |
1827 | ||
1828 | code = INDEX_TO_CODE_POINT (charset, code_index); | |
3f62427c | 1829 | } |
8ac5a9cc | 1830 | |
3263d5a2 | 1831 | return code; |
4ed46869 KH |
1832 | } |
1833 | ||
4ed46869 | 1834 | |
3263d5a2 KH |
1835 | DEFUN ("decode-char", Fdecode_char, Sdecode_char, 2, 3, 0, |
1836 | doc: /* Decode the pair of CHARSET and CODE-POINT into a character. | |
1837 | Return nil if CODE-POINT is not valid in CHARSET. | |
4ed46869 | 1838 | |
3263d5a2 | 1839 | CODE-POINT may be a cons (HIGHER-16-BIT-VALUE . LOWER-16-BIT-VALUE). |
859f2b3c | 1840 | |
3263d5a2 | 1841 | Optional argument RESTRICTION specifies a way to map the pair of CCS |
d0cf2d48 | 1842 | and CODE-POINT to a character. Currently not supported and just ignored. */) |
5842a27b | 1843 | (Lisp_Object charset, Lisp_Object code_point, Lisp_Object restriction) |
4ed46869 | 1844 | { |
3263d5a2 KH |
1845 | int c, id; |
1846 | unsigned code; | |
1847 | struct charset *charsetp; | |
859f2b3c | 1848 | |
3263d5a2 | 1849 | CHECK_CHARSET_GET_ID (charset, id); |
be44ca6c | 1850 | code = cons_to_unsigned (code_point, UINT_MAX); |
3263d5a2 KH |
1851 | charsetp = CHARSET_FROM_ID (id); |
1852 | c = DECODE_CHAR (charsetp, code); | |
1853 | return (c >= 0 ? make_number (c) : Qnil); | |
4ed46869 KH |
1854 | } |
1855 | ||
859f2b3c | 1856 | |
3263d5a2 KH |
1857 | DEFUN ("encode-char", Fencode_char, Sencode_char, 2, 3, 0, |
1858 | doc: /* Encode the character CH into a code-point of CHARSET. | |
1859 | Return nil if CHARSET doesn't include CH. | |
17e7ef1b | 1860 | |
d0cf2d48 | 1861 | Optional argument RESTRICTION specifies a way to map CH to a |
3263d5a2 | 1862 | code-point in CCS. Currently not supported and just ignored. */) |
5842a27b | 1863 | (Lisp_Object ch, Lisp_Object charset, Lisp_Object restriction) |
4ed46869 | 1864 | { |
16fed1fc | 1865 | int id; |
3263d5a2 KH |
1866 | unsigned code; |
1867 | struct charset *charsetp; | |
046b1f03 | 1868 | |
3263d5a2 KH |
1869 | CHECK_CHARSET_GET_ID (charset, id); |
1870 | CHECK_NATNUM (ch); | |
3263d5a2 | 1871 | charsetp = CHARSET_FROM_ID (id); |
16fed1fc | 1872 | code = ENCODE_CHAR (charsetp, XINT (ch)); |
3263d5a2 KH |
1873 | if (code == CHARSET_INVALID_CODE (charsetp)) |
1874 | return Qnil; | |
be44ca6c | 1875 | return INTEGER_TO_CONS (code); |
beeedaad KH |
1876 | } |
1877 | ||
beeedaad | 1878 | |
b121a744 KH |
1879 | DEFUN ("make-char", Fmake_char, Smake_char, 1, 5, 0, |
1880 | doc: | |
1881 | /* Return a character of CHARSET whose position codes are CODEn. | |
1882 | ||
1883 | CODE1 through CODE4 are optional, but if you don't supply sufficient | |
1884 | position codes, it is assumed that the minimum code in each dimension | |
04c2f2c5 | 1885 | is specified. */) |
5842a27b | 1886 | (Lisp_Object charset, Lisp_Object code1, Lisp_Object code2, Lisp_Object code3, Lisp_Object code4) |
beeedaad | 1887 | { |
3263d5a2 KH |
1888 | int id, dimension; |
1889 | struct charset *charsetp; | |
b121a744 KH |
1890 | unsigned code; |
1891 | int c; | |
87b089ad | 1892 | |
3263d5a2 KH |
1893 | CHECK_CHARSET_GET_ID (charset, id); |
1894 | charsetp = CHARSET_FROM_ID (id); | |
4ed46869 | 1895 | |
b121a744 KH |
1896 | dimension = CHARSET_DIMENSION (charsetp); |
1897 | if (NILP (code1)) | |
d47073ca KH |
1898 | code = (CHARSET_ASCII_COMPATIBLE_P (charsetp) |
1899 | ? 0 : CHARSET_MIN_CODE (charsetp)); | |
3263d5a2 | 1900 | else |
859f2b3c | 1901 | { |
b121a744 KH |
1902 | CHECK_NATNUM (code1); |
1903 | if (XFASTINT (code1) >= 0x100) | |
1904 | args_out_of_range (make_number (0xFF), code1); | |
1905 | code = XFASTINT (code1); | |
859f2b3c | 1906 | |
b0a1e45e | 1907 | if (dimension > 1) |
beeedaad | 1908 | { |
b121a744 | 1909 | code <<= 8; |
b0a1e45e KH |
1910 | if (NILP (code2)) |
1911 | code |= charsetp->code_space[(dimension - 2) * 4]; | |
beeedaad | 1912 | else |
b121a744 | 1913 | { |
b0a1e45e KH |
1914 | CHECK_NATNUM (code2); |
1915 | if (XFASTINT (code2) >= 0x100) | |
1916 | args_out_of_range (make_number (0xFF), code2); | |
1917 | code |= XFASTINT (code2); | |
b121a744 | 1918 | } |
99529c2c | 1919 | |
b0a1e45e | 1920 | if (dimension > 2) |
b121a744 KH |
1921 | { |
1922 | code <<= 8; | |
b0a1e45e KH |
1923 | if (NILP (code3)) |
1924 | code |= charsetp->code_space[(dimension - 3) * 4]; | |
b121a744 KH |
1925 | else |
1926 | { | |
b0a1e45e KH |
1927 | CHECK_NATNUM (code3); |
1928 | if (XFASTINT (code3) >= 0x100) | |
1929 | args_out_of_range (make_number (0xFF), code3); | |
1930 | code |= XFASTINT (code3); | |
1931 | } | |
1932 | ||
1933 | if (dimension > 3) | |
1934 | { | |
1935 | code <<= 8; | |
1936 | if (NILP (code4)) | |
1937 | code |= charsetp->code_space[0]; | |
1938 | else | |
1939 | { | |
1940 | CHECK_NATNUM (code4); | |
1941 | if (XFASTINT (code4) >= 0x100) | |
1942 | args_out_of_range (make_number (0xFF), code4); | |
1943 | code |= XFASTINT (code4); | |
1944 | } | |
b121a744 KH |
1945 | } |
1946 | } | |
beeedaad | 1947 | } |
859f2b3c | 1948 | } |
beeedaad | 1949 | |
b121a744 KH |
1950 | if (CHARSET_ISO_FINAL (charsetp) >= 0) |
1951 | code &= 0x7F7F7F7F; | |
1952 | c = DECODE_CHAR (charsetp, code); | |
1953 | if (c < 0) | |
1954 | error ("Invalid code(s)"); | |
3263d5a2 | 1955 | return make_number (c); |
4ed46869 KH |
1956 | } |
1957 | ||
beeedaad | 1958 | |
3263d5a2 KH |
1959 | /* Return the first charset in CHARSET_LIST that contains C. |
1960 | CHARSET_LIST is a list of charset IDs. If it is nil, use | |
1961 | Vcharset_ordered_list. */ | |
beeedaad | 1962 | |
3263d5a2 | 1963 | struct charset * |
971de7fb | 1964 | char_charset (int c, Lisp_Object charset_list, unsigned int *code_return) |
2e344af3 | 1965 | { |
8a1816bb KH |
1966 | int maybe_null = 0; |
1967 | ||
3263d5a2 KH |
1968 | if (NILP (charset_list)) |
1969 | charset_list = Vcharset_ordered_list; | |
8a1816bb KH |
1970 | else |
1971 | maybe_null = 1; | |
beeedaad | 1972 | |
6c652beb | 1973 | while (CONSP (charset_list)) |
2e344af3 | 1974 | { |
3263d5a2 KH |
1975 | struct charset *charset = CHARSET_FROM_ID (XINT (XCAR (charset_list))); |
1976 | unsigned code = ENCODE_CHAR (charset, c); | |
beeedaad | 1977 | |
3263d5a2 | 1978 | if (code != CHARSET_INVALID_CODE (charset)) |
beeedaad | 1979 | { |
3263d5a2 KH |
1980 | if (code_return) |
1981 | *code_return = code; | |
1982 | return charset; | |
3f62427c | 1983 | } |
3263d5a2 | 1984 | charset_list = XCDR (charset_list); |
c0be27fd KH |
1985 | if (! maybe_null |
1986 | && c <= MAX_UNICODE_CHAR | |
1987 | && EQ (charset_list, Vcharset_non_preferred_head)) | |
6c652beb | 1988 | return CHARSET_FROM_ID (charset_unicode); |
3f62427c | 1989 | } |
8a1816bb KH |
1990 | return (maybe_null ? NULL |
1991 | : c <= MAX_5_BYTE_CHAR ? CHARSET_FROM_ID (charset_emacs) | |
6a9c90ec | 1992 | : CHARSET_FROM_ID (charset_eight_bit)); |
3f62427c KH |
1993 | } |
1994 | ||
2e344af3 | 1995 | |
3263d5a2 | 1996 | DEFUN ("split-char", Fsplit_char, Ssplit_char, 1, 1, 0, |
3cc67a4d | 1997 | doc: |
d0cf2d48 | 1998 | /*Return list of charset and one to four position-codes of CH. |
3cc67a4d KH |
1999 | The charset is decided by the current priority order of charsets. |
2000 | A position-code is a byte value of each dimension of the code-point of | |
d0cf2d48 | 2001 | CH in the charset. */) |
5842a27b | 2002 | (Lisp_Object ch) |
4ed46869 | 2003 | { |
3263d5a2 KH |
2004 | struct charset *charset; |
2005 | int c, dimension; | |
2006 | unsigned code; | |
4ed46869 KH |
2007 | Lisp_Object val; |
2008 | ||
3263d5a2 KH |
2009 | CHECK_CHARACTER (ch); |
2010 | c = XFASTINT (ch); | |
2011 | charset = CHAR_CHARSET (c); | |
2012 | if (! charset) | |
3cc67a4d | 2013 | abort (); |
3263d5a2 KH |
2014 | code = ENCODE_CHAR (charset, c); |
2015 | if (code == CHARSET_INVALID_CODE (charset)) | |
2016 | abort (); | |
2017 | dimension = CHARSET_DIMENSION (charset); | |
3cc67a4d KH |
2018 | for (val = Qnil; dimension > 0; dimension--) |
2019 | { | |
2020 | val = Fcons (make_number (code & 0xFF), val); | |
2021 | code >>= 8; | |
2022 | } | |
3263d5a2 | 2023 | return Fcons (CHARSET_NAME (charset), val); |
4ed46869 KH |
2024 | } |
2025 | ||
740f080d | 2026 | |
4cb75c4b KH |
2027 | DEFUN ("char-charset", Fchar_charset, Schar_charset, 1, 2, 0, |
2028 | doc: /* Return the charset of highest priority that contains CH. | |
2029 | If optional 2nd arg RESTRICTION is non-nil, it is a list of charsets | |
2030 | from which to find the charset. It may also be a coding system. In | |
2031 | that case, find the charset from what supported by that coding system. */) | |
5842a27b | 2032 | (Lisp_Object ch, Lisp_Object restriction) |
4ed46869 | 2033 | { |
3263d5a2 | 2034 | struct charset *charset; |
4ed46869 | 2035 | |
3263d5a2 | 2036 | CHECK_CHARACTER (ch); |
4cb75c4b KH |
2037 | if (NILP (restriction)) |
2038 | charset = CHAR_CHARSET (XINT (ch)); | |
2039 | else | |
2040 | { | |
4cb75c4b KH |
2041 | if (CONSP (restriction)) |
2042 | { | |
7a84eee5 KH |
2043 | int c = XFASTINT (ch); |
2044 | ||
2045 | for (; CONSP (restriction); restriction = XCDR (restriction)) | |
4cb75c4b | 2046 | { |
f6095868 | 2047 | struct charset *rcharset; |
4cb75c4b | 2048 | |
f6095868 PE |
2049 | CHECK_CHARSET_GET_CHARSET (XCAR (restriction), rcharset); |
2050 | if (ENCODE_CHAR (rcharset, c) != CHARSET_INVALID_CODE (rcharset)) | |
7a84eee5 | 2051 | return XCAR (restriction); |
4cb75c4b | 2052 | } |
7a84eee5 | 2053 | return Qnil; |
4cb75c4b | 2054 | } |
7a84eee5 KH |
2055 | restriction = coding_system_charset_list (restriction); |
2056 | charset = char_charset (XINT (ch), restriction, NULL); | |
4cb75c4b KH |
2057 | if (! charset) |
2058 | return Qnil; | |
2059 | } | |
3263d5a2 | 2060 | return (CHARSET_NAME (charset)); |
4ed46869 KH |
2061 | } |
2062 | ||
17e7ef1b | 2063 | |
3263d5a2 KH |
2064 | DEFUN ("charset-after", Fcharset_after, Scharset_after, 0, 1, 0, |
2065 | doc: /* | |
2066 | Return charset of a character in the current buffer at position POS. | |
2067 | If POS is nil, it defauls to the current point. | |
2068 | If POS is out of range, the value is nil. */) | |
5842a27b | 2069 | (Lisp_Object pos) |
2e344af3 | 2070 | { |
3263d5a2 KH |
2071 | Lisp_Object ch; |
2072 | struct charset *charset; | |
046b1f03 | 2073 | |
3263d5a2 KH |
2074 | ch = Fchar_after (pos); |
2075 | if (! INTEGERP (ch)) | |
2076 | return ch; | |
2077 | charset = CHAR_CHARSET (XINT (ch)); | |
2078 | return (CHARSET_NAME (charset)); | |
6ae1f27e | 2079 | } |
9036eb45 | 2080 | |
87b089ad | 2081 | |
3263d5a2 KH |
2082 | DEFUN ("iso-charset", Fiso_charset, Siso_charset, 3, 3, 0, |
2083 | doc: /* | |
2084 | Return charset of ISO's specification DIMENSION, CHARS, and FINAL-CHAR. | |
2085 | ||
2086 | ISO 2022's designation sequence (escape sequence) distinguishes charsets | |
2087 | by their DIMENSION, CHARS, and FINAL-CHAR, | |
d0cf2d48 | 2088 | whereas Emacs distinguishes them by charset symbol. |
3263d5a2 KH |
2089 | See the documentation of the function `charset-info' for the meanings of |
2090 | DIMENSION, CHARS, and FINAL-CHAR. */) | |
5842a27b | 2091 | (Lisp_Object dimension, Lisp_Object chars, Lisp_Object final_char) |
6ae1f27e | 2092 | { |
3263d5a2 | 2093 | int id; |
82215ce9 | 2094 | int chars_flag; |
a8a35e61 | 2095 | |
3263d5a2 | 2096 | check_iso_charset_parameter (dimension, chars, final_char); |
82215ce9 KH |
2097 | chars_flag = XFASTINT (chars) == 96; |
2098 | id = ISO_CHARSET_TABLE (XFASTINT (dimension), chars_flag, | |
3263d5a2 KH |
2099 | XFASTINT (final_char)); |
2100 | return (id >= 0 ? CHARSET_NAME (CHARSET_FROM_ID (id)) : Qnil); | |
046b1f03 RS |
2101 | } |
2102 | ||
87b089ad | 2103 | |
3263d5a2 KH |
2104 | DEFUN ("clear-charset-maps", Fclear_charset_maps, Sclear_charset_maps, |
2105 | 0, 0, 0, | |
2106 | doc: /* | |
ecca2aad KH |
2107 | Internal use only. |
2108 | Clear temporary charset mapping tables. | |
2109 | It should be called only from temacs invoked for dumping. */) | |
5842a27b | 2110 | (void) |
87b089ad | 2111 | { |
ecca2aad | 2112 | if (temp_charset_work) |
87b089ad | 2113 | { |
baad03f0 | 2114 | xfree (temp_charset_work); |
ecca2aad | 2115 | temp_charset_work = NULL; |
2e344af3 | 2116 | } |
2e344af3 | 2117 | |
ecca2aad KH |
2118 | if (CHAR_TABLE_P (Vchar_unify_table)) |
2119 | Foptimize_char_table (Vchar_unify_table, Qnil); | |
740f080d | 2120 | |
3263d5a2 | 2121 | return Qnil; |
740f080d KH |
2122 | } |
2123 | ||
8ddf5e57 DL |
2124 | DEFUN ("charset-priority-list", Fcharset_priority_list, |
2125 | Scharset_priority_list, 0, 1, 0, | |
2126 | doc: /* Return the list of charsets ordered by priority. | |
2127 | HIGHESTP non-nil means just return the highest priority one. */) | |
5842a27b | 2128 | (Lisp_Object highestp) |
2e344af3 | 2129 | { |
8ddf5e57 | 2130 | Lisp_Object val = Qnil, list = Vcharset_ordered_list; |
2e344af3 | 2131 | |
8ddf5e57 | 2132 | if (!NILP (highestp)) |
16fed1fc | 2133 | return CHARSET_NAME (CHARSET_FROM_ID (XINT (Fcar (list)))); |
2e344af3 | 2134 | |
8ddf5e57 | 2135 | while (!NILP (list)) |
2e344af3 | 2136 | { |
16fed1fc | 2137 | val = Fcons (CHARSET_NAME (CHARSET_FROM_ID (XINT (XCAR (list)))), val); |
8ddf5e57 | 2138 | list = XCDR (list); |
2e344af3 | 2139 | } |
8ddf5e57 | 2140 | return Fnreverse (val); |
2e344af3 KH |
2141 | } |
2142 | ||
8ddf5e57 DL |
2143 | DEFUN ("set-charset-priority", Fset_charset_priority, Sset_charset_priority, |
2144 | 1, MANY, 0, | |
2145 | doc: /* Assign higher priority to the charsets given as arguments. | |
2146 | usage: (set-charset-priority &rest charsets) */) | |
c5101a77 | 2147 | (size_t nargs, Lisp_Object *args) |
4ed46869 | 2148 | { |
af7c60ca | 2149 | Lisp_Object new_head, old_list, arglist[2]; |
321c819c | 2150 | Lisp_Object list_2022, list_emacs_mule; |
c5101a77 PE |
2151 | size_t i; |
2152 | int id; | |
4ed46869 | 2153 | |
8ddf5e57 | 2154 | old_list = Fcopy_sequence (Vcharset_ordered_list); |
af7c60ca | 2155 | new_head = Qnil; |
8ddf5e57 | 2156 | for (i = 0; i < nargs; i++) |
4ed46869 | 2157 | { |
8ddf5e57 | 2158 | CHECK_CHARSET_GET_ID (args[i], id); |
af7c60ca KH |
2159 | if (! NILP (Fmemq (make_number (id), old_list))) |
2160 | { | |
2161 | old_list = Fdelq (make_number (id), old_list); | |
2162 | new_head = Fcons (make_number (id), new_head); | |
2163 | } | |
5729c92f | 2164 | } |
8ddf5e57 | 2165 | arglist[0] = Fnreverse (new_head); |
6a9c90ec | 2166 | arglist[1] = Vcharset_non_preferred_head = old_list; |
8ddf5e57 | 2167 | Vcharset_ordered_list = Fnconc (2, arglist); |
dbbb237d | 2168 | charset_ordered_list_tick++; |
5729c92f | 2169 | |
6809ca75 | 2170 | charset_unibyte = -1; |
321c819c | 2171 | for (old_list = Vcharset_ordered_list, list_2022 = list_emacs_mule = Qnil; |
d017b41e | 2172 | CONSP (old_list); old_list = XCDR (old_list)) |
5729c92f | 2173 | { |
e77415b0 | 2174 | if (! NILP (Fmemq (XCAR (old_list), Viso_2022_charset_list))) |
321c819c KH |
2175 | list_2022 = Fcons (XCAR (old_list), list_2022); |
2176 | if (! NILP (Fmemq (XCAR (old_list), Vemacs_mule_charset_list))) | |
2177 | list_emacs_mule = Fcons (XCAR (old_list), list_emacs_mule); | |
6809ca75 KH |
2178 | if (charset_unibyte < 0) |
2179 | { | |
2180 | struct charset *charset = CHARSET_FROM_ID (XINT (XCAR (old_list))); | |
2181 | ||
2182 | if (CHARSET_DIMENSION (charset) == 1 | |
2183 | && CHARSET_ASCII_COMPATIBLE_P (charset) | |
2184 | && CHARSET_MAX_CHAR (charset) >= 0x80) | |
2185 | charset_unibyte = CHARSET_ID (charset); | |
2186 | } | |
4ed46869 | 2187 | } |
321c819c KH |
2188 | Viso_2022_charset_list = Fnreverse (list_2022); |
2189 | Vemacs_mule_charset_list = Fnreverse (list_emacs_mule); | |
6809ca75 KH |
2190 | if (charset_unibyte < 0) |
2191 | charset_unibyte = charset_iso_8859_1; | |
4ed46869 | 2192 | |
8ddf5e57 | 2193 | return Qnil; |
4ed46869 KH |
2194 | } |
2195 | ||
d5b33309 KH |
2196 | DEFUN ("charset-id-internal", Fcharset_id_internal, Scharset_id_internal, |
2197 | 0, 1, 0, | |
2198 | doc: /* Internal use only. | |
2199 | Return charset identification number of CHARSET. */) | |
5842a27b | 2200 | (Lisp_Object charset) |
4ed46869 | 2201 | { |
d5b33309 | 2202 | int id; |
4ed46869 | 2203 | |
d5b33309 KH |
2204 | CHECK_CHARSET_GET_ID (charset, id); |
2205 | return make_number (id); | |
4ed46869 KH |
2206 | } |
2207 | ||
7a84eee5 KH |
2208 | struct charset_sort_data |
2209 | { | |
2210 | Lisp_Object charset; | |
2211 | int id; | |
2212 | int priority; | |
2213 | }; | |
2214 | ||
2215 | static int | |
2216 | charset_compare (const void *d1, const void *d2) | |
2217 | { | |
2218 | const struct charset_sort_data *data1 = d1, *data2 = d2; | |
2219 | return (data1->priority - data2->priority); | |
2220 | } | |
2221 | ||
2222 | DEFUN ("sort-charsets", Fsort_charsets, Ssort_charsets, 1, 1, 0, | |
2223 | doc: /* Sort charset list CHARSETS by a priority of each charset. | |
2224 | Return the sorted list. CHARSETS is modified by side effects. | |
2225 | See also `charset-priority-list' and `set-charset-priority'. */) | |
2226 | (Lisp_Object charsets) | |
2227 | { | |
2228 | Lisp_Object len = Flength (charsets); | |
2229 | int n = XFASTINT (len), i, j, done; | |
2230 | Lisp_Object tail, elt, attrs; | |
2231 | struct charset_sort_data *sort_data; | |
726929c4 | 2232 | int id, min_id = INT_MAX, max_id = INT_MIN; |
7a84eee5 KH |
2233 | USE_SAFE_ALLOCA; |
2234 | ||
2235 | if (n == 0) | |
2236 | return Qnil; | |
2237 | SAFE_ALLOCA (sort_data, struct charset_sort_data *, sizeof (*sort_data) * n); | |
2238 | for (tail = charsets, i = 0; CONSP (tail); tail = XCDR (tail), i++) | |
2239 | { | |
2240 | elt = XCAR (tail); | |
2241 | CHECK_CHARSET_GET_ATTR (elt, attrs); | |
2242 | sort_data[i].charset = elt; | |
2243 | sort_data[i].id = id = XINT (CHARSET_ATTR_ID (attrs)); | |
726929c4 | 2244 | if (id < min_id) |
7a84eee5 | 2245 | min_id = id; |
726929c4 | 2246 | if (id > max_id) |
7a84eee5 KH |
2247 | max_id = id; |
2248 | } | |
2249 | for (done = 0, tail = Vcharset_ordered_list, i = 0; | |
2250 | done < n && CONSP (tail); tail = XCDR (tail), i++) | |
2251 | { | |
2252 | elt = XCAR (tail); | |
2253 | id = XFASTINT (elt); | |
2254 | if (id >= min_id && id <= max_id) | |
2255 | for (j = 0; j < n; j++) | |
2256 | if (sort_data[j].id == id) | |
2257 | { | |
2258 | sort_data[j].priority = i; | |
2259 | done++; | |
2260 | } | |
2261 | } | |
2262 | qsort (sort_data, n, sizeof *sort_data, charset_compare); | |
2263 | for (i = 0, tail = charsets; CONSP (tail); tail = XCDR (tail), i++) | |
2264 | XSETCAR (tail, sort_data[i].charset); | |
2265 | SAFE_FREE (); | |
2266 | return charsets; | |
2267 | } | |
2268 | ||
4ed46869 | 2269 | \f |
3263d5a2 | 2270 | void |
971de7fb | 2271 | init_charset (void) |
4ed46869 | 2272 | { |
c8f94403 GM |
2273 | Lisp_Object tempdir; |
2274 | tempdir = Fexpand_file_name (build_string ("charsets"), Vdata_directory); | |
51b59d79 | 2275 | if (access (SSDATA (tempdir), 0) < 0) |
c8f94403 GM |
2276 | { |
2277 | dir_warning ("Error: charsets directory (%s) does not exist.\n\ | |
2278 | Emacs will not function correctly without the character map files.\n\ | |
2279 | Please check your installation!\n", | |
2280 | tempdir); | |
2281 | /* TODO should this be a fatal error? (Bug#909) */ | |
2282 | } | |
2283 | ||
2284 | Vcharset_map_path = Fcons (tempdir, Qnil); | |
4ed46869 KH |
2285 | } |
2286 | ||
4ed46869 | 2287 | |
dfcf069d | 2288 | void |
971de7fb | 2289 | init_charset_once (void) |
4ed46869 KH |
2290 | { |
2291 | int i, j, k; | |
2292 | ||
3263d5a2 KH |
2293 | for (i = 0; i < ISO_MAX_DIMENSION; i++) |
2294 | for (j = 0; j < ISO_MAX_CHARS; j++) | |
2295 | for (k = 0; k < ISO_MAX_FINAL; k++) | |
2296 | iso_charset_table[i][j][k] = -1; | |
4ed46869 | 2297 | |
60383934 | 2298 | for (i = 0; i < 256; i++) |
b84ae584 | 2299 | emacs_mule_charset[i] = -1; |
4ed46869 | 2300 | |
7c7dceee KH |
2301 | charset_jisx0201_roman = -1; |
2302 | charset_jisx0208_1978 = -1; | |
2303 | charset_jisx0208 = -1; | |
d32320c4 | 2304 | charset_ksc5601 = -1; |
4ed46869 KH |
2305 | } |
2306 | ||
2307 | #ifdef emacs | |
2308 | ||
dfcf069d | 2309 | void |
971de7fb | 2310 | syms_of_charset (void) |
4ed46869 | 2311 | { |
3263d5a2 KH |
2312 | DEFSYM (Qcharsetp, "charsetp"); |
2313 | ||
2314 | DEFSYM (Qascii, "ascii"); | |
2315 | DEFSYM (Qunicode, "unicode"); | |
6c652beb | 2316 | DEFSYM (Qemacs, "emacs"); |
2fe1edd1 | 2317 | DEFSYM (Qeight_bit, "eight-bit"); |
3263d5a2 KH |
2318 | DEFSYM (Qiso_8859_1, "iso-8859-1"); |
2319 | ||
2320 | DEFSYM (Qgl, "gl"); | |
2321 | DEFSYM (Qgr, "gr"); | |
2322 | ||
3263d5a2 KH |
2323 | staticpro (&Vcharset_ordered_list); |
2324 | Vcharset_ordered_list = Qnil; | |
2325 | ||
2326 | staticpro (&Viso_2022_charset_list); | |
2327 | Viso_2022_charset_list = Qnil; | |
2328 | ||
2329 | staticpro (&Vemacs_mule_charset_list); | |
2330 | Vemacs_mule_charset_list = Qnil; | |
2331 | ||
3943ed76 | 2332 | /* Don't staticpro them here. It's done in syms_of_fns. */ |
088dcc3e DN |
2333 | QCtest = intern_c_string (":test"); |
2334 | Qeq = intern_c_string ("eq"); | |
3943ed76 | 2335 | |
3263d5a2 | 2336 | staticpro (&Vcharset_hash_table); |
8f924df7 KH |
2337 | { |
2338 | Lisp_Object args[2]; | |
2339 | args[0] = QCtest; | |
2340 | args[1] = Qeq; | |
2341 | Vcharset_hash_table = Fmake_hash_table (2, args); | |
2342 | } | |
3263d5a2 KH |
2343 | |
2344 | charset_table_size = 128; | |
2345 | charset_table = ((struct charset *) | |
2346 | xmalloc (sizeof (struct charset) * charset_table_size)); | |
2347 | charset_table_used = 0; | |
2348 | ||
3263d5a2 KH |
2349 | defsubr (&Scharsetp); |
2350 | defsubr (&Smap_charset_chars); | |
2351 | defsubr (&Sdefine_charset_internal); | |
2352 | defsubr (&Sdefine_charset_alias); | |
3263d5a2 KH |
2353 | defsubr (&Scharset_plist); |
2354 | defsubr (&Sset_charset_plist); | |
2355 | defsubr (&Sunify_charset); | |
3fac5a51 | 2356 | defsubr (&Sget_unused_iso_final_char); |
4ed46869 KH |
2357 | defsubr (&Sdeclare_equiv_charset); |
2358 | defsubr (&Sfind_charset_region); | |
2359 | defsubr (&Sfind_charset_string); | |
3263d5a2 KH |
2360 | defsubr (&Sdecode_char); |
2361 | defsubr (&Sencode_char); | |
4ed46869 | 2362 | defsubr (&Ssplit_char); |
3263d5a2 | 2363 | defsubr (&Smake_char); |
4ed46869 | 2364 | defsubr (&Schar_charset); |
90d7b74e | 2365 | defsubr (&Scharset_after); |
4ed46869 | 2366 | defsubr (&Siso_charset); |
3263d5a2 | 2367 | defsubr (&Sclear_charset_maps); |
8ddf5e57 DL |
2368 | defsubr (&Scharset_priority_list); |
2369 | defsubr (&Sset_charset_priority); | |
d5b33309 | 2370 | defsubr (&Scharset_id_internal); |
7a84eee5 | 2371 | defsubr (&Ssort_charsets); |
3263d5a2 | 2372 | |
29208e82 | 2373 | DEFVAR_LISP ("charset-map-path", Vcharset_map_path, |
6f3122a7 | 2374 | doc: /* *List of directories to search for charset map files. */); |
4beef065 | 2375 | Vcharset_map_path = Qnil; |
4ed46869 | 2376 | |
29208e82 | 2377 | DEFVAR_BOOL ("inhibit-load-charset-map", inhibit_load_charset_map, |
ecca2aad KH |
2378 | doc: /* Inhibit loading of charset maps. Used when dumping Emacs. */); |
2379 | inhibit_load_charset_map = 0; | |
2380 | ||
29208e82 | 2381 | DEFVAR_LISP ("charset-list", Vcharset_list, |
528623a0 | 2382 | doc: /* List of all charsets ever defined. */); |
3263d5a2 KH |
2383 | Vcharset_list = Qnil; |
2384 | ||
29208e82 | 2385 | DEFVAR_LISP ("current-iso639-language", Vcurrent_iso639_language, |
6a9c90ec KH |
2386 | doc: /* ISO639 language mnemonic symbol for the current language environment. |
2387 | If the current language environment is for multiple languages (e.g. "Latin-1"), | |
2388 | the value may be a list of mnemonics. */); | |
2389 | Vcurrent_iso639_language = Qnil; | |
2390 | ||
2fe1edd1 KH |
2391 | charset_ascii |
2392 | = define_charset_internal (Qascii, 1, "\x00\x7F\x00\x00\x00\x00", | |
2393 | 0, 127, 'B', -1, 0, 1, 0, 0); | |
14e3d523 KH |
2394 | charset_iso_8859_1 |
2395 | = define_charset_internal (Qiso_8859_1, 1, "\x00\xFF\x00\x00\x00\x00", | |
2396 | 0, 255, -1, -1, -1, 1, 0, 0); | |
2fe1edd1 | 2397 | charset_unicode |
73fbf2d9 | 2398 | = define_charset_internal (Qunicode, 3, "\x00\xFF\x00\xFF\x00\x10", |
2fe1edd1 | 2399 | 0, MAX_UNICODE_CHAR, -1, 0, -1, 1, 0, 0); |
6c652beb KH |
2400 | charset_emacs |
2401 | = define_charset_internal (Qemacs, 3, "\x00\xFF\x00\xFF\x00\x3F", | |
2402 | 0, MAX_5_BYTE_CHAR, -1, 0, -1, 1, 1, 0); | |
2fe1edd1 KH |
2403 | charset_eight_bit |
2404 | = define_charset_internal (Qeight_bit, 1, "\x80\xFF\x00\x00\x00\x00", | |
880820fe | 2405 | 128, 255, -1, 0, -1, 0, 1, |
2fe1edd1 | 2406 | MAX_5_BYTE_CHAR + 1); |
6809ca75 | 2407 | charset_unibyte = charset_iso_8859_1; |
4ed46869 KH |
2408 | } |
2409 | ||
2410 | #endif /* emacs */ |