]>
Commit | Line | Data |
---|---|---|
c1382a3d | 1 | /* |
317b8492 | 2 | * fontconfig/fc-lang/fc-lang.c |
c1382a3d | 3 | * |
46b51147 | 4 | * Copyright © 2002 Keith Packard |
c1382a3d KP |
5 | * |
6 | * Permission to use, copy, modify, distribute, and sell this software and its | |
7 | * documentation for any purpose is hereby granted without fee, provided that | |
8 | * the above copyright notice appear in all copies and that both that | |
9 | * copyright notice and this permission notice appear in supporting | |
10 | * documentation, and that the name of Keith Packard not be used in | |
11 | * advertising or publicity pertaining to distribution of the software without | |
12 | * specific, written prior permission. Keith Packard makes no | |
13 | * representations about the suitability of this software for any purpose. It | |
14 | * is provided "as is" without express or implied warranty. | |
15 | * | |
3074a73b | 16 | * THE AUTHOR(S) DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, |
c1382a3d | 17 | * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO |
3074a73b | 18 | * EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY SPECIAL, INDIRECT OR |
c1382a3d KP |
19 | * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, |
20 | * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER | |
21 | * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR | |
22 | * PERFORMANCE OF THIS SOFTWARE. | |
23 | */ | |
24 | ||
c647f6f1 KP |
25 | #include "fccharset.c" |
26 | #include "fcstr.c" | |
e3096d90 | 27 | #include "fcserialize.c" |
c1382a3d KP |
28 | |
29 | /* | |
30 | * fc-lang | |
31 | * | |
32 | * Read a set of language orthographies and build C declarations for | |
33 | * charsets which can then be used to identify which languages are | |
c647f6f1 KP |
34 | * supported by a given font. Note that this uses some utilities |
35 | * from the fontconfig library, so the necessary file is simply | |
36 | * included in this compilation. A couple of extra utility | |
37 | * functions are also needed in slightly modified form | |
c1382a3d KP |
38 | */ |
39 | ||
c647f6f1 KP |
40 | void |
41 | FcMemAlloc (int kind, int size) | |
42 | { | |
43 | } | |
44 | ||
45 | void | |
46 | FcMemFree (int kind, int size) | |
47 | { | |
48 | } | |
49 | ||
afe5a671 KP |
50 | FcPrivate void |
51 | FcCacheObjectReference (void *object) | |
52 | { | |
53 | } | |
54 | ||
55 | FcPrivate void | |
56 | FcCacheObjectDereference (void *object) | |
57 | { | |
58 | } | |
59 | ||
18b6857c KP |
60 | int FcDebugVal; |
61 | ||
ff3f1f98 KP |
62 | FcChar8 * |
63 | FcConfigHome (void) | |
64 | { | |
8245771d | 65 | return (FcChar8 *) getenv ("HOME"); |
ff3f1f98 KP |
66 | } |
67 | ||
c1382a3d | 68 | static void |
67accef4 | 69 | fatal (const char *file, int lineno, const char *msg) |
c1382a3d | 70 | { |
67accef4 PL |
71 | if (lineno) |
72 | fprintf (stderr, "%s:%d: %s\n", file, lineno, msg); | |
73 | else | |
c7beacf9 | 74 | fprintf (stderr, "%s: %s\n", file, msg); |
c1382a3d KP |
75 | exit (1); |
76 | } | |
77 | ||
78 | static char * | |
79 | get_line (FILE *f, char *line, int *lineno) | |
80 | { | |
81 | char *hash; | |
cf5cf4ca | 82 | int end; |
c1382a3d KP |
83 | if (!fgets (line, 1024, f)) |
84 | return 0; | |
85 | ++(*lineno); | |
86 | hash = strchr (line, '#'); | |
87 | if (hash) | |
88 | *hash = '\0'; | |
cf5cf4ca PL |
89 | |
90 | end = strlen (line); | |
91 | while (end > 0 && isspace (line[end-1])) | |
92 | line[--end] = '\0'; | |
93 | ||
c1382a3d KP |
94 | if (line[0] == '\0' || line[0] == '\n' || line[0] == '\032' || line[0] == '\r') |
95 | return get_line (f, line, lineno); | |
96 | return line; | |
97 | } | |
98 | ||
0d745819 | 99 | static char *dir = 0; |
394b2bf0 | 100 | |
6ae6acf3 | 101 | static FILE * |
394b2bf0 KP |
102 | scanopen (char *file) |
103 | { | |
104 | FILE *f; | |
105 | ||
106 | f = fopen (file, "r"); | |
107 | if (!f && dir) | |
108 | { | |
109 | char path[1024]; | |
110 | ||
111 | strcpy (path, dir); | |
112 | strcat (path, "/"); | |
113 | strcat (path, file); | |
114 | f = fopen (path, "r"); | |
115 | } | |
116 | return f; | |
117 | } | |
118 | ||
c1382a3d KP |
119 | /* |
120 | * build a single charset from a source file | |
121 | * | |
122 | * The file format is quite simple, either | |
123 | * a single hex value or a pair separated with a dash | |
124 | * | |
125 | * Comments begin with '#' | |
126 | */ | |
127 | ||
18b6857c KP |
128 | static const FcCharSet * |
129 | scan (FILE *f, char *file, FcCharSetFreezer *freezer) | |
c1382a3d | 130 | { |
18b6857c KP |
131 | FcCharSet *c = 0; |
132 | const FcCharSet *n; | |
133 | int start, end, ucs4; | |
134 | char line[1024]; | |
135 | int lineno = 0; | |
c1382a3d KP |
136 | |
137 | while (get_line (f, line, &lineno)) | |
138 | { | |
139 | if (!strncmp (line, "include", 7)) | |
140 | { | |
f6993c88 BE |
141 | FILE *included_f; |
142 | char *included_file; | |
143 | included_file = strchr (line, ' '); | |
144 | if (!included_file) | |
145 | fatal (file, lineno, | |
04f7d3e7 | 146 | "invalid syntax, expected: include filename"); |
f6993c88 BE |
147 | while (isspace(*included_file)) |
148 | included_file++; | |
149 | included_f = scanopen (included_file); | |
150 | if (!included_f) | |
151 | fatal (included_file, 0, "can't open"); | |
152 | n = scan (included_f, included_file, freezer); | |
153 | fclose (included_f); | |
154 | if (!c) | |
155 | c = FcCharSetCreate (); | |
156 | if (!FcCharSetMerge (c, n, NULL)) | |
157 | fatal (file, lineno, "out of memory"); | |
158 | FcCharSetDestroy (n); | |
159 | continue; | |
c1382a3d KP |
160 | } |
161 | if (strchr (line, '-')) | |
162 | { | |
163 | if (sscanf (line, "%x-%x", &start, &end) != 2) | |
164 | fatal (file, lineno, "parse error"); | |
165 | } | |
166 | else | |
167 | { | |
168 | if (sscanf (line, "%x", &start) != 1) | |
169 | fatal (file, lineno, "parse error"); | |
170 | end = start; | |
171 | } | |
172 | if (!c) | |
173 | c = FcCharSetCreate (); | |
174 | for (ucs4 = start; ucs4 <= end; ucs4++) | |
175 | { | |
176 | if (!FcCharSetAddChar (c, ucs4)) | |
177 | fatal (file, lineno, "out of memory"); | |
178 | } | |
179 | } | |
18b6857c | 180 | n = FcCharSetFreeze (freezer, c); |
c1382a3d KP |
181 | FcCharSetDestroy (c); |
182 | return n; | |
183 | } | |
184 | ||
185 | /* | |
186 | * Convert a file name into a name suitable for C declarations | |
187 | */ | |
188 | static char * | |
189 | get_name (char *file) | |
190 | { | |
191 | char *name; | |
192 | char *dot; | |
193 | ||
194 | dot = strchr (file, '.'); | |
195 | if (!dot) | |
196 | dot = file + strlen(file); | |
197 | name = malloc (dot - file + 1); | |
198 | strncpy (name, file, dot - file); | |
199 | name[dot-file] = '\0'; | |
200 | return name; | |
201 | } | |
202 | ||
203 | /* | |
204 | * Convert a C name into a language name | |
205 | */ | |
206 | static char * | |
207 | get_lang (char *name) | |
208 | { | |
209 | char *lang = malloc (strlen (name) + 1); | |
210 | char *l = lang; | |
211 | char c; | |
212 | ||
213 | while ((c = *name++)) | |
214 | { | |
996580dc KP |
215 | if (isupper ((int) (unsigned char) c)) |
216 | c = tolower ((int) (unsigned char) c); | |
c1382a3d KP |
217 | if (c == '_') |
218 | c = '-'; | |
219 | if (c == ' ') | |
220 | continue; | |
221 | *l++ = c; | |
222 | } | |
223 | *l++ = '\0'; | |
224 | return lang; | |
225 | } | |
226 | ||
d8d73958 KP |
227 | static int compare (const void *a, const void *b) |
228 | { | |
229 | const FcChar8 *const *as = a, *const *bs = b; | |
230 | return FcStrCmpIgnoreCase (*as, *bs); | |
231 | } | |
232 | ||
234397b4 DD |
233 | #define MAX_LANG 1024 |
234 | #define MAX_LANG_SET_MAP ((MAX_LANG + 31) / 32) | |
235 | ||
236 | #define BitSet(map, id) ((map)[(id)>>5] |= ((FcChar32) 1 << ((id) & 0x1f))) | |
237 | #define BitGet(map, id) ((map)[(id)>>5] >> ((id) & 0x1f)) & 1) | |
238 | ||
c1382a3d KP |
239 | int |
240 | main (int argc, char **argv) | |
241 | { | |
69a3fc78 | 242 | static char *files[MAX_LANG]; |
18b6857c | 243 | static const FcCharSet *sets[MAX_LANG]; |
69a3fc78 | 244 | static int duplicate[MAX_LANG]; |
69a3fc78 PL |
245 | static int country[MAX_LANG]; |
246 | static char *names[MAX_LANG]; | |
247 | static char *langs[MAX_LANG]; | |
7ce19673 | 248 | static int off[MAX_LANG]; |
c1382a3d | 249 | FILE *f; |
234397b4 | 250 | int ncountry = 0; |
c1382a3d | 251 | int i = 0; |
7ce19673 | 252 | int nsets = 0; |
67accef4 | 253 | int argi; |
cd2ec1a9 | 254 | FcCharLeaf **leaves; |
c1382a3d | 255 | int total_leaves = 0; |
7ce19673 | 256 | int l, sl, tl, tn; |
69a3fc78 PL |
257 | static char line[1024]; |
258 | static FcChar32 map[MAX_LANG_SET_MAP]; | |
234397b4 | 259 | int num_lang_set_map; |
0eadb052 KP |
260 | int setRangeStart[26]; |
261 | int setRangeEnd[26]; | |
262 | FcChar8 setRangeChar; | |
18b6857c | 263 | FcCharSetFreezer *freezer; |
c1382a3d | 264 | |
18b6857c KP |
265 | freezer = FcCharSetFreezerCreate (); |
266 | if (!freezer) | |
267 | fatal (argv[0], 0, "out of memory"); | |
67accef4 PL |
268 | argi = 1; |
269 | while (argv[argi]) | |
234397b4 | 270 | { |
67accef4 | 271 | if (!strcmp (argv[argi], "-d")) |
394b2bf0 | 272 | { |
67accef4 PL |
273 | argi++; |
274 | dir = argv[argi++]; | |
394b2bf0 KP |
275 | continue; |
276 | } | |
234397b4 | 277 | if (i == MAX_LANG) |
67accef4 PL |
278 | fatal (argv[0], 0, "Too many languages"); |
279 | files[i++] = argv[argi++]; | |
234397b4 | 280 | } |
d8d73958 KP |
281 | files[i] = 0; |
282 | qsort (files, i, sizeof (char *), compare); | |
283 | i = 0; | |
284 | while (files[i]) | |
c1382a3d | 285 | { |
394b2bf0 | 286 | f = scanopen (files[i]); |
c1382a3d | 287 | if (!f) |
d8d73958 | 288 | fatal (files[i], 0, strerror (errno)); |
18b6857c | 289 | sets[i] = scan (f, files[i], freezer); |
d8d73958 | 290 | names[i] = get_name (files[i]); |
234397b4 DD |
291 | langs[i] = get_lang(names[i]); |
292 | if (strchr (langs[i], '-')) | |
293 | country[ncountry++] = i; | |
294 | ||
c1382a3d KP |
295 | total_leaves += sets[i]->num; |
296 | i++; | |
297 | fclose (f); | |
298 | } | |
7ce19673 | 299 | nsets = i; |
c1382a3d KP |
300 | sets[i] = 0; |
301 | leaves = malloc (total_leaves * sizeof (FcCharLeaf *)); | |
302 | tl = 0; | |
303 | /* | |
304 | * Find unique leaves | |
305 | */ | |
306 | for (i = 0; sets[i]; i++) | |
307 | { | |
c1382a3d KP |
308 | for (sl = 0; sl < sets[i]->num; sl++) |
309 | { | |
310 | for (l = 0; l < tl; l++) | |
7ce19673 | 311 | if (leaves[l] == FcCharSetLeaf(sets[i], sl)) |
c1382a3d KP |
312 | break; |
313 | if (l == tl) | |
7ce19673 | 314 | leaves[tl++] = FcCharSetLeaf(sets[i], sl); |
c1382a3d KP |
315 | } |
316 | } | |
317 | ||
318 | /* | |
319 | * Scan the input until the marker is found | |
320 | */ | |
321 | ||
322 | while (fgets (line, sizeof (line), stdin)) | |
323 | { | |
324 | if (!strncmp (line, "@@@", 3)) | |
325 | break; | |
326 | fputs (line, stdout); | |
327 | } | |
328 | ||
329 | printf ("/* total size: %d unique leaves: %d */\n\n", | |
330 | total_leaves, tl); | |
2903c146 KP |
331 | |
332 | /* | |
333 | * Find duplicate charsets | |
334 | */ | |
335 | duplicate[0] = -1; | |
336 | for (i = 1; sets[i]; i++) | |
337 | { | |
338 | int j; | |
339 | ||
340 | duplicate[i] = -1; | |
341 | for (j = 0; j < i; j++) | |
342 | if (sets[j] == sets[i]) | |
343 | { | |
344 | duplicate[i] = j; | |
345 | break; | |
346 | } | |
347 | } | |
348 | ||
7ce19673 KP |
349 | tn = 0; |
350 | for (i = 0; sets[i]; i++) { | |
351 | if (duplicate[i] >= 0) | |
352 | continue; | |
353 | off[i] = tn; | |
354 | tn += sets[i]->num; | |
355 | } | |
356 | ||
357 | printf ("#define LEAF0 (%d * sizeof (FcLangCharSet))\n", nsets); | |
358 | printf ("#define OFF0 (LEAF0 + %d * sizeof (FcCharLeaf))\n", tl); | |
359 | printf ("#define NUM0 (OFF0 + %d * sizeof (intptr_t))\n", tn); | |
360 | printf ("#define SET(n) (n * sizeof (FcLangCharSet) + offsetof (FcLangCharSet, charset))\n"); | |
361 | printf ("#define OFF(s,o) (OFF0 + o * sizeof (intptr_t) - SET(s))\n"); | |
362 | printf ("#define NUM(s,n) (NUM0 + n * sizeof (FcChar16) - SET(s))\n"); | |
363 | printf ("#define LEAF(o,l) (LEAF0 + l * sizeof (FcCharLeaf) - (OFF0 + o * sizeof (intptr_t)))\n"); | |
364 | printf ("#define fcLangCharSets (fcLangData.langCharSets)\n"); | |
365 | printf ("\n"); | |
366 | ||
367 | printf ("static const struct {\n" | |
368 | " FcLangCharSet langCharSets[%d];\n" | |
369 | " FcCharLeaf leaves[%d];\n" | |
370 | " intptr_t leaf_offsets[%d];\n" | |
371 | " FcChar16 numbers[%d];\n" | |
372 | "} fcLangData = {\n", | |
373 | nsets, tl, tn, tn); | |
374 | ||
0eadb052 | 375 | /* |
7ce19673 | 376 | * Dump sets |
0eadb052 | 377 | */ |
7ce19673 KP |
378 | |
379 | printf ("{\n"); | |
0eadb052 KP |
380 | for (i = 0; sets[i]; i++) |
381 | { | |
7ce19673 KP |
382 | int j = duplicate[i]; |
383 | ||
384 | if (j < 0) | |
385 | j = i; | |
386 | ||
e85789a9 | 387 | printf (" { \"%s\", " |
7ce19673 KP |
388 | " { FC_REF_CONSTANT, %d, OFF(%d,%d), NUM(%d,%d) } }, /* %d */\n", |
389 | langs[i], | |
390 | sets[j]->num, i, off[j], i, off[j], i); | |
0eadb052 | 391 | } |
7ce19673 | 392 | printf ("},\n"); |
0eadb052 | 393 | |
c1382a3d | 394 | /* |
7ce19673 | 395 | * Dump leaves |
c1382a3d | 396 | */ |
7ce19673 KP |
397 | printf ("{\n"); |
398 | for (l = 0; l < tl; l++) | |
c1382a3d | 399 | { |
7ce19673 KP |
400 | printf (" { { /* %d */", l); |
401 | for (i = 0; i < 256/32; i++) | |
82f35f8b | 402 | { |
7ce19673 KP |
403 | if (i % 4 == 0) |
404 | printf ("\n "); | |
405 | printf (" 0x%08x,", leaves[l]->map[i]); | |
82f35f8b | 406 | } |
7ce19673 | 407 | printf ("\n } },\n"); |
82f35f8b | 408 | } |
7ce19673 | 409 | printf ("},\n"); |
82f35f8b | 410 | |
7ce19673 KP |
411 | /* |
412 | * Dump leaves | |
413 | */ | |
414 | printf ("{\n"); | |
82f35f8b PL |
415 | for (i = 0; sets[i]; i++) |
416 | { | |
417 | int n; | |
418 | ||
419 | if (duplicate[i] >= 0) | |
420 | continue; | |
7ce19673 | 421 | printf (" /* %s */\n", names[i]); |
c1382a3d KP |
422 | for (n = 0; n < sets[i]->num; n++) |
423 | { | |
7ce19673 | 424 | if (n % 4 == 0) |
c1382a3d KP |
425 | printf (" "); |
426 | for (l = 0; l < tl; l++) | |
7ce19673 | 427 | if (leaves[l] == FcCharSetLeaf(sets[i], n)) |
c1382a3d KP |
428 | break; |
429 | if (l == tl) | |
430 | fatal (names[i], 0, "can't find leaf"); | |
7ce19673 KP |
431 | printf (" LEAF(%3d,%3d),", off[i], l); |
432 | if (n % 4 == 3) | |
c1382a3d KP |
433 | printf ("\n"); |
434 | } | |
7ce19673 | 435 | if (n % 4 != 0) |
c1382a3d | 436 | printf ("\n"); |
82f35f8b | 437 | } |
7ce19673 KP |
438 | printf ("},\n"); |
439 | ||
82f35f8b | 440 | |
7ce19673 | 441 | printf ("{\n"); |
82f35f8b PL |
442 | for (i = 0; sets[i]; i++) |
443 | { | |
444 | int n; | |
7ce19673 | 445 | |
a151aced PL |
446 | if (duplicate[i] >= 0) |
447 | continue; | |
7ce19673 | 448 | printf (" /* %s */\n", names[i]); |
c1382a3d KP |
449 | for (n = 0; n < sets[i]->num; n++) |
450 | { | |
451 | if (n % 8 == 0) | |
452 | printf (" "); | |
7ce19673 | 453 | printf (" 0x%04x,", FcCharSetNumbers (sets[i])[n]); |
c1382a3d KP |
454 | if (n % 8 == 7) |
455 | printf ("\n"); | |
456 | } | |
457 | if (n % 8 != 0) | |
458 | printf ("\n"); | |
c1382a3d | 459 | } |
7ce19673 | 460 | printf ("}\n"); |
0eadb052 | 461 | |
c1382a3d | 462 | printf ("};\n\n"); |
7ce19673 | 463 | |
234397b4 DD |
464 | printf ("#define NUM_LANG_CHAR_SET %d\n", i); |
465 | num_lang_set_map = (i + 31) / 32; | |
466 | printf ("#define NUM_LANG_SET_MAP %d\n", num_lang_set_map); | |
467 | /* | |
468 | * Dump indices with country codes | |
469 | */ | |
470 | if (ncountry) | |
471 | { | |
0d745819 | 472 | int c; |
234397b4 DD |
473 | int ncountry_ent = 0; |
474 | printf ("\n"); | |
475 | printf ("static const FcChar32 fcLangCountrySets[][NUM_LANG_SET_MAP] = {\n"); | |
476 | for (c = 0; c < ncountry; c++) | |
477 | { | |
478 | i = country[c]; | |
479 | if (i >= 0) | |
480 | { | |
0d745819 | 481 | int lang = strchr (langs[i], '-') - langs[i]; |
234397b4 DD |
482 | int d, k; |
483 | ||
484 | for (k = 0; k < num_lang_set_map; k++) | |
485 | map[k] = 0; | |
486 | ||
487 | BitSet (map, i); | |
488 | for (d = c + 1; d < ncountry; d++) | |
489 | { | |
490 | int j = country[d]; | |
bb36e676 | 491 | if (j >= 0 && !strncmp (langs[j], langs[i], lang + 1)) |
234397b4 DD |
492 | { |
493 | BitSet(map, j); | |
494 | country[d] = -1; | |
495 | } | |
496 | } | |
497 | printf (" {"); | |
498 | for (k = 0; k < num_lang_set_map; k++) | |
499 | printf (" 0x%08x,", map[k]); | |
500 | printf (" }, /* %*.*s */\n", | |
0d745819 | 501 | lang, lang, langs[i]); |
234397b4 DD |
502 | ++ncountry_ent; |
503 | } | |
504 | } | |
505 | printf ("};\n\n"); | |
506 | printf ("#define NUM_COUNTRY_SET %d\n", ncountry_ent); | |
507 | } | |
508 | ||
0eadb052 | 509 | |
7ce19673 KP |
510 | /* |
511 | * Find ranges for each letter for faster searching | |
512 | */ | |
513 | setRangeChar = 'a'; | |
514 | memset(setRangeStart, '\0', sizeof (setRangeStart)); | |
515 | memset(setRangeEnd, '\0', sizeof (setRangeEnd)); | |
516 | for (i = 0; sets[i]; i++) | |
517 | { | |
518 | char c = names[i][0]; | |
519 | ||
520 | while (setRangeChar <= c && c <= 'z') | |
521 | setRangeStart[setRangeChar++ - 'a'] = i; | |
522 | } | |
523 | for (setRangeChar = 'a'; setRangeChar < 'z'; setRangeChar++) | |
524 | setRangeEnd[setRangeChar - 'a'] = setRangeStart[setRangeChar+1-'a'] - 1; | |
525 | setRangeEnd[setRangeChar - 'a'] = i - 1; | |
526 | ||
0eadb052 KP |
527 | /* |
528 | * Dump sets start/finish for the fastpath | |
529 | */ | |
530 | printf ("static const FcLangCharSetRange fcLangCharSetRanges[] = {\n"); | |
531 | for (setRangeChar = 'a'; setRangeChar <= 'z' ; setRangeChar++) | |
532 | { | |
533 | printf (" { %d, %d }, /* %c */\n", | |
534 | setRangeStart[setRangeChar - 'a'], | |
535 | setRangeEnd[setRangeChar - 'a'], setRangeChar); | |
536 | } | |
537 | printf ("};\n\n"); | |
538 | ||
c1382a3d KP |
539 | while (fgets (line, sizeof (line), stdin)) |
540 | fputs (line, stdout); | |
541 | ||
542 | fflush (stdout); | |
543 | exit (ferror (stdout)); | |
544 | } |