1 /*
2 * Copyright © 2009 Red Hat, Inc.
3 * Copyright © 2011 Google, Inc.
4 *
5 * This is part of HarfBuzz, a text shaping library.
6 *
7 * Permission is hereby granted, without written agreement and without
8 * license or royalty fees, to use, copy, modify, and distribute this
9 * software and its documentation for any purpose, provided that the
10 * above copyright notice and the following two paragraphs appear in
11 * all copies of this software.
12 *
13 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
14 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
15 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
16 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
17 * DAMAGE.
18 *
19 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
20 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
21 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
22 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
23 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
24 *
25 * Red Hat Author(s): Behdad Esfahbod
26 * Google Author(s): Behdad Esfahbod, Roozbeh Pournader
27 */
28
29 #include "hb.hh"
30
31 #ifndef HB_NO_OT_TAG
32
33
34 /* hb_script_t */
35
36 static hb_tag_t
hb_ot_old_tag_from_script(hb_script_t script)37 hb_ot_old_tag_from_script (hb_script_t script)
38 {
39 /* This seems to be accurate as of end of 2012. */
40
41 switch ((hb_tag_t) script)
42 {
43 case HB_SCRIPT_INVALID: return HB_OT_TAG_DEFAULT_SCRIPT;
44
45 /* KATAKANA and HIRAGANA both map to 'kana' */
46 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
47
48 /* Spaces at the end are preserved, unlike ISO 15924 */
49 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
50 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
51 /* Unicode-5.0 additions */
52 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
53 /* Unicode-5.1 additions */
54 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
55 }
56
57 /* Else, just change first char to lowercase and return */
58 return ((hb_tag_t) script) | 0x20000000u;
59 }
60
61 static hb_script_t
hb_ot_old_tag_to_script(hb_tag_t tag)62 hb_ot_old_tag_to_script (hb_tag_t tag)
63 {
64 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
65 return HB_SCRIPT_INVALID;
66
67 /* This side of the conversion is fully algorithmic. */
68
69 /* Any spaces at the end of the tag are replaced by repeating the last
70 * letter. Eg 'nko ' -> 'Nkoo' */
71 if (unlikely ((tag & 0x0000FF00u) == 0x00002000u))
72 tag |= (tag >> 8) & 0x0000FF00u; /* Copy second letter to third */
73 if (unlikely ((tag & 0x000000FFu) == 0x00000020u))
74 tag |= (tag >> 8) & 0x000000FFu; /* Copy third letter to fourth */
75
76 /* Change first char to uppercase and return */
77 return (hb_script_t) (tag & ~0x20000000u);
78 }
79
80 static hb_tag_t
hb_ot_new_tag_from_script(hb_script_t script)81 hb_ot_new_tag_from_script (hb_script_t script)
82 {
83 switch ((hb_tag_t) script) {
84 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
85 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
86 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
87 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
88 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
89 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
90 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
91 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
92 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
93 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
94 }
95
96 return HB_OT_TAG_DEFAULT_SCRIPT;
97 }
98
99 static hb_script_t
hb_ot_new_tag_to_script(hb_tag_t tag)100 hb_ot_new_tag_to_script (hb_tag_t tag)
101 {
102 switch (tag) {
103 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
104 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
105 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
106 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
107 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
108 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
109 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
110 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
111 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
112 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
113 }
114
115 return HB_SCRIPT_UNKNOWN;
116 }
117
118 #ifndef HB_DISABLE_DEPRECATED
119 void
hb_ot_tags_from_script(hb_script_t script,hb_tag_t * script_tag_1,hb_tag_t * script_tag_2)120 hb_ot_tags_from_script (hb_script_t script,
121 hb_tag_t *script_tag_1,
122 hb_tag_t *script_tag_2)
123 {
124 unsigned int count = 2;
125 hb_tag_t tags[2];
126 hb_ot_tags_from_script_and_language (script, HB_LANGUAGE_INVALID, &count, tags, nullptr, nullptr);
127 *script_tag_1 = count > 0 ? tags[0] : HB_OT_TAG_DEFAULT_SCRIPT;
128 *script_tag_2 = count > 1 ? tags[1] : HB_OT_TAG_DEFAULT_SCRIPT;
129 }
130 #endif
131
132 /*
133 * Complete list at:
134 * https://docs.microsoft.com/en-us/typography/opentype/spec/scripttags
135 *
136 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
137 * So we just do that, and handle the exceptional cases in a switch.
138 */
139
140 static void
hb_ot_all_tags_from_script(hb_script_t script,unsigned int * count,hb_tag_t * tags)141 hb_ot_all_tags_from_script (hb_script_t script,
142 unsigned int *count /* IN/OUT */,
143 hb_tag_t *tags /* OUT */)
144 {
145 unsigned int i = 0;
146
147 hb_tag_t new_tag = hb_ot_new_tag_from_script (script);
148 if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT))
149 {
150 /* HB_SCRIPT_MYANMAR maps to 'mym2', but there is no 'mym3'. */
151 if (new_tag != HB_TAG('m','y','m','2'))
152 tags[i++] = new_tag | '3';
153 if (*count > i)
154 tags[i++] = new_tag;
155 }
156
157 if (*count > i)
158 {
159 hb_tag_t old_tag = hb_ot_old_tag_from_script (script);
160 if (old_tag != HB_OT_TAG_DEFAULT_SCRIPT)
161 tags[i++] = old_tag;
162 }
163
164 *count = i;
165 }
166
167 /**
168 * hb_ot_tag_to_script:
169 * @tag: a script tag
170 *
171 * Converts a script tag to an #hb_script_t.
172 *
173 * Return value: The #hb_script_t corresponding to @tag.
174 *
175 **/
176 hb_script_t
hb_ot_tag_to_script(hb_tag_t tag)177 hb_ot_tag_to_script (hb_tag_t tag)
178 {
179 unsigned char digit = tag & 0x000000FFu;
180 if (unlikely (digit == '2' || digit == '3'))
181 return hb_ot_new_tag_to_script (tag & 0xFFFFFF32);
182
183 return hb_ot_old_tag_to_script (tag);
184 }
185
186
187 /* hb_language_t */
188
189 static bool
subtag_matches(const char * lang_str,const char * limit,const char * subtag)190 subtag_matches (const char *lang_str,
191 const char *limit,
192 const char *subtag)
193 {
194 do {
195 const char *s = strstr (lang_str, subtag);
196 if (!s || s >= limit)
197 return false;
198 if (!ISALNUM (s[strlen (subtag)]))
199 return true;
200 lang_str = s + strlen (subtag);
201 } while (true);
202 }
203
204 static hb_bool_t
lang_matches(const char * lang_str,const char * spec)205 lang_matches (const char *lang_str, const char *spec)
206 {
207 unsigned int len = strlen (spec);
208
209 return strncmp (lang_str, spec, len) == 0 &&
210 (lang_str[len] == '\0' || lang_str[len] == '-');
211 }
212
213 struct LangTag
214 {
215 char language[4];
216 hb_tag_t tag;
217
cmpLangTag218 int cmp (const char *a) const
219 {
220 const char *b = this->language;
221 unsigned int da, db;
222 const char *p;
223
224 p = strchr (a, '-');
225 da = p ? (unsigned int) (p - a) : strlen (a);
226
227 p = strchr (b, '-');
228 db = p ? (unsigned int) (p - b) : strlen (b);
229
230 return strncmp (a, b, hb_max (da, db));
231 }
cmpLangTag232 int cmp (const LangTag *that) const
233 { return cmp (that->language); }
234 };
235
236 #include "hb-ot-tag-table.hh"
237
238 /* The corresponding languages IDs for the following IDs are unclear,
239 * overlap, or are architecturally weird. Needs more research. */
240
241 /*{"??", {HB_TAG('B','C','R',' ')}},*/ /* Bible Cree */
242 /*{"zh?", {HB_TAG('C','H','N',' ')}},*/ /* Chinese (seen in Microsoft fonts) */
243 /*{"ar-Syrc?", {HB_TAG('G','A','R',' ')}},*/ /* Garshuni */
244 /*{"??", {HB_TAG('N','G','R',' ')}},*/ /* Nagari */
245 /*{"??", {HB_TAG('Y','I','C',' ')}},*/ /* Yi Classic */
246 /*{"zh?", {HB_TAG('Z','H','P',' ')}},*/ /* Chinese Phonetic */
247
248 #ifndef HB_DISABLE_DEPRECATED
249 hb_tag_t
hb_ot_tag_from_language(hb_language_t language)250 hb_ot_tag_from_language (hb_language_t language)
251 {
252 unsigned int count = 1;
253 hb_tag_t tags[1];
254 hb_ot_tags_from_script_and_language (HB_SCRIPT_UNKNOWN, language, nullptr, nullptr, &count, tags);
255 return count > 0 ? tags[0] : HB_OT_TAG_DEFAULT_LANGUAGE;
256 }
257 #endif
258
259 static void
hb_ot_tags_from_language(const char * lang_str,const char * limit,unsigned int * count,hb_tag_t * tags)260 hb_ot_tags_from_language (const char *lang_str,
261 const char *limit,
262 unsigned int *count,
263 hb_tag_t *tags)
264 {
265 const char *s;
266 unsigned int tag_idx;
267
268 /* Check for matches of multiple subtags. */
269 if (hb_ot_tags_from_complex_language (lang_str, limit, count, tags))
270 return;
271
272 /* Find a language matching in the first component. */
273 s = strchr (lang_str, '-');
274 {
275 if (s && limit - lang_str >= 6)
276 {
277 const char *extlang_end = strchr (s + 1, '-');
278 /* If there is an extended language tag, use it. */
279 if (3 == (extlang_end ? extlang_end - s - 1 : strlen (s + 1)) &&
280 ISALPHA (s[1]))
281 lang_str = s + 1;
282 }
283 if (hb_sorted_array (ot_languages).bfind (lang_str, &tag_idx))
284 {
285 unsigned int i;
286 while (tag_idx != 0 &&
287 0 == strcmp (ot_languages[tag_idx].language, ot_languages[tag_idx - 1].language))
288 tag_idx--;
289 for (i = 0;
290 i < *count &&
291 tag_idx + i < ARRAY_LENGTH (ot_languages) &&
292 ot_languages[tag_idx + i].tag != HB_TAG_NONE &&
293 0 == strcmp (ot_languages[tag_idx + i].language, ot_languages[tag_idx].language);
294 i++)
295 tags[i] = ot_languages[tag_idx + i].tag;
296 *count = i;
297 return;
298 }
299 }
300
301 if (!s)
302 s = lang_str + strlen (lang_str);
303 if (s - lang_str == 3) {
304 /* Assume it's ISO-639-3 and upper-case and use it. */
305 tags[0] = hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
306 *count = 1;
307 return;
308 }
309
310 *count = 0;
311 }
312
313 static bool
parse_private_use_subtag(const char * private_use_subtag,unsigned int * count,hb_tag_t * tags,const char * prefix,unsigned char (* normalize)(unsigned char))314 parse_private_use_subtag (const char *private_use_subtag,
315 unsigned int *count,
316 hb_tag_t *tags,
317 const char *prefix,
318 unsigned char (*normalize) (unsigned char))
319 {
320 #ifdef HB_NO_LANGUAGE_PRIVATE_SUBTAG
321 return false;
322 #endif
323
324 if (!(private_use_subtag && count && tags && *count)) return false;
325
326 const char *s = strstr (private_use_subtag, prefix);
327 if (!s) return false;
328
329 char tag[4];
330 int i;
331 s += strlen (prefix);
332 if (s[0] == '-') {
333 s += 1;
334 char c;
335 for (i = 0; i < 8 && ISHEX (s[i]); i++)
336 {
337 c = FROMHEX (s[i]);
338 if (i % 2 == 0)
339 tag[i / 2] = c << 4;
340 else
341 tag[i / 2] += c;
342 }
343 if (i != 8) return false;
344 } else {
345 for (i = 0; i < 4 && ISALNUM (s[i]); i++)
346 tag[i] = normalize (s[i]);
347 if (!i) return false;
348
349 for (; i < 4; i++)
350 tag[i] = ' ';
351 }
352 tags[0] = HB_TAG (tag[0], tag[1], tag[2], tag[3]);
353 if ((tags[0] & 0xDFDFDFDF) == HB_OT_TAG_DEFAULT_SCRIPT)
354 tags[0] ^= ~0xDFDFDFDF;
355 *count = 1;
356 return true;
357 }
358
359 /**
360 * hb_ot_tags_from_script_and_language:
361 * @script: an #hb_script_t to convert.
362 * @language: an #hb_language_t to convert.
363 * @script_count: (inout) (optional): maximum number of script tags to retrieve (IN)
364 * and actual number of script tags retrieved (OUT)
365 * @script_tags: (out) (optional): array of size at least @script_count to store the
366 * script tag results
367 * @language_count: (inout) (optional): maximum number of language tags to retrieve
368 * (IN) and actual number of language tags retrieved (OUT)
369 * @language_tags: (out) (optional): array of size at least @language_count to store
370 * the language tag results
371 *
372 * Converts an #hb_script_t and an #hb_language_t to script and language tags.
373 *
374 * Since: 2.0.0
375 **/
376 void
hb_ot_tags_from_script_and_language(hb_script_t script,hb_language_t language,unsigned int * script_count,hb_tag_t * script_tags,unsigned int * language_count,hb_tag_t * language_tags)377 hb_ot_tags_from_script_and_language (hb_script_t script,
378 hb_language_t language,
379 unsigned int *script_count /* IN/OUT */,
380 hb_tag_t *script_tags /* OUT */,
381 unsigned int *language_count /* IN/OUT */,
382 hb_tag_t *language_tags /* OUT */)
383 {
384 bool needs_script = true;
385
386 if (language == HB_LANGUAGE_INVALID)
387 {
388 if (language_count && language_tags && *language_count)
389 *language_count = 0;
390 }
391 else
392 {
393 const char *lang_str, *s, *limit, *private_use_subtag;
394 bool needs_language;
395
396 lang_str = hb_language_to_string (language);
397 limit = nullptr;
398 private_use_subtag = nullptr;
399 if (lang_str[0] == 'x' && lang_str[1] == '-')
400 {
401 private_use_subtag = lang_str;
402 } else {
403 for (s = lang_str + 1; *s; s++)
404 {
405 if (s[-1] == '-' && s[1] == '-')
406 {
407 if (s[0] == 'x')
408 {
409 private_use_subtag = s;
410 if (!limit)
411 limit = s - 1;
412 break;
413 } else if (!limit)
414 {
415 limit = s - 1;
416 }
417 }
418 }
419 if (!limit)
420 limit = s;
421 }
422
423 needs_script = !parse_private_use_subtag (private_use_subtag, script_count, script_tags, "-hbsc", TOLOWER);
424 needs_language = !parse_private_use_subtag (private_use_subtag, language_count, language_tags, "-hbot", TOUPPER);
425
426 if (needs_language && language_count && language_tags && *language_count)
427 hb_ot_tags_from_language (lang_str, limit, language_count, language_tags);
428 }
429
430 if (needs_script && script_count && script_tags && *script_count)
431 hb_ot_all_tags_from_script (script, script_count, script_tags);
432 }
433
434 /**
435 * hb_ot_tag_to_language:
436 * @tag: an language tag
437 *
438 * Converts a language tag to an #hb_language_t.
439 *
440 * Return value: (transfer none) (nullable):
441 * The #hb_language_t corresponding to @tag.
442 *
443 * Since: 0.9.2
444 **/
445 hb_language_t
hb_ot_tag_to_language(hb_tag_t tag)446 hb_ot_tag_to_language (hb_tag_t tag)
447 {
448 unsigned int i;
449
450 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
451 return nullptr;
452
453 {
454 hb_language_t disambiguated_tag = hb_ot_ambiguous_tag_to_language (tag);
455 if (disambiguated_tag != HB_LANGUAGE_INVALID)
456 return disambiguated_tag;
457 }
458
459 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
460 if (ot_languages[i].tag == tag)
461 return hb_language_from_string (ot_languages[i].language, -1);
462
463 /* Return a custom language in the form of "x-hbot-AABBCCDD".
464 * If it's three letters long, also guess it's ISO 639-3 and lower-case and
465 * prepend it (if it's not a registered tag, the private use subtags will
466 * ensure that calling hb_ot_tag_from_language on the result will still return
467 * the same tag as the original tag).
468 */
469 {
470 char buf[20];
471 char *str = buf;
472 if (ISALPHA (tag >> 24)
473 && ISALPHA ((tag >> 16) & 0xFF)
474 && ISALPHA ((tag >> 8) & 0xFF)
475 && (tag & 0xFF) == ' ')
476 {
477 buf[0] = TOLOWER (tag >> 24);
478 buf[1] = TOLOWER ((tag >> 16) & 0xFF);
479 buf[2] = TOLOWER ((tag >> 8) & 0xFF);
480 buf[3] = '-';
481 str += 4;
482 }
483 snprintf (str, 16, "x-hbot-%08x", tag);
484 return hb_language_from_string (&*buf, -1);
485 }
486 }
487
488 /**
489 * hb_ot_tags_to_script_and_language:
490 * @script_tag: a script tag
491 * @language_tag: a language tag
492 * @script: (out) (optional): the #hb_script_t corresponding to @script_tag.
493 * @language: (out) (optional): the #hb_language_t corresponding to @script_tag and
494 * @language_tag.
495 *
496 * Converts a script tag and a language tag to an #hb_script_t and an
497 * #hb_language_t.
498 *
499 * Since: 2.0.0
500 **/
501 void
hb_ot_tags_to_script_and_language(hb_tag_t script_tag,hb_tag_t language_tag,hb_script_t * script,hb_language_t * language)502 hb_ot_tags_to_script_and_language (hb_tag_t script_tag,
503 hb_tag_t language_tag,
504 hb_script_t *script /* OUT */,
505 hb_language_t *language /* OUT */)
506 {
507 hb_script_t script_out = hb_ot_tag_to_script (script_tag);
508 if (script)
509 *script = script_out;
510 if (language)
511 {
512 unsigned int script_count = 1;
513 hb_tag_t primary_script_tag[1];
514 hb_ot_tags_from_script_and_language (script_out,
515 HB_LANGUAGE_INVALID,
516 &script_count,
517 primary_script_tag,
518 nullptr, nullptr);
519 *language = hb_ot_tag_to_language (language_tag);
520 if (script_count == 0 || primary_script_tag[0] != script_tag)
521 {
522 unsigned char *buf;
523 const char *lang_str = hb_language_to_string (*language);
524 size_t len = strlen (lang_str);
525 buf = (unsigned char *) malloc (len + 16);
526 if (unlikely (!buf))
527 {
528 *language = nullptr;
529 }
530 else
531 {
532 int shift;
533 memcpy (buf, lang_str, len);
534 if (lang_str[0] != 'x' || lang_str[1] != '-') {
535 buf[len++] = '-';
536 buf[len++] = 'x';
537 }
538 buf[len++] = '-';
539 buf[len++] = 'h';
540 buf[len++] = 'b';
541 buf[len++] = 's';
542 buf[len++] = 'c';
543 buf[len++] = '-';
544 for (shift = 28; shift >= 0; shift -= 4)
545 buf[len++] = TOHEX (script_tag >> shift);
546 *language = hb_language_from_string ((char *) buf, len);
547 free (buf);
548 }
549 }
550 }
551 }
552
553 #ifdef MAIN
554 static inline void
test_langs_sorted()555 test_langs_sorted ()
556 {
557 for (unsigned int i = 1; i < ARRAY_LENGTH (ot_languages); i++)
558 {
559 int c = ot_languages[i].cmp (&ot_languages[i - 1]);
560 if (c > 0)
561 {
562 fprintf (stderr, "ot_languages not sorted at index %d: %s %d %s\n",
563 i, ot_languages[i-1].language, c, ot_languages[i].language);
564 abort();
565 }
566 }
567 }
568
569 int
main()570 main ()
571 {
572 test_langs_sorted ();
573 return 0;
574 }
575
576 #endif
577
578
579 #endif
580