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 case HB_SCRIPT_MATH: return HB_OT_TAG_MATH_SCRIPT;
45
46 /* KATAKANA and HIRAGANA both map to 'kana' */
47 case HB_SCRIPT_HIRAGANA: return HB_TAG('k','a','n','a');
48
49 /* Spaces at the end are preserved, unlike ISO 15924 */
50 case HB_SCRIPT_LAO: return HB_TAG('l','a','o',' ');
51 case HB_SCRIPT_YI: return HB_TAG('y','i',' ',' ');
52 /* Unicode-5.0 additions */
53 case HB_SCRIPT_NKO: return HB_TAG('n','k','o',' ');
54 /* Unicode-5.1 additions */
55 case HB_SCRIPT_VAI: return HB_TAG('v','a','i',' ');
56 }
57
58 /* Else, just change first char to lowercase and return */
59 return ((hb_tag_t) script) | 0x20000000u;
60 }
61
62 static hb_script_t
hb_ot_old_tag_to_script(hb_tag_t tag)63 hb_ot_old_tag_to_script (hb_tag_t tag)
64 {
65 if (unlikely (tag == HB_OT_TAG_DEFAULT_SCRIPT))
66 return HB_SCRIPT_INVALID;
67 if (unlikely (tag == HB_OT_TAG_MATH_SCRIPT))
68 return HB_SCRIPT_MATH;
69
70 /* This side of the conversion is fully algorithmic. */
71
72 /* Any spaces at the end of the tag are replaced by repeating the last
73 * letter. Eg 'nko ' -> 'Nkoo' */
74 if (unlikely ((tag & 0x0000FF00u) == 0x00002000u))
75 tag |= (tag >> 8) & 0x0000FF00u; /* Copy second letter to third */
76 if (unlikely ((tag & 0x000000FFu) == 0x00000020u))
77 tag |= (tag >> 8) & 0x000000FFu; /* Copy third letter to fourth */
78
79 /* Change first char to uppercase and return */
80 return (hb_script_t) (tag & ~0x20000000u);
81 }
82
83 static hb_tag_t
hb_ot_new_tag_from_script(hb_script_t script)84 hb_ot_new_tag_from_script (hb_script_t script)
85 {
86 switch ((hb_tag_t) script) {
87 case HB_SCRIPT_BENGALI: return HB_TAG('b','n','g','2');
88 case HB_SCRIPT_DEVANAGARI: return HB_TAG('d','e','v','2');
89 case HB_SCRIPT_GUJARATI: return HB_TAG('g','j','r','2');
90 case HB_SCRIPT_GURMUKHI: return HB_TAG('g','u','r','2');
91 case HB_SCRIPT_KANNADA: return HB_TAG('k','n','d','2');
92 case HB_SCRIPT_MALAYALAM: return HB_TAG('m','l','m','2');
93 case HB_SCRIPT_ORIYA: return HB_TAG('o','r','y','2');
94 case HB_SCRIPT_TAMIL: return HB_TAG('t','m','l','2');
95 case HB_SCRIPT_TELUGU: return HB_TAG('t','e','l','2');
96 case HB_SCRIPT_MYANMAR: return HB_TAG('m','y','m','2');
97 }
98
99 return HB_OT_TAG_DEFAULT_SCRIPT;
100 }
101
102 static hb_script_t
hb_ot_new_tag_to_script(hb_tag_t tag)103 hb_ot_new_tag_to_script (hb_tag_t tag)
104 {
105 switch (tag) {
106 case HB_TAG('b','n','g','2'): return HB_SCRIPT_BENGALI;
107 case HB_TAG('d','e','v','2'): return HB_SCRIPT_DEVANAGARI;
108 case HB_TAG('g','j','r','2'): return HB_SCRIPT_GUJARATI;
109 case HB_TAG('g','u','r','2'): return HB_SCRIPT_GURMUKHI;
110 case HB_TAG('k','n','d','2'): return HB_SCRIPT_KANNADA;
111 case HB_TAG('m','l','m','2'): return HB_SCRIPT_MALAYALAM;
112 case HB_TAG('o','r','y','2'): return HB_SCRIPT_ORIYA;
113 case HB_TAG('t','m','l','2'): return HB_SCRIPT_TAMIL;
114 case HB_TAG('t','e','l','2'): return HB_SCRIPT_TELUGU;
115 case HB_TAG('m','y','m','2'): return HB_SCRIPT_MYANMAR;
116 }
117
118 return HB_SCRIPT_UNKNOWN;
119 }
120
121 #ifndef HB_DISABLE_DEPRECATED
122 void
hb_ot_tags_from_script(hb_script_t script,hb_tag_t * script_tag_1,hb_tag_t * script_tag_2)123 hb_ot_tags_from_script (hb_script_t script,
124 hb_tag_t *script_tag_1,
125 hb_tag_t *script_tag_2)
126 {
127 unsigned int count = 2;
128 hb_tag_t tags[2];
129 hb_ot_tags_from_script_and_language (script, HB_LANGUAGE_INVALID, &count, tags, nullptr, nullptr);
130 *script_tag_1 = count > 0 ? tags[0] : HB_OT_TAG_DEFAULT_SCRIPT;
131 *script_tag_2 = count > 1 ? tags[1] : HB_OT_TAG_DEFAULT_SCRIPT;
132 }
133 #endif
134
135 /*
136 * Complete list at:
137 * https://docs.microsoft.com/en-us/typography/opentype/spec/scripttags
138 *
139 * Most of the script tags are the same as the ISO 15924 tag but lowercased.
140 * So we just do that, and handle the exceptional cases in a switch.
141 */
142
143 static void
hb_ot_all_tags_from_script(hb_script_t script,unsigned int * count,hb_tag_t * tags)144 hb_ot_all_tags_from_script (hb_script_t script,
145 unsigned int *count /* IN/OUT */,
146 hb_tag_t *tags /* OUT */)
147 {
148 unsigned int i = 0;
149
150 hb_tag_t new_tag = hb_ot_new_tag_from_script (script);
151 if (unlikely (new_tag != HB_OT_TAG_DEFAULT_SCRIPT))
152 {
153 /* HB_SCRIPT_MYANMAR maps to 'mym2', but there is no 'mym3'. */
154 if (new_tag != HB_TAG('m','y','m','2'))
155 tags[i++] = new_tag | '3';
156 if (*count > i)
157 tags[i++] = new_tag;
158 }
159
160 if (*count > i)
161 {
162 hb_tag_t old_tag = hb_ot_old_tag_from_script (script);
163 if (old_tag != HB_OT_TAG_DEFAULT_SCRIPT)
164 tags[i++] = old_tag;
165 }
166
167 *count = i;
168 }
169
170 /**
171 * hb_ot_tag_to_script:
172 * @tag: a script tag
173 *
174 * Converts a script tag to an #hb_script_t.
175 *
176 * Return value: The #hb_script_t corresponding to @tag.
177 *
178 **/
179 hb_script_t
hb_ot_tag_to_script(hb_tag_t tag)180 hb_ot_tag_to_script (hb_tag_t tag)
181 {
182 unsigned char digit = tag & 0x000000FFu;
183 if (unlikely (digit == '2' || digit == '3'))
184 return hb_ot_new_tag_to_script (tag & 0xFFFFFF32);
185
186 return hb_ot_old_tag_to_script (tag);
187 }
188
189
190 /* hb_language_t */
191
192 static bool
subtag_matches(const char * lang_str,const char * limit,const char * subtag)193 subtag_matches (const char *lang_str,
194 const char *limit,
195 const char *subtag)
196 {
197 do {
198 const char *s = strstr (lang_str, subtag);
199 if (!s || s >= limit)
200 return false;
201 if (!ISALNUM (s[strlen (subtag)]))
202 return true;
203 lang_str = s + strlen (subtag);
204 } while (true);
205 }
206
207 static hb_bool_t
lang_matches(const char * lang_str,const char * spec)208 lang_matches (const char *lang_str, const char *spec)
209 {
210 unsigned int len = strlen (spec);
211
212 return strncmp (lang_str, spec, len) == 0 &&
213 (lang_str[len] == '\0' || lang_str[len] == '-');
214 }
215
216 struct LangTag
217 {
218 char language[4];
219 hb_tag_t tag;
220
cmpLangTag221 int cmp (const char *a) const
222 {
223 const char *b = this->language;
224 unsigned int da, db;
225 const char *p;
226
227 p = strchr (a, '-');
228 da = p ? (unsigned int) (p - a) : strlen (a);
229
230 p = strchr (b, '-');
231 db = p ? (unsigned int) (p - b) : strlen (b);
232
233 return strncmp (a, b, hb_max (da, db));
234 }
cmpLangTag235 int cmp (const LangTag *that) const
236 { return cmp (that->language); }
237 };
238
239 #include "hb-ot-tag-table.hh"
240
241 /* The corresponding languages IDs for the following IDs are unclear,
242 * overlap, or are architecturally weird. Needs more research. */
243
244 /*{"??", {HB_TAG('B','C','R',' ')}},*/ /* Bible Cree */
245 /*{"zh?", {HB_TAG('C','H','N',' ')}},*/ /* Chinese (seen in Microsoft fonts) */
246 /*{"ar-Syrc?", {HB_TAG('G','A','R',' ')}},*/ /* Garshuni */
247 /*{"??", {HB_TAG('N','G','R',' ')}},*/ /* Nagari */
248 /*{"??", {HB_TAG('Y','I','C',' ')}},*/ /* Yi Classic */
249 /*{"zh?", {HB_TAG('Z','H','P',' ')}},*/ /* Chinese Phonetic */
250
251 #ifndef HB_DISABLE_DEPRECATED
252 hb_tag_t
hb_ot_tag_from_language(hb_language_t language)253 hb_ot_tag_from_language (hb_language_t language)
254 {
255 unsigned int count = 1;
256 hb_tag_t tags[1];
257 hb_ot_tags_from_script_and_language (HB_SCRIPT_UNKNOWN, language, nullptr, nullptr, &count, tags);
258 return count > 0 ? tags[0] : HB_OT_TAG_DEFAULT_LANGUAGE;
259 }
260 #endif
261
262 static void
hb_ot_tags_from_language(const char * lang_str,const char * limit,unsigned int * count,hb_tag_t * tags)263 hb_ot_tags_from_language (const char *lang_str,
264 const char *limit,
265 unsigned int *count,
266 hb_tag_t *tags)
267 {
268 const char *s;
269 unsigned int tag_idx;
270
271 /* Check for matches of multiple subtags. */
272 if (hb_ot_tags_from_complex_language (lang_str, limit, count, tags))
273 return;
274
275 /* Find a language matching in the first component. */
276 s = strchr (lang_str, '-');
277 {
278 if (s && limit - lang_str >= 6)
279 {
280 const char *extlang_end = strchr (s + 1, '-');
281 /* If there is an extended language tag, use it. */
282 if (3 == (extlang_end ? extlang_end - s - 1 : strlen (s + 1)) &&
283 ISALPHA (s[1]))
284 lang_str = s + 1;
285 }
286 if (hb_sorted_array (ot_languages).bfind (lang_str, &tag_idx))
287 {
288 unsigned int i;
289 while (tag_idx != 0 &&
290 0 == strcmp (ot_languages[tag_idx].language, ot_languages[tag_idx - 1].language))
291 tag_idx--;
292 for (i = 0;
293 i < *count &&
294 tag_idx + i < ARRAY_LENGTH (ot_languages) &&
295 ot_languages[tag_idx + i].tag != HB_TAG_NONE &&
296 0 == strcmp (ot_languages[tag_idx + i].language, ot_languages[tag_idx].language);
297 i++)
298 tags[i] = ot_languages[tag_idx + i].tag;
299 *count = i;
300 return;
301 }
302 }
303
304 if (!s)
305 s = lang_str + strlen (lang_str);
306 if (s - lang_str == 3) {
307 /* Assume it's ISO-639-3 and upper-case and use it. */
308 tags[0] = hb_tag_from_string (lang_str, s - lang_str) & ~0x20202000u;
309 *count = 1;
310 return;
311 }
312
313 *count = 0;
314 }
315
316 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))317 parse_private_use_subtag (const char *private_use_subtag,
318 unsigned int *count,
319 hb_tag_t *tags,
320 const char *prefix,
321 unsigned char (*normalize) (unsigned char))
322 {
323 #ifdef HB_NO_LANGUAGE_PRIVATE_SUBTAG
324 return false;
325 #endif
326
327 if (!(private_use_subtag && count && tags && *count)) return false;
328
329 const char *s = strstr (private_use_subtag, prefix);
330 if (!s) return false;
331
332 char tag[4];
333 int i;
334 s += strlen (prefix);
335 if (s[0] == '-') {
336 s += 1;
337 char c;
338 for (i = 0; i < 8 && ISHEX (s[i]); i++)
339 {
340 c = FROMHEX (s[i]);
341 if (i % 2 == 0)
342 tag[i / 2] = c << 4;
343 else
344 tag[i / 2] += c;
345 }
346 if (i != 8) return false;
347 } else {
348 for (i = 0; i < 4 && ISALNUM (s[i]); i++)
349 tag[i] = normalize (s[i]);
350 if (!i) return false;
351
352 for (; i < 4; i++)
353 tag[i] = ' ';
354 }
355 tags[0] = HB_TAG (tag[0], tag[1], tag[2], tag[3]);
356 if ((tags[0] & 0xDFDFDFDF) == HB_OT_TAG_DEFAULT_SCRIPT)
357 tags[0] ^= ~0xDFDFDFDF;
358 *count = 1;
359 return true;
360 }
361
362 /**
363 * hb_ot_tags_from_script_and_language:
364 * @script: an #hb_script_t to convert.
365 * @language: an #hb_language_t to convert.
366 * @script_count: (inout) (optional): maximum number of script tags to retrieve (IN)
367 * and actual number of script tags retrieved (OUT)
368 * @script_tags: (out) (optional): array of size at least @script_count to store the
369 * script tag results
370 * @language_count: (inout) (optional): maximum number of language tags to retrieve
371 * (IN) and actual number of language tags retrieved (OUT)
372 * @language_tags: (out) (optional): array of size at least @language_count to store
373 * the language tag results
374 *
375 * Converts an #hb_script_t and an #hb_language_t to script and language tags.
376 *
377 * Since: 2.0.0
378 **/
379 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)380 hb_ot_tags_from_script_and_language (hb_script_t script,
381 hb_language_t language,
382 unsigned int *script_count /* IN/OUT */,
383 hb_tag_t *script_tags /* OUT */,
384 unsigned int *language_count /* IN/OUT */,
385 hb_tag_t *language_tags /* OUT */)
386 {
387 bool needs_script = true;
388
389 if (language == HB_LANGUAGE_INVALID)
390 {
391 if (language_count && language_tags && *language_count)
392 *language_count = 0;
393 }
394 else
395 {
396 const char *lang_str, *s, *limit, *private_use_subtag;
397 bool needs_language;
398
399 lang_str = hb_language_to_string (language);
400 limit = nullptr;
401 private_use_subtag = nullptr;
402 if (lang_str[0] == 'x' && lang_str[1] == '-')
403 {
404 private_use_subtag = lang_str;
405 } else {
406 for (s = lang_str + 1; *s; s++)
407 {
408 if (s[-1] == '-' && s[1] == '-')
409 {
410 if (s[0] == 'x')
411 {
412 private_use_subtag = s;
413 if (!limit)
414 limit = s - 1;
415 break;
416 } else if (!limit)
417 {
418 limit = s - 1;
419 }
420 }
421 }
422 if (!limit)
423 limit = s;
424 }
425
426 needs_script = !parse_private_use_subtag (private_use_subtag, script_count, script_tags, "-hbsc", TOLOWER);
427 needs_language = !parse_private_use_subtag (private_use_subtag, language_count, language_tags, "-hbot", TOUPPER);
428
429 if (needs_language && language_count && language_tags && *language_count)
430 hb_ot_tags_from_language (lang_str, limit, language_count, language_tags);
431 }
432
433 if (needs_script && script_count && script_tags && *script_count)
434 hb_ot_all_tags_from_script (script, script_count, script_tags);
435 }
436
437 /**
438 * hb_ot_tag_to_language:
439 * @tag: an language tag
440 *
441 * Converts a language tag to an #hb_language_t.
442 *
443 * Return value: (transfer none) (nullable):
444 * The #hb_language_t corresponding to @tag.
445 *
446 * Since: 0.9.2
447 **/
448 hb_language_t
hb_ot_tag_to_language(hb_tag_t tag)449 hb_ot_tag_to_language (hb_tag_t tag)
450 {
451 unsigned int i;
452
453 if (tag == HB_OT_TAG_DEFAULT_LANGUAGE)
454 return nullptr;
455
456 {
457 hb_language_t disambiguated_tag = hb_ot_ambiguous_tag_to_language (tag);
458 if (disambiguated_tag != HB_LANGUAGE_INVALID)
459 return disambiguated_tag;
460 }
461
462 for (i = 0; i < ARRAY_LENGTH (ot_languages); i++)
463 if (ot_languages[i].tag == tag)
464 return hb_language_from_string (ot_languages[i].language, -1);
465
466 /* Return a custom language in the form of "x-hbot-AABBCCDD".
467 * If it's three letters long, also guess it's ISO 639-3 and lower-case and
468 * prepend it (if it's not a registered tag, the private use subtags will
469 * ensure that calling hb_ot_tag_from_language on the result will still return
470 * the same tag as the original tag).
471 */
472 {
473 char buf[20];
474 char *str = buf;
475 if (ISALPHA (tag >> 24)
476 && ISALPHA ((tag >> 16) & 0xFF)
477 && ISALPHA ((tag >> 8) & 0xFF)
478 && (tag & 0xFF) == ' ')
479 {
480 buf[0] = TOLOWER (tag >> 24);
481 buf[1] = TOLOWER ((tag >> 16) & 0xFF);
482 buf[2] = TOLOWER ((tag >> 8) & 0xFF);
483 buf[3] = '-';
484 str += 4;
485 }
486 snprintf (str, 16, "x-hbot-%08x", tag);
487 return hb_language_from_string (&*buf, -1);
488 }
489 }
490
491 /**
492 * hb_ot_tags_to_script_and_language:
493 * @script_tag: a script tag
494 * @language_tag: a language tag
495 * @script: (out) (optional): the #hb_script_t corresponding to @script_tag.
496 * @language: (out) (optional): the #hb_language_t corresponding to @script_tag and
497 * @language_tag.
498 *
499 * Converts a script tag and a language tag to an #hb_script_t and an
500 * #hb_language_t.
501 *
502 * Since: 2.0.0
503 **/
504 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)505 hb_ot_tags_to_script_and_language (hb_tag_t script_tag,
506 hb_tag_t language_tag,
507 hb_script_t *script /* OUT */,
508 hb_language_t *language /* OUT */)
509 {
510 hb_script_t script_out = hb_ot_tag_to_script (script_tag);
511 if (script)
512 *script = script_out;
513 if (language)
514 {
515 unsigned int script_count = 1;
516 hb_tag_t primary_script_tag[1];
517 hb_ot_tags_from_script_and_language (script_out,
518 HB_LANGUAGE_INVALID,
519 &script_count,
520 primary_script_tag,
521 nullptr, nullptr);
522 *language = hb_ot_tag_to_language (language_tag);
523 if (script_count == 0 || primary_script_tag[0] != script_tag)
524 {
525 unsigned char *buf;
526 const char *lang_str = hb_language_to_string (*language);
527 size_t len = strlen (lang_str);
528 buf = (unsigned char *) hb_malloc (len + 16);
529 if (unlikely (!buf))
530 {
531 *language = nullptr;
532 }
533 else
534 {
535 int shift;
536 memcpy (buf, lang_str, len);
537 if (lang_str[0] != 'x' || lang_str[1] != '-') {
538 buf[len++] = '-';
539 buf[len++] = 'x';
540 }
541 buf[len++] = '-';
542 buf[len++] = 'h';
543 buf[len++] = 'b';
544 buf[len++] = 's';
545 buf[len++] = 'c';
546 buf[len++] = '-';
547 for (shift = 28; shift >= 0; shift -= 4)
548 buf[len++] = TOHEX (script_tag >> shift);
549 *language = hb_language_from_string ((char *) buf, len);
550 hb_free (buf);
551 }
552 }
553 }
554 }
555
556 #ifdef MAIN
557 static inline void
test_langs_sorted()558 test_langs_sorted ()
559 {
560 for (unsigned int i = 1; i < ARRAY_LENGTH (ot_languages); i++)
561 {
562 int c = ot_languages[i].cmp (&ot_languages[i - 1]);
563 if (c > 0)
564 {
565 fprintf (stderr, "ot_languages not sorted at index %d: %s %d %s\n",
566 i, ot_languages[i-1].language, c, ot_languages[i].language);
567 abort();
568 }
569 }
570 }
571
572 int
main()573 main ()
574 {
575 test_langs_sorted ();
576 return 0;
577 }
578
579 #endif
580
581
582 #endif
583