1 /*
2 * Copyright © 2011 Google, Inc.
3 *
4 * This is part of HarfBuzz, a text shaping library.
5 *
6 * Permission is hereby granted, without written agreement and without
7 * license or royalty fees, to use, copy, modify, and distribute this
8 * software and its documentation for any purpose, provided that the
9 * above copyright notice and the following two paragraphs appear in
10 * all copies of this software.
11 *
12 * IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
13 * DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
14 * ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
15 * IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
16 * DAMAGE.
17 *
18 * THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
19 * BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
20 * FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
21 * ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
22 * PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
23 *
24 * Google Author(s): Behdad Esfahbod
25 */
26
27 #include "hb-test.h"
28
29 #include <hb-ot.h>
30
31 /* Unit tests for hb-ot-tag.h */
32
33
34 /* https://docs.microsoft.com/en-us/typography/opentype/spec/scripttags */
35
36 static void
test_simple_tags(const char * s,hb_script_t script)37 test_simple_tags (const char *s, hb_script_t script)
38 {
39 hb_script_t tag;
40 unsigned int count = 2;
41 hb_tag_t t[2];
42
43 g_test_message ("Testing script %c%c%c%c: tag %s", HB_UNTAG (hb_script_to_iso15924_tag (script)), s);
44 tag = hb_tag_from_string (s, -1);
45
46 hb_ot_tags_from_script_and_language (script,
47 HB_LANGUAGE_INVALID,
48 &count, t, NULL, NULL);
49
50 if (count)
51 g_assert_cmphex (t[0], ==, tag);
52 else
53 g_assert_cmphex (HB_TAG_CHAR4 ("DFLT"), ==, tag);
54
55 g_assert_cmphex (hb_ot_tag_to_script (tag), ==, script);
56 }
57
58 static void
test_script_tags_from_language(const char * s,const char * lang_s,hb_script_t script)59 test_script_tags_from_language (const char *s, const char *lang_s, hb_script_t script)
60 {
61 hb_script_t tag;
62 unsigned int count = 1;
63 hb_tag_t t;
64
65 g_test_message ("Testing script %c%c%c%c: script tag %s, language tag %s", HB_UNTAG (hb_script_to_iso15924_tag (script)), s, lang_s);
66 tag = hb_tag_from_string (s, -1);
67
68 hb_ot_tags_from_script_and_language (script, hb_language_from_string (lang_s, -1), &count, &t, NULL, NULL);
69
70 if (count != 0)
71 {
72 g_assert_cmpuint (count, ==, 1);
73 g_assert_cmphex (t, ==, tag);
74 }
75 }
76
77 static void
test_indic_tags(const char * s1,const char * s2,const char * s3,hb_script_t script)78 test_indic_tags (const char *s1, const char *s2, const char *s3, hb_script_t script)
79 {
80 hb_script_t tag1, tag2, tag3;
81 hb_tag_t t[3];
82 unsigned int count = 3;
83
84 g_test_message ("Testing script %c%c%c%c: USE tag %s, new tag %s, old tag %s", HB_UNTAG (hb_script_to_iso15924_tag (script)), s1, s2, s3);
85 tag1 = hb_tag_from_string (s1, -1);
86 tag2 = hb_tag_from_string (s2, -1);
87 tag3 = hb_tag_from_string (s3, -1);
88
89 hb_ot_tags_from_script_and_language (script,
90 HB_LANGUAGE_INVALID,
91 &count, t, NULL, NULL);
92
93 g_assert_cmpuint (count, ==, 3);
94 g_assert_cmphex (t[0], ==, tag1);
95 g_assert_cmphex (t[1], ==, tag2);
96 g_assert_cmphex (t[2], ==, tag3);
97
98 g_assert_cmphex (hb_ot_tag_to_script (tag1), ==, script);
99 g_assert_cmphex (hb_ot_tag_to_script (tag2), ==, script);
100 g_assert_cmphex (hb_ot_tag_to_script (tag3), ==, script);
101 }
102
103 static void
test_ot_tag_script_degenerate(void)104 test_ot_tag_script_degenerate (void)
105 {
106 hb_tag_t t[2];
107 unsigned int count = 2;
108
109 g_assert_cmphex (HB_TAG_CHAR4 ("DFLT"), ==, HB_OT_TAG_DEFAULT_SCRIPT);
110
111 /* HIRAGANA and KATAKANA both map to 'kana' */
112 test_simple_tags ("kana", HB_SCRIPT_KATAKANA);
113
114 hb_ot_tags_from_script_and_language (HB_SCRIPT_HIRAGANA,
115 HB_LANGUAGE_INVALID,
116 &count, t, NULL, NULL);
117
118 g_assert_cmpuint (count, ==, 1);
119 g_assert_cmphex (t[0], ==, HB_TAG_CHAR4 ("kana"));
120
121 test_simple_tags ("DFLT", HB_SCRIPT_INVALID);
122
123 /* Spaces are replaced */
124 g_assert_cmphex (hb_ot_tag_to_script (HB_TAG_CHAR4 ("be ")), ==, hb_script_from_string ("Beee", -1));
125 }
126
127 static void
test_ot_tag_script_simple(void)128 test_ot_tag_script_simple (void)
129 {
130 /* Arbitrary non-existent script */
131 test_simple_tags ("wwyz", hb_script_from_string ("wWyZ", -1));
132
133 /* These we don't really care about */
134 test_simple_tags ("zyyy", HB_SCRIPT_COMMON);
135 test_simple_tags ("zinh", HB_SCRIPT_INHERITED);
136 test_simple_tags ("zzzz", HB_SCRIPT_UNKNOWN);
137
138 test_simple_tags ("arab", HB_SCRIPT_ARABIC);
139 test_simple_tags ("copt", HB_SCRIPT_COPTIC);
140 test_simple_tags ("kana", HB_SCRIPT_KATAKANA);
141 test_simple_tags ("latn", HB_SCRIPT_LATIN);
142
143 test_simple_tags ("math", HB_SCRIPT_MATH);
144
145 /* These are trickier since their OT script tags have space. */
146 test_simple_tags ("lao ", HB_SCRIPT_LAO);
147 test_simple_tags ("yi ", HB_SCRIPT_YI);
148 /* Unicode-5.0 additions */
149 test_simple_tags ("nko ", HB_SCRIPT_NKO);
150 /* Unicode-5.1 additions */
151 test_simple_tags ("vai ", HB_SCRIPT_VAI);
152
153 /* https://docs.microsoft.com/en-us/typography/opentype/spec/scripttags */
154
155 /* Unicode-5.2 additions */
156 test_simple_tags ("mtei", HB_SCRIPT_MEETEI_MAYEK);
157 /* Unicode-6.0 additions */
158 test_simple_tags ("mand", HB_SCRIPT_MANDAIC);
159 }
160
161 static void
test_ot_tag_script_from_language(void)162 test_ot_tag_script_from_language (void)
163 {
164 test_script_tags_from_language (NULL, NULL, HB_SCRIPT_INVALID);
165 test_script_tags_from_language (NULL, "en", HB_SCRIPT_INVALID);
166 test_script_tags_from_language ("copt", "en", HB_SCRIPT_COPTIC);
167 test_script_tags_from_language (NULL, "x-hbsc", HB_SCRIPT_INVALID);
168 test_script_tags_from_language ("copt", "x-hbsc", HB_SCRIPT_COPTIC);
169 test_script_tags_from_language (NULL, "x-hbsc-", HB_SCRIPT_INVALID);
170 test_script_tags_from_language (NULL, "x-hbsc-1", HB_SCRIPT_INVALID);
171 test_script_tags_from_language (NULL, "x-hbsc-1a", HB_SCRIPT_INVALID);
172 test_script_tags_from_language (NULL, "x-hbsc-1a2b3c4x", HB_SCRIPT_INVALID);
173 test_script_tags_from_language ("2lon", "x-hbsc-326c6f6e67", HB_SCRIPT_INVALID);
174 test_script_tags_from_language ("abc ", "x-hbscabc", HB_SCRIPT_INVALID);
175 test_script_tags_from_language ("deva", "x-hbscdeva", HB_SCRIPT_INVALID);
176 test_script_tags_from_language ("dev2", "x-hbscdev2", HB_SCRIPT_INVALID);
177 test_script_tags_from_language ("dev3", "x-hbscdev3", HB_SCRIPT_INVALID);
178 test_script_tags_from_language ("dev3", "x-hbsc-64657633", HB_SCRIPT_INVALID);
179 test_script_tags_from_language ("copt", "x-hbotpap0-hbsccopt", HB_SCRIPT_INVALID);
180 test_script_tags_from_language (NULL, "en-x-hbsc", HB_SCRIPT_INVALID);
181 test_script_tags_from_language ("copt", "en-x-hbsc", HB_SCRIPT_COPTIC);
182 test_script_tags_from_language ("abc ", "en-x-hbscabc", HB_SCRIPT_INVALID);
183 test_script_tags_from_language ("deva", "en-x-hbscdeva", HB_SCRIPT_INVALID);
184 test_script_tags_from_language ("dev2", "en-x-hbscdev2", HB_SCRIPT_INVALID);
185 test_script_tags_from_language ("dev3", "en-x-hbscdev3", HB_SCRIPT_INVALID);
186 test_script_tags_from_language ("dev3", "en-x-hbsc-64657633", HB_SCRIPT_INVALID);
187 test_script_tags_from_language ("copt", "en-x-hbotpap0-hbsccopt", HB_SCRIPT_INVALID);
188 }
189
190 static void
test_ot_tag_script_indic(void)191 test_ot_tag_script_indic (void)
192 {
193 test_indic_tags ("bng3", "bng2", "beng", HB_SCRIPT_BENGALI);
194 test_indic_tags ("dev3", "dev2", "deva", HB_SCRIPT_DEVANAGARI);
195 test_indic_tags ("gjr3", "gjr2", "gujr", HB_SCRIPT_GUJARATI);
196 test_indic_tags ("gur3", "gur2", "guru", HB_SCRIPT_GURMUKHI);
197 test_indic_tags ("knd3", "knd2", "knda", HB_SCRIPT_KANNADA);
198 test_indic_tags ("mlm3", "mlm2", "mlym", HB_SCRIPT_MALAYALAM);
199 test_indic_tags ("ory3", "ory2", "orya", HB_SCRIPT_ORIYA);
200 test_indic_tags ("tml3", "tml2", "taml", HB_SCRIPT_TAMIL);
201 test_indic_tags ("tel3", "tel2", "telu", HB_SCRIPT_TELUGU);
202 }
203
204
205
206 /* https://docs.microsoft.com/en-us/typography/opentype/spec/languagetags */
207
208 static void
test_language_two_way(const char * tag_s,const char * lang_s)209 test_language_two_way (const char *tag_s, const char *lang_s)
210 {
211 hb_language_t lang = hb_language_from_string (lang_s, -1);
212 hb_tag_t tag = hb_tag_from_string (tag_s, -1);
213 hb_tag_t tag2;
214 unsigned int count = 1;
215
216 g_test_message ("Testing language %s <-> tag %s", lang_s, tag_s);
217
218 hb_ot_tags_from_script_and_language (HB_SCRIPT_INVALID,
219 lang,
220 NULL, NULL, &count, &tag2);
221
222 if (count)
223 g_assert_cmphex (tag, ==, tag2);
224 else
225 g_assert_cmphex (tag, ==, HB_TAG_CHAR4 ("dflt"));
226 g_assert (lang == hb_ot_tag_to_language (tag));
227 }
228
229 static void
test_tag_from_language(const char * tag_s,const char * lang_s)230 test_tag_from_language (const char *tag_s, const char *lang_s)
231 {
232 hb_language_t lang = hb_language_from_string (lang_s, -1);
233 hb_tag_t tag = hb_tag_from_string (tag_s, -1);
234 hb_tag_t tag2;
235 unsigned int count = 1;
236
237 g_test_message ("Testing language %s -> tag %s", lang_s, tag_s);
238
239 hb_ot_tags_from_script_and_language (HB_SCRIPT_INVALID,
240 lang,
241 NULL, NULL, &count, &tag2);
242
243 if (count)
244 g_assert_cmphex (tag, ==, tag2);
245 else
246 g_assert_cmphex (tag, ==, HB_TAG_CHAR4 ("dflt"));
247 }
248
249 static void
test_tag_to_language(const char * tag_s,const char * lang_s)250 test_tag_to_language (const char *tag_s, const char *lang_s)
251 {
252 hb_language_t lang = hb_language_from_string (lang_s, -1);
253 hb_tag_t tag = hb_tag_from_string (tag_s, -1);
254
255 g_test_message ("Testing tag %s -> language %s", tag_s, lang_s);
256
257 g_assert (lang == hb_ot_tag_to_language (tag));
258 }
259
260 static void
test_tags_to_script_and_language(const char * script_tag_s,const char * lang_tag_s,const char * script_s,const char * lang_s)261 test_tags_to_script_and_language (const char *script_tag_s,
262 const char *lang_tag_s,
263 const char *script_s,
264 const char *lang_s)
265 {
266 hb_script_t actual_script[1];
267 hb_language_t actual_lang[1];
268 hb_tag_t script_tag = hb_tag_from_string (script_tag_s, -1);
269 hb_tag_t lang_tag = hb_tag_from_string (lang_tag_s, -1);
270 hb_ot_tags_to_script_and_language (script_tag, lang_tag, actual_script, actual_lang);
271 g_assert_cmphex (*actual_script, ==, hb_tag_from_string (script_s, -1));
272 g_assert_cmpstr (hb_language_to_string (*actual_lang), ==, lang_s);
273 }
274
275 static void
test_ot_tags_to_script_and_language(void)276 test_ot_tags_to_script_and_language (void)
277 {
278 test_tags_to_script_and_language ("DFLT", "ENG", "", "en-x-hbsc-44464c54");
279 test_tags_to_script_and_language ("latn", "ENG", "Latn", "en");
280 test_tags_to_script_and_language ("deva", "MAR", "Deva", "mr-x-hbsc-64657661");
281 test_tags_to_script_and_language ("dev2", "MAR", "Deva", "mr-x-hbsc-64657632");
282 test_tags_to_script_and_language ("dev3", "MAR", "Deva", "mr");
283 test_tags_to_script_and_language ("qaa", "QTZ0", "Qaaa", "x-hbot-51545a30-hbsc-71616120");
284 }
285
286 static void
test_ot_tag_language(void)287 test_ot_tag_language (void)
288 {
289 g_assert_cmphex (HB_TAG_CHAR4 ("dflt"), ==, HB_OT_TAG_DEFAULT_LANGUAGE);
290 test_language_two_way ("dflt", NULL);
291
292 test_language_two_way ("ALT", "alt");
293
294 test_language_two_way ("ARA", "ar");
295
296 test_language_two_way ("AZE", "az");
297 test_tag_from_language ("AZE", "az-ir");
298 test_tag_from_language ("AZE", "az-az");
299
300 test_language_two_way ("ENG", "en");
301 test_tag_from_language ("ENG", "en_US");
302
303 test_language_two_way ("CJA", "cja-x-hbot-434a4120"); /* Western Cham */
304 test_language_two_way ("CJM", "cjm-x-hbot-434a4d20"); /* Eastern Cham */
305 test_tag_from_language ("CJM", "cjm");
306 test_language_two_way ("EVN", "eve");
307
308 test_language_two_way ("HAL", "cfm"); /* BCP47 and current ISO639-3 code for Halam/Falam Chin */
309 test_tag_from_language ("HAL", "flm"); /* Retired ISO639-3 code for Halam/Falam Chin */
310
311 test_language_two_way ("HYE0", "hy");
312 test_language_two_way ("HYE", "hyw");
313
314 test_tag_from_language ("QIN", "bgr"); /* Bawm Chin */
315 test_tag_from_language ("QIN", "cbl"); /* Bualkhaw Chin */
316 test_tag_from_language ("QIN", "cka"); /* Khumi Awa Chin */
317 test_tag_from_language ("QIN", "cmr"); /* Mro-Khimi Chin */
318 test_tag_from_language ("QIN", "cnb"); /* Chinbon Chin */
319 test_tag_from_language ("QIN", "cnh"); /* Hakha Chin */
320 test_tag_from_language ("QIN", "cnk"); /* Khumi Chin */
321 test_tag_from_language ("QIN", "cnw"); /* Ngawn Chin */
322 test_tag_from_language ("QIN", "csh"); /* Asho Chin */
323 test_tag_from_language ("QIN", "csy"); /* Siyin Chin */
324 test_tag_from_language ("QIN", "ctd"); /* Tedim Chin */
325 test_tag_from_language ("QIN", "czt"); /* Zotung Chin */
326 test_tag_from_language ("QIN", "dao"); /* Daai Chin */
327 test_tag_from_language ("QIN", "hlt"); /* Matu Chin */
328 test_tag_from_language ("QIN", "mrh"); /* Mara Chin */
329 test_tag_from_language ("QIN", "pck"); /* Paite Chin */
330 test_tag_from_language ("QIN", "sez"); /* Senthang Chin */
331 test_tag_from_language ("QIN", "tcp"); /* Tawr Chin */
332 test_tag_from_language ("QIN", "tcz"); /* Thado Chin */
333 test_tag_from_language ("QIN", "yos"); /* Yos, deprecated by IANA in favor of Zou [zom] */
334 test_tag_from_language ("QIN", "zom"); /* Zou */
335 test_tag_to_language ("QIN", "bgr"); /* no single BCP47 tag for Chin; picking Bawm Chin */
336
337 test_language_two_way ("FAR", "fa");
338 test_tag_from_language ("FAR", "fa_IR");
339
340 test_language_two_way ("MNK", "man"); /* Mandingo [macrolanguage] */
341
342 test_language_two_way ("SWA", "aii"); /* Swadaya Aramaic */
343
344 test_language_two_way ("SYR", "syr"); /* Syriac [macrolanguage] */
345 test_tag_from_language ("SYR", "amw"); /* Western Neo-Aramaic */
346 test_tag_from_language ("SYR", "cld"); /* Chaldean Neo-Aramaic */
347 test_tag_from_language ("SYR", "syc"); /* Classical Syriac */
348
349 test_language_two_way ("TUA", "tru"); /* Turoyo Aramaic */
350
351 test_tag_from_language ("ZHS", "zh"); /* Chinese */
352 test_tag_from_language ("ZHS", "zh-cn"); /* Chinese (China) */
353 test_tag_from_language ("ZHS", "zh-sg"); /* Chinese (Singapore) */
354 test_tag_from_language ("ZHTM", "zh-mo"); /* Chinese (Macao) */
355 test_tag_from_language ("ZHTM", "zh-hant-mo"); /* Chinese (Macao) */
356 test_tag_from_language ("ZHS", "zh-hans-mo"); /* Chinese (Simplified, Macao) */
357 test_language_two_way ("ZHH", "zh-HK"); /* Chinese (Hong Kong) */
358 test_tag_from_language ("ZHH", "zH-HanT-hK"); /* Chinese (Hong Kong) */
359 test_tag_from_language ("ZHS", "zH-HanS-hK"); /* Chinese (Simplified, Hong Kong) */
360 test_tag_from_language ("ZHT", "zh-tw"); /* Chinese (Taiwan) */
361 test_language_two_way ("ZHS", "zh-Hans"); /* Chinese (Simplified) */
362 test_language_two_way ("ZHT", "zh-Hant"); /* Chinese (Traditional) */
363 test_tag_from_language ("ZHS", "zh-xx"); /* Chinese (Other) */
364
365 test_tag_from_language ("ZHS", "zh-Hans-TW");
366
367 test_tag_from_language ("ZHH", "yue");
368 test_tag_from_language ("ZHH", "yue-Hant");
369 test_tag_from_language ("ZHS", "yue-Hans");
370
371 test_language_two_way ("ABC", "abc-x-hbot-41424320");
372 test_language_two_way ("ABCD", "x-hbot-41424344");
373 test_tag_from_language ("ABC", "asdf-asdf-wer-x-hbotabc-zxc");
374 test_tag_from_language ("ABC", "asdf-asdf-wer-x-hbotabc");
375 test_tag_from_language ("ABCD", "asdf-asdf-wer-x-hbotabcd");
376 test_tag_from_language ("ABC", "asdf-asdf-wer-x-hbot-41424320-zxc");
377 test_tag_from_language ("ABC", "asdf-asdf-wer-x-hbot-41424320");
378 test_tag_from_language ("ABCD", "asdf-asdf-wer-x-hbot-41424344");
379
380 test_tag_from_language ("dflt", "asdf-asdf-wer-x-hbot");
381 test_tag_from_language ("dflt", "asdf-asdf-wer-x-hbot-zxc");
382 test_tag_from_language ("dflt", "asdf-asdf-wer-x-hbot-zxc-414243");
383 test_tag_from_language ("dflt", "asdf-asdf-wer-x-hbot-414243");
384 test_tag_from_language ("dflt", "asdf-asdf-wer-x-hbot-4142432");
385
386 test_tag_from_language ("dflt", "xy");
387 test_tag_from_language ("XYZ", "xyz"); /* Unknown ISO 639-3 */
388 test_tag_from_language ("XYZ", "xyz-qw"); /* Unknown ISO 639-3 */
389
390 /*
391 * Invalid input. The precise answer does not matter, as long as it
392 * does not crash or get into an infinite loop.
393 */
394 test_tag_from_language ("IPPH", "-fonipa");
395
396 /*
397 * Tags that contain "-fonipa" as a substring but which do not contain
398 * the subtag "fonipa".
399 */
400 test_tag_from_language ("ENG", "en-fonipax");
401 test_tag_from_language ("ENG", "en-x-fonipa");
402 test_tag_from_language ("ENG", "en-a-fonipa");
403 test_tag_from_language ("ENG", "en-a-qwe-b-fonipa");
404
405 /* International Phonetic Alphabet */
406 test_tag_from_language ("IPPH", "en-fonipa");
407 test_tag_from_language ("IPPH", "en-fonipax-fonipa");
408 test_tag_from_language ("IPPH", "rm-CH-fonipa-sursilv-x-foobar");
409 test_language_two_way ("IPPH", "und-fonipa");
410 test_tag_from_language ("IPPH", "zh-fonipa");
411
412 /* North American Phonetic Alphabet (Americanist Phonetic Notation) */
413 test_tag_from_language ("APPH", "en-fonnapa");
414 test_tag_from_language ("APPH", "chr-fonnapa");
415 test_language_two_way ("APPH", "und-fonnapa");
416
417 /* Khutsuri Georgian */
418 test_tag_from_language ("KGE", "ka-Geok");
419 test_language_two_way ("KGE", "und-Geok");
420
421 /* Irish Traditional */
422 test_language_two_way ("IRT", "ga-Latg");
423
424 /* Moldavian */
425 test_language_two_way ("MOL", "ro-MD");
426
427 /* Polytonic Greek */
428 test_language_two_way ("PGR", "el-polyton");
429 test_tag_from_language ("PGR", "el-CY-polyton");
430
431 /* Estrangela Syriac */
432 test_tag_from_language ("SYRE", "aii-Syre");
433 test_tag_from_language ("SYRE", "de-Syre");
434 test_tag_from_language ("SYRE", "syr-Syre");
435 test_language_two_way ("SYRE", "und-Syre");
436
437 /* Western Syriac */
438 test_tag_from_language ("SYRJ", "aii-Syrj");
439 test_tag_from_language ("SYRJ", "de-Syrj");
440 test_tag_from_language ("SYRJ", "syr-Syrj");
441 test_language_two_way ("SYRJ", "und-Syrj");
442
443 /* Eastern Syriac */
444 test_tag_from_language ("SYRN", "aii-Syrn");
445 test_tag_from_language ("SYRN", "de-Syrn");
446 test_tag_from_language ("SYRN", "syr-Syrn");
447 test_language_two_way ("SYRN", "und-Syrn");
448
449 /* Test that x-hbot overrides the base language */
450 test_tag_from_language ("ABC", "fa-x-hbotabc-hbot-41686121-zxc");
451 test_tag_from_language ("ABC", "fa-ir-x-hbotabc-hbot-41686121-zxc");
452 test_tag_from_language ("ABC", "zh-x-hbotabc-hbot-41686121-zxc");
453 test_tag_from_language ("ABC", "zh-cn-x-hbotabc-hbot-41686121-zxc");
454 test_tag_from_language ("ABC", "zh-xy-x-hbotabc-hbot-41686121-zxc");
455 test_tag_from_language ("ABC", "xyz-xy-x-hbotabc-hbot-41686121-zxc");
456
457 test_tag_from_language ("Aha!", "fa-x-hbot-41686121-hbotabc-zxc");
458 test_tag_from_language ("Aha!", "fa-ir-x-hbot-41686121-hbotabc-zxc");
459 test_tag_from_language ("Aha!", "zh-x-hbot-41686121-hbotabc-zxc");
460 test_tag_from_language ("Aha!", "zh-cn-x-hbot-41686121-hbotabc-zxc");
461 test_tag_from_language ("Aha!", "zh-xy-x-hbot-41686121-hbotabc-zxc");
462 test_tag_from_language ("Aha!", "xyz-xy-x-hbot-41686121-hbotabc-zxc");
463
464 /* Invalid x-hbot */
465 test_tag_from_language ("dflt", "x-hbot");
466 test_tag_from_language ("dflt", "x-hbot-");
467 test_tag_from_language ("dflt", "x-hbot-1");
468 test_tag_from_language ("dflt", "x-hbot-1a");
469 test_tag_from_language ("dflt", "x-hbot-1a2b3c4x");
470 test_tag_from_language ("2lon", "x-hbot-326c6f6e67");
471
472 /* Unnormalized BCP 47 tags */
473 test_tag_from_language ("ARA", "ar-aao");
474 test_tag_from_language ("JBO", "art-lojban");
475 test_tag_from_language ("KOK", "kok-gom");
476 test_tag_from_language ("LTZ", "i-lux");
477 test_tag_from_language ("MNG", "drh");
478 test_tag_from_language ("MOR", "ar-ary");
479 test_tag_from_language ("MOR", "ar-ary-DZ");
480 test_tag_from_language ("NOR", "no-bok");
481 test_tag_from_language ("NYN", "no-nyn");
482 test_tag_from_language ("ZHS", "i-hak");
483 test_tag_from_language ("ZHS", "zh-guoyu");
484 test_tag_from_language ("ZHS", "zh-min");
485 test_tag_from_language ("ZHS", "zh-min-nan");
486 test_tag_from_language ("ZHS", "zh-xiang");
487
488 /* BCP 47 tags that look similar to unrelated language system tags */
489 test_tag_from_language ("SQI", "als");
490 test_tag_from_language ("dflt", "far");
491
492 /* A UN M.49 region code, not an extended language subtag */
493 test_tag_from_language ("ARA", "ar-001");
494
495 /* An invalid tag */
496 test_tag_from_language ("TRK", "tr@foo=bar");
497 }
498
499 static void
test_tags(hb_script_t script,const char * lang_s,unsigned int script_count,unsigned int language_count,unsigned int expected_script_count,unsigned int expected_language_count,...)500 test_tags (hb_script_t script,
501 const char *lang_s,
502 unsigned int script_count,
503 unsigned int language_count,
504 unsigned int expected_script_count,
505 unsigned int expected_language_count,
506 ...)
507 {
508 va_list expected_tags;
509 unsigned int i;
510 hb_tag_t *script_tags = malloc (script_count * sizeof (hb_tag_t));
511 hb_tag_t *language_tags = malloc (language_count * sizeof (hb_tag_t));
512 hb_language_t lang;
513 g_assert (script_tags);
514 g_assert (language_tags);
515 lang = hb_language_from_string (lang_s, -1);
516 va_start (expected_tags, expected_language_count);
517
518 hb_ot_tags_from_script_and_language (script, lang, &script_count, script_tags, &language_count, language_tags);
519
520 g_assert_cmpuint (script_count, ==, expected_script_count);
521 g_assert_cmpuint (language_count, ==, expected_language_count);
522
523 for (i = 0; i < script_count + language_count; i++)
524 {
525 hb_tag_t expected_tag = hb_tag_from_string (va_arg (expected_tags, const char *), -1);
526 hb_tag_t actual_tag = i < script_count ? script_tags[i] : language_tags[i - script_count];
527 g_assert_cmphex (actual_tag, ==, expected_tag);
528 }
529
530 free (script_tags);
531 free (language_tags);
532 va_end (expected_tags);
533 }
534
535 static void
test_ot_tag_full(void)536 test_ot_tag_full (void)
537 {
538 test_tags (HB_SCRIPT_INVALID, "en", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 0, 1, "ENG");
539 test_tags (HB_SCRIPT_INVALID, "en-x-hbscdflt", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 1, 1, "DFLT", "ENG");
540 test_tags (HB_SCRIPT_LATIN, "en", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 1, 1, "latn", "ENG");
541 test_tags (HB_SCRIPT_LATIN, "en", 0, 0, 0, 0);
542 test_tags (HB_SCRIPT_INVALID, "und-fonnapa", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 0, 1, "APPH");
543 test_tags (HB_SCRIPT_INVALID, "en-fonnapa", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 0, 1, "APPH");
544 test_tags (HB_SCRIPT_INVALID, "x-hbot1234-hbsc5678", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 1, 1, "5678", "1234");
545 test_tags (HB_SCRIPT_INVALID, "x-hbsc5678-hbot1234", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 1, 1, "5678", "1234");
546 test_tags (HB_SCRIPT_MALAYALAM, "ml", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 3, 2, "mlm3", "mlm2", "mlym", "MAL", "MLR");
547 test_tags (HB_SCRIPT_MALAYALAM, "ml", 1, 1, 1, 1, "mlm3", "MAL");
548 test_tags (HB_SCRIPT_MYANMAR, "und", HB_OT_MAX_TAGS_PER_SCRIPT, 0, 2, 0, "mym2", "mymr");
549 test_tags (HB_SCRIPT_INVALID, "xyz", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 0, 1, "XYZ");
550 test_tags (HB_SCRIPT_INVALID, "xy", HB_OT_MAX_TAGS_PER_SCRIPT, HB_OT_MAX_TAGS_PER_LANGUAGE, 0, 0);
551 }
552
553 int
main(int argc,char ** argv)554 main (int argc, char **argv)
555 {
556 hb_test_init (&argc, &argv);
557
558 hb_test_add (test_ot_tag_script_degenerate);
559 hb_test_add (test_ot_tag_script_simple);
560 hb_test_add (test_ot_tag_script_from_language);
561 hb_test_add (test_ot_tag_script_indic);
562
563 hb_test_add (test_ot_tags_to_script_and_language);
564
565 hb_test_add (test_ot_tag_language);
566
567 hb_test_add (test_ot_tag_full);
568
569 return hb_test_run();
570 }
571