SkOTTable_name.cpp 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586
  1. /*
  2. * Copyright 2013 Google Inc.
  3. *
  4. * Use of this source code is governed by a BSD-style license that can be
  5. * found in the LICENSE file.
  6. */
  7. #include "src/sfnt/SkOTTable_name.h"
  8. #include "include/private/SkTemplates.h"
  9. #include "src/core/SkEndian.h"
  10. #include "src/core/SkStringUtils.h"
  11. #include "src/core/SkTSearch.h"
  12. #include "src/core/SkUtils.h"
  13. static SkUnichar next_unichar_UTF16BE(const uint8_t** srcPtr, size_t* length) {
  14. SkASSERT(srcPtr && *srcPtr && length);
  15. SkASSERT(*length > 0);
  16. uint16_t leading;
  17. if (*length < sizeof(leading)) {
  18. *length = 0;
  19. return 0xFFFD;
  20. }
  21. memcpy(&leading, *srcPtr, sizeof(leading));
  22. *srcPtr += sizeof(leading);
  23. *length -= sizeof(leading);
  24. SkUnichar c = SkEndian_SwapBE16(leading);
  25. if (SkUTF16_IsTrailingSurrogate(c)) {
  26. return 0xFFFD;
  27. }
  28. if (SkUTF16_IsLeadingSurrogate(c)) {
  29. uint16_t trailing;
  30. if (*length < sizeof(trailing)) {
  31. *length = 0;
  32. return 0xFFFD;
  33. }
  34. memcpy(&trailing, *srcPtr, sizeof(trailing));
  35. SkUnichar c2 = SkEndian_SwapBE16(trailing);
  36. if (!SkUTF16_IsTrailingSurrogate(c2)) {
  37. return 0xFFFD;
  38. }
  39. *srcPtr += sizeof(trailing);
  40. *length -= sizeof(trailing);
  41. c = (c << 10) + c2 + (0x10000 - (0xD800 << 10) - 0xDC00);
  42. }
  43. return c;
  44. }
  45. static void SkString_from_UTF16BE(const uint8_t* utf16be, size_t length, SkString& utf8) {
  46. // Note that utf16be may not be 2-byte aligned.
  47. SkASSERT(utf16be != nullptr);
  48. utf8.reset();
  49. while (length) {
  50. utf8.appendUnichar(next_unichar_UTF16BE(&utf16be, &length));
  51. }
  52. }
  53. /** UnicodeFromMacRoman[macRomanPoint - 0x80] -> unicodeCodePoint.
  54. * Derived from http://www.unicode.org/Public/MAPPINGS/VENDORS/APPLE/ROMAN.TXT .
  55. * In MacRoman the first 128 code points match ASCII code points.
  56. * This maps the second 128 MacRoman code points to unicode code points.
  57. */
  58. static const uint16_t UnicodeFromMacRoman[0x80] = {
  59. 0x00C4, 0x00C5, 0x00C7, 0x00C9, 0x00D1, 0x00D6, 0x00DC, 0x00E1,
  60. 0x00E0, 0x00E2, 0x00E4, 0x00E3, 0x00E5, 0x00E7, 0x00E9, 0x00E8,
  61. 0x00EA, 0x00EB, 0x00ED, 0x00EC, 0x00EE, 0x00EF, 0x00F1, 0x00F3,
  62. 0x00F2, 0x00F4, 0x00F6, 0x00F5, 0x00FA, 0x00F9, 0x00FB, 0x00FC,
  63. 0x2020, 0x00B0, 0x00A2, 0x00A3, 0x00A7, 0x2022, 0x00B6, 0x00DF,
  64. 0x00AE, 0x00A9, 0x2122, 0x00B4, 0x00A8, 0x2260, 0x00C6, 0x00D8,
  65. 0x221E, 0x00B1, 0x2264, 0x2265, 0x00A5, 0x00B5, 0x2202, 0x2211,
  66. 0x220F, 0x03C0, 0x222B, 0x00AA, 0x00BA, 0x03A9, 0x00E6, 0x00F8,
  67. 0x00BF, 0x00A1, 0x00AC, 0x221A, 0x0192, 0x2248, 0x2206, 0x00AB,
  68. 0x00BB, 0x2026, 0x00A0, 0x00C0, 0x00C3, 0x00D5, 0x0152, 0x0153,
  69. 0x2013, 0x2014, 0x201C, 0x201D, 0x2018, 0x2019, 0x00F7, 0x25CA,
  70. 0x00FF, 0x0178, 0x2044, 0x20AC, 0x2039, 0x203A, 0xFB01, 0xFB02,
  71. 0x2021, 0x00B7, 0x201A, 0x201E, 0x2030, 0x00C2, 0x00CA, 0x00C1,
  72. 0x00CB, 0x00C8, 0x00CD, 0x00CE, 0x00CF, 0x00CC, 0x00D3, 0x00D4,
  73. 0xF8FF, 0x00D2, 0x00DA, 0x00DB, 0x00D9, 0x0131, 0x02C6, 0x02DC,
  74. 0x00AF, 0x02D8, 0x02D9, 0x02DA, 0x00B8, 0x02DD, 0x02DB, 0x02C7,
  75. };
  76. static void SkStringFromMacRoman(const uint8_t* macRoman, size_t length, SkString& utf8) {
  77. utf8.reset();
  78. for (size_t i = 0; i < length; ++i) {
  79. utf8.appendUnichar(macRoman[i] < 0x80 ? macRoman[i]
  80. : UnicodeFromMacRoman[macRoman[i] - 0x80]);
  81. }
  82. }
  83. static const struct BCP47FromLanguageId {
  84. uint16_t languageID;
  85. const char* bcp47;
  86. }
  87. /** The Mac and Windows values do not conflict, so this is currently one single table. */
  88. BCP47FromLanguageID[] = {
  89. /** A mapping from Mac Language Designators to BCP 47 codes.
  90. * The following list was constructed more or less manually.
  91. * Apple now uses BCP 47 (post OSX10.4), so there will be no new entries.
  92. */
  93. {0, "en"}, //English
  94. {1, "fr"}, //French
  95. {2, "de"}, //German
  96. {3, "it"}, //Italian
  97. {4, "nl"}, //Dutch
  98. {5, "sv"}, //Swedish
  99. {6, "es"}, //Spanish
  100. {7, "da"}, //Danish
  101. {8, "pt"}, //Portuguese
  102. {9, "nb"}, //Norwegian
  103. {10, "he"}, //Hebrew
  104. {11, "ja"}, //Japanese
  105. {12, "ar"}, //Arabic
  106. {13, "fi"}, //Finnish
  107. {14, "el"}, //Greek
  108. {15, "is"}, //Icelandic
  109. {16, "mt"}, //Maltese
  110. {17, "tr"}, //Turkish
  111. {18, "hr"}, //Croatian
  112. {19, "zh-Hant"}, //Chinese (Traditional)
  113. {20, "ur"}, //Urdu
  114. {21, "hi"}, //Hindi
  115. {22, "th"}, //Thai
  116. {23, "ko"}, //Korean
  117. {24, "lt"}, //Lithuanian
  118. {25, "pl"}, //Polish
  119. {26, "hu"}, //Hungarian
  120. {27, "et"}, //Estonian
  121. {28, "lv"}, //Latvian
  122. {29, "se"}, //Sami
  123. {30, "fo"}, //Faroese
  124. {31, "fa"}, //Farsi (Persian)
  125. {32, "ru"}, //Russian
  126. {33, "zh-Hans"}, //Chinese (Simplified)
  127. {34, "nl"}, //Dutch
  128. {35, "ga"}, //Irish(Gaelic)
  129. {36, "sq"}, //Albanian
  130. {37, "ro"}, //Romanian
  131. {38, "cs"}, //Czech
  132. {39, "sk"}, //Slovak
  133. {40, "sl"}, //Slovenian
  134. {41, "yi"}, //Yiddish
  135. {42, "sr"}, //Serbian
  136. {43, "mk"}, //Macedonian
  137. {44, "bg"}, //Bulgarian
  138. {45, "uk"}, //Ukrainian
  139. {46, "be"}, //Byelorussian
  140. {47, "uz"}, //Uzbek
  141. {48, "kk"}, //Kazakh
  142. {49, "az-Cyrl"}, //Azerbaijani (Cyrillic)
  143. {50, "az-Arab"}, //Azerbaijani (Arabic)
  144. {51, "hy"}, //Armenian
  145. {52, "ka"}, //Georgian
  146. {53, "mo"}, //Moldavian
  147. {54, "ky"}, //Kirghiz
  148. {55, "tg"}, //Tajiki
  149. {56, "tk"}, //Turkmen
  150. {57, "mn-Mong"}, //Mongolian (Traditional)
  151. {58, "mn-Cyrl"}, //Mongolian (Cyrillic)
  152. {59, "ps"}, //Pashto
  153. {60, "ku"}, //Kurdish
  154. {61, "ks"}, //Kashmiri
  155. {62, "sd"}, //Sindhi
  156. {63, "bo"}, //Tibetan
  157. {64, "ne"}, //Nepali
  158. {65, "sa"}, //Sanskrit
  159. {66, "mr"}, //Marathi
  160. {67, "bn"}, //Bengali
  161. {68, "as"}, //Assamese
  162. {69, "gu"}, //Gujarati
  163. {70, "pa"}, //Punjabi
  164. {71, "or"}, //Oriya
  165. {72, "ml"}, //Malayalam
  166. {73, "kn"}, //Kannada
  167. {74, "ta"}, //Tamil
  168. {75, "te"}, //Telugu
  169. {76, "si"}, //Sinhalese
  170. {77, "my"}, //Burmese
  171. {78, "km"}, //Khmer
  172. {79, "lo"}, //Lao
  173. {80, "vi"}, //Vietnamese
  174. {81, "id"}, //Indonesian
  175. {82, "tl"}, //Tagalog
  176. {83, "ms-Latn"}, //Malay (Roman)
  177. {84, "ms-Arab"}, //Malay (Arabic)
  178. {85, "am"}, //Amharic
  179. {86, "ti"}, //Tigrinya
  180. {87, "om"}, //Oromo
  181. {88, "so"}, //Somali
  182. {89, "sw"}, //Swahili
  183. {90, "rw"}, //Kinyarwanda/Ruanda
  184. {91, "rn"}, //Rundi
  185. {92, "ny"}, //Nyanja/Chewa
  186. {93, "mg"}, //Malagasy
  187. {94, "eo"}, //Esperanto
  188. {128, "cy"}, //Welsh
  189. {129, "eu"}, //Basque
  190. {130, "ca"}, //Catalan
  191. {131, "la"}, //Latin
  192. {132, "qu"}, //Quechua
  193. {133, "gn"}, //Guarani
  194. {134, "ay"}, //Aymara
  195. {135, "tt"}, //Tatar
  196. {136, "ug"}, //Uighur
  197. {137, "dz"}, //Dzongkha
  198. {138, "jv-Latn"}, //Javanese (Roman)
  199. {139, "su-Latn"}, //Sundanese (Roman)
  200. {140, "gl"}, //Galician
  201. {141, "af"}, //Afrikaans
  202. {142, "br"}, //Breton
  203. {143, "iu"}, //Inuktitut
  204. {144, "gd"}, //Scottish (Gaelic)
  205. {145, "gv"}, //Manx (Gaelic)
  206. {146, "ga"}, //Irish (Gaelic with Lenition)
  207. {147, "to"}, //Tongan
  208. {148, "el"}, //Greek (Polytonic) Note: ISO 15924 does not have an equivalent script name.
  209. {149, "kl"}, //Greenlandic
  210. {150, "az-Latn"}, //Azerbaijani (Roman)
  211. {151, "nn"}, //Nynorsk
  212. /** A mapping from Windows LCID to BCP 47 codes.
  213. * This list is the sorted, curated output of tools/win_lcid.cpp.
  214. * Note that these are sorted by value for quick binary lookup, and not logically by lsb.
  215. * The 'bare' language ids (e.g. 0x0001 for Arabic) are ommitted
  216. * as they do not appear as valid language ids in the OpenType specification.
  217. */
  218. { 0x0401, "ar-SA" }, //Arabic
  219. { 0x0402, "bg-BG" }, //Bulgarian
  220. { 0x0403, "ca-ES" }, //Catalan
  221. { 0x0404, "zh-TW" }, //Chinese (Traditional)
  222. { 0x0405, "cs-CZ" }, //Czech
  223. { 0x0406, "da-DK" }, //Danish
  224. { 0x0407, "de-DE" }, //German
  225. { 0x0408, "el-GR" }, //Greek
  226. { 0x0409, "en-US" }, //English
  227. { 0x040a, "es-ES_tradnl" }, //Spanish
  228. { 0x040b, "fi-FI" }, //Finnish
  229. { 0x040c, "fr-FR" }, //French
  230. { 0x040d, "he-IL" }, //Hebrew
  231. { 0x040d, "he" }, //Hebrew
  232. { 0x040e, "hu-HU" }, //Hungarian
  233. { 0x040e, "hu" }, //Hungarian
  234. { 0x040f, "is-IS" }, //Icelandic
  235. { 0x0410, "it-IT" }, //Italian
  236. { 0x0411, "ja-JP" }, //Japanese
  237. { 0x0412, "ko-KR" }, //Korean
  238. { 0x0413, "nl-NL" }, //Dutch
  239. { 0x0414, "nb-NO" }, //Norwegian (Bokmål)
  240. { 0x0415, "pl-PL" }, //Polish
  241. { 0x0416, "pt-BR" }, //Portuguese
  242. { 0x0417, "rm-CH" }, //Romansh
  243. { 0x0418, "ro-RO" }, //Romanian
  244. { 0x0419, "ru-RU" }, //Russian
  245. { 0x041a, "hr-HR" }, //Croatian
  246. { 0x041b, "sk-SK" }, //Slovak
  247. { 0x041c, "sq-AL" }, //Albanian
  248. { 0x041d, "sv-SE" }, //Swedish
  249. { 0x041e, "th-TH" }, //Thai
  250. { 0x041f, "tr-TR" }, //Turkish
  251. { 0x0420, "ur-PK" }, //Urdu
  252. { 0x0421, "id-ID" }, //Indonesian
  253. { 0x0422, "uk-UA" }, //Ukrainian
  254. { 0x0423, "be-BY" }, //Belarusian
  255. { 0x0424, "sl-SI" }, //Slovenian
  256. { 0x0425, "et-EE" }, //Estonian
  257. { 0x0426, "lv-LV" }, //Latvian
  258. { 0x0427, "lt-LT" }, //Lithuanian
  259. { 0x0428, "tg-Cyrl-TJ" }, //Tajik (Cyrillic)
  260. { 0x0429, "fa-IR" }, //Persian
  261. { 0x042a, "vi-VN" }, //Vietnamese
  262. { 0x042b, "hy-AM" }, //Armenian
  263. { 0x042c, "az-Latn-AZ" }, //Azeri (Latin)
  264. { 0x042d, "eu-ES" }, //Basque
  265. { 0x042e, "hsb-DE" }, //Upper Sorbian
  266. { 0x042f, "mk-MK" }, //Macedonian (FYROM)
  267. { 0x0432, "tn-ZA" }, //Setswana
  268. { 0x0434, "xh-ZA" }, //isiXhosa
  269. { 0x0435, "zu-ZA" }, //isiZulu
  270. { 0x0436, "af-ZA" }, //Afrikaans
  271. { 0x0437, "ka-GE" }, //Georgian
  272. { 0x0438, "fo-FO" }, //Faroese
  273. { 0x0439, "hi-IN" }, //Hindi
  274. { 0x043a, "mt-MT" }, //Maltese
  275. { 0x043b, "se-NO" }, //Sami (Northern)
  276. { 0x043e, "ms-MY" }, //Malay
  277. { 0x043f, "kk-KZ" }, //Kazakh
  278. { 0x0440, "ky-KG" }, //Kyrgyz
  279. { 0x0441, "sw-KE" }, //Kiswahili
  280. { 0x0442, "tk-TM" }, //Turkmen
  281. { 0x0443, "uz-Latn-UZ" }, //Uzbek (Latin)
  282. { 0x0443, "uz" }, //Uzbek
  283. { 0x0444, "tt-RU" }, //Tatar
  284. { 0x0445, "bn-IN" }, //Bengali
  285. { 0x0446, "pa-IN" }, //Punjabi
  286. { 0x0447, "gu-IN" }, //Gujarati
  287. { 0x0448, "or-IN" }, //Oriya
  288. { 0x0449, "ta-IN" }, //Tamil
  289. { 0x044a, "te-IN" }, //Telugu
  290. { 0x044b, "kn-IN" }, //Kannada
  291. { 0x044c, "ml-IN" }, //Malayalam
  292. { 0x044d, "as-IN" }, //Assamese
  293. { 0x044e, "mr-IN" }, //Marathi
  294. { 0x044f, "sa-IN" }, //Sanskrit
  295. { 0x0450, "mn-Cyrl" }, //Mongolian (Cyrillic)
  296. { 0x0451, "bo-CN" }, //Tibetan
  297. { 0x0452, "cy-GB" }, //Welsh
  298. { 0x0453, "km-KH" }, //Khmer
  299. { 0x0454, "lo-LA" }, //Lao
  300. { 0x0456, "gl-ES" }, //Galician
  301. { 0x0457, "kok-IN" }, //Konkani
  302. { 0x045a, "syr-SY" }, //Syriac
  303. { 0x045b, "si-LK" }, //Sinhala
  304. { 0x045d, "iu-Cans-CA" }, //Inuktitut (Syllabics)
  305. { 0x045e, "am-ET" }, //Amharic
  306. { 0x0461, "ne-NP" }, //Nepali
  307. { 0x0462, "fy-NL" }, //Frisian
  308. { 0x0463, "ps-AF" }, //Pashto
  309. { 0x0464, "fil-PH" }, //Filipino
  310. { 0x0465, "dv-MV" }, //Divehi
  311. { 0x0468, "ha-Latn-NG" }, //Hausa (Latin)
  312. { 0x046a, "yo-NG" }, //Yoruba
  313. { 0x046b, "quz-BO" }, //Quechua
  314. { 0x046c, "nso-ZA" }, //Sesotho sa Leboa
  315. { 0x046d, "ba-RU" }, //Bashkir
  316. { 0x046e, "lb-LU" }, //Luxembourgish
  317. { 0x046f, "kl-GL" }, //Greenlandic
  318. { 0x0470, "ig-NG" }, //Igbo
  319. { 0x0478, "ii-CN" }, //Yi
  320. { 0x047a, "arn-CL" }, //Mapudungun
  321. { 0x047c, "moh-CA" }, //Mohawk
  322. { 0x047e, "br-FR" }, //Breton
  323. { 0x0480, "ug-CN" }, //Uyghur
  324. { 0x0481, "mi-NZ" }, //Maori
  325. { 0x0482, "oc-FR" }, //Occitan
  326. { 0x0483, "co-FR" }, //Corsican
  327. { 0x0484, "gsw-FR" }, //Alsatian
  328. { 0x0485, "sah-RU" }, //Yakut
  329. { 0x0486, "qut-GT" }, //K'iche
  330. { 0x0487, "rw-RW" }, //Kinyarwanda
  331. { 0x0488, "wo-SN" }, //Wolof
  332. { 0x048c, "prs-AF" }, //Dari
  333. { 0x0491, "gd-GB" }, //Scottish Gaelic
  334. { 0x0801, "ar-IQ" }, //Arabic
  335. { 0x0804, "zh-Hans" }, //Chinese (Simplified)
  336. { 0x0807, "de-CH" }, //German
  337. { 0x0809, "en-GB" }, //English
  338. { 0x080a, "es-MX" }, //Spanish
  339. { 0x080c, "fr-BE" }, //French
  340. { 0x0810, "it-CH" }, //Italian
  341. { 0x0813, "nl-BE" }, //Dutch
  342. { 0x0814, "nn-NO" }, //Norwegian (Nynorsk)
  343. { 0x0816, "pt-PT" }, //Portuguese
  344. { 0x081a, "sr-Latn-CS" }, //Serbian (Latin)
  345. { 0x081d, "sv-FI" }, //Swedish
  346. { 0x082c, "az-Cyrl-AZ" }, //Azeri (Cyrillic)
  347. { 0x082e, "dsb-DE" }, //Lower Sorbian
  348. { 0x082e, "dsb" }, //Lower Sorbian
  349. { 0x083b, "se-SE" }, //Sami (Northern)
  350. { 0x083c, "ga-IE" }, //Irish
  351. { 0x083e, "ms-BN" }, //Malay
  352. { 0x0843, "uz-Cyrl-UZ" }, //Uzbek (Cyrillic)
  353. { 0x0845, "bn-BD" }, //Bengali
  354. { 0x0850, "mn-Mong-CN" }, //Mongolian (Traditional Mongolian)
  355. { 0x085d, "iu-Latn-CA" }, //Inuktitut (Latin)
  356. { 0x085f, "tzm-Latn-DZ" }, //Tamazight (Latin)
  357. { 0x086b, "quz-EC" }, //Quechua
  358. { 0x0c01, "ar-EG" }, //Arabic
  359. { 0x0c04, "zh-Hant" }, //Chinese (Traditional)
  360. { 0x0c07, "de-AT" }, //German
  361. { 0x0c09, "en-AU" }, //English
  362. { 0x0c0a, "es-ES" }, //Spanish
  363. { 0x0c0c, "fr-CA" }, //French
  364. { 0x0c1a, "sr-Cyrl-CS" }, //Serbian (Cyrillic)
  365. { 0x0c3b, "se-FI" }, //Sami (Northern)
  366. { 0x0c6b, "quz-PE" }, //Quechua
  367. { 0x1001, "ar-LY" }, //Arabic
  368. { 0x1004, "zh-SG" }, //Chinese (Simplified)
  369. { 0x1007, "de-LU" }, //German
  370. { 0x1009, "en-CA" }, //English
  371. { 0x100a, "es-GT" }, //Spanish
  372. { 0x100c, "fr-CH" }, //French
  373. { 0x101a, "hr-BA" }, //Croatian (Latin)
  374. { 0x103b, "smj-NO" }, //Sami (Lule)
  375. { 0x1401, "ar-DZ" }, //Arabic
  376. { 0x1404, "zh-MO" }, //Chinese (Traditional)
  377. { 0x1407, "de-LI" }, //German
  378. { 0x1409, "en-NZ" }, //English
  379. { 0x140a, "es-CR" }, //Spanish
  380. { 0x140c, "fr-LU" }, //French
  381. { 0x141a, "bs-Latn-BA" }, //Bosnian (Latin)
  382. { 0x141a, "bs" }, //Bosnian
  383. { 0x143b, "smj-SE" }, //Sami (Lule)
  384. { 0x143b, "smj" }, //Sami (Lule)
  385. { 0x1801, "ar-MA" }, //Arabic
  386. { 0x1809, "en-IE" }, //English
  387. { 0x180a, "es-PA" }, //Spanish
  388. { 0x180c, "fr-MC" }, //French
  389. { 0x181a, "sr-Latn-BA" }, //Serbian (Latin)
  390. { 0x183b, "sma-NO" }, //Sami (Southern)
  391. { 0x1c01, "ar-TN" }, //Arabic
  392. { 0x1c09, "en-ZA" }, //English
  393. { 0x1c0a, "es-DO" }, //Spanish
  394. { 0x1c1a, "sr-Cyrl-BA" }, //Serbian (Cyrillic)
  395. { 0x1c3b, "sma-SE" }, //Sami (Southern)
  396. { 0x1c3b, "sma" }, //Sami (Southern)
  397. { 0x2001, "ar-OM" }, //Arabic
  398. { 0x2009, "en-JM" }, //English
  399. { 0x200a, "es-VE" }, //Spanish
  400. { 0x201a, "bs-Cyrl-BA" }, //Bosnian (Cyrillic)
  401. { 0x201a, "bs-Cyrl" }, //Bosnian (Cyrillic)
  402. { 0x203b, "sms-FI" }, //Sami (Skolt)
  403. { 0x203b, "sms" }, //Sami (Skolt)
  404. { 0x2401, "ar-YE" }, //Arabic
  405. { 0x2409, "en-029" }, //English
  406. { 0x240a, "es-CO" }, //Spanish
  407. { 0x241a, "sr-Latn-RS" }, //Serbian (Latin)
  408. { 0x243b, "smn-FI" }, //Sami (Inari)
  409. { 0x2801, "ar-SY" }, //Arabic
  410. { 0x2809, "en-BZ" }, //English
  411. { 0x280a, "es-PE" }, //Spanish
  412. { 0x281a, "sr-Cyrl-RS" }, //Serbian (Cyrillic)
  413. { 0x2c01, "ar-JO" }, //Arabic
  414. { 0x2c09, "en-TT" }, //English
  415. { 0x2c0a, "es-AR" }, //Spanish
  416. { 0x2c1a, "sr-Latn-ME" }, //Serbian (Latin)
  417. { 0x3001, "ar-LB" }, //Arabic
  418. { 0x3009, "en-ZW" }, //English
  419. { 0x300a, "es-EC" }, //Spanish
  420. { 0x301a, "sr-Cyrl-ME" }, //Serbian (Cyrillic)
  421. { 0x3401, "ar-KW" }, //Arabic
  422. { 0x3409, "en-PH" }, //English
  423. { 0x340a, "es-CL" }, //Spanish
  424. { 0x3801, "ar-AE" }, //Arabic
  425. { 0x380a, "es-UY" }, //Spanish
  426. { 0x3c01, "ar-BH" }, //Arabic
  427. { 0x3c0a, "es-PY" }, //Spanish
  428. { 0x4001, "ar-QA" }, //Arabic
  429. { 0x4009, "en-IN" }, //English
  430. { 0x400a, "es-BO" }, //Spanish
  431. { 0x4409, "en-MY" }, //English
  432. { 0x440a, "es-SV" }, //Spanish
  433. { 0x4809, "en-SG" }, //English
  434. { 0x480a, "es-HN" }, //Spanish
  435. { 0x4c0a, "es-NI" }, //Spanish
  436. { 0x500a, "es-PR" }, //Spanish
  437. { 0x540a, "es-US" }, //Spanish
  438. };
  439. namespace {
  440. bool BCP47FromLanguageIdLess(const BCP47FromLanguageId& a, const BCP47FromLanguageId& b) {
  441. return a.languageID < b.languageID;
  442. }
  443. }
  444. bool SkOTTableName::Iterator::next(SkOTTableName::Iterator::Record& record) {
  445. SkOTTableName nameTable;
  446. if (fNameTableSize < sizeof(nameTable)) {
  447. return false;
  448. }
  449. memcpy(&nameTable, fNameTable, sizeof(nameTable));
  450. const uint8_t* nameRecords = fNameTable + sizeof(nameTable);
  451. const size_t nameRecordsSize = fNameTableSize - sizeof(nameTable);
  452. const size_t stringTableOffset = SkEndian_SwapBE16(nameTable.stringOffset);
  453. if (fNameTableSize < stringTableOffset) {
  454. return false;
  455. }
  456. const uint8_t* stringTable = fNameTable + stringTableOffset;
  457. const size_t stringTableSize = fNameTableSize - stringTableOffset;
  458. // Find the next record which matches the requested type.
  459. SkOTTableName::Record nameRecord;
  460. const size_t nameRecordsCount = SkEndian_SwapBE16(nameTable.count);
  461. const size_t nameRecordsMax = SkTMin(nameRecordsCount, nameRecordsSize / sizeof(nameRecord));
  462. do {
  463. if (fIndex >= nameRecordsMax) {
  464. return false;
  465. }
  466. memcpy(&nameRecord, nameRecords + sizeof(nameRecord)*fIndex, sizeof(nameRecord));
  467. ++fIndex;
  468. } while (fType != -1 && nameRecord.nameID.fontSpecific != fType);
  469. record.type = nameRecord.nameID.fontSpecific;
  470. // Decode the name into UTF-8.
  471. const size_t nameOffset = SkEndian_SwapBE16(nameRecord.offset);
  472. const size_t nameLength = SkEndian_SwapBE16(nameRecord.length);
  473. if (stringTableSize < nameOffset + nameLength) {
  474. return false; // continue?
  475. }
  476. const uint8_t* nameString = stringTable + nameOffset;
  477. switch (nameRecord.platformID.value) {
  478. case SkOTTableName::Record::PlatformID::Windows:
  479. if (SkOTTableName::Record::EncodingID::Windows::UnicodeBMPUCS2
  480. != nameRecord.encodingID.windows.value
  481. && SkOTTableName::Record::EncodingID::Windows::UnicodeUCS4
  482. != nameRecord.encodingID.windows.value
  483. && SkOTTableName::Record::EncodingID::Windows::Symbol
  484. != nameRecord.encodingID.windows.value)
  485. {
  486. record.name.reset();
  487. break; // continue?
  488. }
  489. case SkOTTableName::Record::PlatformID::Unicode:
  490. case SkOTTableName::Record::PlatformID::ISO:
  491. SkString_from_UTF16BE(nameString, nameLength, record.name);
  492. break;
  493. case SkOTTableName::Record::PlatformID::Macintosh:
  494. // TODO: need better decoding, especially on Mac.
  495. if (SkOTTableName::Record::EncodingID::Macintosh::Roman
  496. != nameRecord.encodingID.macintosh.value)
  497. {
  498. record.name.reset();
  499. break; // continue?
  500. }
  501. SkStringFromMacRoman(nameString, nameLength, record.name);
  502. break;
  503. case SkOTTableName::Record::PlatformID::Custom:
  504. // These should never appear in a 'name' table.
  505. default:
  506. SkASSERT(false);
  507. record.name.reset();
  508. break; // continue?
  509. }
  510. // Determine the language.
  511. const uint16_t languageID = SkEndian_SwapBE16(nameRecord.languageID.languageTagID);
  512. // Handle format 1 languages.
  513. if (SkOTTableName::format_1 == nameTable.format && languageID >= 0x8000) {
  514. const uint16_t languageTagRecordIndex = languageID - 0x8000;
  515. if (nameRecordsSize < sizeof(nameRecord)*nameRecordsCount) {
  516. return false; //"und" or break?
  517. }
  518. const uint8_t* format1extData = nameRecords + sizeof(nameRecord)*nameRecordsCount;
  519. size_t format1extSize = nameRecordsSize - sizeof(nameRecord)*nameRecordsCount;
  520. SkOTTableName::Format1Ext format1ext;
  521. if (format1extSize < sizeof(format1ext)) {
  522. return false; // "und" or break?
  523. }
  524. memcpy(&format1ext, format1extData, sizeof(format1ext));
  525. const uint8_t* languageTagRecords = format1extData + sizeof(format1ext);
  526. size_t languageTagRecordsSize = format1extSize - sizeof(format1ext);
  527. if (languageTagRecordIndex < SkEndian_SwapBE16(format1ext.langTagCount)) {
  528. SkOTTableName::Format1Ext::LangTagRecord languageTagRecord;
  529. if (languageTagRecordsSize < sizeof(languageTagRecord)*(languageTagRecordIndex+1)) {
  530. return false; // "und"?
  531. }
  532. const uint8_t* languageTagData = languageTagRecords
  533. + sizeof(languageTagRecord)*languageTagRecordIndex;
  534. memcpy(&languageTagRecord, languageTagData, sizeof(languageTagRecord));
  535. uint16_t languageOffset = SkEndian_SwapBE16(languageTagRecord.offset);
  536. uint16_t languageLength = SkEndian_SwapBE16(languageTagRecord.length);
  537. if (fNameTableSize < stringTableOffset + languageOffset + languageLength) {
  538. return false; // "und"?
  539. }
  540. const uint8_t* languageString = stringTable + languageOffset;
  541. SkString_from_UTF16BE(languageString, languageLength, record.language);
  542. return true;
  543. }
  544. }
  545. // Handle format 0 languages, translating them into BCP 47.
  546. const BCP47FromLanguageId target = { languageID, "" };
  547. int languageIndex = SkTSearch<BCP47FromLanguageId, BCP47FromLanguageIdLess>(
  548. BCP47FromLanguageID, SK_ARRAY_COUNT(BCP47FromLanguageID), target, sizeof(target));
  549. if (languageIndex >= 0) {
  550. record.language = BCP47FromLanguageID[languageIndex].bcp47;
  551. return true;
  552. }
  553. // Unknown language, return the BCP 47 code 'und' for 'undetermined'.
  554. record.language = "und";
  555. return true;
  556. }