15821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Copyright (c) 2009 The Chromium Authors. All rights reserved. 25821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Use of this source code is governed by a BSD-style license that can be 35821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// found in the LICENSE file. 45821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 55821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#ifndef LANGUAGES_PUBLIC_LANGUAGES_H_ 65821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#define LANGUAGES_PUBLIC_LANGUAGES_H_ 75821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 85821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This interface defines the Language enum and functions that depend 95821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// only on Language values. 105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// A hash-function for Language, hash<Language>, is defined in 125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// i18n/languages/public/languages-hash.h 135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#ifndef SWIG 155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Language enum defined in languages.proto 165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Also description on how to add languages. 175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#include "languages/proto/languages.pb.h" 185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// We need this for compatibility: 205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - The Language enum in the default namespace. 215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// COMMENTED OUT TO REDUCE DEPENDENCIES ON GOOGLE3 CODE 225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//using namespace i18n::languages; 235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#else 255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// And we must have a swig-compatible enum. 265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This one is a simple cleaned up version of language.proto, making the enum 275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// compatible with C++. 285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#include "i18n/languages/internal/languages_proto_wrapper.h" 295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#endif 315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// COMMENTED OUT TO REDUCE DEPENDENCIES ON GOOGLE3 CODE 335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//#include "util/utf8/proptables/script_enum.h" 345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const int kNumLanguages = NUM_LANGUAGES; 365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return the default language (ENGLISH). 385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)Language default_language(); 395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Language predicates 435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsValidLanguage() 445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IS_LANGUAGE_UNKNOWN() 455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsCJKLanguage() 465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsChineseLanguage() 475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsNorwegianLanguage() 485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsPortugueseLanguage() 495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsRightToLeftLanguage() 505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsMaybeRightToLeftLanguage() 515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsSameLanguage() 525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsScriptRequiringLongerSnippets() 535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsValidLanguage 565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// =============== 575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Function to check if the input is within range of the Language enum. If 595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsValidLanguage(lang) returns true, it is safe to call 605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// static_cast<Language>(lang). 615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)inline bool IsValidLanguage(int lang) { 635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) return ((lang >= 0) && (lang < kNumLanguages)); 645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)} 655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return true if the language is "unknown". (This function was 675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// previously a macro, hence the spelling in all caps.) 685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)inline bool IS_LANGUAGE_UNKNOWN(Language lang) { 705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) return lang == TG_UNKNOWN_LANGUAGE || lang == UNKNOWN_LANGUAGE; 715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)} 725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsCJKLanguage 745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ------------- 755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language is either Chinese 775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// (simplified or traditional), Japanese, or Korean. 785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsCJKLanguage(Language lang); 795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 805821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsChineseLanguage 815821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ----------------- 825821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 835821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language is either Chinese 845821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// (simplified or traditional) 855821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsChineseLanguage(Language lang); 865821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 875821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsNorwegianLanguage 885821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 895821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 905821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language is any of the Norwegian 915821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// (regular or Nynorsk). 925821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsNorwegianLanguage(Language lang); 935821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 945821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsPortugueseLanguage 955821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 965821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 975821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language is any of the Portuguese 985821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// languages (regular, Portugal or Brazil) 995821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsPortugueseLanguage(Language lang); 1005821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1015821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsSameLanguage 1025821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------- 1035821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1045821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// WARNING: This function provides only a simple test on the values of 1055821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// the two Language arguments. It returns false if either language is 1065821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// invalid. It returns true if the language arguments are equal, or 1075821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// if they are both Chinese languages, both Norwegian languages, or 1085821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// both Portuguese languages, as defined by IsChineseLanguage, 1095821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsNorwegianLanguage, and IsPortugueseLanguage. Otherwise it returns 1105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// false. 1115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsSameLanguage(Language lang1, Language lang2); 1125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsRightToLeftLanguage 1155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// --------------------- 1165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language is only written right-to-left 1185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// (E.g., Hebrew, Arabic, Persian etc.) 1195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IMPORTANT NOTE: Technically we're talking about scripts, not languages. 1215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// There are languages that can be written in more than one script. 1225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Examples: 1235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - Kurdish and Azeri ('AZERBAIJANI') can be written left-to-right in 1245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Latin or Cyrillic script, and right-to-left in Arabic script. 1255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - Sindhi and Punjabi are written in different scripts, depending on 1265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// region and dialect. 1275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - Turkmen used an Arabic script historically, but not any more. 1285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - Pashto and Uyghur can use Arabic script, but use a Roman script 1295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// on the Internet. 1305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// - Kashmiri and Urdu are written either with Arabic or Devanagari script. 1315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function only returns true for languages that are always, unequivocally 1335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// written in right-to-left script. 1345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// TODO(benjy): If we want to do anything special with multi-script languages 1365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// we should create new 'languages' for each language+script, as we do for 1375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// traditional vs. simplified Chinese. However most such languages are rare in 1385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// use and even rarer on the web, so this is unlikely to be something we'll 1395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// be concerned with for a while. 1405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsRightToLeftLanguage(Language lang); 1415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsMaybeRightToLeftLanguage 1435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------------- 1445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the language may appear on the web in a 1465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// right-to-left script (E.g., Hebrew, Arabic, Persian, Urdu, Kurdish, etc.) 1475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// NOTE: See important notes under IsRightToLeftLanguage(...). 1495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true for languages that *may* appear on the web in a 1515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// right-to-left script, even if they may also appear in a left-to-right 1525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// script. 1535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function should typically be used in cases where doing some work on 1555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// left-to-right text would be OK (usually a no-op), and this function is used 1565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// just to cut down on unnecessary work on regular, LTR text. 1575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool IsMaybeRightToLeftLanguage(Language lang); 1585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IsScriptRequiringLongerSnippets 1605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 1615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This function returns true if the script chracteristics require longer 1635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// snippet length (Devanagari, Bengali, Gurmukhi, 1645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Gujarati, Oriya, Tamil, Telugu, Kannada, Malayalam). 1655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// COMMENTED OUT TO REDUCE DEPENDENCIES ON GOOGLE3 CODE 1665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// bool IsScriptRequiringLongerSnippets(UnicodeScript script); 1675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 1705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LANGUAGE NAMES 1715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 1725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This interface defines a standard name for each valid Language, 1735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// and a standard name for invalid languages. Some language names use all 1745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// uppercase letters, but others use mixed case. 1755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageName() [Language to name] 1765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageEnumName() [language to enum name] 1775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageFromName() [name to Language] 1785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// default_language_name() 1795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// invalid_language_name() 1805821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 1815821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1825821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Given a Language, returns its standard name. 1835821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return invalid_language_name() if the language is invalid. 1845821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageName(Language lang); 1855821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1865821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Given a Language, return the name of the enum constant for that 1875821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// language. In all but a few cases, this is the same as its standard 1885821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// name. For example, LanguageName(CHINESE) returns "Chinese", but 1895821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageEnumName(CHINESE) returns "CHINESE". This is intended for 1905821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// code that is generating C++ code, where the enum constant is more 1915821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// useful than its integer value. Return "NUM_LANGUAGES" if 1925821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// the language is invalid. 1935821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageEnumName(Language lang); 1945821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1955821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The maximum length of a standard language name. 1965821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const int kMaxLanguageNameSize = 50; 1975821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 1985821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The standard name for the default language. 1995821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* default_language_name(); 2005821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2015821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The standard name for all invalid languages. 2025821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* invalid_language_name(); 2035821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2045821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If lang_name matches the standard name of a Language, using a 2055821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// case-insensitive comparison, set *language to that Language and 2065821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// return true. 2075821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise, set *language to UNKNOWN_LANGUAGE and return false. 2085821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2095821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For backwards compatibility, "HATIAN_CREOLE" is allowed as a name 2105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// for HAITIAN_CREOLE, and "QUECHAU" is allowed as a name for QUECHUA. 2115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For compatibility with LanguageEnumName, "UNKNOWN_LANGUAGE" is allowed 2125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// as a name for UNKNOWN_LANGUAGE (the return value is true in this case, 2135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// as it is for "Unknown"), and "CHINESE_T" is allowed as a name for 2145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// CHINESE_T (i.e., a synonym for "ChineseT"). 2155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// REQUIRES: language must not be NULL. 2175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool LanguageFromName(const char* lang_name, Language *language); 2195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 2235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LANGUAGE CODES 2245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This interface defines a standard code for each valid language, and 2265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// a standard code for invalid languages. These are derived from ISO codes, 2275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// with some Google additions. 2285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode() 2295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// default_language_code() 2305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// invalid_language_code() 2315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeWithDialects() 2325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeISO639_1() 2335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeISO639_2() 2345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ******************************************* 2355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Given a Language, return its standard code. There are Google-specific codes: 2375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For CHINESE_T, return "zh-TW". 2385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For TG_UNKNOWN_LANGUAGE, return "ut". 2395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For UNKNOWN_LANGUAGE, return "un". 2405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For PORTUGUESE_P, return "pt-PT". 2415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For PORTUGUESE_B, return "pt-BR". 2425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For LIMBU, return "sit-NP". 2435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For CHEROKEE, return "chr". 2445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For SYRIAC, return "syr". 2455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise return the ISO 639-1 two-letter language code for lang. 2465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If lang is invalid, return invalid_language_code(). 2475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// NOTE: See the note below about the codes for Chinese languages. 2495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageCode(Language lang); 2515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The maximum length of a language code. 2535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const int kMaxLanguageCodeSize = 50; 2545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The standard code for the default language. 2565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* default_language_code(); 2575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The standard code for all invalid languages. 2595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* invalid_language_code(); 2605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------------------------------- 2635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// NOTE: CHINESE LANGUAGE CODES 2645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// There are three functions that return codes for Chinese languages. 2665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(lang) and LanguageCodeWithDialects(lang) are defined here. 2675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(lang, encoding) is defined in i18n/encodings.lang_enc.h. 2685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The following list shows the different results. 2695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(CHINESE) returns "zh" 2715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(CHINESE_T) returns "zh-TW". 2725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeWithDialects(CHINESE) returns "zh-CN". 2745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeWithDialects(CHINESE_T) returns "zh-TW". 2755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(CHINESE_T, <any encoding>) returns "zh-TW". 2775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(CHINESE, CHINESE_BIG5) returns "zh-TW". 2785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCode(CHINESE, <any other encoding>) returns "zh-CN". 2795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2805821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------------------------------- 2815821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2825821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeWithDialects 2835821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ------------------------ 2845821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2855821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If lang is CHINESE, return "zh-CN". Otherwise return LanguageCode(lang). 2865821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageCodeWithDialects(Language lang); 2875821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2885821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeISO639_1 2895821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 2905821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2915821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return the ISO 639-1 two-letter language code for lang. 2925821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return invalid_language_code() if lang is invalid or does not have 2935821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// an ISO 639-1 two-letter language code. 2945821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageCodeISO639_1(Language lang); 2955821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 2965821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeISO639_2 2975821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 2985821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 2995821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return the ISO 639-2 three-letter language for lang. 3005821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Return invalid_language_code() if lang is invalid or does not have 3015821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// an ISO 639-2 three-letter language code. 3025821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageCodeISO639_2(Language lang); 3035821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3045821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageFromCode 3055821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ---------------- 3065821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3075821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If lang_code matches the code for a Language, using a case-insensitive 3085821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// comparison, set *lang to that Language and return true. 3095821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise, set *lang to UNKNOWN_LANGUAGE and return false. 3105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// lang_code can be an ISO 639-1 (two-letter) code, an ISO 639-2 3125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// (three-letter) code, or a Google-specific code (see LanguageCode). 3135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Certain language-code aliases are also allowed: 3155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "zh-cn" and "zh_cn", set *lang to CHINESE. 3165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "zh-tw" and "zh_tw", set *lang to CHINESE_T. 3175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "he", set *lang to HEBREW. 3185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "in", set *lang to INDONESIAN. 3195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "ji", set *lang to YIDDISH. 3205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For "fil", set *lang to TAGALOG. 3215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// REQUIRES: 'lang' must not be NULL. 3235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool LanguageFromCode(const char* lang_code, Language *language); 3245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageFromCodeOrName 3275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ---------------------- 3285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If lang_code_or_name is a language code or a language name. 3305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// set *language to the corresponding Language and return true. 3315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise set *language to UNKNOWN_LANGUAGE and return false. 3325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool LanguageFromCodeOrName(const char* lang_code_or_name, 3345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) Language* language); 3355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageNameFromCode 3375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// -------------------- 3385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If language_code is the code for a Language (see LanguageFromCode), 3405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// return the standard name of that language (see LanguageName). 3415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise return invalid_language_name(). 3425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)const char* LanguageNameFromCode(const char* language_code); 3445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Miscellany 3475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// LanguageCodeToUnderscoreForm 3495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// ---------------------------- 3505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Given a language code, convert the dash "-" to underscore "_". 3525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Specifically, if result_length <= strlen(lang_code), set result[0] 3545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// to '\0' and return false. Otherwise, copy lang_code to result, 3555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// converting every dash to an underscore, converting every character 3565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// before the first dash or underscore to lower case, and converting 3575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// every character after the first dash or underscore to upper 3585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// case. If there is no dash or underscore, convert the entire string 3595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// to lower case. 3605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// REQUIRES: 'lang_code' must not be NULL. 'result' must not be NULL. 3625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool LanguageCodeToUnderscoreForm(const char* lang_code, 3645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) char* result, 3655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) int result_length); 3665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// AlwaysPutInExpectedRestrict 3695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// --------------------------- 3705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// 3715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// For Web pages in certain top-level domains, Web Search always 3725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// applies a "country restrict". If 'tld' matches one of those, using 3735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// a case-SENSITIVE comparison, set *expected_language to the Language 3745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// most commonly found in that top-level domain and return true. 3755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Otherwise, set *expected_language to UNKNOWN_LANGUAGE and return false. 3765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)bool AlwaysPutInExpectedRestrict(const char *tld, Language *expected_language); 3775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) 3795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#endif // LANGUAGES_PUBLIC_LANGUAGES_H_ 380