chrome/common/spellcheck_common.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "chrome/common/spellcheck_common.h"
   6
   7 #include "base/files/file_path.h"
   8 #include "base/logging.h"
   9 #include "third_party/icu/source/common/unicode/uloc.h"
  10
  11 namespace chrome {
  12 namespace spellcheck_common {
  13
  14 struct LanguageRegion {
  15   const char* language;  // The language.
  16   const char* language_region;  // language & region, used by dictionaries.
  17 };
  18
  19 struct LanguageVersion {
  20   const char* language;  // The language input.
  21   const char* version;   // The corresponding version.
  22 };
  23
  24 static const LanguageRegion g_supported_spellchecker_languages[] = {
  25   // Several languages are not to be included in the spellchecker list:
  26   // th-TH, vi-VI.
  27   {"af", "af-ZA"},
  28   {"bg", "bg-BG"},
  29   {"ca", "ca-ES"},
  30   {"cs", "cs-CZ"},
  31   {"da", "da-DK"},
  32   {"de", "de-DE"},
  33   {"el", "el-GR"},
  34   {"en-AU", "en-AU"},
  35   {"en-CA", "en-CA"},
  36   {"en-GB", "en-GB"},
  37   {"en-US", "en-US"},
  38   {"es", "es-ES"},
  39   {"et", "et-EE"},
  40   {"fo", "fo-FO"},
  41   {"fr", "fr-FR"},
  42   {"he", "he-IL"},
  43   {"hi", "hi-IN"},
  44   {"hr", "hr-HR"},
  45   {"hu", "hu-HU"},
  46   {"id", "id-ID"},
  47   {"it", "it-IT"},
  48   {"ko", "ko"},
  49   {"lt", "lt-LT"},
  50   {"lv", "lv-LV"},
  51   {"nb", "nb-NO"},
  52   {"nl", "nl-NL"},
  53   {"pl", "pl-PL"},
  54   {"pt-BR", "pt-BR"},
  55   {"pt-PT", "pt-PT"},
  56   {"ro", "ro-RO"},
  57   {"ru", "ru-RU"},
  58   {"sh", "sh"},
  59   {"sk", "sk-SK"},
  60   {"sl", "sl-SI"},
  61   {"sq", "sq"},
  62   {"sr", "sr"},
  63   {"sv", "sv-SE"},
  64   {"ta", "ta-IN"},
  65   {"tg", "tg-TG"},
  66   {"tr", "tr-TR"},
  67   {"uk", "uk-UA"},
  68   {"vi", "vi-VN"},
  69 };
  70
  71 bool IsValidRegion(const std::string& region) {
  72   for (size_t i = 0; i < arraysize(g_supported_spellchecker_languages);
  73        ++i) {
  74     if (g_supported_spellchecker_languages[i].language_region == region)
  75       return true;
  76   }
  77   return false;
  78 }
  79
  80 // This function returns the language-region version of language name.
  81 // e.g. returns hi-IN for hi.
  82 std::string GetSpellCheckLanguageRegion(const std::string& input_language) {
  83   for (size_t i = 0; i < arraysize(g_supported_spellchecker_languages);
  84        ++i) {
  85     if (g_supported_spellchecker_languages[i].language == input_language) {
  86       return std::string(
  87           g_supported_spellchecker_languages[i].language_region);
  88     }
  89   }
  90
  91   return input_language;
  92 }
  93
  94 base::FilePath GetVersionedFileName(const std::string& input_language,
  95                                     const base::FilePath& dict_dir) {
  96   // The default dictionary version is 3-0. This version indicates that the bdic
  97   // file contains a checksum.
  98   static const char kDefaultVersionString[] = "-3-0";
  99
 100   // Add non-default version strings here. Use the same version for all the
 101   // dictionaries that you add at the same time. Increment the major version
 102   // number if you're updating either dic or aff files. Increment the minor
 103   // version number if you're updating only dic_delta files.
 104   static LanguageVersion special_version_string[] = {
 105     {"tr-TR", "-4-0"},  // Jan 9, 2013: Add "FLAG num" to aff to avoid heapcheck
 106                         // crash.
 107     {"tg-TG", "-5-0"},  // Mar 4, 2014: Add Tajik dictionary.
 108
 109     // Oct 28, 2014: Update from upstream, add new words.
 110     {"en-AU", "-4-0"},
 111     {"en-GB", "-4-0"},
 112
 113     // March 10, 2015: Update from upstream, enable typographical apostrophe.
 114     {"en-CA", "-6-1"},
 115     {"en-US", "-6-1"},
 116   };
 117
 118   // Generate the bdict file name using default version string or special
 119   // version string, depending on the language.
 120   std::string language = GetSpellCheckLanguageRegion(input_language);
 121   std::string versioned_bdict_file_name(language + kDefaultVersionString +
 122                                         ".bdic");
 123   for (size_t i = 0; i < arraysize(special_version_string); ++i) {
 124     if (language == special_version_string[i].language) {
 125       versioned_bdict_file_name =
 126           language + special_version_string[i].version + ".bdic";
 127       break;
 128     }
 129   }
 130
 131   return dict_dir.AppendASCII(versioned_bdict_file_name);
 132 }
 133
 134 std::string GetCorrespondingSpellCheckLanguage(const std::string& language) {
 135   // Look for exact match in the Spell Check language list.
 136   for (size_t i = 0; i < arraysize(g_supported_spellchecker_languages);
 137        ++i) {
 138     // First look for exact match in the language region of the list.
 139     std::string spellcheck_language(
 140         g_supported_spellchecker_languages[i].language);
 141     if (spellcheck_language == language)
 142       return language;
 143
 144     // Next, look for exact match in the language_region part of the list.
 145     std::string spellcheck_language_region(
 146         g_supported_spellchecker_languages[i].language_region);
 147     if (spellcheck_language_region == language)
 148       return g_supported_spellchecker_languages[i].language;
 149   }
 150
 151   // No match found - return blank.
 152   return std::string();
 153 }
 154
 155 void SpellCheckLanguages(std::vector<std::string>* languages) {
 156   for (size_t i = 0; i < arraysize(g_supported_spellchecker_languages);
 157        ++i) {
 158     languages->push_back(g_supported_spellchecker_languages[i].language);
 159   }
 160 }
 161
 162 void GetISOLanguageCountryCodeFromLocale(const std::string& locale,
 163                                          std::string* language_code,
 164                                          std::string* country_code) {
 165   DCHECK(language_code);
 166   DCHECK(country_code);
 167   char language[ULOC_LANG_CAPACITY] = ULOC_ENGLISH;
 168   const char* country = "USA";
 169   if (!locale.empty()) {
 170     UErrorCode error = U_ZERO_ERROR;
 171     char id[ULOC_LANG_CAPACITY + ULOC_SCRIPT_CAPACITY + ULOC_COUNTRY_CAPACITY];
 172     uloc_addLikelySubtags(locale.c_str(), id, arraysize(id), &error);
 173     error = U_ZERO_ERROR;
 174     uloc_getLanguage(id, language, arraysize(language), &error);
 175     country = uloc_getISO3Country(id);
 176   }
 177   *language_code = std::string(language);
 178   *country_code = std::string(country);
 179 }
 180
 181 }  // namespace spellcheck_common
 182 }  // namespace chrome