ui/base/l10n/l10n_util.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 "ui/base/l10n/l10n_util.h"
   6
   7 #include <algorithm>
   8 #include <cstdlib>
   9 #include <iterator>
  10 #include <string>
  11
  12 #include "base/command_line.h"
  13 #include "base/compiler_specific.h"
  14 #include "base/file_util.h"
  15 #include "base/i18n/file_util_icu.h"
  16 #include "base/i18n/rtl.h"
  17 #include "base/i18n/string_compare.h"
  18 #include "base/lazy_instance.h"
  19 #include "base/memory/scoped_ptr.h"
  20 #include "base/path_service.h"
  21 #include "base/strings/string_number_conversions.h"
  22 #include "base/strings/string_split.h"
  23 #include "base/strings/string_util.h"
  24 #include "base/strings/stringprintf.h"
  25 #include "base/strings/sys_string_conversions.h"
  26 #include "base/strings/utf_string_conversions.h"
  27 #include "build/build_config.h"
  28 #include "third_party/icu/source/common/unicode/rbbi.h"
  29 #include "third_party/icu/source/common/unicode/uloc.h"
  30 #include "ui/base/l10n/l10n_util_collator.h"
  31 #include "ui/base/l10n/l10n_util_plurals.h"
  32 #include "ui/base/resource/resource_bundle.h"
  33 #include "ui/base/ui_base_paths.h"
  34
  35 #if defined(OS_ANDROID)
  36 #include "ui/base/l10n/l10n_util_android.h"
  37 #endif
  38
  39 #if defined(USE_GLIB)
  40 #include <glib.h>
  41 #endif
  42
  43 #if defined(OS_WIN)
  44 #include "ui/base/l10n/l10n_util_win.h"
  45 #endif  // OS_WIN
  46
  47 namespace {
  48
  49 static const char* const kAcceptLanguageList[] = {
  50   "af",     // Afrikaans
  51   "am",     // Amharic
  52   "ar",     // Arabic
  53   "az",     // Azerbaijani
  54   "be",     // Belarusian
  55   "bg",     // Bulgarian
  56   "bh",     // Bihari
  57   "bn",     // Bengali
  58   "br",     // Breton
  59   "bs",     // Bosnian
  60   "ca",     // Catalan
  61   "co",     // Corsican
  62   "cs",     // Czech
  63   "cy",     // Welsh
  64   "da",     // Danish
  65   "de",     // German
  66   "de-AT",  // German (Austria)
  67   "de-CH",  // German (Switzerland)
  68   "de-DE",  // German (Germany)
  69   "el",     // Greek
  70   "en",     // English
  71   "en-AU",  // English (Australia)
  72   "en-CA",  // English (Canada)
  73   "en-GB",  // English (UK)
  74   "en-NZ",  // English (New Zealand)
  75   "en-US",  // English (US)
  76   "en-ZA",  // English (South Africa)
  77   "eo",     // Esperanto
  78   // TODO(jungshik) : Do we want to list all es-Foo for Latin-American
  79   // Spanish speaking countries?
  80   "es",     // Spanish
  81   "es-419", // Spanish (Latin America)
  82   "et",     // Estonian
  83   "eu",     // Basque
  84   "fa",     // Persian
  85   "fi",     // Finnish
  86   "fil",    // Filipino
  87   "fo",     // Faroese
  88   "fr",     // French
  89   "fr-CA",  // French (Canada)
  90   "fr-CH",  // French (Switzerland)
  91   "fr-FR",  // French (France)
  92   "fy",     // Frisian
  93   "ga",     // Irish
  94   "gd",     // Scots Gaelic
  95   "gl",     // Galician
  96   "gn",     // Guarani
  97   "gu",     // Gujarati
  98   "ha",     // Hausa
  99   "haw",    // Hawaiian
 100   "he",     // Hebrew
 101   "hi",     // Hindi
 102   "hr",     // Croatian
 103   "hu",     // Hungarian
 104   "hy",     // Armenian
 105   "ia",     // Interlingua
 106   "id",     // Indonesian
 107   "is",     // Icelandic
 108   "it",     // Italian
 109   "it-CH",  // Italian (Switzerland)
 110   "it-IT",  // Italian (Italy)
 111   "ja",     // Japanese
 112   "jw",     // Javanese
 113   "ka",     // Georgian
 114   "kk",     // Kazakh
 115   "km",     // Cambodian
 116   "kn",     // Kannada
 117   "ko",     // Korean
 118   "ku",     // Kurdish
 119   "ky",     // Kyrgyz
 120   "la",     // Latin
 121   "ln",     // Lingala
 122   "lo",     // Laothian
 123   "lt",     // Lithuanian
 124   "lv",     // Latvian
 125   "mk",     // Macedonian
 126   "ml",     // Malayalam
 127   "mn",     // Mongolian
 128   "mo",     // Moldavian
 129   "mr",     // Marathi
 130   "ms",     // Malay
 131   "mt",     // Maltese
 132   "nb",     // Norwegian (Bokmal)
 133   "ne",     // Nepali
 134   "nl",     // Dutch
 135   "nn",     // Norwegian (Nynorsk)
 136   "no",     // Norwegian
 137   "oc",     // Occitan
 138   "om",     // Oromo
 139   "or",     // Oriya
 140   "pa",     // Punjabi
 141   "pl",     // Polish
 142   "ps",     // Pashto
 143   "pt",     // Portuguese
 144   "pt-BR",  // Portuguese (Brazil)
 145   "pt-PT",  // Portuguese (Portugal)
 146   "qu",     // Quechua
 147   "rm",     // Romansh
 148   "ro",     // Romanian
 149   "ru",     // Russian
 150   "sd",     // Sindhi
 151   "sh",     // Serbo-Croatian
 152   "si",     // Sinhalese
 153   "sk",     // Slovak
 154   "sl",     // Slovenian
 155   "sn",     // Shona
 156   "so",     // Somali
 157   "sq",     // Albanian
 158   "sr",     // Serbian
 159   "st",     // Sesotho
 160   "su",     // Sundanese
 161   "sv",     // Swedish
 162   "sw",     // Swahili
 163   "ta",     // Tamil
 164   "te",     // Telugu
 165   "tg",     // Tajik
 166   "th",     // Thai
 167   "ti",     // Tigrinya
 168   "tk",     // Turkmen
 169   "to",     // Tonga
 170   "tr",     // Turkish
 171   "tt",     // Tatar
 172   "tw",     // Twi
 173   "ug",     // Uighur
 174   "uk",     // Ukrainian
 175   "ur",     // Urdu
 176   "uz",     // Uzbek
 177   "vi",     // Vietnamese
 178   "xh",     // Xhosa
 179   "yi",     // Yiddish
 180   "yo",     // Yoruba
 181   "zh",     // Chinese
 182   "zh-CN",  // Chinese (Simplified)
 183   "zh-TW",  // Chinese (Traditional)
 184   "zu",     // Zulu
 185 };
 186
 187 // Returns true if |locale_name| has an alias in the ICU data file.
 188 bool IsDuplicateName(const std::string& locale_name) {
 189   static const char* const kDuplicateNames[] = {
 190     "en",
 191     "pt",
 192     "zh",
 193     "zh_hans_cn",
 194     "zh_hant_hk",
 195     "zh_hant_mo",
 196     "zh_hans_sg",
 197     "zh_hant_tw"
 198   };
 199
 200   // Skip all 'es_RR'. Currently, we use 'es' for es-ES (Spanish in Spain).
 201   // 'es-419' (Spanish in Latin America) is not available in ICU so that it
 202   // has to be added manually in GetAvailableLocales().
 203   if (LowerCaseEqualsASCII(locale_name.substr(0, 3),  "es_"))
 204     return true;
 205   for (size_t i = 0; i < arraysize(kDuplicateNames); ++i) {
 206     if (base::strcasecmp(kDuplicateNames[i], locale_name.c_str()) == 0)
 207       return true;
 208   }
 209   return false;
 210 }
 211
 212 // We added 30+ minimally populated locales with only a few entries
 213 // (exemplar character set, script, writing direction and its own
 214 // lanaguage name). These locales have to be distinguished from the
 215 // fully populated locales to which Chrome is localized.
 216 bool IsLocalePartiallyPopulated(const std::string& locale_name) {
 217   // For partially populated locales, even the translation for "English"
 218   // is not available. A more robust/elegant way to check is to add a special
 219   // field (say, 'isPartial' to our version of ICU locale files) and
 220   // check its value, but this hack seems to work well.
 221   return !l10n_util::IsLocaleNameTranslated("en", locale_name);
 222 }
 223
 224 #if !defined(OS_MACOSX)
 225 bool IsLocaleAvailable(const std::string& locale) {
 226   // If locale has any illegal characters in it, we don't want to try to
 227   // load it because it may be pointing outside the locale data file directory.
 228   if (!base::i18n::IsFilenameLegal(base::ASCIIToUTF16(locale)))
 229     return false;
 230
 231   // IsLocalePartiallyPopulated() can be called here for an early return w/o
 232   // checking the resource availability below. It'd help when Chrome is run
 233   // under a system locale Chrome is not localized to (e.g.Farsi on Linux),
 234   // but it'd slow down the start up time a little bit for locales Chrome is
 235   // localized to. So, we don't call it here.
 236   if (!l10n_util::IsLocaleSupportedByOS(locale))
 237     return false;
 238
 239   // If the ResourceBundle is not yet initialized, return false to avoid the
 240   // CHECK failure in ResourceBundle::GetSharedInstance().
 241   if (!ResourceBundle::HasSharedInstance())
 242     return false;
 243
 244   // TODO(hshi): make ResourceBundle::LocaleDataPakExists() a static function
 245   // so that this can be invoked without initializing the global instance.
 246   // See crbug.com/230432: CHECK failure in GetUserDataDir().
 247   return ResourceBundle::GetSharedInstance().LocaleDataPakExists(locale);
 248 }
 249 #endif
 250
 251 // On Linux, the text layout engine Pango determines paragraph directionality
 252 // by looking at the first strongly-directional character in the text. This
 253 // means text such as "Google Chrome foo bar..." will be layed out LTR even
 254 // if "foo bar" is RTL. So this function prepends the necessary RLM in such
 255 // cases.
 256 void AdjustParagraphDirectionality(base::string16* paragraph) {
 257 #if defined(OS_POSIX) && !defined(OS_MACOSX) && !defined(OS_ANDROID)
 258   if (base::i18n::IsRTL() &&
 259       base::i18n::StringContainsStrongRTLChars(*paragraph)) {
 260     paragraph->insert(0, 1,
 261                       static_cast<base::char16>(base::i18n::kRightToLeftMark));
 262   }
 263 #endif
 264 }
 265
 266 struct AvailableLocalesTraits
 267     : base::DefaultLazyInstanceTraits<std::vector<std::string> > {
 268   static std::vector<std::string>* New(void* instance) {
 269     std::vector<std::string>* locales =
 270         base::DefaultLazyInstanceTraits<std::vector<std::string> >::New(
 271             instance);
 272     int num_locales = uloc_countAvailable();
 273     for (int i = 0; i < num_locales; ++i) {
 274       std::string locale_name = uloc_getAvailable(i);
 275       // Filter out the names that have aliases.
 276       if (IsDuplicateName(locale_name))
 277         continue;
 278       // Filter out locales for which we have only partially populated data
 279       // and to which Chrome is not localized.
 280       if (IsLocalePartiallyPopulated(locale_name))
 281         continue;
 282       if (!l10n_util::IsLocaleSupportedByOS(locale_name))
 283         continue;
 284       // Normalize underscores to hyphens because that's what our locale files
 285       // use.
 286       std::replace(locale_name.begin(), locale_name.end(), '_', '-');
 287
 288       // Map the Chinese locale names over to zh-CN and zh-TW.
 289       if (LowerCaseEqualsASCII(locale_name, "zh-hans")) {
 290         locale_name = "zh-CN";
 291       } else if (LowerCaseEqualsASCII(locale_name, "zh-hant")) {
 292         locale_name = "zh-TW";
 293       }
 294       locales->push_back(locale_name);
 295     }
 296
 297     // Manually add 'es-419' to the list. See the comment in IsDuplicateName().
 298     locales->push_back("es-419");
 299     return locales;
 300   }
 301 };
 302
 303 base::LazyInstance<std::vector<std::string>, AvailableLocalesTraits>
 304     g_available_locales = LAZY_INSTANCE_INITIALIZER;
 305
 306 }  // namespace
 307
 308 namespace l10n_util {
 309
 310 std::string GetCanonicalLocale(const std::string& locale) {
 311   return base::i18n::GetCanonicalLocale(locale.c_str());
 312 }
 313
 314 std::string GetLanguage(const std::string& locale) {
 315   const std::string::size_type hyphen_pos = locale.find('-');
 316   return std::string(locale, 0, hyphen_pos);
 317 }
 318
 319 bool CheckAndResolveLocale(const std::string& locale,
 320                            std::string* resolved_locale) {
 321 #if defined(OS_MACOSX)
 322   NOTIMPLEMENTED();
 323   return false;
 324 #else
 325   if (IsLocaleAvailable(locale)) {
 326     *resolved_locale = locale;
 327     return true;
 328   }
 329
 330   // If there's a variant, skip over it so we can try without the region
 331   // code.  For example, ca_ES@valencia should cause us to try ca@valencia
 332   // before ca.
 333   std::string::size_type variant_pos = locale.find('@');
 334   if (variant_pos != std::string::npos)
 335     return false;
 336
 337   // If the locale matches language but not country, use that instead.
 338   // TODO(jungshik) : Nothing is done about languages that Chrome
 339   // does not support but available on Windows. We fall
 340   // back to en-US in GetApplicationLocale so that it's a not critical,
 341   // but we can do better.
 342   const std::string lang(GetLanguage(locale));
 343   if (lang.size() < locale.size()) {
 344     std::string region(locale, lang.size() + 1);
 345     std::string tmp_locale(lang);
 346     // Map es-RR other than es-ES to es-419 (Chrome's Latin American
 347     // Spanish locale).
 348     if (LowerCaseEqualsASCII(lang, "es") &&
 349         !LowerCaseEqualsASCII(region, "es")) {
 350       tmp_locale.append("-419");
 351     } else if (LowerCaseEqualsASCII(lang, "zh")) {
 352       // Map zh-HK and zh-MO to zh-TW. Otherwise, zh-FOO is mapped to zh-CN.
 353       if (LowerCaseEqualsASCII(region, "hk") ||
 354           LowerCaseEqualsASCII(region, "mo")) { // Macao
 355         tmp_locale.append("-TW");
 356       } else {
 357         tmp_locale.append("-CN");
 358       }
 359     } else if (LowerCaseEqualsASCII(lang, "en")) {
 360       // Map Australian, Canadian, New Zealand and South African English
 361       // to British English for now.
 362       // TODO(jungshik): en-CA may have to change sides once
 363       // we have OS locale separate from app locale (Chrome's UI language).
 364       if (LowerCaseEqualsASCII(region, "au") ||
 365           LowerCaseEqualsASCII(region, "ca") ||
 366           LowerCaseEqualsASCII(region, "nz") ||
 367           LowerCaseEqualsASCII(region, "za")) {
 368         tmp_locale.append("-GB");
 369       } else {
 370         tmp_locale.append("-US");
 371       }
 372     }
 373     if (IsLocaleAvailable(tmp_locale)) {
 374       resolved_locale->swap(tmp_locale);
 375       return true;
 376     }
 377   }
 378
 379   // Google updater uses no, tl, iw and en for our nb, fil, he, and en-US.
 380   struct {
 381     const char* source;
 382     const char* dest;
 383   } alias_map[] = {
 384       {"no", "nb"},
 385       {"tl", "fil"},
 386       {"iw", "he"},
 387       {"en", "en-US"},
 388   };
 389
 390   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(alias_map); ++i) {
 391     if (LowerCaseEqualsASCII(lang, alias_map[i].source)) {
 392       std::string tmp_locale(alias_map[i].dest);
 393       if (IsLocaleAvailable(tmp_locale)) {
 394         resolved_locale->swap(tmp_locale);
 395         return true;
 396       }
 397     }
 398   }
 399
 400   return false;
 401 #endif
 402 }
 403
 404 std::string GetApplicationLocaleInternal(const std::string& pref_locale) {
 405 #if defined(OS_MACOSX)
 406
 407   // Use any override (Cocoa for the browser), otherwise use the preference
 408   // passed to the function.
 409   std::string app_locale = l10n_util::GetLocaleOverride();
 410   if (app_locale.empty())
 411     app_locale = pref_locale;
 412
 413   // The above should handle all of the cases Chrome normally hits, but for some
 414   // unit tests, we need something to fall back too.
 415   if (app_locale.empty())
 416     app_locale = "en-US";
 417
 418   return app_locale;
 419
 420 #else
 421
 422   std::string resolved_locale;
 423   std::vector<std::string> candidates;
 424
 425   // We only use --lang and the app pref on Windows.  On Linux, we only
 426   // look at the LC_*/LANG environment variables.  We do, however, pass --lang
 427   // to renderer and plugin processes so they know what language the parent
 428   // process decided to use.
 429
 430 #if defined(OS_WIN)
 431
 432   // First, try the preference value.
 433   if (!pref_locale.empty())
 434     candidates.push_back(GetCanonicalLocale(pref_locale));
 435
 436   // Next, try the overridden locale.
 437   const std::vector<std::string>& languages = l10n_util::GetLocaleOverrides();
 438   if (!languages.empty()) {
 439     candidates.reserve(candidates.size() + languages.size());
 440     std::transform(languages.begin(), languages.end(),
 441                    std::back_inserter(candidates), &GetCanonicalLocale);
 442   } else {
 443     // If no override was set, defer to ICU
 444     candidates.push_back(base::i18n::GetConfiguredLocale());
 445   }
 446
 447 #elif defined(OS_ANDROID)
 448
 449   // On Android, query java.util.Locale for the default locale.
 450   candidates.push_back(GetDefaultLocale());
 451
 452 #elif defined(USE_GLIB) && !defined(OS_CHROMEOS)
 453
 454   // GLib implements correct environment variable parsing with
 455   // the precedence order: LANGUAGE, LC_ALL, LC_MESSAGES and LANG.
 456   // We used to use our custom parsing code along with ICU for this purpose.
 457   // If we have a port that does not depend on GTK, we have to
 458   // restore our custom code for that port.
 459   const char* const* languages = g_get_language_names();
 460   DCHECK(languages);  // A valid pointer is guaranteed.
 461   DCHECK(*languages);  // At least one entry, "C", is guaranteed.
 462
 463   for (; *languages != NULL; ++languages) {
 464     candidates.push_back(base::i18n::GetCanonicalLocale(*languages));
 465   }
 466
 467 #else
 468
 469   // By default, use the application locale preference. This applies to ChromeOS
 470   // and linux systems without glib.
 471   if (!pref_locale.empty())
 472     candidates.push_back(pref_locale);
 473
 474 #endif
 475
 476   std::vector<std::string>::const_iterator i = candidates.begin();
 477   for (; i != candidates.end(); ++i) {
 478     if (CheckAndResolveLocale(*i, &resolved_locale)) {
 479       return resolved_locale;
 480     }
 481   }
 482
 483   // Fallback on en-US.
 484   const std::string fallback_locale("en-US");
 485   if (IsLocaleAvailable(fallback_locale)) {
 486     return fallback_locale;
 487   }
 488
 489   return std::string();
 490
 491 #endif
 492 }
 493
 494 std::string GetApplicationLocale(const std::string& pref_locale,
 495                                  bool set_icu_locale) {
 496   const std::string locale = GetApplicationLocaleInternal(pref_locale);
 497   if (set_icu_locale && !locale.empty())
 498     base::i18n::SetICUDefaultLocale(locale);
 499   return locale;
 500 }
 501
 502 std::string GetApplicationLocale(const std::string& pref_locale) {
 503   return GetApplicationLocale(pref_locale, true /* set_icu_locale */);
 504 }
 505
 506 bool IsLocaleNameTranslated(const char* locale,
 507                             const std::string& display_locale) {
 508   base::string16 display_name =
 509       l10n_util::GetDisplayNameForLocale(locale, display_locale, false);
 510   // Because ICU sets the error code to U_USING_DEFAULT_WARNING whether or not
 511   // uloc_getDisplayName returns the actual translation or the default
 512   // value (locale code), we have to rely on this hack to tell whether
 513   // the translation is available or not.  If ICU doesn't have a translated
 514   // name for this locale, GetDisplayNameForLocale will just return the
 515   // locale code.
 516   return !base::IsStringASCII(display_name) ||
 517       base::UTF16ToASCII(display_name) != locale;
 518 }
 519
 520 base::string16 GetDisplayNameForLocale(const std::string& locale,
 521                                        const std::string& display_locale,
 522                                        bool is_for_ui) {
 523   std::string locale_code = locale;
 524   // Internally, we use the language code of zh-CN and zh-TW, but we want the
 525   // display names to be Chinese (Simplified) and Chinese (Traditional) instead
 526   // of Chinese (China) and Chinese (Taiwan).
 527   // Translate uses "tl" (Tagalog) to mean "fil" (Filipino) until Google
 528   // translate is changed to understand "fil". Make "tl" alias to "fil".
 529   if (locale_code == "zh-CN")
 530     locale_code = "zh-Hans";
 531   else if (locale_code == "zh-TW")
 532     locale_code = "zh-Hant";
 533   else if (locale_code == "tl")
 534     locale_code = "fil";
 535
 536   base::string16 display_name;
 537 #if defined(OS_ANDROID)
 538   // Use Java API to get locale display name so that we can remove most of
 539   // the lang data from icu data to reduce binary size, except for zh-Hans and
 540   // zh-Hant because the current Android Java API doesn't support scripts.
 541   // TODO(wangxianzhu): remove the special handling of zh-Hans and zh-Hant once
 542   // Android Java API supports scripts.
 543   if (!StartsWithASCII(locale_code, "zh-Han", true)) {
 544     display_name = GetDisplayNameForLocale(locale_code, display_locale);
 545   } else
 546 #endif
 547   {
 548     UErrorCode error = U_ZERO_ERROR;
 549     const int kBufferSize = 1024;
 550
 551     int actual_size = uloc_getDisplayName(
 552         locale_code.c_str(), display_locale.c_str(),
 553         WriteInto(&display_name, kBufferSize), kBufferSize - 1, &error);
 554     DCHECK(U_SUCCESS(error));
 555     display_name.resize(actual_size);
 556   }
 557
 558   // Add directional markup so parentheses are properly placed.
 559   if (is_for_ui && base::i18n::IsRTL())
 560     base::i18n::AdjustStringForLocaleDirection(&display_name);
 561   return display_name;
 562 }
 563
 564 base::string16 GetDisplayNameForCountry(const std::string& country_code,
 565                                         const std::string& display_locale) {
 566   return GetDisplayNameForLocale("_" + country_code, display_locale, false);
 567 }
 568
 569 std::string NormalizeLocale(const std::string& locale) {
 570   std::string normalized_locale(locale);
 571   std::replace(normalized_locale.begin(), normalized_locale.end(), '-', '_');
 572
 573   return normalized_locale;
 574 }
 575
 576 void GetParentLocales(const std::string& current_locale,
 577                       std::vector<std::string>* parent_locales) {
 578   std::string locale(NormalizeLocale(current_locale));
 579
 580   const int kNameCapacity = 256;
 581   char parent[kNameCapacity];
 582   base::strlcpy(parent, locale.c_str(), kNameCapacity);
 583   parent_locales->push_back(parent);
 584   UErrorCode err = U_ZERO_ERROR;
 585   while (uloc_getParent(parent, parent, kNameCapacity, &err) > 0) {
 586     if (U_FAILURE(err))
 587       break;
 588     parent_locales->push_back(parent);
 589   }
 590 }
 591
 592 bool IsValidLocaleSyntax(const std::string& locale) {
 593   // Check that the length is plausible.
 594   if (locale.size() < 2 || locale.size() >= ULOC_FULLNAME_CAPACITY)
 595     return false;
 596
 597   // Strip off the part after an '@' sign, which might contain keywords,
 598   // as in en_IE@currency=IEP or fr@collation=phonebook;calendar=islamic-civil.
 599   // We don't validate that part much, just check that there's at least one
 600   // equals sign in a plausible place. Normalize the prefix so that hyphens
 601   // are changed to underscores.
 602   std::string prefix = NormalizeLocale(locale);
 603   size_t split_point = locale.find("@");
 604   if (split_point != std::string::npos) {
 605     std::string keywords = locale.substr(split_point + 1);
 606     prefix = locale.substr(0, split_point);
 607
 608     size_t equals_loc = keywords.find("=");
 609     if (equals_loc == std::string::npos ||
 610         equals_loc < 1 || equals_loc > keywords.size() - 2)
 611       return false;
 612   }
 613
 614   // Check that all characters before the at-sign are alphanumeric or
 615   // underscore.
 616   for (size_t i = 0; i < prefix.size(); i++) {
 617     char ch = prefix[i];
 618     if (!IsAsciiAlpha(ch) && !IsAsciiDigit(ch) && ch != '_')
 619       return false;
 620   }
 621
 622   // Check that the initial token (before the first hyphen/underscore)
 623   // is 1 - 3 alphabetical characters (a language tag).
 624   for (size_t i = 0; i < prefix.size(); i++) {
 625     char ch = prefix[i];
 626     if (ch == '_') {
 627       if (i < 1 || i > 3)
 628         return false;
 629       break;
 630     }
 631     if (!IsAsciiAlpha(ch))
 632       return false;
 633   }
 634
 635   // Check that the all tokens after the initial token are 1 - 8 characters.
 636   // (Tokenize/StringTokenizer don't work here, they collapse multiple
 637   // delimiters into one.)
 638   int token_len = 0;
 639   int token_index = 0;
 640   for (size_t i = 0; i < prefix.size(); i++) {
 641     if (prefix[i] != '_') {
 642       token_len++;
 643       continue;
 644     }
 645
 646     if (token_index > 0 && (token_len < 1 || token_len > 8)) {
 647       return false;
 648     }
 649     token_index++;
 650     token_len = 0;
 651   }
 652   if (token_index == 0 && (token_len < 1 || token_len > 3)) {
 653     return false;
 654   } else if (token_len < 1 || token_len > 8) {
 655     return false;
 656   }
 657
 658   return true;
 659 }
 660
 661 std::string GetStringUTF8(int message_id) {
 662   return base::UTF16ToUTF8(GetStringUTF16(message_id));
 663 }
 664
 665 base::string16 GetStringUTF16(int message_id) {
 666   ResourceBundle& rb = ResourceBundle::GetSharedInstance();
 667   base::string16 str = rb.GetLocalizedString(message_id);
 668   AdjustParagraphDirectionality(&str);
 669
 670   return str;
 671 }
 672
 673 base::string16 GetStringFUTF16(int message_id,
 674                                const std::vector<base::string16>& replacements,
 675                                std::vector<size_t>* offsets) {
 676   // TODO(tc): We could save a string copy if we got the raw string as
 677   // a StringPiece and were able to call ReplaceStringPlaceholders with
 678   // a StringPiece format string and base::string16 substitution strings.  In
 679   // practice, the strings should be relatively short.
 680   ResourceBundle& rb = ResourceBundle::GetSharedInstance();
 681   const base::string16& format_string = rb.GetLocalizedString(message_id);
 682
 683 #ifndef NDEBUG
 684   // Make sure every replacement string is being used, so we don't just
 685   // silently fail to insert one. If |offsets| is non-NULL, then don't do this
 686   // check as the code may simply want to find the placeholders rather than
 687   // actually replacing them.
 688   if (!offsets) {
 689     std::string utf8_string = base::UTF16ToUTF8(format_string);
 690
 691     // $9 is the highest allowed placeholder.
 692     for (size_t i = 0; i < 9; ++i) {
 693       bool placeholder_should_exist = replacements.size() > i;
 694
 695       std::string placeholder =
 696           base::StringPrintf("$%d", static_cast<int>(i + 1));
 697       size_t pos = utf8_string.find(placeholder.c_str());
 698       if (placeholder_should_exist) {
 699         DCHECK_NE(std::string::npos, pos) <<
 700             " Didn't find a " << placeholder << " placeholder in " <<
 701             utf8_string;
 702       } else {
 703         DCHECK_EQ(std::string::npos, pos) <<
 704             " Unexpectedly found a " << placeholder << " placeholder in " <<
 705             utf8_string;
 706       }
 707     }
 708   }
 709 #endif
 710
 711   base::string16 formatted = ReplaceStringPlaceholders(
 712       format_string, replacements, offsets);
 713   AdjustParagraphDirectionality(&formatted);
 714
 715   return formatted;
 716 }
 717
 718 std::string GetStringFUTF8(int message_id,
 719                            const base::string16& a) {
 720   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a));
 721 }
 722
 723 std::string GetStringFUTF8(int message_id,
 724                            const base::string16& a,
 725                            const base::string16& b) {
 726   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b));
 727 }
 728
 729 std::string GetStringFUTF8(int message_id,
 730                            const base::string16& a,
 731                            const base::string16& b,
 732                            const base::string16& c) {
 733   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c));
 734 }
 735
 736 std::string GetStringFUTF8(int message_id,
 737                            const base::string16& a,
 738                            const base::string16& b,
 739                            const base::string16& c,
 740                            const base::string16& d) {
 741   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c, d));
 742 }
 743
 744 base::string16 GetStringFUTF16(int message_id,
 745                                const base::string16& a) {
 746   std::vector<base::string16> replacements;
 747   replacements.push_back(a);
 748   return GetStringFUTF16(message_id, replacements, NULL);
 749 }
 750
 751 base::string16 GetStringFUTF16(int message_id,
 752                                const base::string16& a,
 753                                const base::string16& b) {
 754   return GetStringFUTF16(message_id, a, b, NULL);
 755 }
 756
 757 base::string16 GetStringFUTF16(int message_id,
 758                                const base::string16& a,
 759                                const base::string16& b,
 760                                const base::string16& c) {
 761   std::vector<base::string16> replacements;
 762   replacements.push_back(a);
 763   replacements.push_back(b);
 764   replacements.push_back(c);
 765   return GetStringFUTF16(message_id, replacements, NULL);
 766 }
 767
 768 base::string16 GetStringFUTF16(int message_id,
 769                                const base::string16& a,
 770                                const base::string16& b,
 771                                const base::string16& c,
 772                                const base::string16& d) {
 773   std::vector<base::string16> replacements;
 774   replacements.push_back(a);
 775   replacements.push_back(b);
 776   replacements.push_back(c);
 777   replacements.push_back(d);
 778   return GetStringFUTF16(message_id, replacements, NULL);
 779 }
 780
 781 base::string16 GetStringFUTF16(int message_id,
 782                                const base::string16& a,
 783                                const base::string16& b,
 784                                const base::string16& c,
 785                                const base::string16& d,
 786                                const base::string16& e) {
 787   std::vector<base::string16> replacements;
 788   replacements.push_back(a);
 789   replacements.push_back(b);
 790   replacements.push_back(c);
 791   replacements.push_back(d);
 792   replacements.push_back(e);
 793   return GetStringFUTF16(message_id, replacements, NULL);
 794 }
 795
 796 base::string16 GetStringFUTF16(int message_id,
 797                                const base::string16& a,
 798                                size_t* offset) {
 799   DCHECK(offset);
 800   std::vector<size_t> offsets;
 801   std::vector<base::string16> replacements;
 802   replacements.push_back(a);
 803   base::string16 result = GetStringFUTF16(message_id, replacements, &offsets);
 804   DCHECK(offsets.size() == 1);
 805   *offset = offsets[0];
 806   return result;
 807 }
 808
 809 base::string16 GetStringFUTF16(int message_id,
 810                                const base::string16& a,
 811                                const base::string16& b,
 812                                std::vector<size_t>* offsets) {
 813   std::vector<base::string16> replacements;
 814   replacements.push_back(a);
 815   replacements.push_back(b);
 816   return GetStringFUTF16(message_id, replacements, offsets);
 817 }
 818
 819 base::string16 GetStringFUTF16Int(int message_id, int a) {
 820   return GetStringFUTF16(message_id, base::UTF8ToUTF16(base::IntToString(a)));
 821 }
 822
 823 base::string16 GetStringFUTF16Int(int message_id, int64 a) {
 824   return GetStringFUTF16(message_id, base::UTF8ToUTF16(base::Int64ToString(a)));
 825 }
 826
 827 // Specialization of operator() method for base::string16 version.
 828 template <>
 829 bool StringComparator<base::string16>::operator()(const base::string16& lhs,
 830                                                   const base::string16& rhs) {
 831   // If we can not get collator instance for specified locale, just do simple
 832   // string compare.
 833   if (!collator_)
 834     return lhs < rhs;
 835   return base::i18n::CompareString16WithCollator(collator_, lhs, rhs) ==
 836       UCOL_LESS;
 837 };
 838
 839 base::string16 GetPluralStringFUTF16(const std::vector<int>& message_ids,
 840                                int number) {
 841   scoped_ptr<icu::PluralFormat> format = BuildPluralFormat(message_ids);
 842   DCHECK(format);
 843
 844   UErrorCode err = U_ZERO_ERROR;
 845   icu::UnicodeString result_files_string = format->format(number, err);
 846   int capacity = result_files_string.length() + 1;
 847   DCHECK_GT(capacity, 1);
 848   base::string16 result;
 849   result_files_string.extract(
 850       static_cast<UChar*>(WriteInto(&result, capacity)), capacity, err);
 851   DCHECK(U_SUCCESS(err));
 852   return result;
 853 }
 854
 855 std::string GetPluralStringFUTF8(const std::vector<int>& message_ids,
 856                                  int number) {
 857   return base::UTF16ToUTF8(GetPluralStringFUTF16(message_ids, number));
 858 }
 859
 860 void SortStrings16(const std::string& locale,
 861                    std::vector<base::string16>* strings) {
 862   SortVectorWithStringKey(locale, strings, false);
 863 }
 864
 865 const std::vector<std::string>& GetAvailableLocales() {
 866   return g_available_locales.Get();
 867 }
 868
 869 void GetAcceptLanguagesForLocale(const std::string& display_locale,
 870                                  std::vector<std::string>* locale_codes) {
 871   for (size_t i = 0; i < arraysize(kAcceptLanguageList); ++i) {
 872     if (!l10n_util::IsLocaleNameTranslated(kAcceptLanguageList[i],
 873                                            display_locale))
 874       // TODO(jungshik) : Put them at the of the list with language codes
 875       // enclosed by brackets instead of skipping.
 876         continue;
 877     locale_codes->push_back(kAcceptLanguageList[i]);
 878   }
 879 }
 880
 881 int GetLocalizedContentsWidthInPixels(int pixel_resource_id) {
 882   int width = 0;
 883   base::StringToInt(l10n_util::GetStringUTF8(pixel_resource_id), &width);
 884   DCHECK_GT(width, 0);
 885   return width;
 886 }
 887
 888 const char* const* GetAcceptLanguageListForTesting() {
 889   return kAcceptLanguageList;
 890 }
 891
 892 size_t GetAcceptLanguageListSizeForTesting() {
 893   return arraysize(kAcceptLanguageList);
 894 }
 895
 896 }  // namespace l10n_util