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/files/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/strings/string_number_conversions.h"
  21 #include "base/strings/string_split.h"
  22 #include "base/strings/string_util.h"
  23 #include "base/strings/stringprintf.h"
  24 #include "base/strings/sys_string_conversions.h"
  25 #include "base/strings/utf_string_conversions.h"
  26 #include "build/build_config.h"
  27 #include "third_party/icu/source/common/unicode/rbbi.h"
  28 #include "third_party/icu/source/common/unicode/uloc.h"
  29 #include "ui/base/l10n/l10n_util_collator.h"
  30 #include "ui/base/l10n/l10n_util_plurals.h"
  31 #include "ui/base/resource/resource_bundle.h"
  32 #include "ui/base/ui_base_paths.h"
  33
  34 #if defined(OS_ANDROID)
  35 #include "base/android/locale_utils.h"
  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   "ckb",    // Kurdish (Arabci),  Sorani
  62   "co",     // Corsican
  63   "cs",     // Czech
  64   "cy",     // Welsh
  65   "da",     // Danish
  66   "de",     // German
  67   "de-AT",  // German (Austria)
  68   "de-CH",  // German (Switzerland)
  69   "de-DE",  // German (Germany)
  70   "de-LI",  // German (Liechtenstein)
  71   "el",     // Greek
  72   "en",     // English
  73   "en-AU",  // English (Australia)
  74   "en-CA",  // English (Canada)
  75   "en-GB",  // English (UK)
  76   "en-NZ",  // English (New Zealand)
  77   "en-US",  // English (US)
  78   "en-ZA",  // English (South Africa)
  79   "eo",     // Esperanto
  80   // TODO(jungshik) : Do we want to list all es-Foo for Latin-American
  81   // Spanish speaking countries?
  82   "es",     // Spanish
  83   "es-419", // Spanish (Latin America)
  84   "et",     // Estonian
  85   "eu",     // Basque
  86   "fa",     // Persian
  87   "fi",     // Finnish
  88   "fil",    // Filipino
  89   "fo",     // Faroese
  90   "fr",     // French
  91   "fr-CA",  // French (Canada)
  92   "fr-CH",  // French (Switzerland)
  93   "fr-FR",  // French (France)
  94   "fy",     // Frisian
  95   "ga",     // Irish
  96   "gd",     // Scots Gaelic
  97   "gl",     // Galician
  98   "gn",     // Guarani
  99   "gu",     // Gujarati
 100   "ha",     // Hausa
 101   "haw",    // Hawaiian
 102   "he",     // Hebrew
 103   "hi",     // Hindi
 104   "hr",     // Croatian
 105   "hu",     // Hungarian
 106   "hy",     // Armenian
 107   "ia",     // Interlingua
 108   "id",     // Indonesian
 109   "is",     // Icelandic
 110   "it",     // Italian
 111   "it-CH",  // Italian (Switzerland)
 112   "it-IT",  // Italian (Italy)
 113   "ja",     // Japanese
 114   "jw",     // Javanese
 115   "ka",     // Georgian
 116   "kk",     // Kazakh
 117   "km",     // Cambodian
 118   "kn",     // Kannada
 119   "ko",     // Korean
 120   "ku",     // Kurdish
 121   "ky",     // Kyrgyz
 122   "la",     // Latin
 123   "ln",     // Lingala
 124   "lo",     // Laothian
 125   "lt",     // Lithuanian
 126   "lv",     // Latvian
 127   "mk",     // Macedonian
 128   "ml",     // Malayalam
 129   "mn",     // Mongolian
 130   "mo",     // Moldavian
 131   "mr",     // Marathi
 132   "ms",     // Malay
 133   "mt",     // Maltese
 134   "nb",     // Norwegian (Bokmal)
 135   "ne",     // Nepali
 136   "nl",     // Dutch
 137   "nn",     // Norwegian (Nynorsk)
 138   "no",     // Norwegian
 139   "oc",     // Occitan
 140   "om",     // Oromo
 141   "or",     // Oriya
 142   "pa",     // Punjabi
 143   "pl",     // Polish
 144   "ps",     // Pashto
 145   "pt",     // Portuguese
 146   "pt-BR",  // Portuguese (Brazil)
 147   "pt-PT",  // Portuguese (Portugal)
 148   "qu",     // Quechua
 149   "rm",     // Romansh
 150   "ro",     // Romanian
 151   "ru",     // Russian
 152   "sd",     // Sindhi
 153   "sh",     // Serbo-Croatian
 154   "si",     // Sinhalese
 155   "sk",     // Slovak
 156   "sl",     // Slovenian
 157   "sn",     // Shona
 158   "so",     // Somali
 159   "sq",     // Albanian
 160   "sr",     // Serbian
 161   "st",     // Sesotho
 162   "su",     // Sundanese
 163   "sv",     // Swedish
 164   "sw",     // Swahili
 165   "ta",     // Tamil
 166   "te",     // Telugu
 167   "tg",     // Tajik
 168   "th",     // Thai
 169   "ti",     // Tigrinya
 170   "tk",     // Turkmen
 171   "to",     // Tonga
 172   "tr",     // Turkish
 173   "tt",     // Tatar
 174   "tw",     // Twi
 175   "ug",     // Uighur
 176   "uk",     // Ukrainian
 177   "ur",     // Urdu
 178   "uz",     // Uzbek
 179   "vi",     // Vietnamese
 180   "xh",     // Xhosa
 181   "yi",     // Yiddish
 182   "yo",     // Yoruba
 183   "zh",     // Chinese
 184   "zh-CN",  // Chinese (Simplified)
 185   "zh-TW",  // Chinese (Traditional)
 186   "zu",     // Zulu
 187 };
 188
 189 // Returns true if |locale_name| has an alias in the ICU data file.
 190 bool IsDuplicateName(const std::string& locale_name) {
 191   static const char* const kDuplicateNames[] = {
 192     "en",
 193     "en_001",
 194     "pt", // pt-BR and pt-PT are used.
 195     "zh",
 196     "zh_hans_cn",
 197     "zh_hant_hk",
 198     "zh_hant_mo",
 199     "zh_hans_sg",
 200     "zh_hant_tw"
 201   };
 202
 203   // Skip all the es_Foo other than es_419 for now.
 204   if (StartsWithASCII(locale_name, "es_", false))
 205     return !EndsWith(locale_name, "419", true);
 206
 207   for (size_t i = 0; i < arraysize(kDuplicateNames); ++i) {
 208     if (base::strcasecmp(kDuplicateNames[i], locale_name.c_str()) == 0)
 209       return true;
 210   }
 211   return false;
 212 }
 213
 214 // We added 30+ minimally populated locales with only a few entries
 215 // (exemplar character set, script, writing direction and its own
 216 // lanaguage name). These locales have to be distinguished from the
 217 // fully populated locales to which Chrome is localized.
 218 bool IsLocalePartiallyPopulated(const std::string& locale_name) {
 219   // For partially populated locales, even the translation for "English"
 220   // is not available. A more robust/elegant way to check is to add a special
 221   // field (say, 'isPartial' to our version of ICU locale files) and
 222   // check its value, but this hack seems to work well.
 223   return !l10n_util::IsLocaleNameTranslated("en", locale_name);
 224 }
 225
 226 #if !defined(OS_MACOSX)
 227 bool IsLocaleAvailable(const std::string& locale) {
 228   // If locale has any illegal characters in it, we don't want to try to
 229   // load it because it may be pointing outside the locale data file directory.
 230   if (!base::i18n::IsFilenameLegal(base::ASCIIToUTF16(locale)))
 231     return false;
 232
 233   // IsLocalePartiallyPopulated() can be called here for an early return w/o
 234   // checking the resource availability below. It'd help when Chrome is run
 235   // under a system locale Chrome is not localized to (e.g.Farsi on Linux),
 236   // but it'd slow down the start up time a little bit for locales Chrome is
 237   // localized to. So, we don't call it here.
 238   if (!l10n_util::IsLocaleSupportedByOS(locale))
 239     return false;
 240
 241   // If the ResourceBundle is not yet initialized, return false to avoid the
 242   // CHECK failure in ResourceBundle::GetSharedInstance().
 243   if (!ResourceBundle::HasSharedInstance())
 244     return false;
 245
 246   // TODO(hshi): make ResourceBundle::LocaleDataPakExists() a static function
 247   // so that this can be invoked without initializing the global instance.
 248   // See crbug.com/230432: CHECK failure in GetUserDataDir().
 249   return ResourceBundle::GetSharedInstance().LocaleDataPakExists(locale);
 250 }
 251 #endif
 252
 253 // On Linux, the text layout engine Pango determines paragraph directionality
 254 // by looking at the first strongly-directional character in the text. This
 255 // means text such as "Google Chrome foo bar..." will be layed out LTR even
 256 // if "foo bar" is RTL. So this function prepends the necessary RLM in such
 257 // cases.
 258 void AdjustParagraphDirectionality(base::string16* paragraph) {
 259 #if defined(OS_POSIX) && !defined(OS_MACOSX) && !defined(OS_ANDROID)
 260   if (base::i18n::IsRTL() &&
 261       base::i18n::StringContainsStrongRTLChars(*paragraph)) {
 262     paragraph->insert(0, 1,
 263                       static_cast<base::char16>(base::i18n::kRightToLeftMark));
 264   }
 265 #endif
 266 }
 267
 268 struct AvailableLocalesTraits
 269     : base::DefaultLazyInstanceTraits<std::vector<std::string> > {
 270   static std::vector<std::string>* New(void* instance) {
 271     std::vector<std::string>* locales =
 272         base::DefaultLazyInstanceTraits<std::vector<std::string> >::New(
 273             instance);
 274     int num_locales = uloc_countAvailable();
 275     for (int i = 0; i < num_locales; ++i) {
 276       std::string locale_name = uloc_getAvailable(i);
 277       // Filter out the names that have aliases.
 278       if (IsDuplicateName(locale_name))
 279         continue;
 280       // Filter out locales for which we have only partially populated data
 281       // and to which Chrome is not localized.
 282       if (IsLocalePartiallyPopulated(locale_name))
 283         continue;
 284       if (!l10n_util::IsLocaleSupportedByOS(locale_name))
 285         continue;
 286       // Normalize underscores to hyphens because that's what our locale files
 287       // use.
 288       std::replace(locale_name.begin(), locale_name.end(), '_', '-');
 289
 290       // Map the Chinese locale names over to zh-CN and zh-TW.
 291       if (LowerCaseEqualsASCII(locale_name, "zh-hans")) {
 292         locale_name = "zh-CN";
 293       } else if (LowerCaseEqualsASCII(locale_name, "zh-hant")) {
 294         locale_name = "zh-TW";
 295       }
 296       locales->push_back(locale_name);
 297     }
 298
 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 GetLanguage(const std::string& locale) {
 311   const std::string::size_type hyphen_pos = locale.find('-');
 312   return std::string(locale, 0, hyphen_pos);
 313 }
 314
 315 bool CheckAndResolveLocale(const std::string& locale,
 316                            std::string* resolved_locale) {
 317 #if defined(OS_MACOSX)
 318   NOTIMPLEMENTED();
 319   return false;
 320 #else
 321   if (IsLocaleAvailable(locale)) {
 322     *resolved_locale = locale;
 323     return true;
 324   }
 325
 326   // If there's a variant, skip over it so we can try without the region
 327   // code.  For example, ca_ES@valencia should cause us to try ca@valencia
 328   // before ca.
 329   std::string::size_type variant_pos = locale.find('@');
 330   if (variant_pos != std::string::npos)
 331     return false;
 332
 333   // If the locale matches language but not country, use that instead.
 334   // TODO(jungshik) : Nothing is done about languages that Chrome
 335   // does not support but available on Windows. We fall
 336   // back to en-US in GetApplicationLocale so that it's a not critical,
 337   // but we can do better.
 338   const std::string lang(GetLanguage(locale));
 339   if (lang.size() < locale.size()) {
 340     std::string region(locale, lang.size() + 1);
 341     std::string tmp_locale(lang);
 342     // Map es-RR other than es-ES to es-419 (Chrome's Latin American
 343     // Spanish locale).
 344     if (LowerCaseEqualsASCII(lang, "es") &&
 345         !LowerCaseEqualsASCII(region, "es")) {
 346       tmp_locale.append("-419");
 347     } else if (LowerCaseEqualsASCII(lang, "zh")) {
 348       // Map zh-HK and zh-MO to zh-TW. Otherwise, zh-FOO is mapped to zh-CN.
 349       if (LowerCaseEqualsASCII(region, "hk") ||
 350           LowerCaseEqualsASCII(region, "mo")) { // Macao
 351         tmp_locale.append("-TW");
 352       } else {
 353         tmp_locale.append("-CN");
 354       }
 355     } else if (LowerCaseEqualsASCII(lang, "en")) {
 356       // Map Australian, Canadian, New Zealand and South African English
 357       // to British English for now.
 358       // TODO(jungshik): en-CA may have to change sides once
 359       // we have OS locale separate from app locale (Chrome's UI language).
 360       if (LowerCaseEqualsASCII(region, "au") ||
 361           LowerCaseEqualsASCII(region, "ca") ||
 362           LowerCaseEqualsASCII(region, "nz") ||
 363           LowerCaseEqualsASCII(region, "za")) {
 364         tmp_locale.append("-GB");
 365       } else {
 366         tmp_locale.append("-US");
 367       }
 368     }
 369     if (IsLocaleAvailable(tmp_locale)) {
 370       resolved_locale->swap(tmp_locale);
 371       return true;
 372     }
 373   }
 374
 375   // Google updater uses no, tl, iw and en for our nb, fil, he, and en-US.
 376   struct {
 377     const char* source;
 378     const char* dest;
 379   } alias_map[] = {
 380       {"no", "nb"},
 381       {"tl", "fil"},
 382       {"iw", "he"},
 383       {"en", "en-US"},
 384   };
 385
 386   for (size_t i = 0; i < arraysize(alias_map); ++i) {
 387     if (LowerCaseEqualsASCII(lang, alias_map[i].source)) {
 388       std::string tmp_locale(alias_map[i].dest);
 389       if (IsLocaleAvailable(tmp_locale)) {
 390         resolved_locale->swap(tmp_locale);
 391         return true;
 392       }
 393     }
 394   }
 395
 396   return false;
 397 #endif
 398 }
 399
 400 std::string GetApplicationLocaleInternal(const std::string& pref_locale) {
 401 #if defined(OS_MACOSX)
 402
 403   // Use any override (Cocoa for the browser), otherwise use the preference
 404   // passed to the function.
 405   std::string app_locale = l10n_util::GetLocaleOverride();
 406   if (app_locale.empty())
 407     app_locale = pref_locale;
 408
 409   // The above should handle all of the cases Chrome normally hits, but for some
 410   // unit tests, we need something to fall back too.
 411   if (app_locale.empty())
 412     app_locale = "en-US";
 413
 414   return app_locale;
 415
 416 #else
 417
 418   std::string resolved_locale;
 419   std::vector<std::string> candidates;
 420
 421   // We only use --lang and the app pref on Windows.  On Linux, we only
 422   // look at the LC_*/LANG environment variables.  We do, however, pass --lang
 423   // to renderer and plugin processes so they know what language the parent
 424   // process decided to use.
 425
 426 #if defined(OS_WIN)
 427
 428   // First, try the preference value.
 429   if (!pref_locale.empty())
 430     candidates.push_back(base::i18n::GetCanonicalLocale(pref_locale));
 431
 432   // Next, try the overridden locale.
 433   const std::vector<std::string>& languages = l10n_util::GetLocaleOverrides();
 434   if (!languages.empty()) {
 435     candidates.reserve(candidates.size() + languages.size());
 436     std::transform(languages.begin(), languages.end(),
 437                    std::back_inserter(candidates),
 438                    &base::i18n::GetCanonicalLocale);
 439   } else {
 440     // If no override was set, defer to ICU
 441     candidates.push_back(base::i18n::GetConfiguredLocale());
 442   }
 443
 444 #elif defined(OS_ANDROID)
 445
 446   // On Android, query java.util.Locale for the default locale.
 447   candidates.push_back(base::android::GetDefaultLocale());
 448
 449 #elif defined(USE_GLIB) && !defined(OS_CHROMEOS)
 450
 451   // GLib implements correct environment variable parsing with
 452   // the precedence order: LANGUAGE, LC_ALL, LC_MESSAGES and LANG.
 453   // We used to use our custom parsing code along with ICU for this purpose.
 454   // If we have a port that does not depend on GTK, we have to
 455   // restore our custom code for that port.
 456   const char* const* languages = g_get_language_names();
 457   DCHECK(languages);  // A valid pointer is guaranteed.
 458   DCHECK(*languages);  // At least one entry, "C", is guaranteed.
 459
 460   for (; *languages != NULL; ++languages) {
 461     candidates.push_back(base::i18n::GetCanonicalLocale(*languages));
 462   }
 463
 464 #else
 465
 466   // By default, use the application locale preference. This applies to ChromeOS
 467   // and linux systems without glib.
 468   if (!pref_locale.empty())
 469     candidates.push_back(pref_locale);
 470
 471 #endif
 472
 473   std::vector<std::string>::const_iterator i = candidates.begin();
 474   for (; i != candidates.end(); ++i) {
 475     if (CheckAndResolveLocale(*i, &resolved_locale)) {
 476       return resolved_locale;
 477     }
 478   }
 479
 480   // Fallback on en-US.
 481   const std::string fallback_locale("en-US");
 482   if (IsLocaleAvailable(fallback_locale)) {
 483     return fallback_locale;
 484   }
 485
 486   return std::string();
 487
 488 #endif
 489 }
 490
 491 std::string GetApplicationLocale(const std::string& pref_locale,
 492                                  bool set_icu_locale) {
 493   const std::string locale = GetApplicationLocaleInternal(pref_locale);
 494   if (set_icu_locale && !locale.empty())
 495     base::i18n::SetICUDefaultLocale(locale);
 496   return locale;
 497 }
 498
 499 std::string GetApplicationLocale(const std::string& pref_locale) {
 500   return GetApplicationLocale(pref_locale, true /* set_icu_locale */);
 501 }
 502
 503 bool IsLocaleNameTranslated(const char* locale,
 504                             const std::string& display_locale) {
 505   base::string16 display_name =
 506       l10n_util::GetDisplayNameForLocale(locale, display_locale, false);
 507   // Because ICU sets the error code to U_USING_DEFAULT_WARNING whether or not
 508   // uloc_getDisplayName returns the actual translation or the default
 509   // value (locale code), we have to rely on this hack to tell whether
 510   // the translation is available or not.  If ICU doesn't have a translated
 511   // name for this locale, GetDisplayNameForLocale will just return the
 512   // locale code.
 513   return !base::IsStringASCII(display_name) ||
 514       base::UTF16ToASCII(display_name) != locale;
 515 }
 516
 517 base::string16 GetDisplayNameForLocale(const std::string& locale,
 518                                        const std::string& display_locale,
 519                                        bool is_for_ui) {
 520   std::string locale_code = locale;
 521   // Internally, we use the language code of zh-CN and zh-TW, but we want the
 522   // display names to be Chinese (Simplified) and Chinese (Traditional) instead
 523   // of Chinese (China) and Chinese (Taiwan).
 524   // Translate uses "tl" (Tagalog) to mean "fil" (Filipino) until Google
 525   // translate is changed to understand "fil". Make "tl" alias to "fil".
 526   if (locale_code == "zh-CN")
 527     locale_code = "zh-Hans";
 528   else if (locale_code == "zh-TW")
 529     locale_code = "zh-Hant";
 530   else if (locale_code == "tl")
 531     locale_code = "fil";
 532   else if (locale_code == "mo")
 533     locale_code = "ro-MD";
 534
 535   base::string16 display_name;
 536 #if defined(OS_ANDROID)
 537   // Use Java API to get locale display name so that we can remove most of
 538   // the lang data from icu data to reduce binary size, except for zh-Hans and
 539   // zh-Hant because the current Android Java API doesn't support scripts.
 540   // TODO(wangxianzhu): remove the special handling of zh-Hans and zh-Hant once
 541   // Android Java API supports scripts.
 542   if (!StartsWithASCII(locale_code, "zh-Han", true)) {
 543     display_name = GetDisplayNameForLocale(locale_code, display_locale);
 544   } else
 545 #endif
 546   {
 547     UErrorCode error = U_ZERO_ERROR;
 548     const int kBufferSize = 1024;
 549
 550     int actual_size = uloc_getDisplayName(
 551         locale_code.c_str(), display_locale.c_str(),
 552         WriteInto(&display_name, kBufferSize), kBufferSize - 1, &error);
 553     DCHECK(U_SUCCESS(error));
 554     display_name.resize(actual_size);
 555   }
 556
 557   // Add directional markup so parentheses are properly placed.
 558   if (is_for_ui && base::i18n::IsRTL())
 559     base::i18n::AdjustStringForLocaleDirection(&display_name);
 560   return display_name;
 561 }
 562
 563 base::string16 GetDisplayNameForCountry(const std::string& country_code,
 564                                         const std::string& display_locale) {
 565   return GetDisplayNameForLocale("_" + country_code, display_locale, false);
 566 }
 567
 568 std::string NormalizeLocale(const std::string& locale) {
 569   std::string normalized_locale(locale);
 570   std::replace(normalized_locale.begin(), normalized_locale.end(), '-', '_');
 571
 572   return normalized_locale;
 573 }
 574
 575 void GetParentLocales(const std::string& current_locale,
 576                       std::vector<std::string>* parent_locales) {
 577   std::string locale(NormalizeLocale(current_locale));
 578
 579   const int kNameCapacity = 256;
 580   char parent[kNameCapacity];
 581   base::strlcpy(parent, locale.c_str(), kNameCapacity);
 582   parent_locales->push_back(parent);
 583   UErrorCode err = U_ZERO_ERROR;
 584   while (uloc_getParent(parent, parent, kNameCapacity, &err) > 0) {
 585     if (U_FAILURE(err))
 586       break;
 587     parent_locales->push_back(parent);
 588   }
 589 }
 590
 591 bool IsValidLocaleSyntax(const std::string& locale) {
 592   // Check that the length is plausible.
 593   if (locale.size() < 2 || locale.size() >= ULOC_FULLNAME_CAPACITY)
 594     return false;
 595
 596   // Strip off the part after an '@' sign, which might contain keywords,
 597   // as in en_IE@currency=IEP or fr@collation=phonebook;calendar=islamic-civil.
 598   // We don't validate that part much, just check that there's at least one
 599   // equals sign in a plausible place. Normalize the prefix so that hyphens
 600   // are changed to underscores.
 601   std::string prefix = NormalizeLocale(locale);
 602   size_t split_point = locale.find("@");
 603   if (split_point != std::string::npos) {
 604     std::string keywords = locale.substr(split_point + 1);
 605     prefix = locale.substr(0, split_point);
 606
 607     size_t equals_loc = keywords.find("=");
 608     if (equals_loc == std::string::npos ||
 609         equals_loc < 1 || equals_loc > keywords.size() - 2)
 610       return false;
 611   }
 612
 613   // Check that all characters before the at-sign are alphanumeric or
 614   // underscore.
 615   for (size_t i = 0; i < prefix.size(); i++) {
 616     char ch = prefix[i];
 617     if (!IsAsciiAlpha(ch) && !IsAsciiDigit(ch) && ch != '_')
 618       return false;
 619   }
 620
 621   // Check that the initial token (before the first hyphen/underscore)
 622   // is 1 - 3 alphabetical characters (a language tag).
 623   for (size_t i = 0; i < prefix.size(); i++) {
 624     char ch = prefix[i];
 625     if (ch == '_') {
 626       if (i < 1 || i > 3)
 627         return false;
 628       break;
 629     }
 630     if (!IsAsciiAlpha(ch))
 631       return false;
 632   }
 633
 634   // Check that the all tokens after the initial token are 1 - 8 characters.
 635   // (Tokenize/StringTokenizer don't work here, they collapse multiple
 636   // delimiters into one.)
 637   int token_len = 0;
 638   int token_index = 0;
 639   for (size_t i = 0; i < prefix.size(); i++) {
 640     if (prefix[i] != '_') {
 641       token_len++;
 642       continue;
 643     }
 644
 645     if (token_index > 0 && (token_len < 1 || token_len > 8)) {
 646       return false;
 647     }
 648     token_index++;
 649     token_len = 0;
 650   }
 651   if (token_index == 0 && (token_len < 1 || token_len > 3)) {
 652     return false;
 653   } else if (token_len < 1 || token_len > 8) {
 654     return false;
 655   }
 656
 657   return true;
 658 }
 659
 660 std::string GetStringUTF8(int message_id) {
 661   return base::UTF16ToUTF8(GetStringUTF16(message_id));
 662 }
 663
 664 base::string16 GetStringUTF16(int message_id) {
 665   ResourceBundle& rb = ResourceBundle::GetSharedInstance();
 666   base::string16 str = rb.GetLocalizedString(message_id);
 667   AdjustParagraphDirectionality(&str);
 668
 669   return str;
 670 }
 671
 672 base::string16 GetStringFUTF16(int message_id,
 673                                const std::vector<base::string16>& replacements,
 674                                std::vector<size_t>* offsets) {
 675   // TODO(tc): We could save a string copy if we got the raw string as
 676   // a StringPiece and were able to call ReplaceStringPlaceholders with
 677   // a StringPiece format string and base::string16 substitution strings.  In
 678   // practice, the strings should be relatively short.
 679   ResourceBundle& rb = ResourceBundle::GetSharedInstance();
 680   const base::string16& format_string = rb.GetLocalizedString(message_id);
 681
 682 #ifndef NDEBUG
 683   // Make sure every replacement string is being used, so we don't just
 684   // silently fail to insert one. If |offsets| is non-NULL, then don't do this
 685   // check as the code may simply want to find the placeholders rather than
 686   // actually replacing them.
 687   if (!offsets) {
 688     std::string utf8_string = base::UTF16ToUTF8(format_string);
 689
 690     // $9 is the highest allowed placeholder.
 691     for (size_t i = 0; i < 9; ++i) {
 692       bool placeholder_should_exist = replacements.size() > i;
 693
 694       std::string placeholder =
 695           base::StringPrintf("$%d", static_cast<int>(i + 1));
 696       size_t pos = utf8_string.find(placeholder.c_str());
 697       if (placeholder_should_exist) {
 698         DCHECK_NE(std::string::npos, pos) <<
 699             " Didn't find a " << placeholder << " placeholder in " <<
 700             utf8_string;
 701       } else {
 702         DCHECK_EQ(std::string::npos, pos) <<
 703             " Unexpectedly found a " << placeholder << " placeholder in " <<
 704             utf8_string;
 705       }
 706     }
 707   }
 708 #endif
 709
 710   base::string16 formatted = ReplaceStringPlaceholders(
 711       format_string, replacements, offsets);
 712   AdjustParagraphDirectionality(&formatted);
 713
 714   return formatted;
 715 }
 716
 717 std::string GetStringFUTF8(int message_id,
 718                            const base::string16& a) {
 719   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a));
 720 }
 721
 722 std::string GetStringFUTF8(int message_id,
 723                            const base::string16& a,
 724                            const base::string16& b) {
 725   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b));
 726 }
 727
 728 std::string GetStringFUTF8(int message_id,
 729                            const base::string16& a,
 730                            const base::string16& b,
 731                            const base::string16& c) {
 732   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c));
 733 }
 734
 735 std::string GetStringFUTF8(int message_id,
 736                            const base::string16& a,
 737                            const base::string16& b,
 738                            const base::string16& c,
 739                            const base::string16& d) {
 740   return base::UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c, d));
 741 }
 742
 743 base::string16 GetStringFUTF16(int message_id,
 744                                const base::string16& a) {
 745   std::vector<base::string16> replacements;
 746   replacements.push_back(a);
 747   return GetStringFUTF16(message_id, replacements, NULL);
 748 }
 749
 750 base::string16 GetStringFUTF16(int message_id,
 751                                const base::string16& a,
 752                                const base::string16& b) {
 753   return GetStringFUTF16(message_id, a, b, NULL);
 754 }
 755
 756 base::string16 GetStringFUTF16(int message_id,
 757                                const base::string16& a,
 758                                const base::string16& b,
 759                                const base::string16& c) {
 760   std::vector<base::string16> replacements;
 761   replacements.push_back(a);
 762   replacements.push_back(b);
 763   replacements.push_back(c);
 764   return GetStringFUTF16(message_id, replacements, NULL);
 765 }
 766
 767 base::string16 GetStringFUTF16(int message_id,
 768                                const base::string16& a,
 769                                const base::string16& b,
 770                                const base::string16& c,
 771                                const base::string16& d) {
 772   std::vector<base::string16> replacements;
 773   replacements.push_back(a);
 774   replacements.push_back(b);
 775   replacements.push_back(c);
 776   replacements.push_back(d);
 777   return GetStringFUTF16(message_id, replacements, NULL);
 778 }
 779
 780 base::string16 GetStringFUTF16(int message_id,
 781                                const base::string16& a,
 782                                const base::string16& b,
 783                                const base::string16& c,
 784                                const base::string16& d,
 785                                const base::string16& e) {
 786   std::vector<base::string16> replacements;
 787   replacements.push_back(a);
 788   replacements.push_back(b);
 789   replacements.push_back(c);
 790   replacements.push_back(d);
 791   replacements.push_back(e);
 792   return GetStringFUTF16(message_id, replacements, NULL);
 793 }
 794
 795 base::string16 GetStringFUTF16(int message_id,
 796                                const base::string16& a,
 797                                size_t* offset) {
 798   DCHECK(offset);
 799   std::vector<size_t> offsets;
 800   std::vector<base::string16> replacements;
 801   replacements.push_back(a);
 802   base::string16 result = GetStringFUTF16(message_id, replacements, &offsets);
 803   DCHECK(offsets.size() == 1);
 804   *offset = offsets[0];
 805   return result;
 806 }
 807
 808 base::string16 GetStringFUTF16(int message_id,
 809                                const base::string16& a,
 810                                const base::string16& b,
 811                                std::vector<size_t>* offsets) {
 812   std::vector<base::string16> replacements;
 813   replacements.push_back(a);
 814   replacements.push_back(b);
 815   return GetStringFUTF16(message_id, replacements, offsets);
 816 }
 817
 818 base::string16 GetStringFUTF16Int(int message_id, int a) {
 819   return GetStringFUTF16(message_id, base::UTF8ToUTF16(base::IntToString(a)));
 820 }
 821
 822 base::string16 GetStringFUTF16Int(int message_id, int64 a) {
 823   return GetStringFUTF16(message_id, base::UTF8ToUTF16(base::Int64ToString(a)));
 824 }
 825
 826 base::string16 GetPluralStringFUTF16(int message_id, int number) {
 827   base::string16 pattern = GetStringUTF16(message_id);
 828   UErrorCode err = U_ZERO_ERROR;
 829   icu::MessageFormat format(
 830       icu::UnicodeString(FALSE, pattern.data(), pattern.length()), err);
 831   icu::UnicodeString result_unistring;
 832   FormatNumberInPlural(format, number, &result_unistring, &err);
 833   int capacity = result_unistring.length() + 1;
 834   DCHECK_GT(capacity, 1);
 835   base::string16 result;
 836   result_unistring.extract(
 837       static_cast<UChar*>(WriteInto(&result, capacity)), capacity, err);
 838   DCHECK(U_SUCCESS(err));
 839   return result;
 840 }
 841
 842 std::string GetPluralStringFUTF8(int message_id, int number) {
 843   return base::UTF16ToUTF8(GetPluralStringFUTF16(message_id, number));
 844 }
 845
 846 void SortStrings16(const std::string& locale,
 847                    std::vector<base::string16>* strings) {
 848   SortVectorWithStringKey(locale, strings, false);
 849 }
 850
 851 const std::vector<std::string>& GetAvailableLocales() {
 852   return g_available_locales.Get();
 853 }
 854
 855 void GetAcceptLanguagesForLocale(const std::string& display_locale,
 856                                  std::vector<std::string>* locale_codes) {
 857   for (size_t i = 0; i < arraysize(kAcceptLanguageList); ++i) {
 858     if (!l10n_util::IsLocaleNameTranslated(kAcceptLanguageList[i],
 859                                            display_locale))
 860       // TODO(jungshik) : Put them at the of the list with language codes
 861       // enclosed by brackets instead of skipping.
 862         continue;
 863     locale_codes->push_back(kAcceptLanguageList[i]);
 864   }
 865 }
 866
 867 int GetLocalizedContentsWidthInPixels(int pixel_resource_id) {
 868   int width = 0;
 869   base::StringToInt(l10n_util::GetStringUTF8(pixel_resource_id), &width);
 870   DCHECK_GT(width, 0);
 871   return width;
 872 }
 873
 874 const char* const* GetAcceptLanguageListForTesting() {
 875   return kAcceptLanguageList;
 876 }
 877
 878 size_t GetAcceptLanguageListSizeForTesting() {
 879   return arraysize(kAcceptLanguageList);
 880 }
 881
 882 }  // namespace l10n_util