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
47namespace {
48
49static 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.
188bool 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.
216bool 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)
225bool 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 (!file_util::IsFilenameLegal(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.
256void 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#if defined(OS_WIN)
267std::string GetCanonicalLocale(const std::string& locale) {
268  return base::i18n::GetCanonicalLocale(locale.c_str());
269}
270#endif
271
272struct AvailableLocalesTraits
273    : base::DefaultLazyInstanceTraits<std::vector<std::string> > {
274  static std::vector<std::string>* New(void* instance) {
275    std::vector<std::string>* locales =
276        base::DefaultLazyInstanceTraits<std::vector<std::string> >::New(
277            instance);
278    int num_locales = uloc_countAvailable();
279    for (int i = 0; i < num_locales; ++i) {
280      std::string locale_name = uloc_getAvailable(i);
281      // Filter out the names that have aliases.
282      if (IsDuplicateName(locale_name))
283        continue;
284      // Filter out locales for which we have only partially populated data
285      // and to which Chrome is not localized.
286      if (IsLocalePartiallyPopulated(locale_name))
287        continue;
288      if (!l10n_util::IsLocaleSupportedByOS(locale_name))
289        continue;
290      // Normalize underscores to hyphens because that's what our locale files
291      // use.
292      std::replace(locale_name.begin(), locale_name.end(), '_', '-');
293
294      // Map the Chinese locale names over to zh-CN and zh-TW.
295      if (LowerCaseEqualsASCII(locale_name, "zh-hans")) {
296        locale_name = "zh-CN";
297      } else if (LowerCaseEqualsASCII(locale_name, "zh-hant")) {
298        locale_name = "zh-TW";
299      }
300      locales->push_back(locale_name);
301    }
302
303    // Manually add 'es-419' to the list. See the comment in IsDuplicateName().
304    locales->push_back("es-419");
305    return locales;
306  }
307};
308
309base::LazyInstance<std::vector<std::string>, AvailableLocalesTraits>
310    g_available_locales = LAZY_INSTANCE_INITIALIZER;
311
312}  // namespace
313
314namespace l10n_util {
315
316bool CheckAndResolveLocale(const std::string& locale,
317                           std::string* resolved_locale) {
318#if defined(OS_MACOSX)
319  NOTIMPLEMENTED();
320  return false;
321#else
322  if (IsLocaleAvailable(locale)) {
323    *resolved_locale = locale;
324    return true;
325  }
326
327  // If there's a variant, skip over it so we can try without the region
328  // code.  For example, ca_ES@valencia should cause us to try ca@valencia
329  // before ca.
330  std::string::size_type variant_pos = locale.find('@');
331  if (variant_pos != std::string::npos)
332    return false;
333
334  // If the locale matches language but not country, use that instead.
335  // TODO(jungshik) : Nothing is done about languages that Chrome
336  // does not support but available on Windows. We fall
337  // back to en-US in GetApplicationLocale so that it's a not critical,
338  // but we can do better.
339  std::string::size_type hyphen_pos = locale.find('-');
340  std::string lang(locale, 0, hyphen_pos);
341  if (hyphen_pos != std::string::npos && hyphen_pos > 0) {
342    std::string region(locale, hyphen_pos + 1);
343    std::string tmp_locale(lang);
344    // Map es-RR other than es-ES to es-419 (Chrome's Latin American
345    // Spanish locale).
346    if (LowerCaseEqualsASCII(lang, "es") &&
347        !LowerCaseEqualsASCII(region, "es")) {
348      tmp_locale.append("-419");
349    } else if (LowerCaseEqualsASCII(lang, "zh")) {
350      // Map zh-HK and zh-MO to zh-TW. Otherwise, zh-FOO is mapped to zh-CN.
351      if (LowerCaseEqualsASCII(region, "hk") ||
352          LowerCaseEqualsASCII(region, "mo")) { // Macao
353        tmp_locale.append("-TW");
354      } else {
355        tmp_locale.append("-CN");
356      }
357    } else if (LowerCaseEqualsASCII(lang, "en")) {
358      // Map Australian, Canadian, New Zealand and South African English
359      // to British English for now.
360      // TODO(jungshik): en-CA may have to change sides once
361      // we have OS locale separate from app locale (Chrome's UI language).
362      if (LowerCaseEqualsASCII(region, "au") ||
363          LowerCaseEqualsASCII(region, "ca") ||
364          LowerCaseEqualsASCII(region, "nz") ||
365          LowerCaseEqualsASCII(region, "za")) {
366        tmp_locale.append("-GB");
367      } else {
368        tmp_locale.append("-US");
369      }
370    }
371    if (IsLocaleAvailable(tmp_locale)) {
372      resolved_locale->swap(tmp_locale);
373      return true;
374    }
375  }
376
377  // Google updater uses no, tl, iw and en for our nb, fil, he, and en-US.
378  struct {
379    const char* source;
380    const char* dest;
381  } alias_map[] = {
382      {"no", "nb"},
383      {"tl", "fil"},
384      {"iw", "he"},
385      {"en", "en-US"},
386  };
387
388  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(alias_map); ++i) {
389    if (LowerCaseEqualsASCII(lang, alias_map[i].source)) {
390      std::string tmp_locale(alias_map[i].dest);
391      if (IsLocaleAvailable(tmp_locale)) {
392        resolved_locale->swap(tmp_locale);
393        return true;
394      }
395    }
396  }
397
398  return false;
399#endif
400}
401
402std::string GetApplicationLocale(const std::string& pref_locale) {
403#if defined(OS_MACOSX)
404
405  // Use any override (Cocoa for the browser), otherwise use the preference
406  // passed to the function.
407  std::string app_locale = l10n_util::GetLocaleOverride();
408  if (app_locale.empty())
409    app_locale = pref_locale;
410
411  // The above should handle all of the cases Chrome normally hits, but for some
412  // unit tests, we need something to fall back too.
413  if (app_locale.empty())
414    app_locale = "en-US";
415
416  // Windows/Linux call SetICUDefaultLocale after determining the actual locale
417  // with CheckAndResolveLocal to make ICU APIs work in that locale.
418  // Mac doesn't use a locale directory tree of resources (it uses Mac style
419  // resources), so mirror the Windows/Linux behavior of calling
420  // SetICUDefaultLocale.
421  base::i18n::SetICUDefaultLocale(app_locale);
422  return app_locale;
423
424#else
425
426  std::string resolved_locale;
427  std::vector<std::string> candidates;
428
429  // We only use --lang and the app pref on Windows.  On Linux, we only
430  // look at the LC_*/LANG environment variables.  We do, however, pass --lang
431  // to renderer and plugin processes so they know what language the parent
432  // process decided to use.
433
434#if defined(OS_WIN)
435
436  // First, try the preference value.
437  if (!pref_locale.empty())
438    candidates.push_back(GetCanonicalLocale(pref_locale));
439
440  // Next, try the overridden locale.
441  const std::vector<std::string>& languages = l10n_util::GetLocaleOverrides();
442  if (!languages.empty()) {
443    candidates.reserve(candidates.size() + languages.size());
444    std::transform(languages.begin(), languages.end(),
445                   std::back_inserter(candidates), &GetCanonicalLocale);
446  } else {
447    // If no override was set, defer to ICU
448    candidates.push_back(base::i18n::GetConfiguredLocale());
449  }
450
451#elif defined(OS_ANDROID)
452
453  // On Android, query java.util.Locale for the default locale.
454  candidates.push_back(GetDefaultLocale());
455
456#elif defined(USE_GLIB) && !defined(OS_CHROMEOS)
457
458  // GLib implements correct environment variable parsing with
459  // the precedence order: LANGUAGE, LC_ALL, LC_MESSAGES and LANG.
460  // We used to use our custom parsing code along with ICU for this purpose.
461  // If we have a port that does not depend on GTK, we have to
462  // restore our custom code for that port.
463  const char* const* languages = g_get_language_names();
464  DCHECK(languages);  // A valid pointer is guaranteed.
465  DCHECK(*languages);  // At least one entry, "C", is guaranteed.
466
467  for (; *languages != NULL; ++languages) {
468    candidates.push_back(base::i18n::GetCanonicalLocale(*languages));
469  }
470
471#else
472
473  // By default, use the application locale preference. This applies to ChromeOS
474  // and linux systems without glib.
475  if (!pref_locale.empty())
476    candidates.push_back(pref_locale);
477
478#endif
479
480  std::vector<std::string>::const_iterator i = candidates.begin();
481  for (; i != candidates.end(); ++i) {
482    if (CheckAndResolveLocale(*i, &resolved_locale)) {
483      base::i18n::SetICUDefaultLocale(resolved_locale);
484      return resolved_locale;
485    }
486  }
487
488  // Fallback on en-US.
489  const std::string fallback_locale("en-US");
490  if (IsLocaleAvailable(fallback_locale)) {
491    base::i18n::SetICUDefaultLocale(fallback_locale);
492    return fallback_locale;
493  }
494
495  return std::string();
496
497#endif
498}
499
500bool IsLocaleNameTranslated(const char* locale,
501                            const std::string& display_locale) {
502  base::string16 display_name =
503      l10n_util::GetDisplayNameForLocale(locale, display_locale, false);
504  // Because ICU sets the error code to U_USING_DEFAULT_WARNING whether or not
505  // uloc_getDisplayName returns the actual translation or the default
506  // value (locale code), we have to rely on this hack to tell whether
507  // the translation is available or not.  If ICU doesn't have a translated
508  // name for this locale, GetDisplayNameForLocale will just return the
509  // locale code.
510  return !IsStringASCII(display_name) || UTF16ToASCII(display_name) != locale;
511}
512
513base::string16 GetDisplayNameForLocale(const std::string& locale,
514                                       const std::string& display_locale,
515                                       bool is_for_ui) {
516  std::string locale_code = locale;
517  // Internally, we use the language code of zh-CN and zh-TW, but we want the
518  // display names to be Chinese (Simplified) and Chinese (Traditional) instead
519  // of Chinese (China) and Chinese (Taiwan).  To do that, we pass zh-Hans
520  // and zh-Hant to ICU. Even with this mapping, we'd get
521  // 'Chinese (Simplified Han)' and 'Chinese (Traditional Han)' in English and
522  // even longer results in other languages. Arguably, they're better than
523  // the current results : Chinese (China) / Chinese (Taiwan).
524  // TODO(jungshik): Do one of the following:
525  // 1. Special-case Chinese by getting the custom-translation for them
526  // 2. Recycle IDS_ENCODING_{SIMP,TRAD}_CHINESE.
527  // 3. Get translations for two directly from the ICU resouce bundle
528  // because they're not accessible with other any API.
529  // 4. Patch ICU to special-case zh-Hans/zh-Hant for us.
530  // #1 and #2 wouldn't work if display_locale != current UI locale although
531  // we can think of additional hack to work around the problem.
532  // #3 can be potentially expensive.
533  if (locale_code == "zh-CN")
534    locale_code = "zh-Hans";
535  else if (locale_code == "zh-TW")
536    locale_code = "zh-Hant";
537
538  base::string16 display_name;
539#if defined(OS_ANDROID)
540  // Use Java API to get locale display name so that we can remove most of
541  // the lang data from icu data to reduce binary size, except for zh-Hans and
542  // zh-Hant because the current Android Java API doesn't support scripts.
543  // TODO(wangxianzhu): remove the special handling of zh-Hans and zh-Hant once
544  // Android Java API supports scripts.
545  if (!StartsWithASCII(locale_code, "zh-Han", true)) {
546    display_name = GetDisplayNameForLocale(locale_code, display_locale);
547  } else
548#endif
549  {
550    UErrorCode error = U_ZERO_ERROR;
551    const int kBufferSize = 1024;
552
553    int actual_size = uloc_getDisplayName(
554        locale_code.c_str(), display_locale.c_str(),
555        WriteInto(&display_name, kBufferSize), kBufferSize - 1, &error);
556    DCHECK(U_SUCCESS(error));
557    display_name.resize(actual_size);
558  }
559
560  // Add directional markup so parentheses are properly placed.
561  if (is_for_ui && base::i18n::IsRTL())
562    base::i18n::AdjustStringForLocaleDirection(&display_name);
563  return display_name;
564}
565
566base::string16 GetDisplayNameForCountry(const std::string& country_code,
567                                        const std::string& display_locale) {
568  return GetDisplayNameForLocale("_" + country_code, display_locale, false);
569}
570
571std::string NormalizeLocale(const std::string& locale) {
572  std::string normalized_locale(locale);
573  std::replace(normalized_locale.begin(), normalized_locale.end(), '-', '_');
574
575  return normalized_locale;
576}
577
578void GetParentLocales(const std::string& current_locale,
579                      std::vector<std::string>* parent_locales) {
580  std::string locale(NormalizeLocale(current_locale));
581
582  const int kNameCapacity = 256;
583  char parent[kNameCapacity];
584  base::strlcpy(parent, locale.c_str(), kNameCapacity);
585  parent_locales->push_back(parent);
586  UErrorCode err = U_ZERO_ERROR;
587  while (uloc_getParent(parent, parent, kNameCapacity, &err) > 0) {
588    if (U_FAILURE(err))
589      break;
590    parent_locales->push_back(parent);
591  }
592}
593
594bool IsValidLocaleSyntax(const std::string& locale) {
595  // Check that the length is plausible.
596  if (locale.size() < 2 || locale.size() >= ULOC_FULLNAME_CAPACITY)
597    return false;
598
599  // Strip off the part after an '@' sign, which might contain keywords,
600  // as in en_IE@currency=IEP or fr@collation=phonebook;calendar=islamic-civil.
601  // We don't validate that part much, just check that there's at least one
602  // equals sign in a plausible place. Normalize the prefix so that hyphens
603  // are changed to underscores.
604  std::string prefix = NormalizeLocale(locale);
605  size_t split_point = locale.find("@");
606  if (split_point != std::string::npos) {
607    std::string keywords = locale.substr(split_point + 1);
608    prefix = locale.substr(0, split_point);
609
610    size_t equals_loc = keywords.find("=");
611    if (equals_loc == std::string::npos ||
612        equals_loc < 1 || equals_loc > keywords.size() - 2)
613      return false;
614  }
615
616  // Check that all characters before the at-sign are alphanumeric or
617  // underscore.
618  for (size_t i = 0; i < prefix.size(); i++) {
619    char ch = prefix[i];
620    if (!IsAsciiAlpha(ch) && !IsAsciiDigit(ch) && ch != '_')
621      return false;
622  }
623
624  // Check that the initial token (before the first hyphen/underscore)
625  // is 1 - 3 alphabetical characters (a language tag).
626  for (size_t i = 0; i < prefix.size(); i++) {
627    char ch = prefix[i];
628    if (ch == '_') {
629      if (i < 1 || i > 3)
630        return false;
631      break;
632    }
633    if (!IsAsciiAlpha(ch))
634      return false;
635  }
636
637  // Check that the all tokens after the initial token are 1 - 8 characters.
638  // (Tokenize/StringTokenizer don't work here, they collapse multiple
639  // delimiters into one.)
640  int token_len = 0;
641  int token_index = 0;
642  for (size_t i = 0; i < prefix.size(); i++) {
643    if (prefix[i] != '_') {
644      token_len++;
645      continue;
646    }
647
648    if (token_index > 0 && (token_len < 1 || token_len > 8)) {
649      return false;
650    }
651    token_index++;
652    token_len = 0;
653  }
654  if (token_index == 0 && (token_len < 1 || token_len > 3)) {
655    return false;
656  } else if (token_len < 1 || token_len > 8) {
657    return false;
658  }
659
660  return true;
661}
662
663std::string GetStringUTF8(int message_id) {
664  return UTF16ToUTF8(GetStringUTF16(message_id));
665}
666
667base::string16 GetStringUTF16(int message_id) {
668  ResourceBundle& rb = ResourceBundle::GetSharedInstance();
669  base::string16 str = rb.GetLocalizedString(message_id);
670  AdjustParagraphDirectionality(&str);
671
672  return str;
673}
674
675base::string16 GetStringFUTF16(int message_id,
676                               const std::vector<base::string16>& replacements,
677                               std::vector<size_t>* offsets) {
678  // TODO(tc): We could save a string copy if we got the raw string as
679  // a StringPiece and were able to call ReplaceStringPlaceholders with
680  // a StringPiece format string and base::string16 substitution strings.  In
681  // practice, the strings should be relatively short.
682  ResourceBundle& rb = ResourceBundle::GetSharedInstance();
683  const base::string16& format_string = rb.GetLocalizedString(message_id);
684
685#ifndef NDEBUG
686  // Make sure every replacement string is being used, so we don't just
687  // silently fail to insert one. If |offsets| is non-NULL, then don't do this
688  // check as the code may simply want to find the placeholders rather than
689  // actually replacing them.
690  if (!offsets) {
691    std::string utf8_string = UTF16ToUTF8(format_string);
692
693    // $9 is the highest allowed placeholder.
694    for (size_t i = 0; i < 9; ++i) {
695      bool placeholder_should_exist = replacements.size() > i;
696
697      std::string placeholder =
698          base::StringPrintf("$%d", static_cast<int>(i + 1));
699      size_t pos = utf8_string.find(placeholder.c_str());
700      if (placeholder_should_exist) {
701        DCHECK_NE(std::string::npos, pos) <<
702            " Didn't find a " << placeholder << " placeholder in " <<
703            utf8_string;
704      } else {
705        DCHECK_EQ(std::string::npos, pos) <<
706            " Unexpectedly found a " << placeholder << " placeholder in " <<
707            utf8_string;
708      }
709    }
710  }
711#endif
712
713  base::string16 formatted = ReplaceStringPlaceholders(
714      format_string, replacements, offsets);
715  AdjustParagraphDirectionality(&formatted);
716
717  return formatted;
718}
719
720std::string GetStringFUTF8(int message_id,
721                           const base::string16& a) {
722  return UTF16ToUTF8(GetStringFUTF16(message_id, a));
723}
724
725std::string GetStringFUTF8(int message_id,
726                           const base::string16& a,
727                           const base::string16& b) {
728  return UTF16ToUTF8(GetStringFUTF16(message_id, a, b));
729}
730
731std::string GetStringFUTF8(int message_id,
732                           const base::string16& a,
733                           const base::string16& b,
734                           const base::string16& c) {
735  return UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c));
736}
737
738std::string GetStringFUTF8(int message_id,
739                           const base::string16& a,
740                           const base::string16& b,
741                           const base::string16& c,
742                           const base::string16& d) {
743  return UTF16ToUTF8(GetStringFUTF16(message_id, a, b, c, d));
744}
745
746base::string16 GetStringFUTF16(int message_id,
747                               const base::string16& a) {
748  std::vector<base::string16> replacements;
749  replacements.push_back(a);
750  return GetStringFUTF16(message_id, replacements, NULL);
751}
752
753base::string16 GetStringFUTF16(int message_id,
754                               const base::string16& a,
755                               const base::string16& b) {
756  return GetStringFUTF16(message_id, a, b, NULL);
757}
758
759base::string16 GetStringFUTF16(int message_id,
760                               const base::string16& a,
761                               const base::string16& b,
762                               const base::string16& c) {
763  std::vector<base::string16> replacements;
764  replacements.push_back(a);
765  replacements.push_back(b);
766  replacements.push_back(c);
767  return GetStringFUTF16(message_id, replacements, NULL);
768}
769
770base::string16 GetStringFUTF16(int message_id,
771                               const base::string16& a,
772                               const base::string16& b,
773                               const base::string16& c,
774                               const base::string16& d) {
775  std::vector<base::string16> replacements;
776  replacements.push_back(a);
777  replacements.push_back(b);
778  replacements.push_back(c);
779  replacements.push_back(d);
780  return GetStringFUTF16(message_id, replacements, NULL);
781}
782
783base::string16 GetStringFUTF16(int message_id,
784                               const base::string16& a,
785                               const base::string16& b,
786                               const base::string16& c,
787                               const base::string16& d,
788                               const base::string16& e) {
789  std::vector<base::string16> replacements;
790  replacements.push_back(a);
791  replacements.push_back(b);
792  replacements.push_back(c);
793  replacements.push_back(d);
794  replacements.push_back(e);
795  return GetStringFUTF16(message_id, replacements, NULL);
796}
797
798base::string16 GetStringFUTF16(int message_id,
799                               const base::string16& a,
800                               size_t* offset) {
801  DCHECK(offset);
802  std::vector<size_t> offsets;
803  std::vector<base::string16> replacements;
804  replacements.push_back(a);
805  base::string16 result = GetStringFUTF16(message_id, replacements, &offsets);
806  DCHECK(offsets.size() == 1);
807  *offset = offsets[0];
808  return result;
809}
810
811base::string16 GetStringFUTF16(int message_id,
812                               const base::string16& a,
813                               const base::string16& b,
814                               std::vector<size_t>* offsets) {
815  std::vector<base::string16> replacements;
816  replacements.push_back(a);
817  replacements.push_back(b);
818  return GetStringFUTF16(message_id, replacements, offsets);
819}
820
821base::string16 GetStringFUTF16Int(int message_id, int a) {
822  return GetStringFUTF16(message_id, UTF8ToUTF16(base::IntToString(a)));
823}
824
825base::string16 GetStringFUTF16Int(int message_id, int64 a) {
826  return GetStringFUTF16(message_id, UTF8ToUTF16(base::Int64ToString(a)));
827}
828
829// Specialization of operator() method for base::string16 version.
830template <>
831bool StringComparator<base::string16>::operator()(const base::string16& lhs,
832                                                  const base::string16& rhs) {
833  // If we can not get collator instance for specified locale, just do simple
834  // string compare.
835  if (!collator_)
836    return lhs < rhs;
837  return base::i18n::CompareString16WithCollator(collator_, lhs, rhs) ==
838      UCOL_LESS;
839};
840
841base::string16 GetPluralStringFUTF16(const std::vector<int>& message_ids,
842                               int number) {
843  scoped_ptr<icu::PluralFormat> format = BuildPluralFormat(message_ids);
844  DCHECK(format);
845
846  UErrorCode err = U_ZERO_ERROR;
847  icu::UnicodeString result_files_string = format->format(number, err);
848  int capacity = result_files_string.length() + 1;
849  DCHECK_GT(capacity, 1);
850  base::string16 result;
851  result_files_string.extract(
852      static_cast<UChar*>(WriteInto(&result, capacity)), capacity, err);
853  DCHECK(U_SUCCESS(err));
854  return result;
855}
856
857std::string GetPluralStringFUTF8(const std::vector<int>& message_ids,
858                                 int number) {
859  return base::UTF16ToUTF8(GetPluralStringFUTF16(message_ids, number));
860}
861
862void SortStrings16(const std::string& locale,
863                   std::vector<base::string16>* strings) {
864  SortVectorWithStringKey(locale, strings, false);
865}
866
867const std::vector<std::string>& GetAvailableLocales() {
868  return g_available_locales.Get();
869}
870
871void GetAcceptLanguagesForLocale(const std::string& display_locale,
872                                 std::vector<std::string>* locale_codes) {
873  for (size_t i = 0; i < arraysize(kAcceptLanguageList); ++i) {
874    if (!l10n_util::IsLocaleNameTranslated(kAcceptLanguageList[i],
875                                           display_locale))
876      // TODO(jungshik) : Put them at the of the list with language codes
877      // enclosed by brackets instead of skipping.
878        continue;
879    locale_codes->push_back(kAcceptLanguageList[i]);
880  }
881}
882
883int GetLocalizedContentsWidthInPixels(int pixel_resource_id) {
884  int width = 0;
885  base::StringToInt(l10n_util::GetStringUTF8(pixel_resource_id), &width);
886  DCHECK_GT(width, 0);
887  return width;
888}
889
890}  // namespace l10n_util
891