Trying to re-land http://codereview.chromium.org/6901141.

src/extensions/experimental/language-matcher.cc

Index: src/extensions/experimental/language-matcher.cc
===================================================================
--- src/extensions/experimental/language-matcher.cc	(revision 0)
+++ src/extensions/experimental/language-matcher.cc	(revision 0)
@@ -0,0 +1,249 @@
+// Copyright 2011 the V8 project authors. All rights reserved.
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+//       notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+//       copyright notice, this list of conditions and the following
+//       disclaimer in the documentation and/or other materials provided
+//       with the distribution.
+//     * Neither the name of Google Inc. nor the names of its
+//       contributors may be used to endorse or promote products derived
+//       from this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// TODO(cira): Remove LanguageMatcher from v8 when ICU implements
+// language matching API.
+
+#include "language-matcher.h"
+
+#include "i18n-utils.h"
+#include "unicode/datefmt.h"  // For getAvailableLocales
+#include "unicode/locid.h"
+#include "unicode/uloc.h"
+#include "utils.h"
+
+namespace v8 {
+namespace internal {
+
+const unsigned int LanguageMatcher::kLanguageWeight = 75;
+const unsigned int LanguageMatcher::kScriptWeight = 20;
+const unsigned int LanguageMatcher::kRegionWeight = 5;
+const unsigned int LanguageMatcher::kThreshold = 50;
+const unsigned int LanguageMatcher::kPositionBonus = 1;
+const char* const LanguageMatcher::kDefaultLocale = "root";
+
+static const char* GetLanguageException(const char*);
+static bool BCP47ToICUFormat(const char*, char*);
+static int CompareLocaleSubtags(const char*, const char*);
+static bool BuildLocaleName(const char*, const char*, LocaleIDMatch*);
+
+LocaleIDMatch::LocaleIDMatch()
+    : score(-1) {
+  I18NUtils::StrNCopy(
+      bcp47_id, ULOC_FULLNAME_CAPACITY, LanguageMatcher::kDefaultLocale);
+
+  I18NUtils::StrNCopy(
+      icu_id, ULOC_FULLNAME_CAPACITY, LanguageMatcher::kDefaultLocale);
+}
+
+LocaleIDMatch& LocaleIDMatch::operator=(const LocaleIDMatch& rhs) {
+  I18NUtils::StrNCopy(this->bcp47_id, ULOC_FULLNAME_CAPACITY, rhs.bcp47_id);
+  I18NUtils::StrNCopy(this->icu_id, ULOC_FULLNAME_CAPACITY, rhs.icu_id);
+  this->score = rhs.score;
+
+  return *this;
+}
+
+// static
+void LanguageMatcher::GetBestMatchForPriorityList(
+    v8::Handle<v8::Array> locales, LocaleIDMatch* result) {
+  v8::HandleScope handle_scope;
+
+  unsigned int position_bonus = locales->Length() * kPositionBonus;
+
+  int max_score = 0;
+  LocaleIDMatch match;
+  for (unsigned int i = 0; i < locales->Length(); ++i) {
+    position_bonus -= kPositionBonus;
+
+    v8::TryCatch try_catch;
+    v8::Local<v8::Value> locale_id = locales->Get(v8::Integer::New(i));
+
+    // Return default if exception is raised when reading parameter.
+    if (try_catch.HasCaught()) break;
+
+    // JavaScript arrays can be heterogenous so check each item
+    // if it's a string.
+    if (!locale_id->IsString()) continue;
+
+    if (!CompareToSupportedLocaleIDList(locale_id->ToString(), &match)) {
+      continue;
+    }
+
+    // Skip items under threshold.
+    if (match.score < kThreshold) continue;
+
+    match.score += position_bonus;
+    if (match.score > max_score) {
+      *result = match;
+
+      max_score = match.score;
+    }
+  }
+}
+
+// static
+void LanguageMatcher::GetBestMatchForString(
+    v8::Handle<v8::String> locale, LocaleIDMatch* result) {
+  LocaleIDMatch match;
+
+  if (CompareToSupportedLocaleIDList(locale, &match) &&
+      match.score >= kThreshold) {
+    *result = match;
+  }
+}
+
+// static
+bool LanguageMatcher::CompareToSupportedLocaleIDList(
+    v8::Handle<v8::String> locale_id, LocaleIDMatch* result) {
+  static int32_t available_count = 0;
+  // Depending on how ICU data is built, locales returned by
+  // Locale::getAvailableLocale() are not guaranteed to support DateFormat,
+  // Collation and other services.  We can call getAvailableLocale() of all the
+  // services we want to support and take the intersection of them all, but
+  // using DateFormat::getAvailableLocales() should suffice.
+  // TODO(cira): Maybe make this thread-safe?
+  static const icu::Locale* available_locales =
+      icu::DateFormat::getAvailableLocales(available_count);
+
+  // Skip this locale_id if it's not in ASCII.
+  static LocaleIDMatch default_match;
+  v8::String::AsciiValue is_ascii(locale_id);
+  if (*is_ascii == NULL) return false;

[jungshik at Google, 2011/05/04 22:45:47]

nit: ascii_value? 

[Nebojša Ćirić, 2011/05/04 23:09:40]

Done.

+
+  char locale[ULOC_FULLNAME_CAPACITY];
+  if (!BCP47ToICUFormat(*is_ascii, locale)) return false;
+
+  icu::Locale input_locale(locale);
+
+  // Position of the best match locale in list of available locales.
+  int position = -1;
+  const char* language = GetLanguageException(input_locale.getLanguage());
+  const char* script = input_locale.getScript();
+  const char* region = input_locale.getCountry();
+  for (int32_t i = 0; i < available_count; ++i) {
+    int current_score = 0;
+    int sign =
+        CompareLocaleSubtags(language, available_locales[i].getLanguage());
+    current_score += sign * kLanguageWeight;
+
+    sign = CompareLocaleSubtags(script, available_locales[i].getScript());
+    current_score += sign * kScriptWeight;
+
+    sign = CompareLocaleSubtags(region, available_locales[i].getCountry());
+    current_score += sign * kRegionWeight;
+
+    if (current_score >= kThreshold && current_score > result->score) {
+      result->score = current_score;
+      position = i;
+    }
+  }
+
+  // Didn't find any good matches so use defaults.
+  if (position == -1) return false;
+
+  return BuildLocaleName(available_locales[position].getBaseName(),
+                         input_locale.getName(), result);
+}
+
+// For some unsupported language subtags it is better to fallback to related
+// language that is supported than to default.
+static const char* GetLanguageException(const char* language) {
+  // Serbo-croatian to Serbian.
+  if (!strcmp(language, "sh")) return "sr";
+
+  // Norweigan to Norweiaan to Norwegian Bokmal.
+  if (!strcmp(language, "no")) return "nb";
+
+  // Moldavian to Romanian.
+  if (!strcmp(language, "mo")) return "ro";
+
+  // Tagalog to Filipino.
+  if (!strcmp(language, "tl")) return "fil";
+
+  return language;
+}
+
+// Converts user input from BCP47 locale id format to ICU compatible format.
+// Returns false if uloc_forLanguageTag call fails or if extension is too long.
+static bool BCP47ToICUFormat(const char* locale_id, char* result) {
+  UErrorCode status = U_ZERO_ERROR;
+  int32_t locale_size = 0;
+
+  char locale[ULOC_FULLNAME_CAPACITY];
+  I18NUtils::StrNCopy(locale, ULOC_FULLNAME_CAPACITY, locale_id);
+
+  // uloc_forLanguageTag has a bug where long extension can crash the code.
+  // We need to check if extension part of language id conforms to the length.

[jungshik at Google, 2011/05/04 22:45:47]

Can you add a URL  to the bug you just filed? Thanks. 

[Nebojša Ćirić, 2011/05/04 23:09:40]

Done.

+  const char* extension = strstr(locale_id, "-u-");
+  if (extension != NULL &&
+      strlen(extension) > ULOC_KEYWORD_AND_VALUES_CAPACITY) {
+    // Truncate to get non-crashing string, but still preserve base language.

[jungshik at Google, 2011/05/04 22:45:47]

How about truncating at the last logical boundary before going over the limit?
Well, perhaps, it's not worth the trouble. If you don't want to do that, you may
want to note that here. 

[Nebojša Ćirić, 2011/05/04 23:09:40]

I've calculated the length of the base language and cutting there, removing
extension completely.

On 2011/05/04 22:45:47, Jungshik Shin wrote:

+    locale[ULOC_LANG_CAPACITY] = '\0';
+  }
+
+  uloc_forLanguageTag(locale, result, ULOC_FULLNAME_CAPACITY,
+                      &locale_size, &status);
+  return !U_FAILURE(status);
+}
+
+// Compares locale id subtags.
+// Returns 1 for match or -1 for mismatch.
+static int CompareLocaleSubtags(const char* lsubtag, const char* rsubtag) {
+  return strcmp(lsubtag, rsubtag) == 0 ? 1 : -1;
+}
+
+// Builds a BCP47 compliant locale id from base name of matched locale and
+// full user specified locale.
+// Returns false if uloc_toLanguageTag failed to convert locale id.
+// Example:
+//   base_name of matched locale (ICU ID): de_DE
+//   input_locale_name (ICU ID): de_AT@collation=phonebk
+//   result (ICU ID): de_DE@collation=phonebk
+//   result (BCP47 ID): de-DE-u-co-phonebk
+static bool BuildLocaleName(const char* base_name,
+                            const char* input_locale_name,
+                            LocaleIDMatch* result) {
+  I18NUtils::StrNCopy(result->icu_id, ULOC_LANG_CAPACITY, base_name);
+
+  // Get extensions (if any) from the original locale.
+  const char* extension = strchr(input_locale_name, ULOC_KEYWORD_SEPARATOR);
+  if (extension != NULL) {
+    I18NUtils::StrNCopy(result->icu_id + strlen(base_name),
+                        ULOC_KEYWORD_AND_VALUES_CAPACITY, extension);
+  } else {
+    I18NUtils::StrNCopy(result->icu_id, ULOC_LANG_CAPACITY, base_name);
+  }
+
+  // Convert ICU locale name into BCP47 format.
+  UErrorCode status = U_ZERO_ERROR;
+  uloc_toLanguageTag(result->icu_id, result->bcp47_id,
+                     ULOC_FULLNAME_CAPACITY, false, &status);
+  return !U_FAILURE(status);
+}
+
+} }  // namespace v8::internal
Property changes on: src/extensions/experimental/language-matcher.cc
___________________________________________________________________
Added: svn:eol-style
   + LF