Change v8Locale to match proposal - constructor is different and I've added

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 +1,210 @@
+// 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 "unicode/locid.h"
+#include "unicode/uloc.h"
+#include "utils.h"  // For Min/Max
+
+namespace v8 {
+namespace internal {
+
+const unsigned int LanguageMatcher::kLanguageDistance = 75;

[jungshik at Google, 2011/05/02 19:57:23]

nit: again, this is not a distance but a score/weight or inverse-distance. 

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+const unsigned int LanguageMatcher::kScriptDistance = 20;
+const unsigned int LanguageMatcher::kRegionDistance = 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 void BCP47ToICUFormat(v8::Handle<v8::String>, char*);
+static int CompareLocaleSubtags(const char*, const char*);
+static void BuildLocaleName(const char*, const char*, LocaleIDMatch*);
+
+LocaleIDMatch::LocaleIDMatch()
+    : rank(-1) {
+  snprintf(bcp47_locale, ULOC_FULLNAME_CAPACITY,
+           "%s", LanguageMatcher::kDefaultLocale);
+  snprintf(icu_locale, ULOC_FULLNAME_CAPACITY,
+           "%s", LanguageMatcher::kDefaultLocale);
+}
+
+LocaleIDMatch& LocaleIDMatch::operator=(const LocaleIDMatch& rhs) {
+  snprintf(this->bcp47_locale, ULOC_FULLNAME_CAPACITY,
+           "%s", rhs.bcp47_locale);
+  snprintf(this->icu_locale, ULOC_FULLNAME_CAPACITY,
+           "%s", rhs.icu_locale);
+  this->rank = rhs.rank;
+}
+
+// static
+void LanguageMatcher::GetBestMatchForPriorityList(
+    v8::Handle<v8::Array> locales, LocaleIDMatch* result) {
+  v8::HandleScope handle_scope;
+
+  unsigned int position_bonus = locales->Length() * kPositionBonus;
+
+  int max_rank = 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;
+
+    CompareToSupportedLocaleIDList(locale_id->ToString(), &match);
+
+    // Skip items under threshold.
+    if (match.rank < kThreshold) continue;
+
+    match.rank += position_bonus;
+    if (match.rank > max_rank) {
+      *result = match;
+
+      max_rank = match.rank;
+    }
+  }
+}
+
+// static
+void LanguageMatcher::GetBestMatchForString(
+    v8::Handle<v8::String> locale, LocaleIDMatch* result) {
+  LocaleIDMatch match;
+  CompareToSupportedLocaleIDList(locale, &match);
+
+  if (match.rank >= kThreshold) {
+    *result = match;
+  }
+}
+
+// static
+void LanguageMatcher::CompareToSupportedLocaleIDList(
+    v8::Handle<v8::String> locale_id, LocaleIDMatch* result) {
+  static int32_t available_count = 0;
+  static const icu::Locale* available_locales =
+      icu::Locale::getAvailableLocales(available_count);

[jungshik at Google, 2011/05/02 19:57:23]

Don't we need any lock around here?

More importantly, what's returned by getAvailableLocale() can be deceptive (at
least in Chrome) because the return value will include 70+ locales for which ICU
data (in Chrome) has only nominal data (the native name, the script and writing
direction of the primary language of a locale) and nothing else. 

See
http://www.google.com/codesearch/p?hl=en#hfE6470xZHk/app/l10n_util.cc&q=IsLoc...
  for a hackish way of detecting this. 

An alternative is to call getAvailableLocale for DateTimeFormatter and use it
instead. (
http://icu-project.org/apiref/icu4c/classDateFormat.html#a5934ceffff9eafdd3ed...
). 

[Nebojša Ćirić, 2011/05/02 22:44:01]

Added a TODO, but I don't think it's an issue. One of the calls would win, and
they all return the same information, so there is no difference which one wins.

As for DateFormat, I agree. Done.

On 2011/05/02 19:57:23, Jungshik Shin wrote:
Done.

+
+  char locale[ULOC_FULLNAME_CAPACITY];
+  BCP47ToICUFormat(locale_id, locale);
+  icu::Locale user_locale(locale);

[jungshik at Google, 2011/05/02 19:57:23]

how about input_locale? 

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+
+  // Position of the best match locale in list of available locales.
+  int position = -1;
+  result->rank = 0;
+  const char* language = GetLanguageException(user_locale.getLanguage());
+  const char* script = user_locale.getScript();
+  const char* region = user_locale.getCountry();
+  for (int32_t i = 0; i < available_count; ++i) {
+    int current_rank = 0;
+    int sign =
+        CompareLocaleSubtags(language, available_locales[i].getLanguage());
+    current_rank += sign * kLanguageDistance;
+
+    sign = CompareLocaleSubtags(script, available_locales[i].getScript());
+    current_rank += sign * kScriptDistance;
+
+    sign = CompareLocaleSubtags(region, available_locales[i].getCountry());
+    current_rank += sign * kRegionDistance;
+
+    if (current_rank > result->rank) {
+      result->rank = current_rank;
+      position = i;
+    }
+  }
+
+  if (result->rank < kThreshold || position == -1) {
+    LocaleIDMatch default_match;
+    *result = default_match;
+    return;
+  }
+
+  BuildLocaleName(available_locales[position].getBaseName(),
+                  user_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) {
+  if (!strcmp(language, "sh")) return "sr";
+  if (!strcmp(language, "no")) return "nb";
+  if (!strcmp(language, "mo")) return "ro";
+  if (!strcmp(language, "tl")) return "fil";

[jungshik at Google, 2011/05/02 19:57:23]

Comment these entries? e.g. Tagalog => Filipino

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+
+  return language;
+}
+
+// Converts user input from BCP47 locale id format to ICU compatible format.
+static void BCP47ToICUFormat(v8::Handle<v8::String> locale_id, char* locale) {
+  UErrorCode status = U_ZERO_ERROR;
+  int32_t locale_size = 0;
+  uloc_forLanguageTag(*v8::String::Utf8Value(locale_id), locale,

[jungshik at Google, 2011/05/02 19:57:23]

ASCIIValue?

[Nebojša Ćirić, 2011/05/02 22:44:01]

Already done in parent method.

On 2011/05/02 19:57:23, Jungshik Shin wrote:

+                      ULOC_FULLNAME_CAPACITY, &locale_size, &status);
+}
+
+// Compares locale id subtags.
+// Returns 1 for match or-1 for mismatch.

[jungshik at Google, 2011/05/02 19:57:23]

nit: "or-1" -> "or -1"

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+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.
+// Example:
+//   base_name: de_DE

[jungshik at Google, 2011/05/02 19:57:23]

base_name of the matched locale (ICU ID): de_DE

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+//   user_locale_name: de_AT@collation=phonebk

[jungshik at Google, 2011/05/02 19:57:23]

input_locale_name (ICU ID)

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+//   ICU compatible result: de_DE@collation=phonebk

[jungshik at Google, 2011/05/02 19:57:23]

result (ICU ID)

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+//   BCP47 compatible result: de-DE-u-co-phonebk

[jungshik at Google, 2011/05/02 19:57:23]

result (BCP 47)

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+static void BuildLocaleName(const char* base_name,
+                            const char* user_locale_name,
+                            LocaleIDMatch* result) {
+  // Get extensions (if any) from the original locale.
+  const char* extension = strchr(user_locale_name, ULOC_KEYWORD_SEPARATOR);
+  if (extension != NULL) {
+    snprintf(result->icu_locale, ULOC_FULLNAME_CAPACITY,
+             "%s%s", base_name, extension);
+  } else {
+    snprintf(result->icu_locale, ULOC_FULLNAME_CAPACITY, "%s", base_name);
+  }
+
+  // Convert ICU locale name into BCP47 format.
+  UErrorCode status = U_ZERO_ERROR;
+  uloc_toLanguageTag(result->icu_locale, result->bcp47_locale,
+                     ULOC_FULLNAME_CAPACITY, false, &status);

[jungshik at Google, 2011/05/02 19:57:23]

Instead of ignoring status, either assert U_SUCCESS() or return the result in a
boolean value so that a caller can do something about it.

[Nebojša Ćirić, 2011/05/02 22:44:01]

Done.

+}
+
+} }  // namespace v8::internal
Property changes on: src/extensions/experimental/language-matcher.cc
___________________________________________________________________
Added: svn:eol-style
   + LF