fix "IME changed" notification uses a wrong field of IME

chrome/browser/chromeos/input_method/input_method_util.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "chrome/browser/chromeos/input_method/input_method_util.h"
 
 #include <algorithm>
 #include <functional>
 #include <map>
 #include <utility>
@@ skipping 28 matching lines @@
   { "english-m", "??" },
   { "xkb:de:neo:ger", "NEO" },
   // To distinguish from "xkb:es::spa"
   { "xkb:es:cat:cat", "CAS" },
   // To distinguish from "xkb:gb::eng"
   { "xkb:gb:dvorak:eng", "DV" },
   // To distinguish from "xkb:jp::jpn"
   { "mozc", "\xe3\x81\x82" },  // U+3042, Japanese Hiragana letter A in UTF-8.
   { "mozc-dv", "\xe3\x81\x82" },
   { "mozc-jp", "\xe3\x81\x82" },
   { "zinnia-japanese", "\xe6\x89\x8b" },  // U+624B, "hand"

[Yusuke Sato, 2012/09/24 22:00:50]

not your fault at all, but please remove this line. zinnia has been removed from
Chrome OS.

[sschmitz, 2012/09/25 01:41:43]

Done.

   // For simplified Chinese input methods
   { "pinyin", "\xe6\x8b\xbc" },  // U+62FC
   { "pinyin-dv", "\xe6\x8b\xbc" },
   // For traditional Chinese input methods
   { "mozc-chewing", "\xe9\x85\xb7" },  // U+9177
   { "m17n:zh:cangjie", "\xe5\x80\x89" },  // U+5009
   { "m17n:zh:quick", "\xe9\x80\x9f" },  // U+901F
   // For Hangul input method.
   { "mozc-hangul", "\xed\x95\x9c" },  // U+D55C
 };
 
 const size_t kMappingFromIdToIndicatorTextLen =
     ARRAYSIZE_UNSAFE(kMappingFromIdToIndicatorText);
 
+// A mapping from an input method id to a medium length language indicator.
+// For those languages that want to display a slightly longer text in the
+// "Your input method has changed to..." bubble than in the status tray.
+// If an entry is not found in this table the short name is used.
+const struct {
+  const char* input_method_id;
+  const char* medium_length_name;
+} kMappingFromIdToMediumLengthName[] = {
+  { "mozc", "\xe6\x97\xa5\xe6\x9c\xac\xe8\xaa\x9e" },

[Yusuke Sato, 2012/09/24 22:00:50]

Two requests:

1. Please use chrome/app/generated_resources.grd database rather than hard-coded
UTF-8 Japanese for this kind of long string. You can use "Japanese IME" or
something similar as a default English string in .grd.

2. Please also handle Chinese and Korean IMEs here since "Your input method has
changed to 拼" is as ugly as "Your input method has changed to あ" (which is
crbug.com/137629)

pinyin, pinyin-dv --> "Simplified Chinese IME"
mozc-chewing, m17n:zh:cangjie, m17n:zh:quick --> "Traditional Chinese IME"
mozc-hangul --> "Korean IME"

[sschmitz, 2012/09/24 22:48:14]

I can do that. Who will translate? And, for each of those we really only want
one language translation besides the placeholder for English. Correct? Do we
indicate that in the  *.grd file somehow?

[Yusuke Sato, 2012/09/24 23:15:24]

L10N folks will take care of that for you:
http://dev.chromium.org/developers/design-documents/ui-localization
hmm I don't think so. For example, when the UI language is set to French, we can
(or should) show "IME japonais", but it should be "IME japonés" when it's
Spanish.

kMappingFromIdToIndicatorText[] uses hard-coded UTF-8 strings because the length
of the string is restricted (up to 4 ASCII chars) and therefore it's hard to
translate them correctly. But for kMappingFromIdToMediumLengthName[], this
doesn't seem to be the case.

[sschmitz, 2012/09/25 01:41:43]

Done.

+  { "mozc-dv", "\xe6\x97\xa5\xe6\x9c\xac\xe8\xaa\x9e" },
+  { "mozc-jp", "\xe6\x97\xa5\xe6\x9c\xac\xe8\xaa\x9e" },
+  { "zinnia-japanese", "\xe6\x97\xa5\xe6\x9c\xac\xe8\xaa\x9e" },

[Jun Mukai, 2012/09/24 21:48:32]

It would be fine for Japanese, but what about other languages?  Whom should we
ask for it?

[Daniel Erat, 2012/09/24 21:50:27]

Is this a suggestion to make all of these instead be translatable strings (along
with the short names)?  If so, that sounds reasonable to me -- seeing these in a
C++ file was a bit unexpected.

[Yusuke Sato, 2012/09/24 22:00:50]

ditto. please remove zinnia.

[Yusuke Sato, 2012/09/24 22:03:44]

+1 for translatable string for kMappingFromIdToMediumLengthName[]. Not sure if
we should do that for the short names though (sicne we can only use up to 4
ASCII characters or 2 CJK characters there).

[sschmitz, 2012/09/25 01:41:43]

I followed Yusuke's advice and left the way for short names, but use resource
IDs for medium length names. 
Done.

+};
+const size_t kMappingFromIdToMediumLengthNameLen =
+    ARRAYSIZE_UNSAFE(kMappingFromIdToMediumLengthName);
+
 string16 GetLanguageName(const std::string& language_code) {
   const string16 language_name = l10n_util::GetDisplayNameForLocale(
       language_code, g_browser_process->GetApplicationLocale(), true);
   return language_name;
 }
 
 }
 
 namespace chromeos {
 
@@ skipping 304 matching lines @@
     const std::string& input_method_id) const {
   string16 display_name;
   if (!IsExtensionInputMethod(input_method_id) &&
       TranslateStringInternal(input_method_id, &display_name)) {
     return UTF16ToUTF8(display_name);
   }
   // Return an empty string if the display name is not found.
   return "";
 }
 
+string16 InputMethodUtil::GetInputMethodMediumName(

[Daniel Erat, 2012/09/24 21:31:29]

nit: make the order here match that in the header (i.e. move medium between
short and long)

[sschmitz, 2012/09/25 01:41:43]

Done.

+    const InputMethodDescriptor& input_method) const {
+  // For the "Your input method has changed to..." bubble. In most cases
+  // it uses the same name as the short name, unless found in a table
+  // for medium length names.
+  for (size_t i = 0; i < kMappingFromIdToMediumLengthNameLen; ++i) {
+    if (kMappingFromIdToMediumLengthName[i].input_method_id ==
+        input_method.id())
+      return
+          UTF8ToUTF16(kMappingFromIdToMediumLengthName[i].medium_length_name);
+  }
+  return GetInputMethodShortName(input_method);
+}
+
 string16 InputMethodUtil::GetInputMethodShortName(
     const InputMethodDescriptor& input_method) const {
   // For the status area, we use two-letter, upper-case language code like
   // "US" and "JP".
   string16 text;
 
   // Check special cases first.
   for (size_t i = 0; i < kMappingFromIdToIndicatorTextLen; ++i) {
     if (kMappingFromIdToIndicatorText[i].input_method_id == input_method.id()) {
       text = UTF8ToUTF16(kMappingFromIdToIndicatorText[i].indicator_text);
@@ skipping 292 matching lines @@
   ReloadInternalMaps();
 }
 
 void InputMethodUtil::SetHardwareInputMethodIdForTesting(
     const std::string& input_method_id) {
   hardware_input_method_id_for_testing_ = input_method_id;
 }
 
 }  // namespace input_method
 }  // namespace chromeos