Extract input_method_whitelist.{h|cc} from c/b/chromeos.

This is part of extraction work for input method from chrome/browser/chromeos.

BUG=164375
TEST=None
TBR=sky

Review URL: https://chromiumcodereview.appspot.com/12438012

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@190596 0039d316-1c4b-4281-b951-d872f2087c98

6 files changed, 498 insertions, 0 deletions

diff --git a/chromeos/ime/gen_input_methods.py b/chromeos/ime/gen_input_methods.py
new file mode 100755
index 0000000..a9f90bf
--- /dev/null
+++ b/chromeos/ime/gen_input_methods.py
@@ -0,0 +1,123 @@
+#!/usr/bin/env python
+# 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.
+
+"""Generate a C++ header from ibus_input_methods.txt.
+
+This program generates a C++ header file containing the information on
+available input methods.  It parses input_methods.txt, and then generates a
+static array definition from the information extracted. The input and output
+file names are specified on the command line.
+
+Run it like:
+  gen_input_methods.py input_methods.txt input_methods.h
+
+It will produce output that looks like:
+
+// This file is automatically generated by gen_input_methods.py
+#ifndef CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+#define CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+
+namespace chromeos {
+namespace input_method {
+
+struct InputMethodsInfo {
+  const char* input_method_id;
+  const char* language_code;
+  const char* xkb_keyboard_id;
+};
+const InputMethodsInfo kInputMethods[] = {
+  {"mozc-chewing", "zh-TW", "us"},
+  {"xkb:us::eng", "en-US", "us"},
+  {"xkb:us:dvorak:eng", "en-US", "us(dvorak)"},
+  {"xkb:be::fra", "fr", "be"},
+  {"xkb:br::por", "pt-BR", "br"},
+};
+
+}  // namespace input_method
+}  // namespace chromeos
+
+#endif  // CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+
+"""
+
+import fileinput
+import re
+import sys
+
+OUTPUT_HEADER = """// Automatically generated by gen_input_methods.py
+#ifndef CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+#define CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+
+namespace chromeos {
+namespace input_method {
+
+struct InputMethodsInfo {
+  const char* input_method_id;
+  const char* language_code;
+  const char* xkb_layout_id;
+};
+const InputMethodsInfo kInputMethods[] = {
+"""
+
+CPP_FORMAT = '#if %s\n'
+ENGINE_FORMAT = ('  {"%(input_method_id)s", "%(language_code)s", ' +
+                 '"%(xkb_layout_id)s"},\n')
+
+OUTPUT_FOOTER = """
+};
+
+}  // namespace input_method
+}  // namespace chromeos
+
+#endif  // CHROME_BROWSER_CHROMEOS_INPUT_METHOD_INPUT_METHODS_H_
+"""
+
+def CreateEngineHeader(engines):
+  """Create the header file from a list of engines.
+
+  Arguments:
+    engines: list of engine objects
+  Returns:
+    The text of a C++ header file containing the engine data.
+  """
+  output = []
+  output.append(OUTPUT_HEADER)
+  for engine in engines:
+    if engine.has_key('if'):
+      output.append(CPP_FORMAT % engine['if'])
+    output.append(ENGINE_FORMAT % engine)
+    if engine.has_key('if'):
+      output.append('#endif\n')
+  output.append(OUTPUT_FOOTER)
+
+  return "".join(output)
+
+
+def main(argv):
+  if len(argv) != 3:
+    print 'Usage: gen_input_methods.py [whitelist] [output]'
+    sys.exit(1)
+  engines = []
+  for line in fileinput.input(sys.argv[1]):
+    line = line.strip()
+    if not line or re.match(r'#', line):
+      continue
+    columns = line.split()
+    assert len(columns) == 3 or len(columns) == 4, "Invalid format: " + line
+    engine = {}
+    engine['input_method_id'] = columns[0]
+    engine['xkb_layout_id'] = columns[1]
+    engine['language_code'] = columns[2]
+    if len(columns) == 4:
+      engine['if'] = columns[3]
+    engines.append(engine)
+
+  output = CreateEngineHeader(engines)
+  output_file = open(sys.argv[2], 'w')
+  output_file.write(output)
+
+
+if __name__ == '__main__':
+  main(sys.argv)
diff --git a/chromeos/ime/input_method.gyp b/chromeos/ime/input_method.gyp
new file mode 100644
index 0000000..1da130e
--- /dev/null
+++ b/chromeos/ime/input_method.gyp
@@ -0,0 +1,40 @@
+# 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.
+
+{
+  'variables': {
+    'input_method_out_dir':
+      '<(SHARED_INTERMEDIATE_DIR)/chromeos/ime',
+  },
+  'targets': [
+    {
+      'target_name': 'gencode',
+      'type': 'none',
+      'actions': [
+        {
+          'inputs': [
+            'input_methods.txt',
+            'gen_input_methods.py',
+          ],
+          'outputs': [
+            '<(input_method_out_dir)/input_methods.h',
+          ],
+          'action_name': 'gen_input_methods',
+          'action': [
+            'python',
+            'gen_input_methods.py',
+            'input_methods.txt',
+            '<(input_method_out_dir)/input_methods.h',
+          ],
+          'message': 'Generating input_methods.h',
+        },
+      ],
+      'direct_dependent_settings': {
+        'include_dirs': [
+          '<(SHARED_INTERMEDIATE_DIR)',
+        ],
+      },
+    },
+  ]
+}
diff --git a/chromeos/ime/input_method_whitelist.cc b/chromeos/ime/input_method_whitelist.cc
new file mode 100644
index 0000000..ba0b7b8
--- /dev/null
+++ b/chromeos/ime/input_method_whitelist.cc
@@ -0,0 +1,45 @@
+// Copyright 2013 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 "chromeos/ime/input_method_whitelist.h"
+
+#include <vector>
+
+#include "chromeos/ime/input_method_descriptor.h"
+#include "chromeos/ime/input_methods.h"
+
+namespace chromeos {
+namespace input_method {
+
+InputMethodWhitelist::InputMethodWhitelist() {
+  for (size_t i = 0; i < arraysize(kInputMethods); ++i) {
+    supported_input_methods_.insert(kInputMethods[i].input_method_id);
+  }
+}
+
+InputMethodWhitelist::~InputMethodWhitelist() {
+}
+
+bool InputMethodWhitelist::InputMethodIdIsWhitelisted(
+    const std::string& input_method_id) const {
+  return supported_input_methods_.count(input_method_id) > 0;
+}
+
+scoped_ptr<InputMethodDescriptors>
+InputMethodWhitelist::GetSupportedInputMethods() const {
+  scoped_ptr<InputMethodDescriptors> input_methods(new InputMethodDescriptors);
+  input_methods->reserve(arraysize(kInputMethods));
+  for (size_t i = 0; i < arraysize(kInputMethods); ++i) {
+    input_methods->push_back(InputMethodDescriptor(
+        kInputMethods[i].input_method_id,
+        "",
+        kInputMethods[i].xkb_layout_id,
+        kInputMethods[i].language_code,
+        false));
+  }
+  return input_methods.Pass();
+}
+
+}  // namespace input_method
+}  // namespace chromeos
diff --git a/chromeos/ime/input_method_whitelist.h b/chromeos/ime/input_method_whitelist.h
new file mode 100644
index 0000000..3a22493
--- /dev/null
+++ b/chromeos/ime/input_method_whitelist.h
@@ -0,0 +1,46 @@
+// Copyright 2013 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.
+
+#ifndef CHROMEOS_IME_INPUT_METHOD_WHITELIST_H_
+#define CHROMEOS_IME_INPUT_METHOD_WHITELIST_H_
+
+#include <set>
+#include <string>
+#include <vector>
+
+#include "base/basictypes.h"
+#include "base/memory/scoped_ptr.h"
+#include "chromeos/chromeos_export.h"
+
+namespace chromeos {
+namespace input_method {
+
+class InputMethodDescriptor;
+typedef std::vector<InputMethodDescriptor> InputMethodDescriptors;
+
+// A class which has white listed input method list. The list is generated by
+// gen_input_methods.py from input_methods.txt.
+class CHROMEOS_EXPORT InputMethodWhitelist {
+ public:
+  InputMethodWhitelist();
+  ~InputMethodWhitelist();
+
+  // Returns true if |input_method_id| is whitelisted.
+  bool InputMethodIdIsWhitelisted(const std::string& input_method_id) const;
+
+  // Returns all input methods that are supported, including ones not active.
+  // This function never returns NULL. Note that input method extensions are not
+  // included in the result.
+  scoped_ptr<InputMethodDescriptors> GetSupportedInputMethods() const;
+
+ private:
+  std::set<std::string> supported_input_methods_;
+
+  DISALLOW_COPY_AND_ASSIGN(InputMethodWhitelist);
+};
+
+}  // namespace input_method
+}  // namespace chromeos
+
+#endif  // CHROMEOS_IME_INPUT_METHOD_WHITELIST_H_
diff --git a/chromeos/ime/input_method_whitelist_unittest.cc b/chromeos/ime/input_method_whitelist_unittest.cc
new file mode 100644
index 0000000..eb4ed28
--- /dev/null
+++ b/chromeos/ime/input_method_whitelist_unittest.cc
@@ -0,0 +1,32 @@
+// 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 "base/logging.h"
+#include "chromeos/ime/input_method_whitelist.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+namespace chromeos {
+namespace input_method {
+
+namespace {
+
+class InputMethodWhitelistTest : public testing::Test {
+ protected:
+  const InputMethodWhitelist whitelist_;
+};
+
+}  // namespace
+
+TEST_F(InputMethodWhitelistTest, TestInputMethodIdIsWhitelisted) {
+  EXPECT_TRUE(whitelist_.InputMethodIdIsWhitelisted("mozc"));
+  EXPECT_TRUE(whitelist_.InputMethodIdIsWhitelisted("xkb:us:dvorak:eng"));
+  EXPECT_FALSE(whitelist_.InputMethodIdIsWhitelisted("mozc,"));
+  EXPECT_FALSE(whitelist_.InputMethodIdIsWhitelisted("mozc,xkb:us:dvorak:eng"));
+  EXPECT_FALSE(whitelist_.InputMethodIdIsWhitelisted("not-supported-id"));
+  EXPECT_FALSE(whitelist_.InputMethodIdIsWhitelisted(","));
+  EXPECT_FALSE(whitelist_.InputMethodIdIsWhitelisted(""));
+}
+
+}  // namespace input_method
+}  // namespace chromeos
diff --git a/chromeos/ime/input_methods.txt b/chromeos/ime/input_methods.txt
new file mode 100644
index 0000000..1d1d9ab
--- /dev/null
+++ b/chromeos/ime/input_methods.txt
@@ -0,0 +1,212 @@
+# The list of input methods that we support. The input methods metadata is
+# also defined here. We use this data for the following purposes.
+#
+# 1. Exclude unnecessary input methods. For example, we don't need
+#    "ja:anthy", and "zh:pinyin" engines in ibus-m17n since we have better
+#    alternatives outside of ibus-m17n. Excluding these input methods from
+#    the IBus engine XML files, such as /usr/share/ibus/component/m17n.xml,
+#    makes the startup time of the IBus daemon faster. This filtering is
+#    done using a python script at build time of ibus-m17n.
+#    See crosbug.com/4161 for details.
+#
+# 2. Provide the input methods metadata to Chrome at build time. Chrome
+#    needs to know about the supported input methods for the input method
+#    features, such as the input method switching and the options page,
+#    to work. Note that we can retrieve the same metadata from the IBus
+#    daemon, but Chrome needs be able to get the metadata without talking
+#    to the IBus daemon. This requirement comes from the fact that the
+#    IBus daemon should not run if the user is just using keyboard layouts
+#    such as US keyboard. The metadata generation is done using a python
+#    script at build time of Chrome. See crosbug.com/16961 for details.
+#
+# The order of the list is important. In Chrome's "Languages and input" options
+# page, we list input methods in this order, when more than one input methods
+# are available for a language (ex. French).
+#
+# Each non-comment line contains the following tab-separated columns.
+#
+# 1) The input method ID used by Chrome. (ex. "xkb:ca::fra") You should *NEVER*
+#    change the ID since the ID might be written in user's ~/Preferences. For
+#    example, three-letter ISO 639-2/B codes are used for IDs start with "xkb:"
+#    for histrical reason, but we should not replace them with two-letter 639-1
+#    codes that are currently in use in the 3rd column.
+# 2) The keyboard layout ID used by XKB. (ex. "us", "us(dvorak)", "ca",
+#    "handwriting-vk,jp"). See also: /usr/share/X11/xkb/symbols.
+# 3) The language code (ex. "fr"). Only one format, ISO 639-1 compliant two-
+#    letter language code which can be recognized by ICU, is allowed. Do not use
+#    three-letter ones (ISO 639-2/T and 639-2/B) here. For "en", "pt", and "zh",
+#    two-letter upper-case country code should be added (ex. "en-US", "zh-TW").
+#    See http://en.wikipedia.org/wiki/List_of_ISO_639-1_codes for details.
+# 4) [optional] The #if condition for the input method.
+#    (ex. "defined(A)", "!defined(A)||defined(B)")
+#
+# Notes:
+#   When adding a line to this list, please also add a mapping from the input
+#   method ID to the keyboard overlay ID to INPUT_METHOD_ID_TO_OVERLAY_ID in
+#
+#    * tools/gen_keyboard_overlay_data/gen_keyboard_overlay_data.py
+#
+#   and update the following files by running this script.
+#
+#    * chrome/app/generated_resources.grd
+#    * chrome/browser/resources/chromeos/keyboard_overlay_data.js
+#    * chrome/browser/ui/webui/chromeos/keyboard_overlay_ui.cc
+#
+#   If you add an XKB layout which depends on X11's Mod3Mask (e.g. Germany Neo2
+#   XKB layout), you should also update IsMod3UsedByCurrentInputMethod() method
+#   in chrome/browser/ui/ash/event_rewriter.cc. Otherwise, Mod3Mask might be
+#   removed unexpectedly by the rewriter.
+
+# U.S. English
+xkb:us::eng	us	en-US
+xkb:us:intl:eng	us(intl)	en-US	
+xkb:us:altgr-intl:eng	us(altgr-intl)	en-US	
+xkb:us:dvorak:eng	us(dvorak)	en-US	
+xkb:us:colemak:eng	us(colemak)	en-US	
+english-m	us	en-US	
+# U.S. English entiries have to be above the Dutch entry so that xkb:us:intl:eng
+# will be selected as the default keyboard when the UI language is set to Dutch.
+
+# Dutch
+xkb:be::nld	be	nl	
+# We don't support xkb:nl::nld. See b/4430951.
+
+# French
+xkb:fr::fra	fr	fr
+xkb:be::fra	be	fr	
+xkb:ca::fra	ca	fr	
+xkb:ch:fr:fra	ch(fr)	fr	
+
+# German
+xkb:de::ger	de	de	
+xkb:de:neo:ger	de(neo)	de	
+xkb:be::ger	be	de	
+xkb:ch::ger	ch	de	
+
+# Japanese
+mozc	us	ja
+mozc-jp	jp	ja	
+mozc-dv	us(dvorak)	ja	
+# |kMozcJaInputMethodIds| in ibus_ui_controller.cc should also be updated when
+# a new Mozc Japanese IME for another keyboard layout is added.
+xkb:jp::jpn	jp	ja	
+
+# Russian
+xkb:ru::rus	ru	ru	
+xkb:ru:phonetic:rus	ru(phonetic)	ru	
+
+# Thai
+m17n:th:kesmanee	us	th
+m17n:th:pattachote	us	th
+m17n:th:tis820	us	th
+
+# Simplified Chinese
+pinyin	us	zh-CN
+pinyin-dv	us(dvorak)	zh-CN	
+
+# Traditional Chinese
+mozc-chewing	us	zh-TW
+m17n:zh:cangjie	us	zh-TW
+m17n:zh:quick	us	zh-TW
+
+# Vietnamese
+m17n:vi:tcvn	us	vi
+m17n:vi:telex	us	vi
+m17n:vi:viqr	us	vi
+m17n:vi:vni	us	vi
+# Note: Since Chrome does not support "get surrounding text" feature yet,
+# Vietnames input methods, except 4 input methods above, in m17n-db should
+# not work fine. The 4 input methods in m17n-db (>= 1.6.0) don't require the
+# feature.
+
+# Korean
+mozc-hangul	kr(kr104)	ko
+xkb:kr:kr104:kor	kr(kr104)	ko
+
+# Arabic
+m17n:ar:kbd	us	ar
+# TODO(yusukes,jshin): Check if we can use ibux-xkb-layouts for Arabic, Hindi,
+# etc.
+
+# Hindi
+m17n:hi:itrans	us	hi
+# Note: the m17n-contrib package has some more Hindi IMEs.
+
+# Farsi (Persian)
+m17n:fa:isiri	us	fa
+
+# Bengali
+m17n:bn:itrans	us	bn
+# TODO(yusukes): Support the following IMEs in m17n-db (unijoy) and
+#   m17n-db-contrib (inscript, probhat).
+# m17n:bn:unijoy	us	bn
+# m17n:bn:inscript	us	bn
+# m17n:bn:probhat	us	bn
+
+# Gujarati
+m17n:gu:itrans	us	gu
+# TODO(yusukes): Support the following IMEs in m17n-db-contrib.
+# m17n:gu:phonetic	us	gu
+# m17n:gu:inscript	us	gu
+
+# Malayalam
+m17n:ml:itrans	us	ml
+# TODO(yusukes): Support the following IMEs in m17n-db-contrib.
+# m17n:ml:inscript	us	ml
+# m17n:ml:mozhi	us	ml
+
+# Marathi
+# Note: m17n:mr:itrans is in m17n-db-contrib.
+m17n:mr:itrans	us	mr
+# TODO(yusukes): Support the following IMEs in m17n-db-contrib.
+# m17n:mr:phonetic	us	mr
+# m17n:mr:inscript	us	mr
+
+# Tamil
+m17n:ta:phonetic	us	ta
+m17n:ta:inscript	us	ta
+m17n:ta:tamil99	us	ta
+m17n:ta:itrans	us	ta
+m17n:ta:typewriter	us	ta
+
+# Telugu
+m17n:te:itrans	us	te
+# Kannada
+m17n:kn:itrans	us	kn
+
+# Amharic
+m17n:am:sera	us	am
+
+# Keyboard layouts.
+xkb:br::por	br	pt-BR	
+xkb:bg::bul	bg	bg	
+xkb:bg:phonetic:bul	bg(phonetic)	bg	
+xkb:ca:eng:eng	ca(eng)	en-CA	
+xkb:cz::cze	cz	cs	
+xkb:ee::est	ee	et	
+xkb:es::spa	es	es	
+xkb:es:cat:cat	es(cat)	ca	
+xkb:dk::dan	dk	da	
+xkb:gr::gre	gr	el	
+xkb:il::heb	il	he	
+xkb:latam::spa	latam	es	
+xkb:lt::lit	lt	lt	
+xkb:lv:apostrophe:lav	lv(apostrophe)	lv	
+xkb:hr::scr	hr	hr	
+xkb:gb:extd:eng	gb(extd)	en-GB	
+xkb:gb:dvorak:eng	gb(dvorak)	en-GB	
+xkb:fi::fin	fi	fi	
+xkb:hu::hun	hu	hu	
+xkb:it::ita	it	it	
+xkb:no::nob	no	nb	
+xkb:pl::pol	pl	pl	
+xkb:pt::por	pt	pt-PT	
+xkb:ro::rum	ro	ro	
+xkb:se::swe	se	sv	
+xkb:sk::slo	sk	sk	
+xkb:si::slv	si	sl	
+xkb:rs::srp	rs	sr	
+xkb:tr::tur	tr	tr	
+xkb:ua::ukr	ua	uk	
+# TODO(yusukes): Support xkb:latam:deadtilde:spa and/or xkb:latam:nodeadkeys:spa
+# if necessary.