Updated *.pak file format to support both UTF8 and UTF16

Inserted a new field in the header that specifies which encoding is to be used for the text resources.

I also upped file format to version 4.

BUG=76281
TEST=unit_tests


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=100973

Review URL: http://codereview.chromium.org/7744017

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@101213 0039d316-1c4b-4281-b951-d872f2087c98

15 files changed, 198 insertions, 101 deletions

diff --git a/chrome/browser/autofill/autofill_regexes.cc b/chrome/browser/autofill/autofill_regexes.cc
index b1ed629..61b4fa0 100644
--- a/chrome/browser/autofill/autofill_regexes.cc
+++ b/chrome/browser/autofill/autofill_regexes.cc
@@ -81,4 +81,3 @@ bool MatchesPattern(const string16& input, const string16& pattern) {
 }
 
 }  // namespace autofill
-
diff --git a/chrome/browser/prefs/pref_service_unittest.cc b/chrome/browser/prefs/pref_service_unittest.cc
index f2f6935..c1ebd2b 100644
--- a/chrome/browser/prefs/pref_service_unittest.cc
+++ b/chrome/browser/prefs/pref_service_unittest.cc
@@ -27,31 +27,6 @@
 using testing::_;
 using testing::Mock;
 
-// TODO(port): port this test to POSIX.
-#if defined(OS_WIN)
-TEST(PrefServiceTest, LocalizedPrefs) {
-  TestingPrefService prefs;
-  const char kBoolean[] = "boolean";
-  const char kInteger[] = "integer";
-  const char kString[] = "string";
-  prefs.RegisterLocalizedBooleanPref(kBoolean, IDS_LOCALE_BOOL);
-  prefs.RegisterLocalizedIntegerPref(kInteger, IDS_LOCALE_INT);
-  prefs.RegisterLocalizedStringPref(kString, IDS_LOCALE_STRING);
-
-  // The locale default should take preference over the user default.
-  EXPECT_FALSE(prefs.GetBoolean(kBoolean));
-  EXPECT_EQ(1, prefs.GetInteger(kInteger));
-  EXPECT_EQ("hello", prefs.GetString(kString));
-
-  prefs.SetBoolean(kBoolean, true);
-  EXPECT_TRUE(prefs.GetBoolean(kBoolean));
-  prefs.SetInteger(kInteger, 5);
-  EXPECT_EQ(5, prefs.GetInteger(kInteger));
-  prefs.SetString(kString, "foo");
-  EXPECT_EQ("foo", prefs.GetString(kString));
-}
-#endif
-
 TEST(PrefServiceTest, NoObserverFire) {
   TestingPrefService prefs;
 
diff --git a/chrome/browser/themes/browser_theme_pack.cc b/chrome/browser/themes/browser_theme_pack.cc
index d348077..d050ebe 100644
--- a/chrome/browser/themes/browser_theme_pack.cc
+++ b/chrome/browser/themes/browser_theme_pack.cc
@@ -29,7 +29,7 @@ namespace {
 
 // Version number of the current theme pack. We just throw out and rebuild
 // theme packs that aren't int-equal to this.
-const int kThemePackVersion = 17;
+const int kThemePackVersion = 18;
 
 // IDs that are in the DataPack won't clash with the positive integer
 // uint16. kHeaderID should always have the maximum value because we want the
@@ -450,7 +450,7 @@ bool BrowserThemePack::WriteToDisk(FilePath path) const {
   RepackImages(prepared_images_, &reencoded_images);
   AddRawImagesTo(reencoded_images, &resources);
 
-  return ui::DataPack::WritePack(path, resources);
+  return ui::DataPack::WritePack(path, resources, ui::DataPack::BINARY);
 }
 
 bool BrowserThemePack::GetTint(int id, color_utils::HSL* hsl) const {
diff --git a/chrome_frame/simple_resource_loader.cc b/chrome_frame/simple_resource_loader.cc
index 2d1e80a..a28a62c 100644
--- a/chrome_frame/simple_resource_loader.cc
+++ b/chrome_frame/simple_resource_loader.cc
@@ -241,10 +241,14 @@ std::wstring SimpleResourceLoader::GetLocalizedResource(int message_id) {
     return std::wstring();
   }
 
-  // Data pack encodes strings as UTF16.
-  DCHECK_EQ(data.length() % 2, 0U);
-  string16 msg(reinterpret_cast<const char16*>(data.data()),
-               data.length() / 2);
+  // Data pack encodes strings as either UTF8 or UTF16.
+  string16 msg;
+  if (data_pack_->GetTextEncodingType() == ui::DataPack::UTF16) {
+    msg = string16(reinterpret_cast<const char16*>(data.data()),
+                   data.length() / 2);
+  } else if (data_pack_->GetTextEncodingType() == ui::DataPack::UTF8) {
+    msg = UTF8ToUTF16(data);
+  }
   return msg;
 }
 
diff --git a/tools/data_pack/data_pack.py b/tools/data_pack/data_pack.py
index 93a6cfd..aeeeb78 100755
--- a/tools/data_pack/data_pack.py
+++ b/tools/data_pack/data_pack.py
@@ -12,25 +12,33 @@ to point to grit scripts.
 
 import struct
 
-FILE_FORMAT_VERSION = 3
-HEADER_LENGTH = 2 * 4  # Two uint32s. (file version and number of entries)
+FILE_FORMAT_VERSION = 4
+HEADER_LENGTH = 2 * 4 + 1  # Two uint32s. (file version, number of entries) and
+                           # one uint8 (encoding of text resources)
+BINARY, UTF8, UTF16 = range(3)
 
 class WrongFileVersion(Exception):
   pass
 
+class DataPackContents:
+  def __init__(self, resources, encoding):
+    self.resources = resources
+    self.encoding = encoding
+
 def ReadDataPack(input_file):
   """Reads a data pack file and returns a dictionary."""
   data = open(input_file, "rb").read()
   original_data = data
 
   # Read the header.
-  version, num_entries = struct.unpack("<II", data[:HEADER_LENGTH])
+  version, num_entries, encoding = struct.unpack("<IIB", data[:HEADER_LENGTH])
   if version != FILE_FORMAT_VERSION:
+    print "Wrong file version in ", input_file
     raise WrongFileVersion
 
   resources = {}
   if num_entries == 0:
-    return resources
+    return DataPackContents(resources, encoding)
 
   # Read the index and data.
   data = data[HEADER_LENGTH:]
@@ -41,15 +49,15 @@ def ReadDataPack(input_file):
     next_id, next_offset = struct.unpack("<HI", data[:kIndexEntrySize])
     resources[id] = original_data[offset:next_offset]
 
-  return resources
+  return DataPackContents(resources, encoding)
 
-def WriteDataPack(resources, output_file):
+def WriteDataPack(resources, output_file, encoding):
   """Write a map of id=>data into output_file as a data pack."""
   ids = sorted(resources.keys())
   file = open(output_file, "wb")
 
   # Write file header.
-  file.write(struct.pack("<II", FILE_FORMAT_VERSION, len(ids)))
+  file.write(struct.pack("<IIB", FILE_FORMAT_VERSION, len(ids), encoding))
 
   # Each entry is a uint16 and a uint32. We have one extra entry for the last
   # item.
@@ -70,9 +78,9 @@ def WriteDataPack(resources, output_file):
 def main():
   # Just write a simple file.
   data = { 1: "", 4: "this is id 4", 6: "this is id 6", 10: "" }
-  WriteDataPack(data, "datapack1.pak")
+  WriteDataPack(data, "datapack1.pak", UTF8)
   data2 = { 1000: "test", 5: "five" }
-  WriteDataPack(data2, "datapack2.pak")
+  WriteDataPack(data2, "datapack2.pak", UTF8)
   print "wrote datapack1 and datapack2 to current directory."
 
 if __name__ == '__main__':
diff --git a/tools/data_pack/repack.py b/tools/data_pack/repack.py
index 2729b10..335fe6f 100755
--- a/tools/data_pack/repack.py
+++ b/tools/data_pack/repack.py
@@ -20,17 +20,29 @@ def RePack(output_file, input_files):
   """Write a new data pack to |output_file| based on a list of filenames
   (|input_files|)"""
   resources = {}
+  encoding = None
   for filename in input_files:
-    new_resources = data_pack.ReadDataPack(filename)
+    new_content = data_pack.ReadDataPack(filename)
 
     # Make sure we have no dups.
-    duplicate_keys = set(new_resources.keys()) & set(resources.keys())
+    duplicate_keys = set(new_content.resources.keys()) & set(resources.keys())
     if len(duplicate_keys) != 0:
       raise exceptions.KeyError("Duplicate keys: " + str(list(duplicate_keys)))
 
-    resources.update(new_resources)
+    # Make sure encoding is consistent.
+    if encoding in (None, data_pack.BINARY):
+      encoding = new_content.encoding
+    elif new_content.encoding not in (data_pack.BINARY, encoding):
+        raise exceptions.KeyError("Inconsistent encodings: " +
+                                  str(encoding) + " vs " +
+                                  str(new_content.encoding))
 
-  data_pack.WriteDataPack(resources, output_file)
+    resources.update(new_content.resources)
+
+  # Encoding is 0 for BINARY, 1 for UTF8 and 2 for UTF16
+  if encoding is None:
+    encoding = data_pack.BINARY
+  data_pack.WriteDataPack(resources, output_file, encoding)
 
 def main(argv):
   if len(argv) < 3:
diff --git a/tools/grit/grit/format/data_pack.py b/tools/grit/grit/format/data_pack.py
index 01c0c9e..87db064 100755
--- a/tools/grit/grit/format/data_pack.py
+++ b/tools/grit/grit/format/data_pack.py
@@ -19,12 +19,19 @@ from grit.node import message
 from grit.node import misc
 
 
-FILE_FORMAT_VERSION = 3
-HEADER_LENGTH = 2 * 4  # Two uint32s. (file version and number of entries)
+PACK_FILE_VERSION = 4
+HEADER_LENGTH = 2 * 4 + 1  # Two uint32s. (file version, number of entries) and
+                           # one uint8 (encoding of text resources)
+BINARY, UTF8, UTF16 = range(3)
 
 class WrongFileVersion(Exception):
   pass
 
+class DataPackContents:
+  def __init__(self, resources, encoding):
+    self.resources = resources
+    self.encoding = encoding
+
 class DataPack(interface.ItemFormatter):
   '''Writes out the data pack file format (platform agnostic resource file).'''
   def Format(self, item, lang='en', begin_item=True, output_dir='.'):
@@ -36,9 +43,9 @@ class DataPack(interface.ItemFormatter):
     nodes = DataPack.GetDataNodes(item)
     data = {}
     for node in nodes:
-      id, value = node.GetDataPackPair(lang)
+      id, value = node.GetDataPackPair(lang, UTF8)
       data[id] = value
-    return DataPack.WriteDataPackToString(data)
+    return DataPack.WriteDataPackToString(data, UTF8)
 
   @staticmethod
   def GetDataNodes(item):
@@ -63,13 +70,15 @@ class DataPack(interface.ItemFormatter):
     original_data = data
 
     # Read the header.
-    version, num_entries = struct.unpack("<II", data[:HEADER_LENGTH])
-    if version != FILE_FORMAT_VERSION:
+    version, num_entries, encoding = struct.unpack("<IIB",
+                                                   data[:HEADER_LENGTH])
+    if version != PACK_FILE_VERSION:
+      print "Wrong file version in ", input_file
       raise WrongFileVersion
 
     resources = {}
     if num_entries == 0:
-      return resources
+      return DataPackContents(resources, encoding)
 
     # Read the index and data.
     data = data[HEADER_LENGTH:]
@@ -80,18 +89,18 @@ class DataPack(interface.ItemFormatter):
       next_id, next_offset = struct.unpack("<HI", data[:kIndexEntrySize])
       resources[id] = original_data[offset:next_offset]
 
-    return resources
+    return DataPackContents(resources, encoding)
 
   @staticmethod
-  def WriteDataPackToString(resources):
+  def WriteDataPackToString(resources, encoding):
     """Write a map of id=>data into a string in the data pack format and return
     it."""
     ids = sorted(resources.keys())
     ret = []
 
     # Write file header.
-    ret.append(struct.pack("<II", FILE_FORMAT_VERSION, len(ids)))
-    HEADER_LENGTH = 2 * 4             # Two uint32s.
+    ret.append(struct.pack("<IIB", PACK_FILE_VERSION, len(ids), encoding))
+    HEADER_LENGTH = 2 * 4 + 1            # Two uint32s and one uint8.
 
     # Each entry is a uint16 + a uint32s. We have one extra entry for the last
     # item.
@@ -111,10 +120,10 @@ class DataPack(interface.ItemFormatter):
     return ''.join(ret)
 
   @staticmethod
-  def WriteDataPack(resources, output_file):
+  def WriteDataPack(resources, output_file, encoding):
     """Write a map of id=>data into output_file as a data pack."""
     file = open(output_file, "wb")
-    content = DataPack.WriteDataPackToString(resources)
+    content = DataPack.WriteDataPackToString(resources, encoding)
     file.write(content)
 
   @staticmethod
@@ -122,25 +131,37 @@ class DataPack(interface.ItemFormatter):
     """Write a new data pack to |output_file| based on a list of filenames
     (|input_files|)"""
     resources = {}
+    encoding = None
     for filename in input_files:
-      new_resources = DataPack.ReadDataPack(filename)
+      new_content = DataPack.ReadDataPack(filename)
 
-      # Make sure we have no duplicates.
-      duplicate_keys = set(new_resources.keys()) & set(resources.keys())
+      # Make sure we have no dups.
+      duplicate_keys = set(new_content.resources.keys()) & set(resources.keys())
       if len(duplicate_keys) != 0:
         raise exceptions.KeyError("Duplicate keys: " +
                                   str(list(duplicate_keys)))
 
-      resources.update(new_resources)
+      # Make sure encoding is consistent.
+      if encoding in (None, BINARY):
+        encoding = new_content.encoding
+      elif new_content.encoding not in (BINARY, encoding):
+          raise exceptions.KeyError("Inconsistent encodings: " +
+                                    str(encoding) + " vs " +
+                                    str(new_content.encoding))
+
+      resources.update(new_content.resources)
 
-    DataPack.WriteDataPack(resources, output_file)
+    # Encoding is 0 for BINARY, 1 for UTF8 and 2 for UTF16
+    if encoding is None:
+      encoding = BINARY
+    DataPack.WriteDataPack(resources, output_file, encoding)
 
 def main():
   # Just write a simple file.
   data = { 1: "", 4: "this is id 4", 6: "this is id 6", 10: "" }
-  WriteDataPack(data, "datapack1.pak")
+  DataPack.WriteDataPack(data, "datapack1.pak", UTF8)
   data2 = { 1000: "test", 5: "five" }
-  WriteDataPack(data2, "datapack2.pak")
+  DataPack.WriteDataPack(data2, "datapack2.pak", UTF8)
   print "wrote datapack1 and datapack2 to current directory."
 
 if __name__ == '__main__':
diff --git a/tools/grit/grit/format/data_pack_unittest.py b/tools/grit/grit/format/data_pack_unittest.py
index 35966639..8de54ef 100644
--- a/tools/grit/grit/format/data_pack_unittest.py
+++ b/tools/grit/grit/format/data_pack_unittest.py
@@ -16,15 +16,17 @@ from grit.format import data_pack
 class FormatDataPackUnittest(unittest.TestCase):
   def testWriteDataPack(self):
     expected = (
-        '\x03\x00\x00\x00\x04\x00\x00\x00'  # header (version, no. entries)
-        '\x01\x00\x26\x00\x00\x00'          # index entry 1
-        '\x04\x00\x26\x00\x00\x00'          # index entry 4
-        '\x06\x00\x32\x00\x00\x00'          # index entry 6
-        '\x0a\x00\x3e\x00\x00\x00'          # index entry 10
-        '\x00\x00\x3e\x00\x00\x00'          # extra entry for the size of last
+        '\x04\x00\x00\x00'                  # header(version
+        '\x04\x00\x00\x00'                  #        no. entries,
+        '\x01'                              #        encoding)
+        '\x01\x00\x27\x00\x00\x00'          # index entry 1
+        '\x04\x00\x27\x00\x00\x00'          # index entry 4
+        '\x06\x00\x33\x00\x00\x00'          # index entry 6
+        '\x0a\x00\x3f\x00\x00\x00'          # index entry 10
+        '\x00\x00\x3f\x00\x00\x00'          # extra entry for the size of last
         'this is id 4this is id 6')         # data
     input = { 1: "", 4: "this is id 4", 6: "this is id 6", 10: "" }
-    output = data_pack.DataPack.WriteDataPackToString(input)
+    output = data_pack.DataPack.WriteDataPackToString(input, data_pack.UTF8)
     self.failUnless(output == expected)
 
 
diff --git a/tools/grit/grit/node/include.py b/tools/grit/grit/node/include.py
index b073bc7..dc29315 100644
--- a/tools/grit/grit/node/include.py
+++ b/tools/grit/grit/node/include.py
@@ -74,7 +74,7 @@ class IncludeNode(base.Node):
     '''
     return self.FilenameToOpen()
 
-  def GetDataPackPair(self, lang):
+  def GetDataPackPair(self, lang, encoding):
     '''Returns a (id, string) pair that represents the resource id and raw
     bytes of the data.  This is used to generate the data pack data file.
     '''
@@ -90,6 +90,8 @@ class IncludeNode(base.Node):
       data = infile.read()
       infile.close()
 
+    # Include does not care about the encoding, because it only returns binary
+    # data.
     return id, data
 
   def Flatten(self, output_dir):
@@ -136,4 +138,3 @@ class IncludeNode(base.Node):
     node.EndParsing()
     return node
   Construct = staticmethod(Construct)
-
diff --git a/tools/grit/grit/node/message.py b/tools/grit/grit/node/message.py
index a48a645..3c5ac64 100644
--- a/tools/grit/grit/node/message.py
+++ b/tools/grit/grit/node/message.py
@@ -1,5 +1,5 @@
 #!/usr/bin/python2.4
-# Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
+# Copyright (c) 2011 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.
 
@@ -19,6 +19,7 @@ from grit import exception
 from grit import tclib
 from grit import util
 
+BINARY, UTF8, UTF16 = range(3)
 
 # Finds whitespace at the start and end of a string which can be multiline.
 _WHITESPACE = re.compile('(?P<start>\s*)(?P<body>.+?)(?P<end>\s*)\Z',
@@ -186,7 +187,7 @@ class MessageNode(base.ContentNode):
     else:
       return self.attrs['offset']
 
-  def GetDataPackPair(self, lang):
+  def GetDataPackPair(self, lang, encoding):
     '''Returns a (id, string) pair that represents the string id and the string
     in utf8.  This is used to generate the data pack data file.
     '''
@@ -199,10 +200,15 @@ class MessageNode(base.ContentNode):
       # Windows automatically translates \n to a new line, but GTK+ doesn't.
       # Manually do the conversion here rather than at run time.
       message = message.replace("\\n", "\n")
-    # |message| is a python unicode string, so convert to a utf16 byte stream
-    # because that's the format of datapacks.  We skip the first 2 bytes
-    # because it is the BOM.
-    return id, message.encode('utf16')[2:]
+    # |message| is a python unicode string, so convert to a byte stream that
+    # has the correct encoding requested for the datapacks. We skip the first
+    # 2 bytes of text resources because it is the BOM.
+    if encoding == UTF8:
+      return id, message.encode('utf8')
+    if encoding == UTF16:
+      return id, message.encode('utf16')[2:]
+    # Default is BINARY
+    return id, message
 
   # static method
   def Construct(parent, message, name, desc='', meaning='', translateable=True):
diff --git a/ui/base/resource/data_pack.cc b/ui/base/resource/data_pack.cc
index c4b1594..5a0dcf1 100644
--- a/ui/base/resource/data_pack.cc
+++ b/ui/base/resource/data_pack.cc
@@ -17,9 +17,9 @@
 
 namespace {
 
-static const uint32 kFileFormatVersion = 3;
-// Length of file header: version and entry count.
-static const size_t kHeaderLength = 2 * sizeof(uint32);
+static const uint32 kFileFormatVersion = 4;
+// Length of file header: version, entry count and text encoding type.
+static const size_t kHeaderLength = 2 * sizeof(uint32) + sizeof(uint8);
 
 #pragma pack(push,2)
 struct DataPackEntry {
@@ -60,7 +60,7 @@ enum LoadErrors {
 namespace ui {
 
 // In .cc for MemoryMappedFile dtor.
-DataPack::DataPack() : resource_count_(0) {
+DataPack::DataPack() : resource_count_(0), text_encoding_type_(BINARY) {
 }
 DataPack::~DataPack() {
 }
@@ -83,7 +83,7 @@ bool DataPack::Load(const FilePath& path) {
   }
 
   // Parse the header of the file.
-  // First uint32: version; second: resource count.
+  // First uint32: version; second: resource count;
   const uint32* ptr = reinterpret_cast<const uint32*>(mmap_->data());
   uint32 version = ptr[0];
   if (version != kFileFormatVersion) {
@@ -96,6 +96,17 @@ bool DataPack::Load(const FilePath& path) {
   }
   resource_count_ = ptr[1];
 
+  // third: text encoding.
+  const uint8* ptr_encoding = reinterpret_cast<const uint8*>(ptr + 2);
+  text_encoding_type_ = static_cast<TextEncodingType>(*ptr_encoding);
+  if (text_encoding_type_ != UTF8 && text_encoding_type_ != UTF16 &&
+      text_encoding_type_ != BINARY) {
+    LOG(ERROR) << "Bad data pack text encoding: got " << text_encoding_type_
+               << ", expected between " << BINARY << " and " << UTF16;
+    mmap_.reset();
+    return false;
+  }
+
   // Sanity check the file.
   // 1) Check we have enough entries.
   if (kHeaderLength + resource_count_ * sizeof(DataPackEntry) >
@@ -163,7 +174,8 @@ RefCountedStaticMemory* DataPack::GetStaticMemory(uint16 resource_id) const {
 
 // static
 bool DataPack::WritePack(const FilePath& path,
-                         const std::map<uint16, base::StringPiece>& resources) {
+                         const std::map<uint16, base::StringPiece>& resources,
+                         TextEncodingType textEncodingType) {
   FILE* file = file_util::OpenFile(path, "wb");
   if (!file)
     return false;
@@ -183,6 +195,21 @@ bool DataPack::WritePack(const FilePath& path,
     return false;
   }
 
+  if (textEncodingType != UTF8 && textEncodingType != UTF16 &&
+      textEncodingType != BINARY) {
+    LOG(ERROR) << "Invalid text encoding type, got " << textEncodingType
+               << ", expected between " << BINARY << " and " << UTF16;
+    file_util::CloseFile(file);
+    return false;
+  }
+
+  uint8 write_buffer = textEncodingType;
+  if (fwrite(&write_buffer, sizeof(uint8), 1, file) != 1) {
+    LOG(ERROR) << "Failed to write file text resources encoding";
+    file_util::CloseFile(file);
+    return false;
+  }
+
   // Each entry is a uint16 + a uint32. We have an extra entry after the last
   // item so we can compute the size of the list item.
   uint32 index_length = (entry_count + 1) * sizeof(DataPackEntry);
diff --git a/ui/base/resource/data_pack.h b/ui/base/resource/data_pack.h
index f6ee78d..0cc3b31 100644
--- a/ui/base/resource/data_pack.h
+++ b/ui/base/resource/data_pack.h
@@ -31,6 +31,13 @@ namespace ui {
 
 class UI_EXPORT DataPack {
  public:
+  // What type of encoding the text resources use.
+  enum TextEncodingType {
+    BINARY,
+    UTF8,
+    UTF16
+  };
+
   DataPack();
   ~DataPack();
 
@@ -47,9 +54,16 @@ class UI_EXPORT DataPack {
   // for localization strings.
   RefCountedStaticMemory* GetStaticMemory(uint16 resource_id) const;
 
-  // Writes a pack file containing |resources| to |path|.
+  // Writes a pack file containing |resources| to |path|. If there are any
+  // text resources to be written, their encoding must already agree to the
+  // |textEncodingType| specified. If no text resources are present, please
+  // indicate BINARY.
   static bool WritePack(const FilePath& path,
-                        const std::map<uint16, base::StringPiece>& resources);
+                        const std::map<uint16, base::StringPiece>& resources,
+                        TextEncodingType textEncodingType);
+
+  // Get the encoding type of text resources.
+  TextEncodingType GetTextEncodingType() const { return text_encoding_type_; }
 
  private:
   // The memory-mapped data.
@@ -58,6 +72,9 @@ class UI_EXPORT DataPack {
   // Number of resources in the data.
   size_t resource_count_;
 
+  // Type of encoding for text resources.
+  TextEncodingType text_encoding_type_;
+
   DISALLOW_COPY_AND_ASSIGN(DataPack);
 };
 
diff --git a/ui/base/resource/data_pack_literal.cc b/ui/base/resource/data_pack_literal.cc
index 4a32fa9..510d07a 100644
--- a/ui/base/resource/data_pack_literal.cc
+++ b/ui/base/resource/data_pack_literal.cc
@@ -7,13 +7,14 @@
 namespace ui {
 
 extern const char kSamplePakContents[] = {
-    0x03, 0x00, 0x00, 0x00,               // header(version
-    0x04, 0x00, 0x00, 0x00,               //        no. entries)
-    0x01, 0x00, 0x26, 0x00, 0x00, 0x00,   // index entry 1
-    0x04, 0x00, 0x26, 0x00, 0x00, 0x00,   // index entry 4
-    0x06, 0x00, 0x32, 0x00, 0x00, 0x00,   // index entry 6
-    0x0a, 0x00, 0x3e, 0x00, 0x00, 0x00,   // index entry 10
-    0x00, 0x00, 0x3e, 0x00, 0x00, 0x00,   // extra entry for the size of last
+    0x04, 0x00, 0x00, 0x00,               // header(version
+    0x04, 0x00, 0x00, 0x00,               //        no. entries
+    0x01,                                 //        encoding)
+    0x01, 0x00, 0x27, 0x00, 0x00, 0x00,   // index entry 1
+    0x04, 0x00, 0x27, 0x00, 0x00, 0x00,   // index entry 4
+    0x06, 0x00, 0x33, 0x00, 0x00, 0x00,   // index entry 6
+    0x0a, 0x00, 0x3f, 0x00, 0x00, 0x00,   // index entry 10
+    0x00, 0x00, 0x3f, 0x00, 0x00, 0x00,   // extra entry for the size of last
     't', 'h', 'i', 's', ' ', 'i', 's', ' ', 'i', 'd', ' ', '4',
     't', 'h', 'i', 's', ' ', 'i', 's', ' ', 'i', 'd', ' ', '6'
 };
diff --git a/ui/base/resource/data_pack_unittest.cc b/ui/base/resource/data_pack_unittest.cc
index fe3613c..1c0bd7a 100644
--- a/ui/base/resource/data_pack_unittest.cc
+++ b/ui/base/resource/data_pack_unittest.cc
@@ -12,6 +12,12 @@
 
 namespace ui {
 
+class DataPackTest
+    : public testing::TestWithParam<DataPack::TextEncodingType> {
+ public:
+  DataPackTest() {}
+};
+
 extern const char kSamplePakContents[];
 extern const size_t kSamplePakSize;
 
@@ -44,6 +50,13 @@ TEST(DataPackTest, Load) {
   ASSERT_FALSE(pack.GetStringPiece(140, &data));
 }
 
+INSTANTIATE_TEST_CASE_P(WriteBINARY, DataPackTest, ::testing::Values(
+    DataPack::BINARY));
+INSTANTIATE_TEST_CASE_P(WriteUTF8, DataPackTest, ::testing::Values(
+    DataPack::UTF8));
+INSTANTIATE_TEST_CASE_P(WriteUTF16, DataPackTest, ::testing::Values(
+    DataPack::UTF16));
+
 TEST(DataPackTest, LoadFileWithTruncatedHeader) {
   FilePath data_path;
   PathService::Get(base::DIR_SOURCE_ROOT, &data_path);
@@ -54,7 +67,7 @@ TEST(DataPackTest, LoadFileWithTruncatedHeader) {
   ASSERT_FALSE(pack.Load(data_path));
 }
 
-TEST(DataPackTest, Write) {
+TEST_P(DataPackTest, Write) {
   ScopedTempDir dir;
   ASSERT_TRUE(dir.CreateUniqueTempDir());
   FilePath file = dir.path().Append(FILE_PATH_LITERAL("data.pak"));
@@ -71,11 +84,12 @@ TEST(DataPackTest, Write) {
   resources.insert(std::make_pair(15, base::StringPiece(fifteen)));
   resources.insert(std::make_pair(3, base::StringPiece(three)));
   resources.insert(std::make_pair(4, base::StringPiece(four)));
-  ASSERT_TRUE(DataPack::WritePack(file, resources));
+  ASSERT_TRUE(DataPack::WritePack(file, resources, GetParam()));
 
   // Now try to read the data back in.
   DataPack pack;
   ASSERT_TRUE(pack.Load(file));
+  EXPECT_EQ(pack.GetTextEncodingType(), GetParam());
 
   base::StringPiece data;
   ASSERT_TRUE(pack.GetStringPiece(1, &data));
diff --git a/ui/base/resource/resource_bundle.cc b/ui/base/resource/resource_bundle.cc
index 7ae0f4e..f1722ff 100644
--- a/ui/base/resource/resource_bundle.cc
+++ b/ui/base/resource/resource_bundle.cc
@@ -11,6 +11,7 @@
 #include "base/stl_util.h"
 #include "base/string_piece.h"
 #include "base/synchronization/lock.h"
+#include "base/utf_string_conversions.h"
 #include "build/build_config.h"
 #include "third_party/skia/include/core/SkBitmap.h"
 #include "ui/base/l10n/l10n_util.h"
@@ -187,10 +188,19 @@ string16 ResourceBundle::GetLocalizedString(int message_id) {
     }
   }
 
-  // Data pack encodes strings as UTF16.
-  DCHECK_EQ(data.length() % 2, 0U);
-  string16 msg(reinterpret_cast<const char16*>(data.data()),
-               data.length() / 2);
+  // Strings should not be loaded from a data pack that contains binary data.
+  DCHECK(locale_resources_data_->GetTextEncodingType() == DataPack::UTF16 ||
+         locale_resources_data_->GetTextEncodingType() == DataPack::UTF8)
+      << "requested localized string from binary pack file";
+
+  // Data pack encodes strings as either UTF8 or UTF16.
+  string16 msg;
+  if (locale_resources_data_->GetTextEncodingType() == DataPack::UTF16) {
+    msg = string16(reinterpret_cast<const char16*>(data.data()),
+                   data.length() / 2);
+  } else if (locale_resources_data_->GetTextEncodingType() == DataPack::UTF8) {
+    msg = UTF8ToUTF16(data);
+  }
   return msg;
 }