Merge Chromium at 10.0.621.0: Initial merge by git.

Change-Id: I070cc91c608dfa4a968a5a54c173260765ac8097

12 files changed, 628 insertions, 327 deletions

diff --git a/base/i18n/break_iterator.cc b/base/i18n/break_iterator.cc
new file mode 100644
index 0000000..e1b5e29
--- /dev/null
+++ b/base/i18n/break_iterator.cc
@@ -0,0 +1,101 @@
+// Copyright (c) 2009 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/i18n/break_iterator.h"
+
+#include "base/logging.h"
+#include "unicode/ubrk.h"
+#include "unicode/uchar.h"
+#include "unicode/ustring.h"
+
+namespace base {
+
+const size_t npos = -1;
+
+BreakIterator::BreakIterator(const string16* str, BreakType break_type)
+    : iter_(NULL),
+      string_(str),
+      break_type_(break_type),
+      prev_(npos),
+      pos_(0) {
+}
+
+BreakIterator::~BreakIterator() {
+  if (iter_)
+    ubrk_close(static_cast<UBreakIterator*>(iter_));
+}
+
+bool BreakIterator::Init() {
+  UErrorCode status = U_ZERO_ERROR;
+  UBreakIteratorType break_type;
+  switch (break_type_) {
+    case BREAK_WORD:
+      break_type = UBRK_WORD;
+      break;
+    case BREAK_SPACE:
+    case BREAK_NEWLINE:
+      break_type = UBRK_LINE;
+      break;
+    default:
+      NOTREACHED() << "invalid break_type_";
+      return false;
+  }
+  iter_ = ubrk_open(break_type, NULL,
+                    string_->data(), static_cast<int32_t>(string_->size()),
+                    &status);
+  if (U_FAILURE(status)) {
+    NOTREACHED() << "ubrk_open failed";
+    return false;
+  }
+  // Move the iterator to the beginning of the string.
+  ubrk_first(static_cast<UBreakIterator*>(iter_));
+  return true;
+}
+
+bool BreakIterator::Advance() {
+  int32_t pos;
+  int32_t status;
+  prev_ = pos_;
+  switch (break_type_) {
+    case BREAK_WORD:
+    case BREAK_SPACE:
+      pos = ubrk_next(static_cast<UBreakIterator*>(iter_));
+      if (pos == UBRK_DONE) {
+        pos_ = npos;
+        return false;
+      }
+      pos_ = static_cast<size_t>(pos);
+      return true;
+    case BREAK_NEWLINE:
+      do {
+        pos = ubrk_next(static_cast<UBreakIterator*>(iter_));
+        if (pos == UBRK_DONE) {
+          break;
+        }
+        pos_ = static_cast<size_t>(pos);
+        status = ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_));
+      } while (status >= UBRK_LINE_SOFT && status < UBRK_LINE_SOFT_LIMIT);
+      if (pos == UBRK_DONE && prev_ == pos_) {
+        pos_ = npos;
+        return false;
+      }
+      return true;
+    default:
+      NOTREACHED() << "invalid break_type_";
+      return false;
+  }
+}
+
+bool BreakIterator::IsWord() const {
+  return (break_type_ == BREAK_WORD &&
+          ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_)) !=
+          UBRK_WORD_NONE);
+}
+
+string16 BreakIterator::GetString() const {
+  DCHECK(prev_ != npos && pos_ != npos);
+  return string_->substr(prev_, pos_ - prev_);
+}
+
+}  // namespace base
diff --git a/base/i18n/break_iterator.h b/base/i18n/break_iterator.h
new file mode 100644
index 0000000..9de7ac7
--- /dev/null
+++ b/base/i18n/break_iterator.h
@@ -0,0 +1,108 @@
+// Copyright (c) 2010 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 BASE_I18N_BREAK_ITERATOR_H_
+#define BASE_I18N_BREAK_ITERATOR_H_
+#pragma once
+
+#include "base/basictypes.h"
+#include "base/string16.h"
+
+// The BreakIterator class iterates through the words, word breaks, and
+// line breaks in a UTF-16 string.
+//
+// It provides several modes, BREAK_WORD, BREAK_SPACE, and BREAK_NEWLINE,
+// which modify how characters are aggregated into the returned string.
+//
+// Under BREAK_WORD mode, once a word is encountered any non-word
+// characters are not included in the returned string (e.g. in the
+// UTF-16 equivalent of the string " foo bar! ", the word breaks are at
+// the periods in ". .foo. .bar.!. .").
+//
+// Under BREAK_SPACE mode, once a word is encountered, any non-word
+// characters are included in the returned string, breaking only when a
+// space-equivalent character is encountered (e.g. in the
+// UTF16-equivalent of the string " foo bar! ", the word breaks are at
+//   the periods in ". .foo .bar! .").
+//
+// Under BREAK_NEWLINE mode, all characters are included in the returned
+// string, breking only when a newline-equivalent character is encountered
+// (eg. in the UTF-16 equivalent of the string "foo\nbar!\n\n", the line
+// breaks are at the periods in ".foo\n.bar\n.\n.").
+//
+// To extract the words from a string, move a BREAK_WORD BreakIterator
+// through the string and test whether IsWord() is true.  E.g.,
+//   BreakIterator iter(&str, BreakIterator::BREAK_WORD);
+//   if (!iter.Init()) return false;
+//   while (iter.Advance()) {
+//     if (iter.IsWord()) {
+//       // region [iter.prev(),iter.pos()) contains a word.
+//       VLOG(1) << "word: " << iter.GetString();
+//     }
+//   }
+
+namespace base {
+
+class BreakIterator {
+ public:
+  enum BreakType {
+    BREAK_WORD,
+    BREAK_SPACE,
+    BREAK_NEWLINE,
+  };
+
+  // Requires |str| to live as long as the BreakIterator does.
+  BreakIterator(const string16* str, BreakType break_type);
+  ~BreakIterator();
+
+  // Init() must be called before any of the iterators are valid.
+  // Returns false if ICU failed to initialize.
+  bool Init();
+
+  // Return the current break position within the string,
+  // or BreakIterator::npos when done.
+  size_t pos() const { return pos_; }
+
+  // Return the value of pos() returned before Advance() was last called.
+  size_t prev() const { return prev_; }
+
+  // Advance to the next break.  Returns false if we've run past the end of
+  // the string.  (Note that the very last "break" is after the final
+  // character in the string, and when we advance to that position it's the
+  // last time Advance() returns true.)
+  bool Advance();
+
+  // Under BREAK_WORD mode, returns true if the break we just hit is the
+  // end of a word. (Otherwise, the break iterator just skipped over e.g.
+  // whitespace or punctuation.)  Under BREAK_SPACE and BREAK_NEWLINE modes,
+  // this distinction doesn't apply and it always retuns false.
+  bool IsWord() const;
+
+  // Return the string between prev() and pos().
+  // Advance() must have been called successfully at least once
+  // for pos() to have advanced to somewhere useful.
+  string16 GetString() const;
+
+ private:
+  // ICU iterator, avoiding ICU ubrk.h dependence.
+  // This is actually an ICU UBreakiterator* type, which turns out to be
+  // a typedef for a void* in the ICU headers. Using void* directly prevents
+  // callers from needing access to the ICU public headers directory.
+  void* iter_;
+
+  // The string we're iterating over.
+  const string16* string_;
+
+  // The breaking style (word/space/newline).
+  BreakType break_type_;
+
+  // Previous and current iterator positions.
+  size_t prev_, pos_;
+
+  DISALLOW_COPY_AND_ASSIGN(BreakIterator);
+};
+
+}  // namespace base
+
+#endif  // BASE_I18N_BREAK_ITERATOR_H__
diff --git a/base/i18n/break_iterator_unittest.cc b/base/i18n/break_iterator_unittest.cc
new file mode 100644
index 0000000..bf4fdc1
--- /dev/null
+++ b/base/i18n/break_iterator_unittest.cc
@@ -0,0 +1,308 @@
+// Copyright (c) 2010 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/i18n/break_iterator.h"
+
+#include "base/string_piece.h"
+#include "base/string_util.h"
+#include "base/utf_string_conversions.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+TEST(BreakIteratorTest, BreakWordEmpty) {
+  string16 empty;
+  base::BreakIterator iter(&empty, base::BreakIterator::BREAK_WORD);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakWord) {
+  string16 space(UTF8ToUTF16(" "));
+  string16 str(UTF8ToUTF16(" foo bar! \npouet boom"));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(space, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("foo"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(space, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("bar"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("!"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(space, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("\n"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("pouet"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(space, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakWide16) {
+  // Two greek words separated by space.
+  const string16 str(WideToUTF16(
+      L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
+      L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2"));
+  const string16 word1(str.substr(0, 10));
+  const string16 word2(str.substr(11, 5));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(word1, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(word2, iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakWide32) {
+  // U+1D49C MATHEMATICAL SCRIPT CAPITAL A
+  const char* very_wide_char = "\xF0\x9D\x92\x9C";
+  const string16 str(
+      UTF8ToUTF16(StringPrintf("%s a", very_wide_char)));
+  const string16 very_wide_word(str.substr(0, 2));
+
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(very_wide_word, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_TRUE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("a"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakSpaceEmpty) {
+  string16 empty;
+  base::BreakIterator iter(&empty, base::BreakIterator::BREAK_SPACE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakSpace) {
+  string16 str(UTF8ToUTF16(" foo bar! \npouet boom"));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_SPACE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("foo "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("bar! \n"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("pouet "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakSpaceSP) {
+  string16 str(UTF8ToUTF16(" foo bar! \npouet boom "));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_SPACE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("foo "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("bar! \n"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("pouet "), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("boom "), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakSpacekWide16) {
+  // Two Greek words.
+  const string16 str(WideToUTF16(
+      L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
+      L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2"));
+  const string16 word1(str.substr(0, 11));
+  const string16 word2(str.substr(11, 5));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_SPACE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(word1, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(word2, iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakSpaceWide32) {
+  // U+1D49C MATHEMATICAL SCRIPT CAPITAL A
+  const char* very_wide_char = "\xF0\x9D\x92\x9C";
+  const string16 str(
+      UTF8ToUTF16(StringPrintf("%s a", very_wide_char)));
+  const string16 very_wide_word(str.substr(0, 3));
+
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_SPACE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(very_wide_word, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("a"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());  // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakLineEmpty) {
+  string16 empty;
+  base::BreakIterator iter(&empty, base::BreakIterator::BREAK_NEWLINE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());   // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakLine) {
+  string16 nl(UTF8ToUTF16("\n"));
+  string16 str(UTF8ToUTF16("\nfoo bar!\n\npouet boom"));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_NEWLINE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(nl, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("foo bar!\n"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(nl, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("pouet boom"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());   // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakLineNL) {
+  string16 nl(UTF8ToUTF16("\n"));
+  string16 str(UTF8ToUTF16("\nfoo bar!\n\npouet boom\n"));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_NEWLINE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(nl, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("foo bar!\n"), iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(nl, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("pouet boom\n"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());   // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakLineWide16) {
+  // Two Greek words separated by newline.
+  const string16 str(WideToUTF16(
+      L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
+      L"\x03bf\x03c2\x000a\x0399\x03c3\x03c4\x03cc\x03c2"));
+  const string16 line1(str.substr(0, 11));
+  const string16 line2(str.substr(11, 5));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_NEWLINE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(line1, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(line2, iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());   // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
+
+TEST(BreakIteratorTest, BreakLineWide32) {
+  // U+1D49C MATHEMATICAL SCRIPT CAPITAL A
+  const char* very_wide_char = "\xF0\x9D\x92\x9C";
+  const string16 str(
+      UTF8ToUTF16(StringPrintf("%s\na", very_wide_char)));
+  const string16 very_wide_line(str.substr(0, 3));
+  base::BreakIterator iter(&str, base::BreakIterator::BREAK_NEWLINE);
+  ASSERT_TRUE(iter.Init());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(very_wide_line, iter.GetString());
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_EQ(UTF8ToUTF16("a"), iter.GetString());
+  EXPECT_FALSE(iter.Advance());
+  EXPECT_FALSE(iter.IsWord());
+  EXPECT_FALSE(iter.Advance());   // Test unexpected advance after end.
+  EXPECT_FALSE(iter.IsWord());
+}
diff --git a/base/i18n/file_util_icu.cc b/base/i18n/file_util_icu.cc
index 0e9c2cd..34eefac 100644
--- a/base/i18n/file_util_icu.cc
+++ b/base/i18n/file_util_icu.cc
@@ -21,6 +21,10 @@ namespace {
 
 class IllegalCharacters {
  public:
+  static IllegalCharacters* GetInstance() {
+    return Singleton<IllegalCharacters>::get();
+  }
+
   bool contains(UChar32 ucs4) {
     return !!set->contains(ucs4);
   }
@@ -76,19 +80,8 @@ IllegalCharacters::IllegalCharacters() {
 
 class LocaleAwareComparator {
  public:
-  LocaleAwareComparator() {
-    UErrorCode error_code = U_ZERO_ERROR;
-    // Use the default collator. The default locale should have been properly
-    // set by the time this constructor is called.
-    collator_.reset(icu::Collator::createInstance(error_code));
-    DCHECK(U_SUCCESS(error_code));
-    // Make it case-sensitive.
-    collator_->setStrength(icu::Collator::TERTIARY);
-    // Note: We do not set UCOL_NORMALIZATION_MODE attribute. In other words, we
-    // do not pay performance penalty to guarantee sort order correctness for
-    // non-FCD (http://unicode.org/notes/tn5/#FCD) file names. This should be a
-    // reasonable tradeoff because such file names should be rare and the sort
-    // order doesn't change much anyway.
+  static LocaleAwareComparator* GetInstance() {
+    return Singleton<LocaleAwareComparator>::get();
   }
 
   // Note: A similar function is available in l10n_util.
@@ -111,6 +104,21 @@ class LocaleAwareComparator {
   }
 
  private:
+  LocaleAwareComparator() {
+    UErrorCode error_code = U_ZERO_ERROR;
+    // Use the default collator. The default locale should have been properly
+    // set by the time this constructor is called.
+    collator_.reset(icu::Collator::createInstance(error_code));
+    DCHECK(U_SUCCESS(error_code));
+    // Make it case-sensitive.
+    collator_->setStrength(icu::Collator::TERTIARY);
+    // Note: We do not set UCOL_NORMALIZATION_MODE attribute. In other words, we
+    // do not pay performance penalty to guarantee sort order correctness for
+    // non-FCD (http://unicode.org/notes/tn5/#FCD) file names. This should be a
+    // reasonable tradeoff because such file names should be rare and the sort
+    // order doesn't change much anyway.
+  }
+
   scoped_ptr<icu::Collator> collator_;
   Lock lock_;
   friend struct DefaultSingletonTraits<LocaleAwareComparator>;
@@ -123,19 +131,19 @@ class LocaleAwareComparator {
 namespace file_util {
 
 bool IsFilenameLegal(const string16& file_name) {
-  return Singleton<IllegalCharacters>()->containsNone(file_name);
+  return IllegalCharacters::GetInstance()->containsNone(file_name);
 }
 
 void ReplaceIllegalCharactersInPath(FilePath::StringType* file_name,
                                     char replace_char) {
   DCHECK(file_name);
 
-  DCHECK(!(Singleton<IllegalCharacters>()->contains(replace_char)));
+  DCHECK(!(IllegalCharacters::GetInstance()->contains(replace_char)));
 
   // Remove leading and trailing whitespace.
   TrimWhitespace(*file_name, TRIM_ALL, file_name);
 
-  IllegalCharacters* illegal = Singleton<IllegalCharacters>::get();
+  IllegalCharacters* illegal = IllegalCharacters::GetInstance();
   int cursor = 0;  // The ICU macros expect an int.
   while (cursor < static_cast<int>(file_name->size())) {
     int char_begin = cursor;
@@ -171,8 +179,8 @@ void ReplaceIllegalCharactersInPath(FilePath::StringType* file_name,
 
 bool LocaleAwareCompareFilenames(const FilePath& a, const FilePath& b) {
 #if defined(OS_WIN)
-  return Singleton<LocaleAwareComparator>()->Compare(a.value().c_str(),
-                                                     b.value().c_str()) < 0;
+  return LocaleAwareComparator::GetInstance()->Compare(a.value().c_str(),
+                                                       b.value().c_str()) < 0;
 
 #elif defined(OS_POSIX)
   // On linux, the file system encoding is not defined. We assume
@@ -181,7 +189,7 @@ bool LocaleAwareCompareFilenames(const FilePath& a, const FilePath& b) {
   // ICU's collator can take strings in OS native encoding. But we convert the
   // strings to UTF-16 ourselves to ensure conversion consistency.
   // TODO(yuzo): Perhaps we should define SysNativeMBToUTF16?
-  return Singleton<LocaleAwareComparator>()->Compare(
+  return LocaleAwareComparator::GetInstance()->Compare(
       WideToUTF16(base::SysNativeMBToWide(a.value().c_str())),
       WideToUTF16(base::SysNativeMBToWide(b.value().c_str()))) < 0;
 #else
diff --git a/base/i18n/number_formatting.cc b/base/i18n/number_formatting.cc
index 7a69294..df6af14 100644
--- a/base/i18n/number_formatting.cc
+++ b/base/i18n/number_formatting.cc
@@ -6,7 +6,8 @@
 
 #include "base/format_macros.h"
 #include "base/logging.h"
-#include "base/singleton.h"
+#include "base/lazy_instance.h"
+#include "base/scoped_ptr.h"
 #include "base/string_util.h"
 #include "base/utf_string_conversions.h"
 #include "unicode/numfmt.h"
@@ -16,25 +17,26 @@ namespace base {
 
 namespace {
 
-struct NumberFormatSingletonTraits
-    : public DefaultSingletonTraits<icu::NumberFormat> {
-  static icu::NumberFormat* New() {
+struct NumberFormatWrapper {
+  NumberFormatWrapper() {
+    // There's no ICU call to destroy a NumberFormat object other than
+    // operator delete, so use the default Delete, which calls operator delete.
+    // This can cause problems if a different allocator is used by this file
+    // than by ICU.
     UErrorCode status = U_ZERO_ERROR;
-    icu::NumberFormat* formatter = icu::NumberFormat::createInstance(status);
+    number_format.reset(icu::NumberFormat::createInstance(status));
     DCHECK(U_SUCCESS(status));
-    return formatter;
   }
-  // There's no ICU call to destroy a NumberFormat object other than
-  // operator delete, so use the default Delete, which calls operator delete.
-  // This can cause problems if a different allocator is used by this file than
-  // by ICU.
+
+  scoped_ptr<icu::NumberFormat> number_format;
 };
 
 }  // namespace
 
+static LazyInstance<NumberFormatWrapper> g_number_format(LINKER_INITIALIZED);
+
 string16 FormatNumber(int64 number) {
-  icu::NumberFormat* number_format =
-      Singleton<icu::NumberFormat, NumberFormatSingletonTraits>::get();
+  icu::NumberFormat* number_format = g_number_format.Get().number_format.get();
 
   if (!number_format) {
     // As a fallback, just return the raw number in a string.
diff --git a/base/i18n/rtl.cc b/base/i18n/rtl.cc
index 6a5d293..12b376d 100644
--- a/base/i18n/rtl.cc
+++ b/base/i18n/rtl.cc
@@ -163,6 +163,7 @@ TextDirection GetFirstStrongCharacterDirection(const std::wstring& text) {
 }
 #endif
 
+#if defined(OS_WIN)
 bool AdjustStringForLocaleDirection(string16* text) {
   if (!IsRTL() || text->empty())
     return false;
@@ -177,6 +178,57 @@ bool AdjustStringForLocaleDirection(string16* text) {
 
   return true;
 }
+#else
+bool AdjustStringForLocaleDirection(string16* text) {
+  // On OS X & GTK the directionality of a label is determined by the first
+  // strongly directional character.
+  // However, we want to make sure that in an LTR-language-UI all strings are
+  // left aligned and vice versa.
+  // A problem can arise if we display a string which starts with user input.
+  // User input may be of the opposite directionality to the UI. So the whole
+  // string will be displayed in the opposite directionality, e.g. if we want to
+  // display in an LTR UI [such as US English]:
+  //
+  // EMAN_NOISNETXE is now installed.
+  //
+  // Since EXTENSION_NAME begins with a strong RTL char, the label's
+  // directionality will be set to RTL and the string will be displayed visually
+  // as:
+  //
+  // .is now installed EMAN_NOISNETXE
+  //
+  // In order to solve this issue, we prepend an LRM to the string. An LRM is a
+  // strongly directional LTR char.
+  // We also append an LRM at the end, which ensures that we're in an LTR
+  // context.
+
+  // Unlike Windows, Linux and OS X can correctly display RTL glyphs out of the
+  // box so there is no issue with displaying zero-width bidi control characters
+  // on any system.  Thus no need for the !IsRTL() check here.
+  if (text->empty())
+    return false;
+
+  bool ui_direction_is_rtl = IsRTL();
+
+  bool has_rtl_chars = StringContainsStrongRTLChars(*text);
+  if (!ui_direction_is_rtl && has_rtl_chars) {
+    WrapStringWithRTLFormatting(text);
+    text->insert(0, 1, kLeftToRightMark);
+    text->push_back(kLeftToRightMark);
+  } else if (ui_direction_is_rtl && has_rtl_chars) {
+    WrapStringWithRTLFormatting(text);
+    text->insert(0, 1, kRightToLeftMark);
+    text->push_back(kRightToLeftMark);
+  } else if (ui_direction_is_rtl) {
+    WrapStringWithLTRFormatting(text);
+    text->insert(0, 1, kRightToLeftMark);
+    text->push_back(kRightToLeftMark);
+  }
+
+  return true;
+}
+
+#endif  // !OS_WIN
 
 #if defined(WCHAR_T_IS_UTF32)
 bool AdjustStringForLocaleDirection(std::wstring* text) {
diff --git a/base/i18n/rtl.h b/base/i18n/rtl.h
index 82ac576..a75ed4f 100644
--- a/base/i18n/rtl.h
+++ b/base/i18n/rtl.h
@@ -84,6 +84,7 @@ TextDirection GetFirstStrongCharacterDirection(const std::wstring& text);
 // string is always treated as a right-to-left string. This is done by
 // inserting certain Unicode formatting marks into the returned string.
 //
+// ** Notes about the Windows version of this function:
 // TODO(idana) bug 6806: this function adjusts the string in question only
 // if the current locale is right-to-left. The function does not take care of
 // the opposite case (an RTL string displayed in an LTR context) since
diff --git a/base/i18n/time_formatting.cc b/base/i18n/time_formatting.cc
index 406145d..3fa984a 100644
--- a/base/i18n/time_formatting.cc
+++ b/base/i18n/time_formatting.cc
@@ -14,24 +14,21 @@ using base::Time;
 
 namespace {
 
-std::wstring TimeFormat(const icu::DateFormat* formatter,
-                        const Time& time) {
+string16 TimeFormat(const icu::DateFormat* formatter,
+                    const Time& time) {
   DCHECK(formatter);
   icu::UnicodeString date_string;
 
   formatter->format(static_cast<UDate>(time.ToDoubleT() * 1000), date_string);
-  std::wstring output;
-  bool success = UTF16ToWide(date_string.getBuffer(), date_string.length(),
-      &output);
-  DCHECK(success);
-  return output;
+  return string16(date_string.getBuffer(),
+                  static_cast<size_t>(date_string.length()));
 }
 
 }  // namespace
 
 namespace base {
 
-std::wstring TimeFormatTimeOfDay(const Time& time) {
+string16 TimeFormatTimeOfDay(const Time& time) {
   // We can omit the locale parameter because the default should match
   // Chrome's application locale.
   scoped_ptr<icu::DateFormat> formatter(
@@ -39,31 +36,31 @@ std::wstring TimeFormatTimeOfDay(const Time& time) {
   return TimeFormat(formatter.get(), time);
 }
 
-std::wstring TimeFormatShortDate(const Time& time) {
+string16 TimeFormatShortDate(const Time& time) {
   scoped_ptr<icu::DateFormat> formatter(
       icu::DateFormat::createDateInstance(icu::DateFormat::kMedium));
   return TimeFormat(formatter.get(), time);
 }
 
-std::wstring TimeFormatShortDateNumeric(const Time& time) {
+string16 TimeFormatShortDateNumeric(const Time& time) {
   scoped_ptr<icu::DateFormat> formatter(
       icu::DateFormat::createDateInstance(icu::DateFormat::kShort));
   return TimeFormat(formatter.get(), time);
 }
 
-std::wstring TimeFormatShortDateAndTime(const Time& time) {
+string16 TimeFormatShortDateAndTime(const Time& time) {
   scoped_ptr<icu::DateFormat> formatter(
       icu::DateFormat::createDateTimeInstance(icu::DateFormat::kShort));
   return TimeFormat(formatter.get(), time);
 }
 
-std::wstring TimeFormatFriendlyDateAndTime(const Time& time) {
+string16 TimeFormatFriendlyDateAndTime(const Time& time) {
   scoped_ptr<icu::DateFormat> formatter(
       icu::DateFormat::createDateTimeInstance(icu::DateFormat::kFull));
   return TimeFormat(formatter.get(), time);
 }
 
-std::wstring TimeFormatFriendlyDate(const Time& time) {
+string16 TimeFormatFriendlyDate(const Time& time) {
   scoped_ptr<icu::DateFormat> formatter(icu::DateFormat::createDateInstance(
       icu::DateFormat::kFull));
   return TimeFormat(formatter.get(), time);
diff --git a/base/i18n/time_formatting.h b/base/i18n/time_formatting.h
index d78ae9b..e70ad3d 100644
--- a/base/i18n/time_formatting.h
+++ b/base/i18n/time_formatting.h
@@ -9,32 +9,32 @@
 #define BASE_I18N_TIME_FORMATTING_H_
 #pragma once
 
-#include <string>
+#include "base/string16.h"
 
 namespace base {
 
 class Time;
 
 // Returns the time of day, e.g., "3:07 PM".
-std::wstring TimeFormatTimeOfDay(const Time& time);
+string16 TimeFormatTimeOfDay(const Time& time);
 
 // Returns a shortened date, e.g. "Nov 7, 2007"
-std::wstring TimeFormatShortDate(const Time& time);
+string16 TimeFormatShortDate(const Time& time);
 
 // Returns a numeric date such as 12/13/52.
-std::wstring TimeFormatShortDateNumeric(const Time& time);
+string16 TimeFormatShortDateNumeric(const Time& time);
 
 // Formats a time in a friendly sentence format, e.g.
 // "Monday, March 6, 2008 2:44:30 PM".
-std::wstring TimeFormatShortDateAndTime(const Time& time);
+string16 TimeFormatShortDateAndTime(const Time& time);
 
 // Formats a time in a friendly sentence format, e.g.
 // "Monday, March 6, 2008 2:44:30 PM".
-std::wstring TimeFormatFriendlyDateAndTime(const Time& time);
+string16 TimeFormatFriendlyDateAndTime(const Time& time);
 
 // Formats a time in a friendly sentence format, e.g.
 // "Monday, March 6, 2008".
-std::wstring TimeFormatFriendlyDate(const Time& time);
+string16 TimeFormatFriendlyDate(const Time& time);
 
 }  // namespace base
 
diff --git a/base/i18n/word_iterator.cc b/base/i18n/word_iterator.cc
deleted file mode 100644
index a9fa4af..0000000
--- a/base/i18n/word_iterator.cc
+++ /dev/null
@@ -1,70 +0,0 @@
-// Copyright (c) 2009 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/i18n/word_iterator.h"
-
-#include "base/logging.h"
-#include "unicode/ubrk.h"
-#include "unicode/ustring.h"
-
-const size_t npos = -1;
-
-WordIterator::WordIterator(const string16* str, BreakType break_type)
-    : iter_(NULL),
-      string_(str),
-      break_type_(break_type),
-      prev_(npos),
-      pos_(0) {
-}
-
-WordIterator::~WordIterator() {
-  if (iter_)
-    ubrk_close(iter_);
-}
-
-bool WordIterator::Init() {
-  UErrorCode status = U_ZERO_ERROR;
-  UBreakIteratorType break_type;
-  switch (break_type_) {
-    case BREAK_WORD:
-      break_type = UBRK_WORD;
-      break;
-    case BREAK_LINE:
-      break_type = UBRK_LINE;
-      break;
-    default:
-      NOTREACHED();
-      break_type = UBRK_LINE;
-  }
-  iter_ = ubrk_open(break_type, NULL,
-                    string_->data(), static_cast<int32_t>(string_->size()),
-                    &status);
-  if (U_FAILURE(status)) {
-    NOTREACHED() << "ubrk_open failed";
-    return false;
-  }
-  ubrk_first(iter_);  // Move the iterator to the beginning of the string.
-  return true;
-}
-
-bool WordIterator::Advance() {
-  prev_ = pos_;
-  const int32_t pos = ubrk_next(iter_);
-  if (pos == UBRK_DONE) {
-    pos_ = npos;
-    return false;
-  } else {
-    pos_ = static_cast<size_t>(pos);
-    return true;
-  }
-}
-
-bool WordIterator::IsWord() const {
-  return (ubrk_getRuleStatus(iter_) != UBRK_WORD_NONE);
-}
-
-string16 WordIterator::GetWord() const {
-  DCHECK(prev_ != npos && pos_ != npos);
-  return string_->substr(prev_, pos_ - prev_);
-}
diff --git a/base/i18n/word_iterator.h b/base/i18n/word_iterator.h
deleted file mode 100644
index b097bc2..0000000
--- a/base/i18n/word_iterator.h
+++ /dev/null
@@ -1,89 +0,0 @@
-// Copyright (c) 2010 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 BASE_I18N_WORD_ITERATOR_H_
-#define BASE_I18N_WORD_ITERATOR_H_
-#pragma once
-
-#include <vector>
-
-#include "unicode/ubrk.h"
-#include "unicode/uchar.h"
-
-#include "base/basictypes.h"
-#include "base/string16.h"
-
-// The WordIterator class iterates through the words and word breaks
-// in a string.  (In the string " foo bar! ", the word breaks are at the
-// periods in ". .foo. .bar.!. .".)
-//
-// To extract the words from a string, move a WordIterator through the
-// string and test whether IsWord() is true.  E.g.,
-//   WordIterator iter(&str, WordIterator::BREAK_WORD);
-//   if (!iter.Init()) return false;
-//   while (iter.Advance()) {
-//     if (iter.IsWord()) {
-//       // region [iter.prev(),iter.pos()) contains a word.
-//       VLOG(1) << "word: " << iter.GetWord();
-//     }
-//   }
-
-
-class WordIterator {
- public:
-  enum BreakType {
-    BREAK_WORD,
-    BREAK_LINE
-  };
-
-  // Requires |str| to live as long as the WordIterator does.
-  WordIterator(const string16* str, BreakType break_type);
-  ~WordIterator();
-
-  // Init() must be called before any of the iterators are valid.
-  // Returns false if ICU failed to initialize.
-  bool Init();
-
-  // Return the current break position within the string,
-  // or WordIterator::npos when done.
-  size_t pos() const { return pos_; }
-  // Return the value of pos() returned before Advance() was last called.
-  size_t prev() const { return prev_; }
-
-  // Advance to the next break.  Returns false if we've run past the end of
-  // the string.  (Note that the very last "word break" is after the final
-  // character in the string, and when we advance to that position it's the
-  // last time Advance() returns true.)
-  bool Advance();
-
-  // Returns true if the break we just hit is the end of a word.
-  // (Otherwise, the break iterator just skipped over e.g. whitespace
-  // or punctuation.)
-  bool IsWord() const;
-
-  // Return the word between prev() and pos().
-  // Advance() must have been called successfully at least once
-  // for pos() to have advanced to somewhere useful.
-  string16 GetWord() const;
-
- private:
-  // ICU iterator.
-  UBreakIterator* iter_;
-#if !defined(WCHAR_T_IS_UTF16)
-  std::vector<UChar> chars_;
-#endif
-
-  // The string we're iterating over.
-  const string16* string_;
-
-  // The breaking style (word/line).
-  BreakType break_type_;
-
-  // Previous and current iterator positions.
-  size_t prev_, pos_;
-
-  DISALLOW_COPY_AND_ASSIGN(WordIterator);
-};
-
-#endif  // BASE_I18N_WORD_ITERATOR_H__
diff --git a/base/i18n/word_iterator_unittest.cc b/base/i18n/word_iterator_unittest.cc
deleted file mode 100644
index 92aff76..0000000
--- a/base/i18n/word_iterator_unittest.cc
+++ /dev/null
@@ -1,117 +0,0 @@
-// Copyright (c) 2010 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/i18n/word_iterator.h"
-
-#include "base/string_piece.h"
-#include "base/string_util.h"
-#include "base/utf_string_conversions.h"
-#include "testing/gtest/include/gtest/gtest.h"
-
-TEST(WordIteratorTest, BreakWord) {
-  string16 space(UTF8ToUTF16(" "));
-
-  string16 str(UTF8ToUTF16(" foo bar! \npouet boom"));
-  WordIterator iter(&str, WordIterator::BREAK_WORD);
-  ASSERT_TRUE(iter.Init());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(space, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("foo"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(space, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("bar"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("!"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(space, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("\n"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("pouet"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(space, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetWord());
-  EXPECT_FALSE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-}
-
-TEST(WordIteratorTest, BreakLine) {
-  string16 str(UTF8ToUTF16(" foo bar! \npouet boom"));
-  WordIterator iter(&str, WordIterator::BREAK_LINE);
-  ASSERT_TRUE(iter.Init());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("foo "), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("bar! \n"), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("pouet "), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetWord());
-  EXPECT_FALSE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-}
-
-TEST(WordIteratorTest, BreakWide16) {
-  //  "Παγκόσμιος Ιστός"
-  const string16 str(WideToUTF16(
-      L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9"
-      L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2"));
-  const string16 word1(str.substr(0, 10));
-  const string16 word2(str.substr(11, 5));
-  WordIterator iter(&str, WordIterator::BREAK_WORD);
-  ASSERT_TRUE(iter.Init());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(word1, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(word2, iter.GetWord());
-  EXPECT_FALSE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-}
-
-TEST(WordIteratorTest, BreakWide32) {
-  // U+1D49C MATHEMATICAL SCRIPT CAPITAL A
-  const char* very_wide_char = "\xF0\x9D\x92\x9C";
-  const string16 str(
-      UTF8ToUTF16(StringPrintf("%s a", very_wide_char)));
-  const string16 very_wide_word(str.substr(0, 2));
-
-  WordIterator iter(&str, WordIterator::BREAK_WORD);
-  ASSERT_TRUE(iter.Init());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(very_wide_word, iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16(" "), iter.GetWord());
-  EXPECT_TRUE(iter.Advance());
-  EXPECT_TRUE(iter.IsWord());
-  EXPECT_EQ(UTF8ToUTF16("a"), iter.GetWord());
-  EXPECT_FALSE(iter.Advance());
-  EXPECT_FALSE(iter.IsWord());
-}