Move string files in base/ to the string subdirectory.

BUG=247723
TEST=no change

Review URL: https://chromiumcodereview.appspot.com/16331011

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@205050 0039d316-1c4b-4281-b951-d872f2087c98

12 files changed, 3708 insertions, 0 deletions

diff --git a/base/strings/string16.cc b/base/strings/string16.cc
new file mode 100644
index 0000000..c802eef
--- /dev/null
+++ b/base/strings/string16.cc
@@ -0,0 +1,82 @@
+// 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 "base/strings/string16.h"
+
+#if defined(WCHAR_T_IS_UTF16)
+
+#error This file should not be used on 2-byte wchar_t systems
+// If this winds up being needed on 2-byte wchar_t systems, either the
+// definitions below can be used, or the host system's wide character
+// functions like wmemcmp can be wrapped.
+
+#elif defined(WCHAR_T_IS_UTF32)
+
+#include <ostream>
+
+#include "base/strings/utf_string_conversions.h"
+
+namespace base {
+
+int c16memcmp(const char16* s1, const char16* s2, size_t n) {
+  // We cannot call memcmp because that changes the semantics.
+  while (n-- > 0) {
+    if (*s1 != *s2) {
+      // We cannot use (*s1 - *s2) because char16 is unsigned.
+      return ((*s1 < *s2) ? -1 : 1);
+    }
+    ++s1;
+    ++s2;
+  }
+  return 0;
+}
+
+size_t c16len(const char16* s) {
+  const char16 *s_orig = s;
+  while (*s) {
+    ++s;
+  }
+  return s - s_orig;
+}
+
+const char16* c16memchr(const char16* s, char16 c, size_t n) {
+  while (n-- > 0) {
+    if (*s == c) {
+      return s;
+    }
+    ++s;
+  }
+  return 0;
+}
+
+char16* c16memmove(char16* s1, const char16* s2, size_t n) {
+  return static_cast<char16*>(memmove(s1, s2, n * sizeof(char16)));
+}
+
+char16* c16memcpy(char16* s1, const char16* s2, size_t n) {
+  return static_cast<char16*>(memcpy(s1, s2, n * sizeof(char16)));
+}
+
+char16* c16memset(char16* s, char16 c, size_t n) {
+  char16 *s_orig = s;
+  while (n-- > 0) {
+    *s = c;
+    ++s;
+  }
+  return s_orig;
+}
+
+std::ostream& operator<<(std::ostream& out, const string16& str) {
+  return out << UTF16ToUTF8(str);
+}
+
+void PrintTo(const string16& str, std::ostream* out) {
+  *out << str;
+}
+
+}  // namespace base
+
+template class std::basic_string<char16, base::string16_char_traits>;
+
+#endif  // WCHAR_T_IS_UTF32
diff --git a/base/strings/string16.h b/base/strings/string16.h
new file mode 100644
index 0000000..fd98f1b
--- /dev/null
+++ b/base/strings/string16.h
@@ -0,0 +1,189 @@
+// 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 BASE_STRINGS_STRING16_H_
+#define BASE_STRINGS_STRING16_H_
+
+// WHAT:
+// A version of std::basic_string that provides 2-byte characters even when
+// wchar_t is not implemented as a 2-byte type. You can access this class as
+// string16. We also define char16, which string16 is based upon.
+//
+// WHY:
+// On Windows, wchar_t is 2 bytes, and it can conveniently handle UTF-16/UCS-2
+// data. Plenty of existing code operates on strings encoded as UTF-16.
+//
+// On many other platforms, sizeof(wchar_t) is 4 bytes by default. We can make
+// it 2 bytes by using the GCC flag -fshort-wchar. But then std::wstring fails
+// at run time, because it calls some functions (like wcslen) that come from
+// the system's native C library -- which was built with a 4-byte wchar_t!
+// It's wasteful to use 4-byte wchar_t strings to carry UTF-16 data, and it's
+// entirely improper on those systems where the encoding of wchar_t is defined
+// as UTF-32.
+//
+// Here, we define string16, which is similar to std::wstring but replaces all
+// libc functions with custom, 2-byte-char compatible routines. It is capable
+// of carrying UTF-16-encoded data.
+
+#include <stdio.h>
+#include <string>
+
+#include "base/base_export.h"
+#include "base/basictypes.h"
+
+#if defined(WCHAR_T_IS_UTF16)
+
+namespace base {
+
+typedef wchar_t char16;
+typedef std::wstring string16;
+typedef std::char_traits<wchar_t> string16_char_traits;
+
+}  // namespace base
+
+#elif defined(WCHAR_T_IS_UTF32)
+
+namespace base {
+
+typedef uint16 char16;
+
+// char16 versions of the functions required by string16_char_traits; these
+// are based on the wide character functions of similar names ("w" or "wcs"
+// instead of "c16").
+BASE_EXPORT int c16memcmp(const char16* s1, const char16* s2, size_t n);
+BASE_EXPORT size_t c16len(const char16* s);
+BASE_EXPORT const char16* c16memchr(const char16* s, char16 c, size_t n);
+BASE_EXPORT char16* c16memmove(char16* s1, const char16* s2, size_t n);
+BASE_EXPORT char16* c16memcpy(char16* s1, const char16* s2, size_t n);
+BASE_EXPORT char16* c16memset(char16* s, char16 c, size_t n);
+
+struct string16_char_traits {
+  typedef char16 char_type;
+  typedef int int_type;
+
+  // int_type needs to be able to hold each possible value of char_type, and in
+  // addition, the distinct value of eof().
+  COMPILE_ASSERT(sizeof(int_type) > sizeof(char_type), unexpected_type_width);
+
+  typedef std::streamoff off_type;
+  typedef mbstate_t state_type;
+  typedef std::fpos<state_type> pos_type;
+
+  static void assign(char_type& c1, const char_type& c2) {
+    c1 = c2;
+  }
+
+  static bool eq(const char_type& c1, const char_type& c2) {
+    return c1 == c2;
+  }
+  static bool lt(const char_type& c1, const char_type& c2) {
+    return c1 < c2;
+  }
+
+  static int compare(const char_type* s1, const char_type* s2, size_t n) {
+    return c16memcmp(s1, s2, n);
+  }
+
+  static size_t length(const char_type* s) {
+    return c16len(s);
+  }
+
+  static const char_type* find(const char_type* s, size_t n,
+                               const char_type& a) {
+    return c16memchr(s, a, n);
+  }
+
+  static char_type* move(char_type* s1, const char_type* s2, int_type n) {
+    return c16memmove(s1, s2, n);
+  }
+
+  static char_type* copy(char_type* s1, const char_type* s2, size_t n) {
+    return c16memcpy(s1, s2, n);
+  }
+
+  static char_type* assign(char_type* s, size_t n, char_type a) {
+    return c16memset(s, a, n);
+  }
+
+  static int_type not_eof(const int_type& c) {
+    return eq_int_type(c, eof()) ? 0 : c;
+  }
+
+  static char_type to_char_type(const int_type& c) {
+    return char_type(c);
+  }
+
+  static int_type to_int_type(const char_type& c) {
+    return int_type(c);
+  }
+
+  static bool eq_int_type(const int_type& c1, const int_type& c2) {
+    return c1 == c2;
+  }
+
+  static int_type eof() {
+    return static_cast<int_type>(EOF);
+  }
+};
+
+typedef std::basic_string<char16, base::string16_char_traits> string16;
+
+BASE_EXPORT extern std::ostream& operator<<(std::ostream& out,
+                                            const string16& str);
+
+// This is required by googletest to print a readable output on test failures.
+BASE_EXPORT extern void PrintTo(const string16& str, std::ostream* out);
+
+}  // namespace base
+
+// The string class will be explicitly instantiated only once, in string16.cc.
+//
+// std::basic_string<> in GNU libstdc++ contains a static data member,
+// _S_empty_rep_storage, to represent empty strings.  When an operation such
+// as assignment or destruction is performed on a string, causing its existing
+// data member to be invalidated, it must not be freed if this static data
+// member is being used.  Otherwise, it counts as an attempt to free static
+// (and not allocated) data, which is a memory error.
+//
+// Generally, due to C++ template magic, _S_empty_rep_storage will be marked
+// as a coalesced symbol, meaning that the linker will combine multiple
+// instances into a single one when generating output.
+//
+// If a string class is used by multiple shared libraries, a problem occurs.
+// Each library will get its own copy of _S_empty_rep_storage.  When strings
+// are passed across a library boundary for alteration or destruction, memory
+// errors will result.  GNU libstdc++ contains a configuration option,
+// --enable-fully-dynamic-string (_GLIBCXX_FULLY_DYNAMIC_STRING), which
+// disables the static data member optimization, but it's a good optimization
+// and non-STL code is generally at the mercy of the system's STL
+// configuration.  Fully-dynamic strings are not the default for GNU libstdc++
+// libstdc++ itself or for the libstdc++ installations on the systems we care
+// about, such as Mac OS X and relevant flavors of Linux.
+//
+// See also http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24196 .
+//
+// To avoid problems, string classes need to be explicitly instantiated only
+// once, in exactly one library.  All other string users see it via an "extern"
+// declaration.  This is precisely how GNU libstdc++ handles
+// std::basic_string<char> (string) and std::basic_string<wchar_t> (wstring).
+//
+// This also works around a Mac OS X linker bug in ld64-85.2.1 (Xcode 3.1.2),
+// in which the linker does not fully coalesce symbols when dead code
+// stripping is enabled.  This bug causes the memory errors described above
+// to occur even when a std::basic_string<> does not cross shared library
+// boundaries, such as in statically-linked executables.
+//
+// TODO(mark): File this bug with Apple and update this note with a bug number.
+
+extern template
+class BASE_EXPORT std::basic_string<base::char16, base::string16_char_traits>;
+
+#endif  // WCHAR_T_IS_UTF32
+
+// TODO(brettw) update users of string16 to use the namespace and remove
+// this "using".
+using base::char16;
+using base::string16;
+
+#endif  // BASE_STRINGS_STRING16_H_
diff --git a/base/strings/string16_unittest.cc b/base/strings/string16_unittest.cc
new file mode 100644
index 0000000..d98b2a9
--- /dev/null
+++ b/base/strings/string16_unittest.cc
@@ -0,0 +1,54 @@
+// 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 <sstream>
+
+#include "base/strings/string16.h"
+
+#include "base/strings/utf_string_conversions.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+#if defined(WCHAR_T_IS_UTF32)
+
+// We define a custom operator<< for string16 so we can use it with logging.
+// This tests that conversion.
+TEST(String16Test, OutputStream) {
+  // Basic stream test.
+  {
+    std::ostringstream stream;
+    stream << "Empty '" << string16() << "' standard '"
+           << string16(ASCIIToUTF16("Hello, world")) << "'";
+    EXPECT_STREQ("Empty '' standard 'Hello, world'",
+                 stream.str().c_str());
+  }
+
+  // Interesting edge cases.
+  {
+    // These should each get converted to the invalid character: EF BF BD.
+    string16 initial_surrogate;
+    initial_surrogate.push_back(0xd800);
+    string16 final_surrogate;
+    final_surrogate.push_back(0xdc00);
+
+    // Old italic A = U+10300, will get converted to: F0 90 8C 80 'z'.
+    string16 surrogate_pair;
+    surrogate_pair.push_back(0xd800);
+    surrogate_pair.push_back(0xdf00);
+    surrogate_pair.push_back('z');
+
+    // Will get converted to the invalid char + 's': EF BF BD 's'.
+    string16 unterminated_surrogate;
+    unterminated_surrogate.push_back(0xd800);
+    unterminated_surrogate.push_back('s');
+
+    std::ostringstream stream;
+    stream << initial_surrogate << "," << final_surrogate << ","
+           << surrogate_pair << "," << unterminated_surrogate;
+
+    EXPECT_STREQ("\xef\xbf\xbd,\xef\xbf\xbd,\xf0\x90\x8c\x80z,\xef\xbf\xbds",
+                 stream.str().c_str());
+  }
+}
+
+#endif
diff --git a/base/strings/string_util.cc b/base/strings/string_util.cc
new file mode 100644
index 0000000..3ed7069
--- /dev/null
+++ b/base/strings/string_util.cc
@@ -0,0 +1,1011 @@
+// 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 "base/strings/string_util.h"
+
+#include <ctype.h>
+#include <errno.h>
+#include <math.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <time.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#include <algorithm>
+#include <vector>
+
+#include "base/basictypes.h"
+#include "base/logging.h"
+#include "base/memory/singleton.h"
+#include "base/strings/utf_string_conversion_utils.h"
+#include "base/strings/utf_string_conversions.h"
+#include "base/third_party/icu/icu_utf.h"
+#include "build/build_config.h"
+
+namespace {
+
+// Force the singleton used by Empty[W]String[16] to be a unique type. This
+// prevents other code that might accidentally use Singleton<string> from
+// getting our internal one.
+struct EmptyStrings {
+  EmptyStrings() {}
+  const std::string s;
+  const std::wstring ws;
+  const string16 s16;
+
+  static EmptyStrings* GetInstance() {
+    return Singleton<EmptyStrings>::get();
+  }
+};
+
+// Used by ReplaceStringPlaceholders to track the position in the string of
+// replaced parameters.
+struct ReplacementOffset {
+  ReplacementOffset(uintptr_t parameter, size_t offset)
+      : parameter(parameter),
+        offset(offset) {}
+
+  // Index of the parameter.
+  uintptr_t parameter;
+
+  // Starting position in the string.
+  size_t offset;
+};
+
+static bool CompareParameter(const ReplacementOffset& elem1,
+                             const ReplacementOffset& elem2) {
+  return elem1.parameter < elem2.parameter;
+}
+
+}  // namespace
+
+namespace base {
+
+bool IsWprintfFormatPortable(const wchar_t* format) {
+  for (const wchar_t* position = format; *position != '\0'; ++position) {
+    if (*position == '%') {
+      bool in_specification = true;
+      bool modifier_l = false;
+      while (in_specification) {
+        // Eat up characters until reaching a known specifier.
+        if (*++position == '\0') {
+          // The format string ended in the middle of a specification.  Call
+          // it portable because no unportable specifications were found.  The
+          // string is equally broken on all platforms.
+          return true;
+        }
+
+        if (*position == 'l') {
+          // 'l' is the only thing that can save the 's' and 'c' specifiers.
+          modifier_l = true;
+        } else if (((*position == 's' || *position == 'c') && !modifier_l) ||
+                   *position == 'S' || *position == 'C' || *position == 'F' ||
+                   *position == 'D' || *position == 'O' || *position == 'U') {
+          // Not portable.
+          return false;
+        }
+
+        if (wcschr(L"diouxXeEfgGaAcspn%", *position)) {
+          // Portable, keep scanning the rest of the format string.
+          in_specification = false;
+        }
+      }
+    }
+  }
+
+  return true;
+}
+
+}  // namespace base
+
+
+const std::string& EmptyString() {
+  return EmptyStrings::GetInstance()->s;
+}
+
+const std::wstring& EmptyWString() {
+  return EmptyStrings::GetInstance()->ws;
+}
+
+const string16& EmptyString16() {
+  return EmptyStrings::GetInstance()->s16;
+}
+
+template<typename STR>
+bool ReplaceCharsT(const STR& input,
+                   const typename STR::value_type replace_chars[],
+                   const STR& replace_with,
+                   STR* output) {
+  bool removed = false;
+  size_t replace_length = replace_with.length();
+
+  *output = input;
+
+  size_t found = output->find_first_of(replace_chars);
+  while (found != STR::npos) {
+    removed = true;
+    output->replace(found, 1, replace_with);
+    found = output->find_first_of(replace_chars, found + replace_length);
+  }
+
+  return removed;
+}
+
+bool ReplaceChars(const string16& input,
+                  const char16 replace_chars[],
+                  const string16& replace_with,
+                  string16* output) {
+  return ReplaceCharsT(input, replace_chars, replace_with, output);
+}
+
+bool ReplaceChars(const std::string& input,
+                  const char replace_chars[],
+                  const std::string& replace_with,
+                  std::string* output) {
+  return ReplaceCharsT(input, replace_chars, replace_with, output);
+}
+
+bool RemoveChars(const string16& input,
+                 const char16 remove_chars[],
+                 string16* output) {
+  return ReplaceChars(input, remove_chars, string16(), output);
+}
+
+bool RemoveChars(const std::string& input,
+                 const char remove_chars[],
+                 std::string* output) {
+  return ReplaceChars(input, remove_chars, std::string(), output);
+}
+
+template<typename STR>
+TrimPositions TrimStringT(const STR& input,
+                          const typename STR::value_type trim_chars[],
+                          TrimPositions positions,
+                          STR* output) {
+  // Find the edges of leading/trailing whitespace as desired.
+  const typename STR::size_type last_char = input.length() - 1;
+  const typename STR::size_type first_good_char = (positions & TRIM_LEADING) ?
+      input.find_first_not_of(trim_chars) : 0;
+  const typename STR::size_type last_good_char = (positions & TRIM_TRAILING) ?
+      input.find_last_not_of(trim_chars) : last_char;
+
+  // When the string was all whitespace, report that we stripped off whitespace
+  // from whichever position the caller was interested in.  For empty input, we
+  // stripped no whitespace, but we still need to clear |output|.
+  if (input.empty() ||
+      (first_good_char == STR::npos) || (last_good_char == STR::npos)) {
+    bool input_was_empty = input.empty();  // in case output == &input
+    output->clear();
+    return input_was_empty ? TRIM_NONE : positions;
+  }
+
+  // Trim the whitespace.
+  *output =
+      input.substr(first_good_char, last_good_char - first_good_char + 1);
+
+  // Return where we trimmed from.
+  return static_cast<TrimPositions>(
+      ((first_good_char == 0) ? TRIM_NONE : TRIM_LEADING) |
+      ((last_good_char == last_char) ? TRIM_NONE : TRIM_TRAILING));
+}
+
+bool TrimString(const std::wstring& input,
+                const wchar_t trim_chars[],
+                std::wstring* output) {
+  return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool TrimString(const string16& input,
+                const char16 trim_chars[],
+                string16* output) {
+  return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
+}
+#endif
+
+bool TrimString(const std::string& input,
+                const char trim_chars[],
+                std::string* output) {
+  return TrimStringT(input, trim_chars, TRIM_ALL, output) != TRIM_NONE;
+}
+
+void TruncateUTF8ToByteSize(const std::string& input,
+                            const size_t byte_size,
+                            std::string* output) {
+  DCHECK(output);
+  if (byte_size > input.length()) {
+    *output = input;
+    return;
+  }
+  DCHECK_LE(byte_size, static_cast<uint32>(kint32max));
+  // Note: This cast is necessary because CBU8_NEXT uses int32s.
+  int32 truncation_length = static_cast<int32>(byte_size);
+  int32 char_index = truncation_length - 1;
+  const char* data = input.data();
+
+  // Using CBU8, we will move backwards from the truncation point
+  // to the beginning of the string looking for a valid UTF8
+  // character.  Once a full UTF8 character is found, we will
+  // truncate the string to the end of that character.
+  while (char_index >= 0) {
+    int32 prev = char_index;
+    uint32 code_point = 0;
+    CBU8_NEXT(data, char_index, truncation_length, code_point);
+    if (!base::IsValidCharacter(code_point) ||
+        !base::IsValidCodepoint(code_point)) {
+      char_index = prev - 1;
+    } else {
+      break;
+    }
+  }
+
+  if (char_index >= 0 )
+    *output = input.substr(0, char_index);
+  else
+    output->clear();
+}
+
+TrimPositions TrimWhitespace(const string16& input,
+                             TrimPositions positions,
+                             string16* output) {
+  return TrimStringT(input, kWhitespaceUTF16, positions, output);
+}
+
+TrimPositions TrimWhitespaceASCII(const std::string& input,
+                                  TrimPositions positions,
+                                  std::string* output) {
+  return TrimStringT(input, kWhitespaceASCII, positions, output);
+}
+
+// This function is only for backward-compatibility.
+// To be removed when all callers are updated.
+TrimPositions TrimWhitespace(const std::string& input,
+                             TrimPositions positions,
+                             std::string* output) {
+  return TrimWhitespaceASCII(input, positions, output);
+}
+
+template<typename STR>
+STR CollapseWhitespaceT(const STR& text,
+                        bool trim_sequences_with_line_breaks) {
+  STR result;
+  result.resize(text.size());
+
+  // Set flags to pretend we're already in a trimmed whitespace sequence, so we
+  // will trim any leading whitespace.
+  bool in_whitespace = true;
+  bool already_trimmed = true;
+
+  int chars_written = 0;
+  for (typename STR::const_iterator i(text.begin()); i != text.end(); ++i) {
+    if (IsWhitespace(*i)) {
+      if (!in_whitespace) {
+        // Reduce all whitespace sequences to a single space.
+        in_whitespace = true;
+        result[chars_written++] = L' ';
+      }
+      if (trim_sequences_with_line_breaks && !already_trimmed &&
+          ((*i == '\n') || (*i == '\r'))) {
+        // Whitespace sequences containing CR or LF are eliminated entirely.
+        already_trimmed = true;
+        --chars_written;
+      }
+    } else {
+      // Non-whitespace chracters are copied straight across.
+      in_whitespace = false;
+      already_trimmed = false;
+      result[chars_written++] = *i;
+    }
+  }
+
+  if (in_whitespace && !already_trimmed) {
+    // Any trailing whitespace is eliminated.
+    --chars_written;
+  }
+
+  result.resize(chars_written);
+  return result;
+}
+
+std::wstring CollapseWhitespace(const std::wstring& text,
+                                bool trim_sequences_with_line_breaks) {
+  return CollapseWhitespaceT(text, trim_sequences_with_line_breaks);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+string16 CollapseWhitespace(const string16& text,
+                            bool trim_sequences_with_line_breaks) {
+  return CollapseWhitespaceT(text, trim_sequences_with_line_breaks);
+}
+#endif
+
+std::string CollapseWhitespaceASCII(const std::string& text,
+                                    bool trim_sequences_with_line_breaks) {
+  return CollapseWhitespaceT(text, trim_sequences_with_line_breaks);
+}
+
+bool ContainsOnlyWhitespaceASCII(const std::string& str) {
+  for (std::string::const_iterator i(str.begin()); i != str.end(); ++i) {
+    if (!IsAsciiWhitespace(*i))
+      return false;
+  }
+  return true;
+}
+
+bool ContainsOnlyWhitespace(const string16& str) {
+  return str.find_first_not_of(kWhitespaceUTF16) == string16::npos;
+}
+
+template<typename STR>
+static bool ContainsOnlyCharsT(const STR& input, const STR& characters) {
+  for (typename STR::const_iterator iter = input.begin();
+       iter != input.end(); ++iter) {
+    if (characters.find(*iter) == STR::npos)
+      return false;
+  }
+  return true;
+}
+
+bool ContainsOnlyChars(const std::wstring& input,
+                       const std::wstring& characters) {
+  return ContainsOnlyCharsT(input, characters);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool ContainsOnlyChars(const string16& input, const string16& characters) {
+  return ContainsOnlyCharsT(input, characters);
+}
+#endif
+
+bool ContainsOnlyChars(const std::string& input,
+                       const std::string& characters) {
+  return ContainsOnlyCharsT(input, characters);
+}
+
+std::string WideToASCII(const std::wstring& wide) {
+  DCHECK(IsStringASCII(wide)) << wide;
+  return std::string(wide.begin(), wide.end());
+}
+
+std::string UTF16ToASCII(const string16& utf16) {
+  DCHECK(IsStringASCII(utf16)) << utf16;
+  return std::string(utf16.begin(), utf16.end());
+}
+
+// Latin1 is just the low range of Unicode, so we can copy directly to convert.
+bool WideToLatin1(const std::wstring& wide, std::string* latin1) {
+  std::string output;
+  output.resize(wide.size());
+  latin1->clear();
+  for (size_t i = 0; i < wide.size(); i++) {
+    if (wide[i] > 255)
+      return false;
+    output[i] = static_cast<char>(wide[i]);
+  }
+  latin1->swap(output);
+  return true;
+}
+
+template<class STR>
+static bool DoIsStringASCII(const STR& str) {
+  for (size_t i = 0; i < str.length(); i++) {
+    typename ToUnsigned<typename STR::value_type>::Unsigned c = str[i];
+    if (c > 0x7F)
+      return false;
+  }
+  return true;
+}
+
+bool IsStringASCII(const std::wstring& str) {
+  return DoIsStringASCII(str);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool IsStringASCII(const string16& str) {
+  return DoIsStringASCII(str);
+}
+#endif
+
+bool IsStringASCII(const base::StringPiece& str) {
+  return DoIsStringASCII(str);
+}
+
+bool IsStringUTF8(const std::string& str) {
+  const char *src = str.data();
+  int32 src_len = static_cast<int32>(str.length());
+  int32 char_index = 0;
+
+  while (char_index < src_len) {
+    int32 code_point;
+    CBU8_NEXT(src, char_index, src_len, code_point);
+    if (!base::IsValidCharacter(code_point))
+      return false;
+  }
+  return true;
+}
+
+template<typename Iter>
+static inline bool DoLowerCaseEqualsASCII(Iter a_begin,
+                                          Iter a_end,
+                                          const char* b) {
+  for (Iter it = a_begin; it != a_end; ++it, ++b) {
+    if (!*b || base::ToLowerASCII(*it) != *b)
+      return false;
+  }
+  return *b == 0;
+}
+
+// Front-ends for LowerCaseEqualsASCII.
+bool LowerCaseEqualsASCII(const std::string& a, const char* b) {
+  return DoLowerCaseEqualsASCII(a.begin(), a.end(), b);
+}
+
+bool LowerCaseEqualsASCII(const std::wstring& a, const char* b) {
+  return DoLowerCaseEqualsASCII(a.begin(), a.end(), b);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool LowerCaseEqualsASCII(const string16& a, const char* b) {
+  return DoLowerCaseEqualsASCII(a.begin(), a.end(), b);
+}
+#endif
+
+bool LowerCaseEqualsASCII(std::string::const_iterator a_begin,
+                          std::string::const_iterator a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+
+bool LowerCaseEqualsASCII(std::wstring::const_iterator a_begin,
+                          std::wstring::const_iterator a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool LowerCaseEqualsASCII(string16::const_iterator a_begin,
+                          string16::const_iterator a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+#endif
+
+// TODO(port): Resolve wchar_t/iterator issues that require OS_ANDROID here.
+#if !defined(OS_ANDROID)
+bool LowerCaseEqualsASCII(const char* a_begin,
+                          const char* a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+
+bool LowerCaseEqualsASCII(const wchar_t* a_begin,
+                          const wchar_t* a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool LowerCaseEqualsASCII(const char16* a_begin,
+                          const char16* a_end,
+                          const char* b) {
+  return DoLowerCaseEqualsASCII(a_begin, a_end, b);
+}
+#endif
+
+#endif  // !defined(OS_ANDROID)
+
+bool EqualsASCII(const string16& a, const base::StringPiece& b) {
+  if (a.length() != b.length())
+    return false;
+  return std::equal(b.begin(), b.end(), a.begin());
+}
+
+bool StartsWithASCII(const std::string& str,
+                     const std::string& search,
+                     bool case_sensitive) {
+  if (case_sensitive)
+    return str.compare(0, search.length(), search) == 0;
+  else
+    return base::strncasecmp(str.c_str(), search.c_str(), search.length()) == 0;
+}
+
+template <typename STR>
+bool StartsWithT(const STR& str, const STR& search, bool case_sensitive) {
+  if (case_sensitive) {
+    return str.compare(0, search.length(), search) == 0;
+  } else {
+    if (search.size() > str.size())
+      return false;
+    return std::equal(search.begin(), search.end(), str.begin(),
+                      base::CaseInsensitiveCompare<typename STR::value_type>());
+  }
+}
+
+bool StartsWith(const std::wstring& str, const std::wstring& search,
+                bool case_sensitive) {
+  return StartsWithT(str, search, case_sensitive);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool StartsWith(const string16& str, const string16& search,
+                bool case_sensitive) {
+  return StartsWithT(str, search, case_sensitive);
+}
+#endif
+
+template <typename STR>
+bool EndsWithT(const STR& str, const STR& search, bool case_sensitive) {
+  typename STR::size_type str_length = str.length();
+  typename STR::size_type search_length = search.length();
+  if (search_length > str_length)
+    return false;
+  if (case_sensitive) {
+    return str.compare(str_length - search_length, search_length, search) == 0;
+  } else {
+    return std::equal(search.begin(), search.end(),
+                      str.begin() + (str_length - search_length),
+                      base::CaseInsensitiveCompare<typename STR::value_type>());
+  }
+}
+
+bool EndsWith(const std::string& str, const std::string& search,
+              bool case_sensitive) {
+  return EndsWithT(str, search, case_sensitive);
+}
+
+bool EndsWith(const std::wstring& str, const std::wstring& search,
+              bool case_sensitive) {
+  return EndsWithT(str, search, case_sensitive);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+bool EndsWith(const string16& str, const string16& search,
+              bool case_sensitive) {
+  return EndsWithT(str, search, case_sensitive);
+}
+#endif
+
+static const char* const kByteStringsUnlocalized[] = {
+  " B",
+  " kB",
+  " MB",
+  " GB",
+  " TB",
+  " PB"
+};
+
+string16 FormatBytesUnlocalized(int64 bytes) {
+  double unit_amount = static_cast<double>(bytes);
+  size_t dimension = 0;
+  const int kKilo = 1024;
+  while (unit_amount >= kKilo &&
+         dimension < arraysize(kByteStringsUnlocalized) - 1) {
+    unit_amount /= kKilo;
+    dimension++;
+  }
+
+  char buf[64];
+  if (bytes != 0 && dimension > 0 && unit_amount < 100) {
+    base::snprintf(buf, arraysize(buf), "%.1lf%s", unit_amount,
+                   kByteStringsUnlocalized[dimension]);
+  } else {
+    base::snprintf(buf, arraysize(buf), "%.0lf%s", unit_amount,
+                   kByteStringsUnlocalized[dimension]);
+  }
+
+  return ASCIIToUTF16(buf);
+}
+
+template<class StringType>
+void DoReplaceSubstringsAfterOffset(StringType* str,
+                                    typename StringType::size_type start_offset,
+                                    const StringType& find_this,
+                                    const StringType& replace_with,
+                                    bool replace_all) {
+  if ((start_offset == StringType::npos) || (start_offset >= str->length()))
+    return;
+
+  DCHECK(!find_this.empty());
+  for (typename StringType::size_type offs(str->find(find_this, start_offset));
+      offs != StringType::npos; offs = str->find(find_this, offs)) {
+    str->replace(offs, find_this.length(), replace_with);
+    offs += replace_with.length();
+
+    if (!replace_all)
+      break;
+  }
+}
+
+void ReplaceFirstSubstringAfterOffset(string16* str,
+                                      string16::size_type start_offset,
+                                      const string16& find_this,
+                                      const string16& replace_with) {
+  DoReplaceSubstringsAfterOffset(str, start_offset, find_this, replace_with,
+                                 false);  // replace first instance
+}
+
+void ReplaceFirstSubstringAfterOffset(std::string* str,
+                                      std::string::size_type start_offset,
+                                      const std::string& find_this,
+                                      const std::string& replace_with) {
+  DoReplaceSubstringsAfterOffset(str, start_offset, find_this, replace_with,
+                                 false);  // replace first instance
+}
+
+void ReplaceSubstringsAfterOffset(string16* str,
+                                  string16::size_type start_offset,
+                                  const string16& find_this,
+                                  const string16& replace_with) {
+  DoReplaceSubstringsAfterOffset(str, start_offset, find_this, replace_with,
+                                 true);  // replace all instances
+}
+
+void ReplaceSubstringsAfterOffset(std::string* str,
+                                  std::string::size_type start_offset,
+                                  const std::string& find_this,
+                                  const std::string& replace_with) {
+  DoReplaceSubstringsAfterOffset(str, start_offset, find_this, replace_with,
+                                 true);  // replace all instances
+}
+
+
+template<typename STR>
+static size_t TokenizeT(const STR& str,
+                        const STR& delimiters,
+                        std::vector<STR>* tokens) {
+  tokens->clear();
+
+  typename STR::size_type start = str.find_first_not_of(delimiters);
+  while (start != STR::npos) {
+    typename STR::size_type end = str.find_first_of(delimiters, start + 1);
+    if (end == STR::npos) {
+      tokens->push_back(str.substr(start));
+      break;
+    } else {
+      tokens->push_back(str.substr(start, end - start));
+      start = str.find_first_not_of(delimiters, end + 1);
+    }
+  }
+
+  return tokens->size();
+}
+
+size_t Tokenize(const std::wstring& str,
+                const std::wstring& delimiters,
+                std::vector<std::wstring>* tokens) {
+  return TokenizeT(str, delimiters, tokens);
+}
+
+#if !defined(WCHAR_T_IS_UTF16)
+size_t Tokenize(const string16& str,
+                const string16& delimiters,
+                std::vector<string16>* tokens) {
+  return TokenizeT(str, delimiters, tokens);
+}
+#endif
+
+size_t Tokenize(const std::string& str,
+                const std::string& delimiters,
+                std::vector<std::string>* tokens) {
+  return TokenizeT(str, delimiters, tokens);
+}
+
+size_t Tokenize(const base::StringPiece& str,
+                const base::StringPiece& delimiters,
+                std::vector<base::StringPiece>* tokens) {
+  return TokenizeT(str, delimiters, tokens);
+}
+
+template<typename STR>
+static STR JoinStringT(const std::vector<STR>& parts, const STR& sep) {
+  if (parts.empty())
+    return STR();
+
+  STR result(parts[0]);
+  typename std::vector<STR>::const_iterator iter = parts.begin();
+  ++iter;
+
+  for (; iter != parts.end(); ++iter) {
+    result += sep;
+    result += *iter;
+  }
+
+  return result;
+}
+
+std::string JoinString(const std::vector<std::string>& parts, char sep) {
+  return JoinStringT(parts, std::string(1, sep));
+}
+
+string16 JoinString(const std::vector<string16>& parts, char16 sep) {
+  return JoinStringT(parts, string16(1, sep));
+}
+
+std::string JoinString(const std::vector<std::string>& parts,
+                       const std::string& separator) {
+  return JoinStringT(parts, separator);
+}
+
+string16 JoinString(const std::vector<string16>& parts,
+                    const string16& separator) {
+  return JoinStringT(parts, separator);
+}
+
+template<class FormatStringType, class OutStringType>
+OutStringType DoReplaceStringPlaceholders(const FormatStringType& format_string,
+    const std::vector<OutStringType>& subst, std::vector<size_t>* offsets) {
+  size_t substitutions = subst.size();
+
+  size_t sub_length = 0;
+  for (typename std::vector<OutStringType>::const_iterator iter = subst.begin();
+       iter != subst.end(); ++iter) {
+    sub_length += iter->length();
+  }
+
+  OutStringType formatted;
+  formatted.reserve(format_string.length() + sub_length);
+
+  std::vector<ReplacementOffset> r_offsets;
+  for (typename FormatStringType::const_iterator i = format_string.begin();
+       i != format_string.end(); ++i) {
+    if ('$' == *i) {
+      if (i + 1 != format_string.end()) {
+        ++i;
+        DCHECK('$' == *i || '1' <= *i) << "Invalid placeholder: " << *i;
+        if ('$' == *i) {
+          while (i != format_string.end() && '$' == *i) {
+            formatted.push_back('$');
+            ++i;
+          }
+          --i;
+        } else {
+          uintptr_t index = 0;
+          while (i != format_string.end() && '0' <= *i && *i <= '9') {
+            index *= 10;
+            index += *i - '0';
+            ++i;
+          }
+          --i;
+          index -= 1;
+          if (offsets) {
+            ReplacementOffset r_offset(index,
+                static_cast<int>(formatted.size()));
+            r_offsets.insert(std::lower_bound(r_offsets.begin(),
+                                              r_offsets.end(),
+                                              r_offset,
+                                              &CompareParameter),
+                             r_offset);
+          }
+          if (index < substitutions)
+            formatted.append(subst.at(index));
+        }
+      }
+    } else {
+      formatted.push_back(*i);
+    }
+  }
+  if (offsets) {
+    for (std::vector<ReplacementOffset>::const_iterator i = r_offsets.begin();
+         i != r_offsets.end(); ++i) {
+      offsets->push_back(i->offset);
+    }
+  }
+  return formatted;
+}
+
+string16 ReplaceStringPlaceholders(const string16& format_string,
+                                   const std::vector<string16>& subst,
+                                   std::vector<size_t>* offsets) {
+  return DoReplaceStringPlaceholders(format_string, subst, offsets);
+}
+
+std::string ReplaceStringPlaceholders(const base::StringPiece& format_string,
+                                      const std::vector<std::string>& subst,
+                                      std::vector<size_t>* offsets) {
+  return DoReplaceStringPlaceholders(format_string, subst, offsets);
+}
+
+string16 ReplaceStringPlaceholders(const string16& format_string,
+                                   const string16& a,
+                                   size_t* offset) {
+  std::vector<size_t> offsets;
+  std::vector<string16> subst;
+  subst.push_back(a);
+  string16 result = ReplaceStringPlaceholders(format_string, subst, &offsets);
+
+  DCHECK(offsets.size() == 1);
+  if (offset) {
+    *offset = offsets[0];
+  }
+  return result;
+}
+
+static bool IsWildcard(base_icu::UChar32 character) {
+  return character == '*' || character == '?';
+}
+
+// Move the strings pointers to the point where they start to differ.
+template <typename CHAR, typename NEXT>
+static void EatSameChars(const CHAR** pattern, const CHAR* pattern_end,
+                         const CHAR** string, const CHAR* string_end,
+                         NEXT next) {
+  const CHAR* escape = NULL;
+  while (*pattern != pattern_end && *string != string_end) {
+    if (!escape && IsWildcard(**pattern)) {
+      // We don't want to match wildcard here, except if it's escaped.
+      return;
+    }
+
+    // Check if the escapement char is found. If so, skip it and move to the
+    // next character.
+    if (!escape && **pattern == '\\') {
+      escape = *pattern;
+      next(pattern, pattern_end);
+      continue;
+    }
+
+    // Check if the chars match, if so, increment the ptrs.
+    const CHAR* pattern_next = *pattern;
+    const CHAR* string_next = *string;
+    base_icu::UChar32 pattern_char = next(&pattern_next, pattern_end);
+    if (pattern_char == next(&string_next, string_end) &&
+        pattern_char != (base_icu::UChar32) CBU_SENTINEL) {
+      *pattern = pattern_next;
+      *string = string_next;
+    } else {
+      // Uh ho, it did not match, we are done. If the last char was an
+      // escapement, that means that it was an error to advance the ptr here,
+      // let's put it back where it was. This also mean that the MatchPattern
+      // function will return false because if we can't match an escape char
+      // here, then no one will.
+      if (escape) {
+        *pattern = escape;
+      }
+      return;
+    }
+
+    escape = NULL;
+  }
+}
+
+template <typename CHAR, typename NEXT>
+static void EatWildcard(const CHAR** pattern, const CHAR* end, NEXT next) {
+  while (*pattern != end) {
+    if (!IsWildcard(**pattern))
+      return;
+    next(pattern, end);
+  }
+}
+
+template <typename CHAR, typename NEXT>
+static bool MatchPatternT(const CHAR* eval, const CHAR* eval_end,
+                          const CHAR* pattern, const CHAR* pattern_end,
+                          int depth,
+                          NEXT next) {
+  const int kMaxDepth = 16;
+  if (depth > kMaxDepth)
+    return false;
+
+  // Eat all the matching chars.
+  EatSameChars(&pattern, pattern_end, &eval, eval_end, next);
+
+  // If the string is empty, then the pattern must be empty too, or contains
+  // only wildcards.
+  if (eval == eval_end) {
+    EatWildcard(&pattern, pattern_end, next);
+    return pattern == pattern_end;
+  }
+
+  // Pattern is empty but not string, this is not a match.
+  if (pattern == pattern_end)
+    return false;
+
+  // If this is a question mark, then we need to compare the rest with
+  // the current string or the string with one character eaten.
+  const CHAR* next_pattern = pattern;
+  next(&next_pattern, pattern_end);
+  if (pattern[0] == '?') {
+    if (MatchPatternT(eval, eval_end, next_pattern, pattern_end,
+                      depth + 1, next))
+      return true;
+    const CHAR* next_eval = eval;
+    next(&next_eval, eval_end);
+    if (MatchPatternT(next_eval, eval_end, next_pattern, pattern_end,
+                      depth + 1, next))
+      return true;
+  }
+
+  // This is a *, try to match all the possible substrings with the remainder
+  // of the pattern.
+  if (pattern[0] == '*') {
+    // Collapse duplicate wild cards (********** into *) so that the
+    // method does not recurse unnecessarily. http://crbug.com/52839
+    EatWildcard(&next_pattern, pattern_end, next);
+
+    while (eval != eval_end) {
+      if (MatchPatternT(eval, eval_end, next_pattern, pattern_end,
+                        depth + 1, next))
+        return true;
+      eval++;
+    }
+
+    // We reached the end of the string, let see if the pattern contains only
+    // wildcards.
+    if (eval == eval_end) {
+      EatWildcard(&pattern, pattern_end, next);
+      if (pattern != pattern_end)
+        return false;
+      return true;
+    }
+  }
+
+  return false;
+}
+
+struct NextCharUTF8 {
+  base_icu::UChar32 operator()(const char** p, const char* end) {
+    base_icu::UChar32 c;
+    int offset = 0;
+    CBU8_NEXT(*p, offset, end - *p, c);
+    *p += offset;
+    return c;
+  }
+};
+
+struct NextCharUTF16 {
+  base_icu::UChar32 operator()(const char16** p, const char16* end) {
+    base_icu::UChar32 c;
+    int offset = 0;
+    CBU16_NEXT(*p, offset, end - *p, c);
+    *p += offset;
+    return c;
+  }
+};
+
+bool MatchPattern(const base::StringPiece& eval,
+                  const base::StringPiece& pattern) {
+  return MatchPatternT(eval.data(), eval.data() + eval.size(),
+                       pattern.data(), pattern.data() + pattern.size(),
+                       0, NextCharUTF8());
+}
+
+bool MatchPattern(const string16& eval, const string16& pattern) {
+  return MatchPatternT(eval.c_str(), eval.c_str() + eval.size(),
+                       pattern.c_str(), pattern.c_str() + pattern.size(),
+                       0, NextCharUTF16());
+}
+
+// The following code is compatible with the OpenBSD lcpy interface.  See:
+//   http://www.gratisoft.us/todd/papers/strlcpy.html
+//   ftp://ftp.openbsd.org/pub/OpenBSD/src/lib/libc/string/{wcs,str}lcpy.c
+
+namespace {
+
+template <typename CHAR>
+size_t lcpyT(CHAR* dst, const CHAR* src, size_t dst_size) {
+  for (size_t i = 0; i < dst_size; ++i) {
+    if ((dst[i] = src[i]) == 0)  // We hit and copied the terminating NULL.
+      return i;
+  }
+
+  // We were left off at dst_size.  We over copied 1 byte.  Null terminate.
+  if (dst_size != 0)
+    dst[dst_size - 1] = 0;
+
+  // Count the rest of the |src|, and return it's length in characters.
+  while (src[dst_size]) ++dst_size;
+  return dst_size;
+}
+
+}  // namespace
+
+size_t base::strlcpy(char* dst, const char* src, size_t dst_size) {
+  return lcpyT<char>(dst, src, dst_size);
+}
+size_t base::wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size) {
+  return lcpyT<wchar_t>(dst, src, dst_size);
+}
diff --git a/base/strings/string_util.h b/base/strings/string_util.h
new file mode 100644
index 0000000..7b4af7d
--- /dev/null
+++ b/base/strings/string_util.h
@@ -0,0 +1,576 @@
+// 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.
+//
+// This file defines utility functions for working with strings.
+
+#ifndef BASE_STRINGS_STRING_UTIL_H_
+#define BASE_STRINGS_STRING_UTIL_H_
+
+#include <ctype.h>
+#include <stdarg.h>   // va_list
+
+#include <string>
+#include <vector>
+
+#include "base/base_export.h"
+#include "base/basictypes.h"
+#include "base/compiler_specific.h"
+#include "base/string16.h"
+#include "base/strings/string_piece.h"  // For implicit conversions.
+
+// Safe standard library wrappers for all platforms.
+
+namespace base {
+
+// C standard-library functions like "strncasecmp" and "snprintf" that aren't
+// cross-platform are provided as "base::strncasecmp", and their prototypes
+// are listed below.  These functions are then implemented as inline calls
+// to the platform-specific equivalents in the platform-specific headers.
+
+// Compares the two strings s1 and s2 without regard to case using
+// the current locale; returns 0 if they are equal, 1 if s1 > s2, and -1 if
+// s2 > s1 according to a lexicographic comparison.
+int strcasecmp(const char* s1, const char* s2);
+
+// Compares up to count characters of s1 and s2 without regard to case using
+// the current locale; returns 0 if they are equal, 1 if s1 > s2, and -1 if
+// s2 > s1 according to a lexicographic comparison.
+int strncasecmp(const char* s1, const char* s2, size_t count);
+
+// Same as strncmp but for char16 strings.
+int strncmp16(const char16* s1, const char16* s2, size_t count);
+
+// Wrapper for vsnprintf that always null-terminates and always returns the
+// number of characters that would be in an untruncated formatted
+// string, even when truncation occurs.
+int vsnprintf(char* buffer, size_t size, const char* format, va_list arguments)
+    PRINTF_FORMAT(3, 0);
+
+// vswprintf always null-terminates, but when truncation occurs, it will either
+// return -1 or the number of characters that would be in an untruncated
+// formatted string.  The actual return value depends on the underlying
+// C library's vswprintf implementation.
+int vswprintf(wchar_t* buffer, size_t size,
+              const wchar_t* format, va_list arguments)
+    WPRINTF_FORMAT(3, 0);
+
+// Some of these implementations need to be inlined.
+
+// We separate the declaration from the implementation of this inline
+// function just so the PRINTF_FORMAT works.
+inline int snprintf(char* buffer, size_t size, const char* format, ...)
+    PRINTF_FORMAT(3, 4);
+inline int snprintf(char* buffer, size_t size, const char* format, ...) {
+  va_list arguments;
+  va_start(arguments, format);
+  int result = vsnprintf(buffer, size, format, arguments);
+  va_end(arguments);
+  return result;
+}
+
+// We separate the declaration from the implementation of this inline
+// function just so the WPRINTF_FORMAT works.
+inline int swprintf(wchar_t* buffer, size_t size, const wchar_t* format, ...)
+    WPRINTF_FORMAT(3, 4);
+inline int swprintf(wchar_t* buffer, size_t size, const wchar_t* format, ...) {
+  va_list arguments;
+  va_start(arguments, format);
+  int result = vswprintf(buffer, size, format, arguments);
+  va_end(arguments);
+  return result;
+}
+
+// BSD-style safe and consistent string copy functions.
+// Copies |src| to |dst|, where |dst_size| is the total allocated size of |dst|.
+// Copies at most |dst_size|-1 characters, and always NULL terminates |dst|, as
+// long as |dst_size| is not 0.  Returns the length of |src| in characters.
+// If the return value is >= dst_size, then the output was truncated.
+// NOTE: All sizes are in number of characters, NOT in bytes.
+BASE_EXPORT size_t strlcpy(char* dst, const char* src, size_t dst_size);
+BASE_EXPORT size_t wcslcpy(wchar_t* dst, const wchar_t* src, size_t dst_size);
+
+// Scan a wprintf format string to determine whether it's portable across a
+// variety of systems.  This function only checks that the conversion
+// specifiers used by the format string are supported and have the same meaning
+// on a variety of systems.  It doesn't check for other errors that might occur
+// within a format string.
+//
+// Nonportable conversion specifiers for wprintf are:
+//  - 's' and 'c' without an 'l' length modifier.  %s and %c operate on char
+//     data on all systems except Windows, which treat them as wchar_t data.
+//     Use %ls and %lc for wchar_t data instead.
+//  - 'S' and 'C', which operate on wchar_t data on all systems except Windows,
+//     which treat them as char data.  Use %ls and %lc for wchar_t data
+//     instead.
+//  - 'F', which is not identified by Windows wprintf documentation.
+//  - 'D', 'O', and 'U', which are deprecated and not available on all systems.
+//     Use %ld, %lo, and %lu instead.
+//
+// Note that there is no portable conversion specifier for char data when
+// working with wprintf.
+//
+// This function is intended to be called from base::vswprintf.
+BASE_EXPORT bool IsWprintfFormatPortable(const wchar_t* format);
+
+// ASCII-specific tolower.  The standard library's tolower is locale sensitive,
+// so we don't want to use it here.
+template <class Char> inline Char ToLowerASCII(Char c) {
+  return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c;
+}
+
+// ASCII-specific toupper.  The standard library's toupper is locale sensitive,
+// so we don't want to use it here.
+template <class Char> inline Char ToUpperASCII(Char c) {
+  return (c >= 'a' && c <= 'z') ? (c + ('A' - 'a')) : c;
+}
+
+// Function objects to aid in comparing/searching strings.
+
+template<typename Char> struct CaseInsensitiveCompare {
+ public:
+  bool operator()(Char x, Char y) const {
+    // TODO(darin): Do we really want to do locale sensitive comparisons here?
+    // See http://crbug.com/24917
+    return tolower(x) == tolower(y);
+  }
+};
+
+template<typename Char> struct CaseInsensitiveCompareASCII {
+ public:
+  bool operator()(Char x, Char y) const {
+    return ToLowerASCII(x) == ToLowerASCII(y);
+  }
+};
+
+}  // namespace base
+
+#if defined(OS_WIN)
+#include "base/string_util_win.h"
+#elif defined(OS_POSIX)
+#include "base/string_util_posix.h"
+#else
+#error Define string operations appropriately for your platform
+#endif
+
+// These threadsafe functions return references to globally unique empty
+// strings.
+//
+// DO NOT USE THESE AS A GENERAL-PURPOSE SUBSTITUTE FOR DEFAULT CONSTRUCTORS.
+// There is only one case where you should use these: functions which need to
+// return a string by reference (e.g. as a class member accessor), and don't
+// have an empty string to use (e.g. in an error case).  These should not be
+// used as initializers, function arguments, or return values for functions
+// which return by value or outparam.
+BASE_EXPORT const std::string& EmptyString();
+BASE_EXPORT const std::wstring& EmptyWString();
+BASE_EXPORT const string16& EmptyString16();
+
+BASE_EXPORT extern const wchar_t kWhitespaceWide[];
+BASE_EXPORT extern const char16 kWhitespaceUTF16[];
+BASE_EXPORT extern const char kWhitespaceASCII[];
+
+BASE_EXPORT extern const char kUtf8ByteOrderMark[];
+
+// Removes characters in |remove_chars| from anywhere in |input|.  Returns true
+// if any characters were removed.  |remove_chars| must be null-terminated.
+// NOTE: Safe to use the same variable for both |input| and |output|.
+BASE_EXPORT bool RemoveChars(const string16& input,
+                             const char16 remove_chars[],
+                             string16* output);
+BASE_EXPORT bool RemoveChars(const std::string& input,
+                             const char remove_chars[],
+                             std::string* output);
+
+// Replaces characters in |replace_chars| from anywhere in |input| with
+// |replace_with|.  Each character in |replace_chars| will be replaced with
+// the |replace_with| string.  Returns true if any characters were replaced.
+// |replace_chars| must be null-terminated.
+// NOTE: Safe to use the same variable for both |input| and |output|.
+BASE_EXPORT bool ReplaceChars(const string16& input,
+                              const char16 replace_chars[],
+                              const string16& replace_with,
+                              string16* output);
+BASE_EXPORT bool ReplaceChars(const std::string& input,
+                              const char replace_chars[],
+                              const std::string& replace_with,
+                              std::string* output);
+
+// Removes characters in |trim_chars| from the beginning and end of |input|.
+// |trim_chars| must be null-terminated.
+// NOTE: Safe to use the same variable for both |input| and |output|.
+BASE_EXPORT bool TrimString(const std::wstring& input,
+                            const wchar_t trim_chars[],
+                            std::wstring* output);
+BASE_EXPORT bool TrimString(const string16& input,
+                            const char16 trim_chars[],
+                            string16* output);
+BASE_EXPORT bool TrimString(const std::string& input,
+                            const char trim_chars[],
+                            std::string* output);
+
+// Truncates a string to the nearest UTF-8 character that will leave
+// the string less than or equal to the specified byte size.
+BASE_EXPORT void TruncateUTF8ToByteSize(const std::string& input,
+                                        const size_t byte_size,
+                                        std::string* output);
+
+// Trims any whitespace from either end of the input string.  Returns where
+// whitespace was found.
+// The non-wide version has two functions:
+// * TrimWhitespaceASCII()
+//   This function is for ASCII strings and only looks for ASCII whitespace;
+// Please choose the best one according to your usage.
+// NOTE: Safe to use the same variable for both input and output.
+enum TrimPositions {
+  TRIM_NONE     = 0,
+  TRIM_LEADING  = 1 << 0,
+  TRIM_TRAILING = 1 << 1,
+  TRIM_ALL      = TRIM_LEADING | TRIM_TRAILING,
+};
+BASE_EXPORT TrimPositions TrimWhitespace(const string16& input,
+                                         TrimPositions positions,
+                                         string16* output);
+BASE_EXPORT TrimPositions TrimWhitespaceASCII(const std::string& input,
+                                              TrimPositions positions,
+                                              std::string* output);
+
+// Deprecated. This function is only for backward compatibility and calls
+// TrimWhitespaceASCII().
+BASE_EXPORT TrimPositions TrimWhitespace(const std::string& input,
+                                         TrimPositions positions,
+                                         std::string* output);
+
+// Searches  for CR or LF characters.  Removes all contiguous whitespace
+// strings that contain them.  This is useful when trying to deal with text
+// copied from terminals.
+// Returns |text|, with the following three transformations:
+// (1) Leading and trailing whitespace is trimmed.
+// (2) If |trim_sequences_with_line_breaks| is true, any other whitespace
+//     sequences containing a CR or LF are trimmed.
+// (3) All other whitespace sequences are converted to single spaces.
+BASE_EXPORT std::wstring CollapseWhitespace(
+    const std::wstring& text,
+    bool trim_sequences_with_line_breaks);
+BASE_EXPORT string16 CollapseWhitespace(
+    const string16& text,
+    bool trim_sequences_with_line_breaks);
+BASE_EXPORT std::string CollapseWhitespaceASCII(
+    const std::string& text,
+    bool trim_sequences_with_line_breaks);
+
+// Returns true if the passed string is empty or contains only white-space
+// characters.
+BASE_EXPORT bool ContainsOnlyWhitespaceASCII(const std::string& str);
+BASE_EXPORT bool ContainsOnlyWhitespace(const string16& str);
+
+// Returns true if |input| is empty or contains only characters found in
+// |characters|.
+BASE_EXPORT bool ContainsOnlyChars(const std::wstring& input,
+                                   const std::wstring& characters);
+BASE_EXPORT bool ContainsOnlyChars(const string16& input,
+                                   const string16& characters);
+BASE_EXPORT bool ContainsOnlyChars(const std::string& input,
+                                   const std::string& characters);
+
+// Converts to 7-bit ASCII by truncating. The result must be known to be ASCII
+// beforehand.
+BASE_EXPORT std::string WideToASCII(const std::wstring& wide);
+BASE_EXPORT std::string UTF16ToASCII(const string16& utf16);
+
+// Converts the given wide string to the corresponding Latin1. This will fail
+// (return false) if any characters are more than 255.
+BASE_EXPORT bool WideToLatin1(const std::wstring& wide, std::string* latin1);
+
+// Returns true if the specified string matches the criteria. How can a wide
+// string be 8-bit or UTF8? It contains only characters that are < 256 (in the
+// first case) or characters that use only 8-bits and whose 8-bit
+// representation looks like a UTF-8 string (the second case).
+//
+// Note that IsStringUTF8 checks not only if the input is structurally
+// valid but also if it doesn't contain any non-character codepoint
+// (e.g. U+FFFE). It's done on purpose because all the existing callers want
+// to have the maximum 'discriminating' power from other encodings. If
+// there's a use case for just checking the structural validity, we have to
+// add a new function for that.
+BASE_EXPORT bool IsStringUTF8(const std::string& str);
+BASE_EXPORT bool IsStringASCII(const std::wstring& str);
+BASE_EXPORT bool IsStringASCII(const base::StringPiece& str);
+BASE_EXPORT bool IsStringASCII(const string16& str);
+
+// Converts the elements of the given string.  This version uses a pointer to
+// clearly differentiate it from the non-pointer variant.
+template <class str> inline void StringToLowerASCII(str* s) {
+  for (typename str::iterator i = s->begin(); i != s->end(); ++i)
+    *i = base::ToLowerASCII(*i);
+}
+
+template <class str> inline str StringToLowerASCII(const str& s) {
+  // for std::string and std::wstring
+  str output(s);
+  StringToLowerASCII(&output);
+  return output;
+}
+
+// Converts the elements of the given string.  This version uses a pointer to
+// clearly differentiate it from the non-pointer variant.
+template <class str> inline void StringToUpperASCII(str* s) {
+  for (typename str::iterator i = s->begin(); i != s->end(); ++i)
+    *i = base::ToUpperASCII(*i);
+}
+
+template <class str> inline str StringToUpperASCII(const str& s) {
+  // for std::string and std::wstring
+  str output(s);
+  StringToUpperASCII(&output);
+  return output;
+}
+
+// Compare the lower-case form of the given string against the given ASCII
+// string.  This is useful for doing checking if an input string matches some
+// token, and it is optimized to avoid intermediate string copies.  This API is
+// borrowed from the equivalent APIs in Mozilla.
+BASE_EXPORT bool LowerCaseEqualsASCII(const std::string& a, const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(const std::wstring& a, const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(const string16& a, const char* b);
+
+// Same thing, but with string iterators instead.
+BASE_EXPORT bool LowerCaseEqualsASCII(std::string::const_iterator a_begin,
+                                      std::string::const_iterator a_end,
+                                      const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(std::wstring::const_iterator a_begin,
+                                      std::wstring::const_iterator a_end,
+                                      const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(string16::const_iterator a_begin,
+                                      string16::const_iterator a_end,
+                                      const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(const char* a_begin,
+                                      const char* a_end,
+                                      const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(const wchar_t* a_begin,
+                                      const wchar_t* a_end,
+                                      const char* b);
+BASE_EXPORT bool LowerCaseEqualsASCII(const char16* a_begin,
+                                      const char16* a_end,
+                                      const char* b);
+
+// Performs a case-sensitive string compare. The behavior is undefined if both
+// strings are not ASCII.
+BASE_EXPORT bool EqualsASCII(const string16& a, const base::StringPiece& b);
+
+// Returns true if str starts with search, or false otherwise.
+BASE_EXPORT bool StartsWithASCII(const std::string& str,
+                                 const std::string& search,
+                                 bool case_sensitive);
+BASE_EXPORT bool StartsWith(const std::wstring& str,
+                            const std::wstring& search,
+                            bool case_sensitive);
+BASE_EXPORT bool StartsWith(const string16& str,
+                            const string16& search,
+                            bool case_sensitive);
+
+// Returns true if str ends with search, or false otherwise.
+BASE_EXPORT bool EndsWith(const std::string& str,
+                          const std::string& search,
+                          bool case_sensitive);
+BASE_EXPORT bool EndsWith(const std::wstring& str,
+                          const std::wstring& search,
+                          bool case_sensitive);
+BASE_EXPORT bool EndsWith(const string16& str,
+                          const string16& search,
+                          bool case_sensitive);
+
+
+// Determines the type of ASCII character, independent of locale (the C
+// library versions will change based on locale).
+template <typename Char>
+inline bool IsAsciiWhitespace(Char c) {
+  return c == ' ' || c == '\r' || c == '\n' || c == '\t';
+}
+template <typename Char>
+inline bool IsAsciiAlpha(Char c) {
+  return ((c >= 'A') && (c <= 'Z')) || ((c >= 'a') && (c <= 'z'));
+}
+template <typename Char>
+inline bool IsAsciiDigit(Char c) {
+  return c >= '0' && c <= '9';
+}
+
+template <typename Char>
+inline bool IsHexDigit(Char c) {
+  return (c >= '0' && c <= '9') ||
+         (c >= 'A' && c <= 'F') ||
+         (c >= 'a' && c <= 'f');
+}
+
+template <typename Char>
+inline Char HexDigitToInt(Char c) {
+  DCHECK(IsHexDigit(c));
+  if (c >= '0' && c <= '9')
+    return c - '0';
+  if (c >= 'A' && c <= 'F')
+    return c - 'A' + 10;
+  if (c >= 'a' && c <= 'f')
+    return c - 'a' + 10;
+  return 0;
+}
+
+// Returns true if it's a whitespace character.
+inline bool IsWhitespace(wchar_t c) {
+  return wcschr(kWhitespaceWide, c) != NULL;
+}
+
+// Return a byte string in human-readable format with a unit suffix. Not
+// appropriate for use in any UI; use of FormatBytes and friends in ui/base is
+// highly recommended instead. TODO(avi): Figure out how to get callers to use
+// FormatBytes instead; remove this.
+BASE_EXPORT string16 FormatBytesUnlocalized(int64 bytes);
+
+// Starting at |start_offset| (usually 0), replace the first instance of
+// |find_this| with |replace_with|.
+BASE_EXPORT void ReplaceFirstSubstringAfterOffset(
+    string16* str,
+    string16::size_type start_offset,
+    const string16& find_this,
+    const string16& replace_with);
+BASE_EXPORT void ReplaceFirstSubstringAfterOffset(
+    std::string* str,
+    std::string::size_type start_offset,
+    const std::string& find_this,
+    const std::string& replace_with);
+
+// Starting at |start_offset| (usually 0), look through |str| and replace all
+// instances of |find_this| with |replace_with|.
+//
+// This does entire substrings; use std::replace in <algorithm> for single
+// characters, for example:
+//   std::replace(str.begin(), str.end(), 'a', 'b');
+BASE_EXPORT void ReplaceSubstringsAfterOffset(
+    string16* str,
+    string16::size_type start_offset,
+    const string16& find_this,
+    const string16& replace_with);
+BASE_EXPORT void ReplaceSubstringsAfterOffset(
+    std::string* str,
+    std::string::size_type start_offset,
+    const std::string& find_this,
+    const std::string& replace_with);
+
+// Reserves enough memory in |str| to accommodate |length_with_null| characters,
+// sets the size of |str| to |length_with_null - 1| characters, and returns a
+// pointer to the underlying contiguous array of characters.  This is typically
+// used when calling a function that writes results into a character array, but
+// the caller wants the data to be managed by a string-like object.  It is
+// convenient in that is can be used inline in the call, and fast in that it
+// avoids copying the results of the call from a char* into a string.
+//
+// |length_with_null| must be at least 2, since otherwise the underlying string
+// would have size 0, and trying to access &((*str)[0]) in that case can result
+// in a number of problems.
+//
+// Internally, this takes linear time because the resize() call 0-fills the
+// underlying array for potentially all
+// (|length_with_null - 1| * sizeof(string_type::value_type)) bytes.  Ideally we
+// could avoid this aspect of the resize() call, as we expect the caller to
+// immediately write over this memory, but there is no other way to set the size
+// of the string, and not doing that will mean people who access |str| rather
+// than str.c_str() will get back a string of whatever size |str| had on entry
+// to this function (probably 0).
+template <class string_type>
+inline typename string_type::value_type* WriteInto(string_type* str,
+                                                   size_t length_with_null) {
+  DCHECK_GT(length_with_null, 1u);
+  str->reserve(length_with_null);
+  str->resize(length_with_null - 1);
+  return &((*str)[0]);
+}
+
+//-----------------------------------------------------------------------------
+
+// Splits a string into its fields delimited by any of the characters in
+// |delimiters|.  Each field is added to the |tokens| vector.  Returns the
+// number of tokens found.
+BASE_EXPORT size_t Tokenize(const std::wstring& str,
+                            const std::wstring& delimiters,
+                            std::vector<std::wstring>* tokens);
+BASE_EXPORT size_t Tokenize(const string16& str,
+                            const string16& delimiters,
+                            std::vector<string16>* tokens);
+BASE_EXPORT size_t Tokenize(const std::string& str,
+                            const std::string& delimiters,
+                            std::vector<std::string>* tokens);
+BASE_EXPORT size_t Tokenize(const base::StringPiece& str,
+                            const base::StringPiece& delimiters,
+                            std::vector<base::StringPiece>* tokens);
+
+// Does the opposite of SplitString().
+BASE_EXPORT string16 JoinString(const std::vector<string16>& parts, char16 s);
+BASE_EXPORT std::string JoinString(
+    const std::vector<std::string>& parts, char s);
+
+// Join |parts| using |separator|.
+BASE_EXPORT std::string JoinString(
+    const std::vector<std::string>& parts,
+    const std::string& separator);
+BASE_EXPORT string16 JoinString(
+    const std::vector<string16>& parts,
+    const string16& separator);
+
+// Replace $1-$2-$3..$9 in the format string with |a|-|b|-|c|..|i| respectively.
+// Additionally, any number of consecutive '$' characters is replaced by that
+// number less one. Eg $$->$, $$$->$$, etc. The offsets parameter here can be
+// NULL. This only allows you to use up to nine replacements.
+BASE_EXPORT string16 ReplaceStringPlaceholders(
+    const string16& format_string,
+    const std::vector<string16>& subst,
+    std::vector<size_t>* offsets);
+
+BASE_EXPORT std::string ReplaceStringPlaceholders(
+    const base::StringPiece& format_string,
+    const std::vector<std::string>& subst,
+    std::vector<size_t>* offsets);
+
+// Single-string shortcut for ReplaceStringHolders. |offset| may be NULL.
+BASE_EXPORT string16 ReplaceStringPlaceholders(const string16& format_string,
+                                               const string16& a,
+                                               size_t* offset);
+
+// Returns true if the string passed in matches the pattern. The pattern
+// string can contain wildcards like * and ?
+// The backslash character (\) is an escape character for * and ?
+// We limit the patterns to having a max of 16 * or ? characters.
+// ? matches 0 or 1 character, while * matches 0 or more characters.
+BASE_EXPORT bool MatchPattern(const base::StringPiece& string,
+                              const base::StringPiece& pattern);
+BASE_EXPORT bool MatchPattern(const string16& string, const string16& pattern);
+
+// Hack to convert any char-like type to its unsigned counterpart.
+// For example, it will convert char, signed char and unsigned char to unsigned
+// char.
+template<typename T>
+struct ToUnsigned {
+  typedef T Unsigned;
+};
+
+template<>
+struct ToUnsigned<char> {
+  typedef unsigned char Unsigned;
+};
+template<>
+struct ToUnsigned<signed char> {
+  typedef unsigned char Unsigned;
+};
+template<>
+struct ToUnsigned<wchar_t> {
+#if defined(WCHAR_T_IS_UTF16)
+  typedef unsigned short Unsigned;
+#elif defined(WCHAR_T_IS_UTF32)
+  typedef uint32 Unsigned;
+#endif
+};
+template<>
+struct ToUnsigned<short> {
+  typedef unsigned short Unsigned;
+};
+
+#endif  // BASE_STRINGS_STRING_UTIL_H_
diff --git a/base/strings/string_util_constants.cc b/base/strings/string_util_constants.cc
new file mode 100644
index 0000000..d92e40c
--- /dev/null
+++ b/base/strings/string_util_constants.cc
@@ -0,0 +1,55 @@
+// 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 "base/strings/string_util.h"
+
+#define WHITESPACE_UNICODE \
+  0x0009, /* <control-0009> to <control-000D> */ \
+  0x000A,                                        \
+  0x000B,                                        \
+  0x000C,                                        \
+  0x000D,                                        \
+  0x0020, /* Space */                            \
+  0x0085, /* <control-0085> */                   \
+  0x00A0, /* No-Break Space */                   \
+  0x1680, /* Ogham Space Mark */                 \
+  0x180E, /* Mongolian Vowel Separator */        \
+  0x2000, /* En Quad to Hair Space */            \
+  0x2001,                                        \
+  0x2002,                                        \
+  0x2003,                                        \
+  0x2004,                                        \
+  0x2005,                                        \
+  0x2006,                                        \
+  0x2007,                                        \
+  0x2008,                                        \
+  0x2009,                                        \
+  0x200A,                                        \
+  0x200C, /* Zero Width Non-Joiner */            \
+  0x2028, /* Line Separator */                   \
+  0x2029, /* Paragraph Separator */              \
+  0x202F, /* Narrow No-Break Space */            \
+  0x205F, /* Medium Mathematical Space */        \
+  0x3000, /* Ideographic Space */                \
+  0
+
+const wchar_t kWhitespaceWide[] = {
+  WHITESPACE_UNICODE
+};
+
+const char16 kWhitespaceUTF16[] = {
+  WHITESPACE_UNICODE
+};
+
+const char kWhitespaceASCII[] = {
+  0x09,    // <control-0009> to <control-000D>
+  0x0A,
+  0x0B,
+  0x0C,
+  0x0D,
+  0x20,    // Space
+  0
+};
+
+const char kUtf8ByteOrderMark[] = "\xEF\xBB\xBF";
diff --git a/base/strings/string_util_posix.h b/base/strings/string_util_posix.h
new file mode 100644
index 0000000..34b14f1
--- /dev/null
+++ b/base/strings/string_util_posix.h
@@ -0,0 +1,53 @@
+// 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 BASE_STRINGS_STRING_UTIL_POSIX_H_
+#define BASE_STRINGS_STRING_UTIL_POSIX_H_
+
+#include <stdarg.h>
+#include <stdio.h>
+#include <string.h>
+#include <wchar.h>
+
+#include "base/logging.h"
+#include "base/strings/string_util.h"
+
+namespace base {
+
+// Chromium code style is to not use malloc'd strings; this is only for use
+// for interaction with APIs that require it.
+inline char* strdup(const char* str) {
+  return ::strdup(str);
+}
+
+inline int strcasecmp(const char* string1, const char* string2) {
+  return ::strcasecmp(string1, string2);
+}
+
+inline int strncasecmp(const char* string1, const char* string2, size_t count) {
+  return ::strncasecmp(string1, string2, count);
+}
+
+inline int vsnprintf(char* buffer, size_t size,
+                     const char* format, va_list arguments) {
+  return ::vsnprintf(buffer, size, format, arguments);
+}
+
+inline int strncmp16(const char16* s1, const char16* s2, size_t count) {
+#if defined(WCHAR_T_IS_UTF16)
+  return ::wcsncmp(s1, s2, count);
+#elif defined(WCHAR_T_IS_UTF32)
+  return c16memcmp(s1, s2, count);
+#endif
+}
+
+inline int vswprintf(wchar_t* buffer, size_t size,
+                     const wchar_t* format, va_list arguments) {
+  DCHECK(IsWprintfFormatPortable(format));
+  return ::vswprintf(buffer, size, format, arguments);
+}
+
+}  // namespace base
+
+#endif  // BASE_STRINGS_STRING_UTIL_POSIX_H_
diff --git a/base/strings/string_util_unittest.cc b/base/strings/string_util_unittest.cc
new file mode 100644
index 0000000..58b7620
--- /dev/null
+++ b/base/strings/string_util_unittest.cc
@@ -0,0 +1,1191 @@
+// 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 "base/strings/string_util.h"
+
+#include <math.h>
+#include <stdarg.h>
+
+#include <limits>
+#include <sstream>
+
+#include "base/basictypes.h"
+#include "base/strings/string16.h"
+#include "base/strings/utf_string_conversions.h"
+#include "testing/gmock/include/gmock/gmock.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+using ::testing::ElementsAre;
+
+namespace base {
+
+static const struct trim_case {
+  const wchar_t* input;
+  const TrimPositions positions;
+  const wchar_t* output;
+  const TrimPositions return_value;
+} trim_cases[] = {
+  {L" Google Video ", TRIM_LEADING, L"Google Video ", TRIM_LEADING},
+  {L" Google Video ", TRIM_TRAILING, L" Google Video", TRIM_TRAILING},
+  {L" Google Video ", TRIM_ALL, L"Google Video", TRIM_ALL},
+  {L"Google Video", TRIM_ALL, L"Google Video", TRIM_NONE},
+  {L"", TRIM_ALL, L"", TRIM_NONE},
+  {L"  ", TRIM_LEADING, L"", TRIM_LEADING},
+  {L"  ", TRIM_TRAILING, L"", TRIM_TRAILING},
+  {L"  ", TRIM_ALL, L"", TRIM_ALL},
+  {L"\t\rTest String\n", TRIM_ALL, L"Test String", TRIM_ALL},
+  {L"\x2002Test String\x00A0\x3000", TRIM_ALL, L"Test String", TRIM_ALL},
+};
+
+static const struct trim_case_ascii {
+  const char* input;
+  const TrimPositions positions;
+  const char* output;
+  const TrimPositions return_value;
+} trim_cases_ascii[] = {
+  {" Google Video ", TRIM_LEADING, "Google Video ", TRIM_LEADING},
+  {" Google Video ", TRIM_TRAILING, " Google Video", TRIM_TRAILING},
+  {" Google Video ", TRIM_ALL, "Google Video", TRIM_ALL},
+  {"Google Video", TRIM_ALL, "Google Video", TRIM_NONE},
+  {"", TRIM_ALL, "", TRIM_NONE},
+  {"  ", TRIM_LEADING, "", TRIM_LEADING},
+  {"  ", TRIM_TRAILING, "", TRIM_TRAILING},
+  {"  ", TRIM_ALL, "", TRIM_ALL},
+  {"\t\rTest String\n", TRIM_ALL, "Test String", TRIM_ALL},
+};
+
+namespace {
+
+// Helper used to test TruncateUTF8ToByteSize.
+bool Truncated(const std::string& input, const size_t byte_size,
+               std::string* output) {
+    size_t prev = input.length();
+    TruncateUTF8ToByteSize(input, byte_size, output);
+    return prev != output->length();
+}
+
+}  // namespace
+
+TEST(StringUtilTest, TruncateUTF8ToByteSize) {
+  std::string output;
+
+  // Empty strings and invalid byte_size arguments
+  EXPECT_FALSE(Truncated(std::string(), 0, &output));
+  EXPECT_EQ(output, "");
+  EXPECT_TRUE(Truncated("\xe1\x80\xbf", 0, &output));
+  EXPECT_EQ(output, "");
+  EXPECT_FALSE(Truncated("\xe1\x80\xbf", -1, &output));
+  EXPECT_FALSE(Truncated("\xe1\x80\xbf", 4, &output));
+
+  // Testing the truncation of valid UTF8 correctly
+  EXPECT_TRUE(Truncated("abc", 2, &output));
+  EXPECT_EQ(output, "ab");
+  EXPECT_TRUE(Truncated("\xc2\x81\xc2\x81", 2, &output));
+  EXPECT_EQ(output.compare("\xc2\x81"), 0);
+  EXPECT_TRUE(Truncated("\xc2\x81\xc2\x81", 3, &output));
+  EXPECT_EQ(output.compare("\xc2\x81"), 0);
+  EXPECT_FALSE(Truncated("\xc2\x81\xc2\x81", 4, &output));
+  EXPECT_EQ(output.compare("\xc2\x81\xc2\x81"), 0);
+
+  {
+    const char array[] = "\x00\x00\xc2\x81\xc2\x81";
+    const std::string array_string(array, arraysize(array));
+    EXPECT_TRUE(Truncated(array_string, 4, &output));
+    EXPECT_EQ(output.compare(std::string("\x00\x00\xc2\x81", 4)), 0);
+  }
+
+  {
+    const char array[] = "\x00\xc2\x81\xc2\x81";
+    const std::string array_string(array, arraysize(array));
+    EXPECT_TRUE(Truncated(array_string, 4, &output));
+    EXPECT_EQ(output.compare(std::string("\x00\xc2\x81", 3)), 0);
+  }
+
+  // Testing invalid UTF8
+  EXPECT_TRUE(Truncated("\xed\xa0\x80\xed\xbf\xbf", 6, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xed\xa0\x8f", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xed\xbf\xbf", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  // Testing invalid UTF8 mixed with valid UTF8
+  EXPECT_FALSE(Truncated("\xe1\x80\xbf", 3, &output));
+  EXPECT_EQ(output.compare("\xe1\x80\xbf"), 0);
+  EXPECT_FALSE(Truncated("\xf1\x80\xa0\xbf", 4, &output));
+  EXPECT_EQ(output.compare("\xf1\x80\xa0\xbf"), 0);
+  EXPECT_FALSE(Truncated("a\xc2\x81\xe1\x80\xbf\xf1\x80\xa0\xbf",
+              10, &output));
+  EXPECT_EQ(output.compare("a\xc2\x81\xe1\x80\xbf\xf1\x80\xa0\xbf"), 0);
+  EXPECT_TRUE(Truncated("a\xc2\x81\xe1\x80\xbf\xf1""a""\x80\xa0",
+              10, &output));
+  EXPECT_EQ(output.compare("a\xc2\x81\xe1\x80\xbf\xf1""a"), 0);
+  EXPECT_FALSE(Truncated("\xef\xbb\xbf" "abc", 6, &output));
+  EXPECT_EQ(output.compare("\xef\xbb\xbf" "abc"), 0);
+
+  // Overlong sequences
+  EXPECT_TRUE(Truncated("\xc0\x80", 2, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xc1\x80\xc1\x81", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xe0\x80\x80", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xe0\x82\x80", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xe0\x9f\xbf", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf0\x80\x80\x8D", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf0\x80\x82\x91", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf0\x80\xa0\x80", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf0\x8f\xbb\xbf", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf8\x80\x80\x80\xbf", 5, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xfc\x80\x80\x80\xa0\xa5", 6, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  // Beyond U+10FFFF (the upper limit of Unicode codespace)
+  EXPECT_TRUE(Truncated("\xf4\x90\x80\x80", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf8\xa0\xbf\x80\xbf", 5, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xfc\x9c\xbf\x80\xbf\x80", 6, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  // BOMs in UTF-16(BE|LE) and UTF-32(BE|LE)
+  EXPECT_TRUE(Truncated("\xfe\xff", 2, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xff\xfe", 2, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  {
+    const char array[] = "\x00\x00\xfe\xff";
+    const std::string array_string(array, arraysize(array));
+    EXPECT_TRUE(Truncated(array_string, 4, &output));
+    EXPECT_EQ(output.compare(std::string("\x00\x00", 2)), 0);
+  }
+
+  // Variants on the previous test
+  {
+    const char array[] = "\xff\xfe\x00\x00";
+    const std::string array_string(array, 4);
+    EXPECT_FALSE(Truncated(array_string, 4, &output));
+    EXPECT_EQ(output.compare(std::string("\xff\xfe\x00\x00", 4)), 0);
+  }
+  {
+    const char array[] = "\xff\x00\x00\xfe";
+    const std::string array_string(array, arraysize(array));
+    EXPECT_TRUE(Truncated(array_string, 4, &output));
+    EXPECT_EQ(output.compare(std::string("\xff\x00\x00", 3)), 0);
+  }
+
+  // Non-characters : U+xxFFF[EF] where xx is 0x00 through 0x10 and <FDD0,FDEF>
+  EXPECT_TRUE(Truncated("\xef\xbf\xbe", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf0\x8f\xbf\xbe", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xf3\xbf\xbf\xbf", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xef\xb7\x90", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_TRUE(Truncated("\xef\xb7\xaf", 3, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  // Strings in legacy encodings that are valid in UTF-8, but
+  // are invalid as UTF-8 in real data.
+  EXPECT_TRUE(Truncated("caf\xe9", 4, &output));
+  EXPECT_EQ(output.compare("caf"), 0);
+  EXPECT_TRUE(Truncated("\xb0\xa1\xb0\xa2", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+  EXPECT_FALSE(Truncated("\xa7\x41\xa6\x6e", 4, &output));
+  EXPECT_EQ(output.compare("\xa7\x41\xa6\x6e"), 0);
+  EXPECT_TRUE(Truncated("\xa7\x41\xa6\x6e\xd9\xee\xe4\xee", 7,
+              &output));
+  EXPECT_EQ(output.compare("\xa7\x41\xa6\x6e"), 0);
+
+  // Testing using the same string as input and output.
+  EXPECT_FALSE(Truncated(output, 4, &output));
+  EXPECT_EQ(output.compare("\xa7\x41\xa6\x6e"), 0);
+  EXPECT_TRUE(Truncated(output, 3, &output));
+  EXPECT_EQ(output.compare("\xa7\x41"), 0);
+
+  // "abc" with U+201[CD] in windows-125[0-8]
+  EXPECT_TRUE(Truncated("\x93" "abc\x94", 5, &output));
+  EXPECT_EQ(output.compare("\x93" "abc"), 0);
+
+  // U+0639 U+064E U+0644 U+064E in ISO-8859-6
+  EXPECT_TRUE(Truncated("\xd9\xee\xe4\xee", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+
+  // U+03B3 U+03B5 U+03B9 U+03AC in ISO-8859-7
+  EXPECT_TRUE(Truncated("\xe3\xe5\xe9\xdC", 4, &output));
+  EXPECT_EQ(output.compare(""), 0);
+}
+
+TEST(StringUtilTest, TrimWhitespace) {
+  string16 output;  // Allow contents to carry over to next testcase
+  for (size_t i = 0; i < arraysize(trim_cases); ++i) {
+    const trim_case& value = trim_cases[i];
+    EXPECT_EQ(value.return_value,
+              TrimWhitespace(WideToUTF16(value.input), value.positions,
+                             &output));
+    EXPECT_EQ(WideToUTF16(value.output), output);
+  }
+
+  // Test that TrimWhitespace() can take the same string for input and output
+  output = ASCIIToUTF16("  This is a test \r\n");
+  EXPECT_EQ(TRIM_ALL, TrimWhitespace(output, TRIM_ALL, &output));
+  EXPECT_EQ(ASCIIToUTF16("This is a test"), output);
+
+  // Once more, but with a string of whitespace
+  output = ASCIIToUTF16("  \r\n");
+  EXPECT_EQ(TRIM_ALL, TrimWhitespace(output, TRIM_ALL, &output));
+  EXPECT_EQ(string16(), output);
+
+  std::string output_ascii;
+  for (size_t i = 0; i < arraysize(trim_cases_ascii); ++i) {
+    const trim_case_ascii& value = trim_cases_ascii[i];
+    EXPECT_EQ(value.return_value,
+              TrimWhitespace(value.input, value.positions, &output_ascii));
+    EXPECT_EQ(value.output, output_ascii);
+  }
+}
+
+static const struct collapse_case {
+  const wchar_t* input;
+  const bool trim;
+  const wchar_t* output;
+} collapse_cases[] = {
+  {L" Google Video ", false, L"Google Video"},
+  {L"Google Video", false, L"Google Video"},
+  {L"", false, L""},
+  {L"  ", false, L""},
+  {L"\t\rTest String\n", false, L"Test String"},
+  {L"\x2002Test String\x00A0\x3000", false, L"Test String"},
+  {L"    Test     \n  \t String    ", false, L"Test String"},
+  {L"\x2002Test\x1680 \x2028 \tString\x00A0\x3000", false, L"Test String"},
+  {L"   Test String", false, L"Test String"},
+  {L"Test String    ", false, L"Test String"},
+  {L"Test String", false, L"Test String"},
+  {L"", true, L""},
+  {L"\n", true, L""},
+  {L"  \r  ", true, L""},
+  {L"\nFoo", true, L"Foo"},
+  {L"\r  Foo  ", true, L"Foo"},
+  {L" Foo bar ", true, L"Foo bar"},
+  {L"  \tFoo  bar  \n", true, L"Foo bar"},
+  {L" a \r b\n c \r\n d \t\re \t f \n ", true, L"abcde f"},
+};
+
+TEST(StringUtilTest, CollapseWhitespace) {
+  for (size_t i = 0; i < arraysize(collapse_cases); ++i) {
+    const collapse_case& value = collapse_cases[i];
+    EXPECT_EQ(value.output, CollapseWhitespace(value.input, value.trim));
+  }
+}
+
+static const struct collapse_case_ascii {
+  const char* input;
+  const bool trim;
+  const char* output;
+} collapse_cases_ascii[] = {
+  {" Google Video ", false, "Google Video"},
+  {"Google Video", false, "Google Video"},
+  {"", false, ""},
+  {"  ", false, ""},
+  {"\t\rTest String\n", false, "Test String"},
+  {"    Test     \n  \t String    ", false, "Test String"},
+  {"   Test String", false, "Test String"},
+  {"Test String    ", false, "Test String"},
+  {"Test String", false, "Test String"},
+  {"", true, ""},
+  {"\n", true, ""},
+  {"  \r  ", true, ""},
+  {"\nFoo", true, "Foo"},
+  {"\r  Foo  ", true, "Foo"},
+  {" Foo bar ", true, "Foo bar"},
+  {"  \tFoo  bar  \n", true, "Foo bar"},
+  {" a \r b\n c \r\n d \t\re \t f \n ", true, "abcde f"},
+};
+
+TEST(StringUtilTest, CollapseWhitespaceASCII) {
+  for (size_t i = 0; i < arraysize(collapse_cases_ascii); ++i) {
+    const collapse_case_ascii& value = collapse_cases_ascii[i];
+    EXPECT_EQ(value.output, CollapseWhitespaceASCII(value.input, value.trim));
+  }
+}
+
+TEST(StringUtilTest, ContainsOnlyWhitespaceASCII) {
+  EXPECT_TRUE(ContainsOnlyWhitespaceASCII(std::string()));
+  EXPECT_TRUE(ContainsOnlyWhitespaceASCII(" "));
+  EXPECT_TRUE(ContainsOnlyWhitespaceASCII("\t"));
+  EXPECT_TRUE(ContainsOnlyWhitespaceASCII("\t \r \n  "));
+  EXPECT_FALSE(ContainsOnlyWhitespaceASCII("a"));
+  EXPECT_FALSE(ContainsOnlyWhitespaceASCII("\thello\r \n  "));
+}
+
+TEST(StringUtilTest, ContainsOnlyWhitespace) {
+  EXPECT_TRUE(ContainsOnlyWhitespace(string16()));
+  EXPECT_TRUE(ContainsOnlyWhitespace(ASCIIToUTF16(" ")));
+  EXPECT_TRUE(ContainsOnlyWhitespace(ASCIIToUTF16("\t")));
+  EXPECT_TRUE(ContainsOnlyWhitespace(ASCIIToUTF16("\t \r \n  ")));
+  EXPECT_FALSE(ContainsOnlyWhitespace(ASCIIToUTF16("a")));
+  EXPECT_FALSE(ContainsOnlyWhitespace(ASCIIToUTF16("\thello\r \n  ")));
+}
+
+TEST(StringUtilTest, IsStringUTF8) {
+  EXPECT_TRUE(IsStringUTF8("abc"));
+  EXPECT_TRUE(IsStringUTF8("\xc2\x81"));
+  EXPECT_TRUE(IsStringUTF8("\xe1\x80\xbf"));
+  EXPECT_TRUE(IsStringUTF8("\xf1\x80\xa0\xbf"));
+  EXPECT_TRUE(IsStringUTF8("a\xc2\x81\xe1\x80\xbf\xf1\x80\xa0\xbf"));
+  EXPECT_TRUE(IsStringUTF8("\xef\xbb\xbf" "abc"));  // UTF-8 BOM
+
+  // surrogate code points
+  EXPECT_FALSE(IsStringUTF8("\xed\xa0\x80\xed\xbf\xbf"));
+  EXPECT_FALSE(IsStringUTF8("\xed\xa0\x8f"));
+  EXPECT_FALSE(IsStringUTF8("\xed\xbf\xbf"));
+
+  // overlong sequences
+  EXPECT_FALSE(IsStringUTF8("\xc0\x80"));  // U+0000
+  EXPECT_FALSE(IsStringUTF8("\xc1\x80\xc1\x81"));  // "AB"
+  EXPECT_FALSE(IsStringUTF8("\xe0\x80\x80"));  // U+0000
+  EXPECT_FALSE(IsStringUTF8("\xe0\x82\x80"));  // U+0080
+  EXPECT_FALSE(IsStringUTF8("\xe0\x9f\xbf"));  // U+07ff
+  EXPECT_FALSE(IsStringUTF8("\xf0\x80\x80\x8D"));  // U+000D
+  EXPECT_FALSE(IsStringUTF8("\xf0\x80\x82\x91"));  // U+0091
+  EXPECT_FALSE(IsStringUTF8("\xf0\x80\xa0\x80"));  // U+0800
+  EXPECT_FALSE(IsStringUTF8("\xf0\x8f\xbb\xbf"));  // U+FEFF (BOM)
+  EXPECT_FALSE(IsStringUTF8("\xf8\x80\x80\x80\xbf"));  // U+003F
+  EXPECT_FALSE(IsStringUTF8("\xfc\x80\x80\x80\xa0\xa5"));  // U+00A5
+
+  // Beyond U+10FFFF (the upper limit of Unicode codespace)
+  EXPECT_FALSE(IsStringUTF8("\xf4\x90\x80\x80"));  // U+110000
+  EXPECT_FALSE(IsStringUTF8("\xf8\xa0\xbf\x80\xbf"));  // 5 bytes
+  EXPECT_FALSE(IsStringUTF8("\xfc\x9c\xbf\x80\xbf\x80"));  // 6 bytes
+
+  // BOMs in UTF-16(BE|LE) and UTF-32(BE|LE)
+  EXPECT_FALSE(IsStringUTF8("\xfe\xff"));
+  EXPECT_FALSE(IsStringUTF8("\xff\xfe"));
+  EXPECT_FALSE(IsStringUTF8(std::string("\x00\x00\xfe\xff", 4)));
+  EXPECT_FALSE(IsStringUTF8("\xff\xfe\x00\x00"));
+
+  // Non-characters : U+xxFFF[EF] where xx is 0x00 through 0x10 and <FDD0,FDEF>
+  EXPECT_FALSE(IsStringUTF8("\xef\xbf\xbe"));  // U+FFFE)
+  EXPECT_FALSE(IsStringUTF8("\xf0\x8f\xbf\xbe"));  // U+1FFFE
+  EXPECT_FALSE(IsStringUTF8("\xf3\xbf\xbf\xbf"));  // U+10FFFF
+  EXPECT_FALSE(IsStringUTF8("\xef\xb7\x90"));  // U+FDD0
+  EXPECT_FALSE(IsStringUTF8("\xef\xb7\xaf"));  // U+FDEF
+  // Strings in legacy encodings. We can certainly make up strings
+  // in a legacy encoding that are valid in UTF-8, but in real data,
+  // most of them are invalid as UTF-8.
+  EXPECT_FALSE(IsStringUTF8("caf\xe9"));  // cafe with U+00E9 in ISO-8859-1
+  EXPECT_FALSE(IsStringUTF8("\xb0\xa1\xb0\xa2"));  // U+AC00, U+AC001 in EUC-KR
+  EXPECT_FALSE(IsStringUTF8("\xa7\x41\xa6\x6e"));  // U+4F60 U+597D in Big5
+  // "abc" with U+201[CD] in windows-125[0-8]
+  EXPECT_FALSE(IsStringUTF8("\x93" "abc\x94"));
+  // U+0639 U+064E U+0644 U+064E in ISO-8859-6
+  EXPECT_FALSE(IsStringUTF8("\xd9\xee\xe4\xee"));
+  // U+03B3 U+03B5 U+03B9 U+03AC in ISO-8859-7
+  EXPECT_FALSE(IsStringUTF8("\xe3\xe5\xe9\xdC"));
+
+  // Check that we support Embedded Nulls. The first uses the canonical UTF-8
+  // representation, and the second uses a 2-byte sequence. The second version
+  // is invalid UTF-8 since UTF-8 states that the shortest encoding for a
+  // given codepoint must be used.
+  static const char kEmbeddedNull[] = "embedded\0null";
+  EXPECT_TRUE(IsStringUTF8(
+      std::string(kEmbeddedNull, sizeof(kEmbeddedNull))));
+  EXPECT_FALSE(IsStringUTF8("embedded\xc0\x80U+0000"));
+}
+
+TEST(StringUtilTest, ConvertASCII) {
+  static const char* char_cases[] = {
+    "Google Video",
+    "Hello, world\n",
+    "0123ABCDwxyz \a\b\t\r\n!+,.~"
+  };
+
+  static const wchar_t* const wchar_cases[] = {
+    L"Google Video",
+    L"Hello, world\n",
+    L"0123ABCDwxyz \a\b\t\r\n!+,.~"
+  };
+
+  for (size_t i = 0; i < arraysize(char_cases); ++i) {
+    EXPECT_TRUE(IsStringASCII(char_cases[i]));
+    std::wstring wide = ASCIIToWide(char_cases[i]);
+    EXPECT_EQ(wchar_cases[i], wide);
+
+    EXPECT_TRUE(IsStringASCII(wchar_cases[i]));
+    std::string ascii = WideToASCII(wchar_cases[i]);
+    EXPECT_EQ(char_cases[i], ascii);
+  }
+
+  EXPECT_FALSE(IsStringASCII("Google \x80Video"));
+  EXPECT_FALSE(IsStringASCII(L"Google \x80Video"));
+
+  // Convert empty strings.
+  std::wstring wempty;
+  std::string empty;
+  EXPECT_EQ(empty, WideToASCII(wempty));
+  EXPECT_EQ(wempty, ASCIIToWide(empty));
+
+  // Convert strings with an embedded NUL character.
+  const char chars_with_nul[] = "test\0string";
+  const int length_with_nul = arraysize(chars_with_nul) - 1;
+  std::string string_with_nul(chars_with_nul, length_with_nul);
+  std::wstring wide_with_nul = ASCIIToWide(string_with_nul);
+  EXPECT_EQ(static_cast<std::wstring::size_type>(length_with_nul),
+            wide_with_nul.length());
+  std::string narrow_with_nul = WideToASCII(wide_with_nul);
+  EXPECT_EQ(static_cast<std::string::size_type>(length_with_nul),
+            narrow_with_nul.length());
+  EXPECT_EQ(0, string_with_nul.compare(narrow_with_nul));
+}
+
+TEST(StringUtilTest, ToUpperASCII) {
+  EXPECT_EQ('C', ToUpperASCII('C'));
+  EXPECT_EQ('C', ToUpperASCII('c'));
+  EXPECT_EQ('2', ToUpperASCII('2'));
+
+  EXPECT_EQ(L'C', ToUpperASCII(L'C'));
+  EXPECT_EQ(L'C', ToUpperASCII(L'c'));
+  EXPECT_EQ(L'2', ToUpperASCII(L'2'));
+
+  std::string in_place_a("Cc2");
+  StringToUpperASCII(&in_place_a);
+  EXPECT_EQ("CC2", in_place_a);
+
+  std::wstring in_place_w(L"Cc2");
+  StringToUpperASCII(&in_place_w);
+  EXPECT_EQ(L"CC2", in_place_w);
+
+  std::string original_a("Cc2");
+  std::string upper_a = StringToUpperASCII(original_a);
+  EXPECT_EQ("CC2", upper_a);
+
+  std::wstring original_w(L"Cc2");
+  std::wstring upper_w = StringToUpperASCII(original_w);
+  EXPECT_EQ(L"CC2", upper_w);
+}
+
+TEST(StringUtilTest, LowerCaseEqualsASCII) {
+  static const struct {
+    const wchar_t* src_w;
+    const char*    src_a;
+    const char*    dst;
+  } lowercase_cases[] = {
+    { L"FoO", "FoO", "foo" },
+    { L"foo", "foo", "foo" },
+    { L"FOO", "FOO", "foo" },
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(lowercase_cases); ++i) {
+    EXPECT_TRUE(LowerCaseEqualsASCII(lowercase_cases[i].src_w,
+                                     lowercase_cases[i].dst));
+    EXPECT_TRUE(LowerCaseEqualsASCII(lowercase_cases[i].src_a,
+                                     lowercase_cases[i].dst));
+  }
+}
+
+TEST(StringUtilTest, FormatBytesUnlocalized) {
+  static const struct {
+    int64 bytes;
+    const char* expected;
+  } cases[] = {
+    // Expected behavior: we show one post-decimal digit when we have
+    // under two pre-decimal digits, except in cases where it makes no
+    // sense (zero or bytes).
+    // Since we switch units once we cross the 1000 mark, this keeps
+    // the display of file sizes or bytes consistently around three
+    // digits.
+    {0, "0 B"},
+    {512, "512 B"},
+    {1024*1024, "1.0 MB"},
+    {1024*1024*1024, "1.0 GB"},
+    {10LL*1024*1024*1024, "10.0 GB"},
+    {99LL*1024*1024*1024, "99.0 GB"},
+    {105LL*1024*1024*1024, "105 GB"},
+    {105LL*1024*1024*1024 + 500LL*1024*1024, "105 GB"},
+    {~(1LL<<63), "8192 PB"},
+
+    {99*1024 + 103, "99.1 kB"},
+    {1024*1024 + 103, "1.0 MB"},
+    {1024*1024 + 205 * 1024, "1.2 MB"},
+    {1024*1024*1024 + (927 * 1024*1024), "1.9 GB"},
+    {10LL*1024*1024*1024, "10.0 GB"},
+    {100LL*1024*1024*1024, "100 GB"},
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(cases); ++i) {
+    EXPECT_EQ(ASCIIToUTF16(cases[i].expected),
+              FormatBytesUnlocalized(cases[i].bytes));
+  }
+}
+TEST(StringUtilTest, ReplaceSubstringsAfterOffset) {
+  static const struct {
+    const char* str;
+    string16::size_type start_offset;
+    const char* find_this;
+    const char* replace_with;
+    const char* expected;
+  } cases[] = {
+    {"aaa", 0, "a", "b", "bbb"},
+    {"abb", 0, "ab", "a", "ab"},
+    {"Removing some substrings inging", 0, "ing", "", "Remov some substrs "},
+    {"Not found", 0, "x", "0", "Not found"},
+    {"Not found again", 5, "x", "0", "Not found again"},
+    {" Making it much longer ", 0, " ", "Four score and seven years ago",
+     "Four score and seven years agoMakingFour score and seven years agoit"
+     "Four score and seven years agomuchFour score and seven years agolonger"
+     "Four score and seven years ago"},
+    {"Invalid offset", 9999, "t", "foobar", "Invalid offset"},
+    {"Replace me only me once", 9, "me ", "", "Replace me only once"},
+    {"abababab", 2, "ab", "c", "abccc"},
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(cases); i++) {
+    string16 str = ASCIIToUTF16(cases[i].str);
+    ReplaceSubstringsAfterOffset(&str, cases[i].start_offset,
+                                 ASCIIToUTF16(cases[i].find_this),
+                                 ASCIIToUTF16(cases[i].replace_with));
+    EXPECT_EQ(ASCIIToUTF16(cases[i].expected), str);
+  }
+}
+
+TEST(StringUtilTest, ReplaceFirstSubstringAfterOffset) {
+  static const struct {
+    const char* str;
+    string16::size_type start_offset;
+    const char* find_this;
+    const char* replace_with;
+    const char* expected;
+  } cases[] = {
+    {"aaa", 0, "a", "b", "baa"},
+    {"abb", 0, "ab", "a", "ab"},
+    {"Removing some substrings inging", 0, "ing", "",
+      "Remov some substrings inging"},
+    {"Not found", 0, "x", "0", "Not found"},
+    {"Not found again", 5, "x", "0", "Not found again"},
+    {" Making it much longer ", 0, " ", "Four score and seven years ago",
+     "Four score and seven years agoMaking it much longer "},
+    {"Invalid offset", 9999, "t", "foobar", "Invalid offset"},
+    {"Replace me only me once", 4, "me ", "", "Replace only me once"},
+    {"abababab", 2, "ab", "c", "abcabab"},
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(cases); i++) {
+    string16 str = ASCIIToUTF16(cases[i].str);
+    ReplaceFirstSubstringAfterOffset(&str, cases[i].start_offset,
+                                     ASCIIToUTF16(cases[i].find_this),
+                                     ASCIIToUTF16(cases[i].replace_with));
+    EXPECT_EQ(ASCIIToUTF16(cases[i].expected), str);
+  }
+}
+
+TEST(StringUtilTest, HexDigitToInt) {
+  EXPECT_EQ(0, HexDigitToInt('0'));
+  EXPECT_EQ(1, HexDigitToInt('1'));
+  EXPECT_EQ(2, HexDigitToInt('2'));
+  EXPECT_EQ(3, HexDigitToInt('3'));
+  EXPECT_EQ(4, HexDigitToInt('4'));
+  EXPECT_EQ(5, HexDigitToInt('5'));
+  EXPECT_EQ(6, HexDigitToInt('6'));
+  EXPECT_EQ(7, HexDigitToInt('7'));
+  EXPECT_EQ(8, HexDigitToInt('8'));
+  EXPECT_EQ(9, HexDigitToInt('9'));
+  EXPECT_EQ(10, HexDigitToInt('A'));
+  EXPECT_EQ(11, HexDigitToInt('B'));
+  EXPECT_EQ(12, HexDigitToInt('C'));
+  EXPECT_EQ(13, HexDigitToInt('D'));
+  EXPECT_EQ(14, HexDigitToInt('E'));
+  EXPECT_EQ(15, HexDigitToInt('F'));
+
+  // Verify the lower case as well.
+  EXPECT_EQ(10, HexDigitToInt('a'));
+  EXPECT_EQ(11, HexDigitToInt('b'));
+  EXPECT_EQ(12, HexDigitToInt('c'));
+  EXPECT_EQ(13, HexDigitToInt('d'));
+  EXPECT_EQ(14, HexDigitToInt('e'));
+  EXPECT_EQ(15, HexDigitToInt('f'));
+}
+
+// This checks where we can use the assignment operator for a va_list. We need
+// a way to do this since Visual C doesn't support va_copy, but assignment on
+// va_list is not guaranteed to be a copy. See StringAppendVT which uses this
+// capability.
+static void VariableArgsFunc(const char* format, ...) {
+  va_list org;
+  va_start(org, format);
+
+  va_list dup;
+  GG_VA_COPY(dup, org);
+  int i1 = va_arg(org, int);
+  int j1 = va_arg(org, int);
+  char* s1 = va_arg(org, char*);
+  double d1 = va_arg(org, double);
+  va_end(org);
+
+  int i2 = va_arg(dup, int);
+  int j2 = va_arg(dup, int);
+  char* s2 = va_arg(dup, char*);
+  double d2 = va_arg(dup, double);
+
+  EXPECT_EQ(i1, i2);
+  EXPECT_EQ(j1, j2);
+  EXPECT_STREQ(s1, s2);
+  EXPECT_EQ(d1, d2);
+
+  va_end(dup);
+}
+
+TEST(StringUtilTest, VAList) {
+  VariableArgsFunc("%d %d %s %lf", 45, 92, "This is interesting", 9.21);
+}
+
+// Test for Tokenize
+template <typename STR>
+void TokenizeTest() {
+  std::vector<STR> r;
+  size_t size;
+
+  size = Tokenize(STR("This is a string"), STR(" "), &r);
+  EXPECT_EQ(4U, size);
+  ASSERT_EQ(4U, r.size());
+  EXPECT_EQ(r[0], STR("This"));
+  EXPECT_EQ(r[1], STR("is"));
+  EXPECT_EQ(r[2], STR("a"));
+  EXPECT_EQ(r[3], STR("string"));
+  r.clear();
+
+  size = Tokenize(STR("one,two,three"), STR(","), &r);
+  EXPECT_EQ(3U, size);
+  ASSERT_EQ(3U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR("two"));
+  EXPECT_EQ(r[2], STR("three"));
+  r.clear();
+
+  size = Tokenize(STR("one,two:three;four"), STR(",:"), &r);
+  EXPECT_EQ(3U, size);
+  ASSERT_EQ(3U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR("two"));
+  EXPECT_EQ(r[2], STR("three;four"));
+  r.clear();
+
+  size = Tokenize(STR("one,two:three;four"), STR(";,:"), &r);
+  EXPECT_EQ(4U, size);
+  ASSERT_EQ(4U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR("two"));
+  EXPECT_EQ(r[2], STR("three"));
+  EXPECT_EQ(r[3], STR("four"));
+  r.clear();
+
+  size = Tokenize(STR("one, two, three"), STR(","), &r);
+  EXPECT_EQ(3U, size);
+  ASSERT_EQ(3U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR(" two"));
+  EXPECT_EQ(r[2], STR(" three"));
+  r.clear();
+
+  size = Tokenize(STR("one, two, three, "), STR(","), &r);
+  EXPECT_EQ(4U, size);
+  ASSERT_EQ(4U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR(" two"));
+  EXPECT_EQ(r[2], STR(" three"));
+  EXPECT_EQ(r[3], STR(" "));
+  r.clear();
+
+  size = Tokenize(STR("one, two, three,"), STR(","), &r);
+  EXPECT_EQ(3U, size);
+  ASSERT_EQ(3U, r.size());
+  EXPECT_EQ(r[0], STR("one"));
+  EXPECT_EQ(r[1], STR(" two"));
+  EXPECT_EQ(r[2], STR(" three"));
+  r.clear();
+
+  size = Tokenize(STR(), STR(","), &r);
+  EXPECT_EQ(0U, size);
+  ASSERT_EQ(0U, r.size());
+  r.clear();
+
+  size = Tokenize(STR(","), STR(","), &r);
+  EXPECT_EQ(0U, size);
+  ASSERT_EQ(0U, r.size());
+  r.clear();
+
+  size = Tokenize(STR(",;:."), STR(".:;,"), &r);
+  EXPECT_EQ(0U, size);
+  ASSERT_EQ(0U, r.size());
+  r.clear();
+
+  size = Tokenize(STR("\t\ta\t"), STR("\t"), &r);
+  EXPECT_EQ(1U, size);
+  ASSERT_EQ(1U, r.size());
+  EXPECT_EQ(r[0], STR("a"));
+  r.clear();
+
+  size = Tokenize(STR("\ta\t\nb\tcc"), STR("\n"), &r);
+  EXPECT_EQ(2U, size);
+  ASSERT_EQ(2U, r.size());
+  EXPECT_EQ(r[0], STR("\ta\t"));
+  EXPECT_EQ(r[1], STR("b\tcc"));
+  r.clear();
+}
+
+TEST(StringUtilTest, TokenizeStdString) {
+  TokenizeTest<std::string>();
+}
+
+TEST(StringUtilTest, TokenizeStringPiece) {
+  TokenizeTest<base::StringPiece>();
+}
+
+// Test for JoinString
+TEST(StringUtilTest, JoinString) {
+  std::vector<std::string> in;
+  EXPECT_EQ("", JoinString(in, ','));
+
+  in.push_back("a");
+  EXPECT_EQ("a", JoinString(in, ','));
+
+  in.push_back("b");
+  in.push_back("c");
+  EXPECT_EQ("a,b,c", JoinString(in, ','));
+
+  in.push_back(std::string());
+  EXPECT_EQ("a,b,c,", JoinString(in, ','));
+  in.push_back(" ");
+  EXPECT_EQ("a|b|c|| ", JoinString(in, '|'));
+}
+
+// Test for JoinString overloaded with std::string separator
+TEST(StringUtilTest, JoinStringWithString) {
+  std::string separator(", ");
+  std::vector<std::string> parts;
+  EXPECT_EQ(std::string(), JoinString(parts, separator));
+
+  parts.push_back("a");
+  EXPECT_EQ("a", JoinString(parts, separator));
+
+  parts.push_back("b");
+  parts.push_back("c");
+  EXPECT_EQ("a, b, c", JoinString(parts, separator));
+
+  parts.push_back(std::string());
+  EXPECT_EQ("a, b, c, ", JoinString(parts, separator));
+  parts.push_back(" ");
+  EXPECT_EQ("a|b|c|| ", JoinString(parts, "|"));
+}
+
+// Test for JoinString overloaded with string16 separator
+TEST(StringUtilTest, JoinStringWithString16) {
+  string16 separator = ASCIIToUTF16(", ");
+  std::vector<string16> parts;
+  EXPECT_EQ(string16(), JoinString(parts, separator));
+
+  parts.push_back(ASCIIToUTF16("a"));
+  EXPECT_EQ(ASCIIToUTF16("a"), JoinString(parts, separator));
+
+  parts.push_back(ASCIIToUTF16("b"));
+  parts.push_back(ASCIIToUTF16("c"));
+  EXPECT_EQ(ASCIIToUTF16("a, b, c"), JoinString(parts, separator));
+
+  parts.push_back(ASCIIToUTF16(""));
+  EXPECT_EQ(ASCIIToUTF16("a, b, c, "), JoinString(parts, separator));
+  parts.push_back(ASCIIToUTF16(" "));
+  EXPECT_EQ(ASCIIToUTF16("a|b|c|| "), JoinString(parts, ASCIIToUTF16("|")));
+}
+
+TEST(StringUtilTest, StartsWith) {
+  EXPECT_TRUE(StartsWithASCII("javascript:url", "javascript", true));
+  EXPECT_FALSE(StartsWithASCII("JavaScript:url", "javascript", true));
+  EXPECT_TRUE(StartsWithASCII("javascript:url", "javascript", false));
+  EXPECT_TRUE(StartsWithASCII("JavaScript:url", "javascript", false));
+  EXPECT_FALSE(StartsWithASCII("java", "javascript", true));
+  EXPECT_FALSE(StartsWithASCII("java", "javascript", false));
+  EXPECT_FALSE(StartsWithASCII(std::string(), "javascript", false));
+  EXPECT_FALSE(StartsWithASCII(std::string(), "javascript", true));
+  EXPECT_TRUE(StartsWithASCII("java", std::string(), false));
+  EXPECT_TRUE(StartsWithASCII("java", std::string(), true));
+
+  EXPECT_TRUE(StartsWith(L"javascript:url", L"javascript", true));
+  EXPECT_FALSE(StartsWith(L"JavaScript:url", L"javascript", true));
+  EXPECT_TRUE(StartsWith(L"javascript:url", L"javascript", false));
+  EXPECT_TRUE(StartsWith(L"JavaScript:url", L"javascript", false));
+  EXPECT_FALSE(StartsWith(L"java", L"javascript", true));
+  EXPECT_FALSE(StartsWith(L"java", L"javascript", false));
+  EXPECT_FALSE(StartsWith(std::wstring(), L"javascript", false));
+  EXPECT_FALSE(StartsWith(std::wstring(), L"javascript", true));
+  EXPECT_TRUE(StartsWith(L"java", std::wstring(), false));
+  EXPECT_TRUE(StartsWith(L"java", std::wstring(), true));
+}
+
+TEST(StringUtilTest, EndsWith) {
+  EXPECT_TRUE(EndsWith(L"Foo.plugin", L".plugin", true));
+  EXPECT_FALSE(EndsWith(L"Foo.Plugin", L".plugin", true));
+  EXPECT_TRUE(EndsWith(L"Foo.plugin", L".plugin", false));
+  EXPECT_TRUE(EndsWith(L"Foo.Plugin", L".plugin", false));
+  EXPECT_FALSE(EndsWith(L".plug", L".plugin", true));
+  EXPECT_FALSE(EndsWith(L".plug", L".plugin", false));
+  EXPECT_FALSE(EndsWith(L"Foo.plugin Bar", L".plugin", true));
+  EXPECT_FALSE(EndsWith(L"Foo.plugin Bar", L".plugin", false));
+  EXPECT_FALSE(EndsWith(std::wstring(), L".plugin", false));
+  EXPECT_FALSE(EndsWith(std::wstring(), L".plugin", true));
+  EXPECT_TRUE(EndsWith(L"Foo.plugin", std::wstring(), false));
+  EXPECT_TRUE(EndsWith(L"Foo.plugin", std::wstring(), true));
+  EXPECT_TRUE(EndsWith(L".plugin", L".plugin", false));
+  EXPECT_TRUE(EndsWith(L".plugin", L".plugin", true));
+  EXPECT_TRUE(EndsWith(std::wstring(), std::wstring(), false));
+  EXPECT_TRUE(EndsWith(std::wstring(), std::wstring(), true));
+}
+
+TEST(StringUtilTest, GetStringFWithOffsets) {
+  std::vector<string16> subst;
+  subst.push_back(ASCIIToUTF16("1"));
+  subst.push_back(ASCIIToUTF16("2"));
+  std::vector<size_t> offsets;
+
+  ReplaceStringPlaceholders(ASCIIToUTF16("Hello, $1. Your number is $2."),
+                            subst,
+                            &offsets);
+  EXPECT_EQ(2U, offsets.size());
+  EXPECT_EQ(7U, offsets[0]);
+  EXPECT_EQ(25U, offsets[1]);
+  offsets.clear();
+
+  ReplaceStringPlaceholders(ASCIIToUTF16("Hello, $2. Your number is $1."),
+                            subst,
+                            &offsets);
+  EXPECT_EQ(2U, offsets.size());
+  EXPECT_EQ(25U, offsets[0]);
+  EXPECT_EQ(7U, offsets[1]);
+  offsets.clear();
+}
+
+TEST(StringUtilTest, ReplaceStringPlaceholdersTooFew) {
+  // Test whether replacestringplaceholders works as expected when there
+  // are fewer inputs than outputs.
+  std::vector<string16> subst;
+  subst.push_back(ASCIIToUTF16("9a"));
+  subst.push_back(ASCIIToUTF16("8b"));
+  subst.push_back(ASCIIToUTF16("7c"));
+
+  string16 formatted =
+      ReplaceStringPlaceholders(
+          ASCIIToUTF16("$1a,$2b,$3c,$4d,$5e,$6f,$1g,$2h,$3i"), subst, NULL);
+
+  EXPECT_EQ(formatted, ASCIIToUTF16("9aa,8bb,7cc,d,e,f,9ag,8bh,7ci"));
+}
+
+TEST(StringUtilTest, ReplaceStringPlaceholders) {
+  std::vector<string16> subst;
+  subst.push_back(ASCIIToUTF16("9a"));
+  subst.push_back(ASCIIToUTF16("8b"));
+  subst.push_back(ASCIIToUTF16("7c"));
+  subst.push_back(ASCIIToUTF16("6d"));
+  subst.push_back(ASCIIToUTF16("5e"));
+  subst.push_back(ASCIIToUTF16("4f"));
+  subst.push_back(ASCIIToUTF16("3g"));
+  subst.push_back(ASCIIToUTF16("2h"));
+  subst.push_back(ASCIIToUTF16("1i"));
+
+  string16 formatted =
+      ReplaceStringPlaceholders(
+          ASCIIToUTF16("$1a,$2b,$3c,$4d,$5e,$6f,$7g,$8h,$9i"), subst, NULL);
+
+  EXPECT_EQ(formatted, ASCIIToUTF16("9aa,8bb,7cc,6dd,5ee,4ff,3gg,2hh,1ii"));
+}
+
+TEST(StringUtilTest, ReplaceStringPlaceholdersMoreThan9Replacements) {
+  std::vector<string16> subst;
+  subst.push_back(ASCIIToUTF16("9a"));
+  subst.push_back(ASCIIToUTF16("8b"));
+  subst.push_back(ASCIIToUTF16("7c"));
+  subst.push_back(ASCIIToUTF16("6d"));
+  subst.push_back(ASCIIToUTF16("5e"));
+  subst.push_back(ASCIIToUTF16("4f"));
+  subst.push_back(ASCIIToUTF16("3g"));
+  subst.push_back(ASCIIToUTF16("2h"));
+  subst.push_back(ASCIIToUTF16("1i"));
+  subst.push_back(ASCIIToUTF16("0j"));
+  subst.push_back(ASCIIToUTF16("-1k"));
+  subst.push_back(ASCIIToUTF16("-2l"));
+  subst.push_back(ASCIIToUTF16("-3m"));
+  subst.push_back(ASCIIToUTF16("-4n"));
+
+  string16 formatted =
+      ReplaceStringPlaceholders(
+          ASCIIToUTF16("$1a,$2b,$3c,$4d,$5e,$6f,$7g,$8h,$9i,"
+                       "$10j,$11k,$12l,$13m,$14n,$1"), subst, NULL);
+
+  EXPECT_EQ(formatted, ASCIIToUTF16("9aa,8bb,7cc,6dd,5ee,4ff,3gg,2hh,"
+                                    "1ii,0jj,-1kk,-2ll,-3mm,-4nn,9a"));
+}
+
+TEST(StringUtilTest, StdStringReplaceStringPlaceholders) {
+  std::vector<std::string> subst;
+  subst.push_back("9a");
+  subst.push_back("8b");
+  subst.push_back("7c");
+  subst.push_back("6d");
+  subst.push_back("5e");
+  subst.push_back("4f");
+  subst.push_back("3g");
+  subst.push_back("2h");
+  subst.push_back("1i");
+
+  std::string formatted =
+      ReplaceStringPlaceholders(
+          "$1a,$2b,$3c,$4d,$5e,$6f,$7g,$8h,$9i", subst, NULL);
+
+  EXPECT_EQ(formatted, "9aa,8bb,7cc,6dd,5ee,4ff,3gg,2hh,1ii");
+}
+
+TEST(StringUtilTest, ReplaceStringPlaceholdersConsecutiveDollarSigns) {
+  std::vector<std::string> subst;
+  subst.push_back("a");
+  subst.push_back("b");
+  subst.push_back("c");
+  EXPECT_EQ(ReplaceStringPlaceholders("$$1 $$$2 $$$$3", subst, NULL),
+            "$1 $$2 $$$3");
+}
+
+TEST(StringUtilTest, MatchPatternTest) {
+  EXPECT_TRUE(MatchPattern("www.google.com", "*.com"));
+  EXPECT_TRUE(MatchPattern("www.google.com", "*"));
+  EXPECT_FALSE(MatchPattern("www.google.com", "www*.g*.org"));
+  EXPECT_TRUE(MatchPattern("Hello", "H?l?o"));
+  EXPECT_FALSE(MatchPattern("www.google.com", "http://*)"));
+  EXPECT_FALSE(MatchPattern("www.msn.com", "*.COM"));
+  EXPECT_TRUE(MatchPattern("Hello*1234", "He??o\\*1*"));
+  EXPECT_FALSE(MatchPattern("", "*.*"));
+  EXPECT_TRUE(MatchPattern("", "*"));
+  EXPECT_TRUE(MatchPattern("", "?"));
+  EXPECT_TRUE(MatchPattern("", ""));
+  EXPECT_FALSE(MatchPattern("Hello", ""));
+  EXPECT_TRUE(MatchPattern("Hello*", "Hello*"));
+  // Stop after a certain recursion depth.
+  EXPECT_FALSE(MatchPattern("123456789012345678", "?????????????????*"));
+
+  // Test UTF8 matching.
+  EXPECT_TRUE(MatchPattern("heart: \xe2\x99\xa0", "*\xe2\x99\xa0"));
+  EXPECT_TRUE(MatchPattern("heart: \xe2\x99\xa0.", "heart: ?."));
+  EXPECT_TRUE(MatchPattern("hearts: \xe2\x99\xa0\xe2\x99\xa0", "*"));
+  // Invalid sequences should be handled as a single invalid character.
+  EXPECT_TRUE(MatchPattern("invalid: \xef\xbf\xbe", "invalid: ?"));
+  // If the pattern has invalid characters, it shouldn't match anything.
+  EXPECT_FALSE(MatchPattern("\xf4\x90\x80\x80", "\xf4\x90\x80\x80"));
+
+  // Test UTF16 character matching.
+  EXPECT_TRUE(MatchPattern(UTF8ToUTF16("www.google.com"),
+                           UTF8ToUTF16("*.com")));
+  EXPECT_TRUE(MatchPattern(UTF8ToUTF16("Hello*1234"),
+                           UTF8ToUTF16("He??o\\*1*")));
+
+  // This test verifies that consecutive wild cards are collapsed into 1
+  // wildcard (when this doesn't occur, MatchPattern reaches it's maximum
+  // recursion depth).
+  EXPECT_TRUE(MatchPattern(UTF8ToUTF16("Hello"),
+                           UTF8ToUTF16("He********************************o")));
+}
+
+TEST(StringUtilTest, LcpyTest) {
+  // Test the normal case where we fit in our buffer.
+  {
+    char dst[10];
+    wchar_t wdst[10];
+    EXPECT_EQ(7U, base::strlcpy(dst, "abcdefg", arraysize(dst)));
+    EXPECT_EQ(0, memcmp(dst, "abcdefg", 8));
+    EXPECT_EQ(7U, base::wcslcpy(wdst, L"abcdefg", arraysize(wdst)));
+    EXPECT_EQ(0, memcmp(wdst, L"abcdefg", sizeof(wchar_t) * 8));
+  }
+
+  // Test dst_size == 0, nothing should be written to |dst| and we should
+  // have the equivalent of strlen(src).
+  {
+    char dst[2] = {1, 2};
+    wchar_t wdst[2] = {1, 2};
+    EXPECT_EQ(7U, base::strlcpy(dst, "abcdefg", 0));
+    EXPECT_EQ(1, dst[0]);
+    EXPECT_EQ(2, dst[1]);
+    EXPECT_EQ(7U, base::wcslcpy(wdst, L"abcdefg", 0));
+#if defined(WCHAR_T_IS_UNSIGNED)
+    EXPECT_EQ(1U, wdst[0]);
+    EXPECT_EQ(2U, wdst[1]);
+#else
+    EXPECT_EQ(1, wdst[0]);
+    EXPECT_EQ(2, wdst[1]);
+#endif
+  }
+
+  // Test the case were we _just_ competely fit including the null.
+  {
+    char dst[8];
+    wchar_t wdst[8];
+    EXPECT_EQ(7U, base::strlcpy(dst, "abcdefg", arraysize(dst)));
+    EXPECT_EQ(0, memcmp(dst, "abcdefg", 8));
+    EXPECT_EQ(7U, base::wcslcpy(wdst, L"abcdefg", arraysize(wdst)));
+    EXPECT_EQ(0, memcmp(wdst, L"abcdefg", sizeof(wchar_t) * 8));
+  }
+
+  // Test the case were we we are one smaller, so we can't fit the null.
+  {
+    char dst[7];
+    wchar_t wdst[7];
+    EXPECT_EQ(7U, base::strlcpy(dst, "abcdefg", arraysize(dst)));
+    EXPECT_EQ(0, memcmp(dst, "abcdef", 7));
+    EXPECT_EQ(7U, base::wcslcpy(wdst, L"abcdefg", arraysize(wdst)));
+    EXPECT_EQ(0, memcmp(wdst, L"abcdef", sizeof(wchar_t) * 7));
+  }
+
+  // Test the case were we are just too small.
+  {
+    char dst[3];
+    wchar_t wdst[3];
+    EXPECT_EQ(7U, base::strlcpy(dst, "abcdefg", arraysize(dst)));
+    EXPECT_EQ(0, memcmp(dst, "ab", 3));
+    EXPECT_EQ(7U, base::wcslcpy(wdst, L"abcdefg", arraysize(wdst)));
+    EXPECT_EQ(0, memcmp(wdst, L"ab", sizeof(wchar_t) * 3));
+  }
+}
+
+TEST(StringUtilTest, WprintfFormatPortabilityTest) {
+  static const struct {
+    const wchar_t* input;
+    bool portable;
+  } cases[] = {
+    { L"%ls", true },
+    { L"%s", false },
+    { L"%S", false },
+    { L"%lS", false },
+    { L"Hello, %s", false },
+    { L"%lc", true },
+    { L"%c", false },
+    { L"%C", false },
+    { L"%lC", false },
+    { L"%ls %s", false },
+    { L"%s %ls", false },
+    { L"%s %ls %s", false },
+    { L"%f", true },
+    { L"%f %F", false },
+    { L"%d %D", false },
+    { L"%o %O", false },
+    { L"%u %U", false },
+    { L"%f %d %o %u", true },
+    { L"%-8d (%02.1f%)", true },
+    { L"% 10s", false },
+    { L"% 10ls", true }
+  };
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(cases); ++i)
+    EXPECT_EQ(cases[i].portable, base::IsWprintfFormatPortable(cases[i].input));
+}
+
+TEST(StringUtilTest, RemoveChars) {
+  const char* kRemoveChars = "-/+*";
+  std::string input = "A-+bc/d!*";
+  EXPECT_TRUE(RemoveChars(input, kRemoveChars, &input));
+  EXPECT_EQ("Abcd!", input);
+
+  // No characters match kRemoveChars.
+  EXPECT_FALSE(RemoveChars(input, kRemoveChars, &input));
+  EXPECT_EQ("Abcd!", input);
+
+  // Empty string.
+  input.clear();
+  EXPECT_FALSE(RemoveChars(input, kRemoveChars, &input));
+  EXPECT_EQ(std::string(), input);
+}
+
+TEST(StringUtilTest, ReplaceChars) {
+  struct TestData {
+    const char* input;
+    const char* replace_chars;
+    const char* replace_with;
+    const char* output;
+    bool result;
+  } cases[] = {
+    { "", "", "", "", false },
+    { "test", "", "", "test", false },
+    { "test", "", "!", "test", false },
+    { "test", "z", "!", "test", false },
+    { "test", "e", "!", "t!st", true },
+    { "test", "e", "!?", "t!?st", true },
+    { "test", "ez", "!", "t!st", true },
+    { "test", "zed", "!?", "t!?st", true },
+    { "test", "t", "!?", "!?es!?", true },
+    { "test", "et", "!>", "!>!>s!>", true },
+    { "test", "zest", "!", "!!!!", true },
+    { "test", "szt", "!", "!e!!", true },
+    { "test", "t", "test", "testestest", true },
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE_UNSAFE(cases); ++i) {
+    std::string output;
+    bool result = ReplaceChars(cases[i].input,
+                               cases[i].replace_chars,
+                               cases[i].replace_with,
+                               &output);
+    EXPECT_EQ(cases[i].result, result);
+    EXPECT_EQ(cases[i].output, output);
+  }
+}
+
+TEST(StringUtilTest, ContainsOnlyChars) {
+  // Providing an empty list of characters should return false but for the empty
+  // string.
+  EXPECT_TRUE(ContainsOnlyChars(std::string(), std::string()));
+  EXPECT_FALSE(ContainsOnlyChars("Hello", std::string()));
+
+  EXPECT_TRUE(ContainsOnlyChars(std::string(), "1234"));
+  EXPECT_TRUE(ContainsOnlyChars("1", "1234"));
+  EXPECT_TRUE(ContainsOnlyChars("1", "4321"));
+  EXPECT_TRUE(ContainsOnlyChars("123", "4321"));
+  EXPECT_FALSE(ContainsOnlyChars("123a", "4321"));
+}
+
+class WriteIntoTest : public testing::Test {
+ protected:
+  static void WritesCorrectly(size_t num_chars) {
+    std::string buffer;
+    char kOriginal[] = "supercali";
+    strncpy(WriteInto(&buffer, num_chars + 1), kOriginal, num_chars);
+    // Using std::string(buffer.c_str()) instead of |buffer| truncates the
+    // string at the first \0.
+    EXPECT_EQ(std::string(kOriginal,
+                          std::min(num_chars, arraysize(kOriginal) - 1)),
+              std::string(buffer.c_str()));
+    EXPECT_EQ(num_chars, buffer.size());
+  }
+};
+
+TEST_F(WriteIntoTest, WriteInto) {
+  // Validate that WriteInto reserves enough space and
+  // sizes a string correctly.
+  WritesCorrectly(1);
+  WritesCorrectly(2);
+  WritesCorrectly(5000);
+
+  // Validate that WriteInto doesn't modify other strings
+  // when using a Copy-on-Write implementation.
+  const char kLive[] = "live";
+  const char kDead[] = "dead";
+  const std::string live = kLive;
+  std::string dead = live;
+  strncpy(WriteInto(&dead, 5), kDead, 4);
+  EXPECT_EQ(kDead, dead);
+  EXPECT_EQ(4u, dead.size());
+  EXPECT_EQ(kLive, live);
+  EXPECT_EQ(4u, live.size());
+}
+
+}  // namespace base
diff --git a/base/strings/string_util_win.h b/base/strings/string_util_win.h
new file mode 100644
index 0000000..602ba27
--- /dev/null
+++ b/base/strings/string_util_win.h
@@ -0,0 +1,61 @@
+// 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 BASE_STRINGS_STRING_UTIL_WIN_H_
+#define BASE_STRINGS_STRING_UTIL_WIN_H_
+
+#include <stdarg.h>
+#include <stdio.h>
+#include <string.h>
+#include <wchar.h>
+
+#include "base/logging.h"
+
+namespace base {
+
+// Chromium code style is to not use malloc'd strings; this is only for use
+// for interaction with APIs that require it.
+inline char* strdup(const char* str) {
+  return _strdup(str);
+}
+
+inline int strcasecmp(const char* s1, const char* s2) {
+  return _stricmp(s1, s2);
+}
+
+inline int strncasecmp(const char* s1, const char* s2, size_t count) {
+  return _strnicmp(s1, s2, count);
+}
+
+inline int strncmp16(const char16* s1, const char16* s2, size_t count) {
+  return ::wcsncmp(s1, s2, count);
+}
+
+inline int vsnprintf(char* buffer, size_t size,
+                     const char* format, va_list arguments) {
+  int length = _vsprintf_p(buffer, size, format, arguments);
+  if (length < 0) {
+    if (size > 0)
+      buffer[0] = 0;
+    return _vscprintf_p(format, arguments);
+  }
+  return length;
+}
+
+inline int vswprintf(wchar_t* buffer, size_t size,
+                     const wchar_t* format, va_list arguments) {
+  DCHECK(IsWprintfFormatPortable(format));
+
+  int length = _vswprintf_p(buffer, size, format, arguments);
+  if (length < 0) {
+    if (size > 0)
+      buffer[0] = 0;
+    return _vscwprintf_p(format, arguments);
+  }
+  return length;
+}
+
+}  // namespace base
+
+#endif  // BASE_STRINGS_STRING_UTIL_WIN_H_
diff --git a/base/strings/stringprintf.cc b/base/strings/stringprintf.cc
new file mode 100644
index 0000000..fe23daa
--- /dev/null
+++ b/base/strings/stringprintf.cc
@@ -0,0 +1,186 @@
+// 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 "base/strings/stringprintf.h"
+
+#include <errno.h>
+
+#include "base/scoped_clear_errno.h"
+#include "base/strings/string_util.h"
+#include "base/strings/utf_string_conversions.h"
+
+namespace base {
+
+namespace {
+
+// Overloaded wrappers around vsnprintf and vswprintf. The buf_size parameter
+// is the size of the buffer. These return the number of characters in the
+// formatted string excluding the NUL terminator. If the buffer is not
+// large enough to accommodate the formatted string without truncation, they
+// return the number of characters that would be in the fully-formatted string
+// (vsnprintf, and vswprintf on Windows), or -1 (vswprintf on POSIX platforms).
+inline int vsnprintfT(char* buffer,
+                      size_t buf_size,
+                      const char* format,
+                      va_list argptr) {
+  return base::vsnprintf(buffer, buf_size, format, argptr);
+}
+
+#if !defined(OS_ANDROID)
+inline int vsnprintfT(wchar_t* buffer,
+                      size_t buf_size,
+                      const wchar_t* format,
+                      va_list argptr) {
+  return base::vswprintf(buffer, buf_size, format, argptr);
+}
+#endif
+
+// Templatized backend for StringPrintF/StringAppendF. This does not finalize
+// the va_list, the caller is expected to do that.
+template <class StringType>
+static void StringAppendVT(StringType* dst,
+                           const typename StringType::value_type* format,
+                           va_list ap) {
+  // First try with a small fixed size buffer.
+  // This buffer size should be kept in sync with StringUtilTest.GrowBoundary
+  // and StringUtilTest.StringPrintfBounds.
+  typename StringType::value_type stack_buf[1024];
+
+  va_list ap_copy;
+  GG_VA_COPY(ap_copy, ap);
+
+#if !defined(OS_WIN)
+  ScopedClearErrno clear_errno;
+#endif
+  int result = vsnprintfT(stack_buf, arraysize(stack_buf), format, ap_copy);
+  va_end(ap_copy);
+
+  if (result >= 0 && result < static_cast<int>(arraysize(stack_buf))) {
+    // It fit.
+    dst->append(stack_buf, result);
+    return;
+  }
+
+  // Repeatedly increase buffer size until it fits.
+  int mem_length = arraysize(stack_buf);
+  while (true) {
+    if (result < 0) {
+#if !defined(OS_WIN)
+      // On Windows, vsnprintfT always returns the number of characters in a
+      // fully-formatted string, so if we reach this point, something else is
+      // wrong and no amount of buffer-doubling is going to fix it.
+      if (errno != 0 && errno != EOVERFLOW)
+#endif
+      {
+        // If an error other than overflow occurred, it's never going to work.
+        DLOG(WARNING) << "Unable to printf the requested string due to error.";
+        return;
+      }
+      // Try doubling the buffer size.
+      mem_length *= 2;
+    } else {
+      // We need exactly "result + 1" characters.
+      mem_length = result + 1;
+    }
+
+    if (mem_length > 32 * 1024 * 1024) {
+      // That should be plenty, don't try anything larger.  This protects
+      // against huge allocations when using vsnprintfT implementations that
+      // return -1 for reasons other than overflow without setting errno.
+      DLOG(WARNING) << "Unable to printf the requested string due to size.";
+      return;
+    }
+
+    std::vector<typename StringType::value_type> mem_buf(mem_length);
+
+    // NOTE: You can only use a va_list once.  Since we're in a while loop, we
+    // need to make a new copy each time so we don't use up the original.
+    GG_VA_COPY(ap_copy, ap);
+    result = vsnprintfT(&mem_buf[0], mem_length, format, ap_copy);
+    va_end(ap_copy);
+
+    if ((result >= 0) && (result < mem_length)) {
+      // It fit.
+      dst->append(&mem_buf[0], result);
+      return;
+    }
+  }
+}
+
+}  // namespace
+
+std::string StringPrintf(const char* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  std::string result;
+  StringAppendV(&result, format, ap);
+  va_end(ap);
+  return result;
+}
+
+#if !defined(OS_ANDROID)
+std::wstring StringPrintf(const wchar_t* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  std::wstring result;
+  StringAppendV(&result, format, ap);
+  va_end(ap);
+  return result;
+}
+#endif
+
+std::string StringPrintV(const char* format, va_list ap) {
+  std::string result;
+  StringAppendV(&result, format, ap);
+  return result;
+}
+
+const std::string& SStringPrintf(std::string* dst, const char* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  dst->clear();
+  StringAppendV(dst, format, ap);
+  va_end(ap);
+  return *dst;
+}
+
+#if !defined(OS_ANDROID)
+const std::wstring& SStringPrintf(std::wstring* dst,
+                                  const wchar_t* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  dst->clear();
+  StringAppendV(dst, format, ap);
+  va_end(ap);
+  return *dst;
+}
+#endif
+
+void StringAppendF(std::string* dst, const char* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  StringAppendV(dst, format, ap);
+  va_end(ap);
+}
+
+#if !defined(OS_ANDROID)
+void StringAppendF(std::wstring* dst, const wchar_t* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  StringAppendV(dst, format, ap);
+  va_end(ap);
+}
+#endif
+
+void StringAppendV(std::string* dst, const char* format, va_list ap) {
+  StringAppendVT(dst, format, ap);
+}
+
+#if !defined(OS_ANDROID)
+void StringAppendV(std::wstring* dst, const wchar_t* format, va_list ap) {
+  StringAppendVT(dst, format, ap);
+}
+#endif
+
+}  // namespace base
diff --git a/base/strings/stringprintf.h b/base/strings/stringprintf.h
new file mode 100644
index 0000000..3c0e399
--- /dev/null
+++ b/base/strings/stringprintf.h
@@ -0,0 +1,62 @@
+// 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 BASE_STRINGS_STRINGPRINTF_H_
+#define BASE_STRINGS_STRINGPRINTF_H_
+
+#include <stdarg.h>   // va_list
+
+#include <string>
+
+#include "base/base_export.h"
+#include "base/compiler_specific.h"
+
+namespace base {
+
+// Return a C++ string given printf-like input.
+BASE_EXPORT std::string StringPrintf(const char* format, ...)
+    PRINTF_FORMAT(1, 2);
+// OS_ANDROID's libc does not support wchar_t, so several overloads are omitted.
+#if !defined(OS_ANDROID)
+BASE_EXPORT std::wstring StringPrintf(const wchar_t* format, ...)
+    WPRINTF_FORMAT(1, 2);
+#endif
+
+// Return a C++ string given vprintf-like input.
+BASE_EXPORT std::string StringPrintV(const char* format, va_list ap)
+    PRINTF_FORMAT(1, 0);
+
+// Store result into a supplied string and return it.
+BASE_EXPORT const std::string& SStringPrintf(std::string* dst,
+                                             const char* format, ...)
+    PRINTF_FORMAT(2, 3);
+#if !defined(OS_ANDROID)
+BASE_EXPORT const std::wstring& SStringPrintf(std::wstring* dst,
+                                              const wchar_t* format, ...)
+    WPRINTF_FORMAT(2, 3);
+#endif
+
+// Append result to a supplied string.
+BASE_EXPORT void StringAppendF(std::string* dst, const char* format, ...)
+    PRINTF_FORMAT(2, 3);
+#if !defined(OS_ANDROID)
+// TODO(evanm): this is only used in a few places in the code;
+// replace with string16 version.
+BASE_EXPORT void StringAppendF(std::wstring* dst, const wchar_t* format, ...)
+    WPRINTF_FORMAT(2, 3);
+#endif
+
+// Lower-level routine that takes a va_list and appends to a specified
+// string.  All other routines are just convenience wrappers around it.
+BASE_EXPORT void StringAppendV(std::string* dst, const char* format, va_list ap)
+    PRINTF_FORMAT(2, 0);
+#if !defined(OS_ANDROID)
+BASE_EXPORT void StringAppendV(std::wstring* dst,
+                               const wchar_t* format, va_list ap)
+    WPRINTF_FORMAT(2, 0);
+#endif
+
+}  // namespace base
+
+#endif  // BASE_STRINGS_STRINGPRINTF_H_
diff --git a/base/strings/stringprintf_unittest.cc b/base/strings/stringprintf_unittest.cc
new file mode 100644
index 0000000..a1bf2da
--- /dev/null
+++ b/base/strings/stringprintf_unittest.cc
@@ -0,0 +1,188 @@
+// 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 "base/strings/stringprintf.h"
+
+#include <errno.h>
+
+#include "base/basictypes.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+namespace base {
+
+namespace {
+
+// A helper for the StringAppendV test that follows.
+//
+// Just forwards its args to StringAppendV.
+static void StringAppendVTestHelper(std::string* out, const char* format, ...) {
+  va_list ap;
+  va_start(ap, format);
+  StringAppendV(out, format, ap);
+  va_end(ap);
+}
+
+}  // namespace
+
+TEST(StringPrintfTest, StringPrintfEmpty) {
+  EXPECT_EQ("", StringPrintf("%s", ""));
+}
+
+TEST(StringPrintfTest, StringPrintfMisc) {
+  EXPECT_EQ("123hello w", StringPrintf("%3d%2s %1c", 123, "hello", 'w'));
+#if !defined(OS_ANDROID)
+  EXPECT_EQ(L"123hello w", StringPrintf(L"%3d%2ls %1lc", 123, L"hello", 'w'));
+#endif
+}
+
+TEST(StringPrintfTest, StringAppendfEmptyString) {
+  std::string value("Hello");
+  StringAppendF(&value, "%s", "");
+  EXPECT_EQ("Hello", value);
+
+#if !defined(OS_ANDROID)
+  std::wstring valuew(L"Hello");
+  StringAppendF(&valuew, L"%ls", L"");
+  EXPECT_EQ(L"Hello", valuew);
+#endif
+}
+
+TEST(StringPrintfTest, StringAppendfString) {
+  std::string value("Hello");
+  StringAppendF(&value, " %s", "World");
+  EXPECT_EQ("Hello World", value);
+
+#if !defined(OS_ANDROID)
+  std::wstring valuew(L"Hello");
+  StringAppendF(&valuew, L" %ls", L"World");
+  EXPECT_EQ(L"Hello World", valuew);
+#endif
+}
+
+TEST(StringPrintfTest, StringAppendfInt) {
+  std::string value("Hello");
+  StringAppendF(&value, " %d", 123);
+  EXPECT_EQ("Hello 123", value);
+
+#if !defined(OS_ANDROID)
+  std::wstring valuew(L"Hello");
+  StringAppendF(&valuew, L" %d", 123);
+  EXPECT_EQ(L"Hello 123", valuew);
+#endif
+}
+
+// Make sure that lengths exactly around the initial buffer size are handled
+// correctly.
+TEST(StringPrintfTest, StringPrintfBounds) {
+  const int kSrcLen = 1026;
+  char src[kSrcLen];
+  for (size_t i = 0; i < arraysize(src); i++)
+    src[i] = 'A';
+
+  wchar_t srcw[kSrcLen];
+  for (size_t i = 0; i < arraysize(srcw); i++)
+    srcw[i] = 'A';
+
+  for (int i = 1; i < 3; i++) {
+    src[kSrcLen - i] = 0;
+    std::string out;
+    SStringPrintf(&out, "%s", src);
+    EXPECT_STREQ(src, out.c_str());
+
+#if !defined(OS_ANDROID)
+    srcw[kSrcLen - i] = 0;
+    std::wstring outw;
+    SStringPrintf(&outw, L"%ls", srcw);
+    EXPECT_STREQ(srcw, outw.c_str());
+#endif
+  }
+}
+
+// Test very large sprintfs that will cause the buffer to grow.
+TEST(StringPrintfTest, Grow) {
+  char src[1026];
+  for (size_t i = 0; i < arraysize(src); i++)
+    src[i] = 'A';
+  src[1025] = 0;
+
+  const char* fmt = "%sB%sB%sB%sB%sB%sB%s";
+
+  std::string out;
+  SStringPrintf(&out, fmt, src, src, src, src, src, src, src);
+
+  const int kRefSize = 320000;
+  char* ref = new char[kRefSize];
+#if defined(OS_WIN)
+  sprintf_s(ref, kRefSize, fmt, src, src, src, src, src, src, src);
+#elif defined(OS_POSIX)
+  snprintf(ref, kRefSize, fmt, src, src, src, src, src, src, src);
+#endif
+
+  EXPECT_STREQ(ref, out.c_str());
+  delete[] ref;
+}
+
+TEST(StringPrintfTest, StringAppendV) {
+  std::string out;
+  StringAppendVTestHelper(&out, "%d foo %s", 1, "bar");
+  EXPECT_EQ("1 foo bar", out);
+}
+
+// Test the boundary condition for the size of the string_util's
+// internal buffer.
+TEST(StringPrintfTest, GrowBoundary) {
+  const int string_util_buf_len = 1024;
+  // Our buffer should be one larger than the size of StringAppendVT's stack
+  // buffer.
+  const int buf_len = string_util_buf_len + 1;
+  char src[buf_len + 1];  // Need extra one for NULL-terminator.
+  for (int i = 0; i < buf_len; ++i)
+    src[i] = 'a';
+  src[buf_len] = 0;
+
+  std::string out;
+  SStringPrintf(&out, "%s", src);
+
+  EXPECT_STREQ(src, out.c_str());
+}
+
+// TODO(evanm): what's the proper cross-platform test here?
+#if defined(OS_WIN)
+// sprintf in Visual Studio fails when given U+FFFF. This tests that the
+// failure case is gracefuly handled.
+TEST(StringPrintfTest, Invalid) {
+  wchar_t invalid[2];
+  invalid[0] = 0xffff;
+  invalid[1] = 0;
+
+  std::wstring out;
+  SStringPrintf(&out, L"%ls", invalid);
+  EXPECT_STREQ(L"", out.c_str());
+}
+#endif
+
+// Test that the positional parameters work.
+TEST(StringPrintfTest, PositionalParameters) {
+  std::string out;
+  SStringPrintf(&out, "%1$s %1$s", "test");
+  EXPECT_STREQ("test test", out.c_str());
+
+#if defined(OS_WIN)
+  std::wstring wout;
+  SStringPrintf(&wout, L"%1$ls %1$ls", L"test");
+  EXPECT_STREQ(L"test test", wout.c_str());
+#endif
+}
+
+// Test that StringPrintf and StringAppendV do not change errno.
+TEST(StringPrintfTest, StringPrintfErrno) {
+  errno = 1;
+  EXPECT_EQ("", StringPrintf("%s", ""));
+  EXPECT_EQ(1, errno);
+  std::string out;
+  StringAppendVTestHelper(&out, "%d foo %s", 1, "bar");
+  EXPECT_EQ(1, errno);
+}
+
+}  // namespace base