Copy the relevant parts of ICU to a new file base/third_party/icu/icu_utf.*

so we can do basic UTF8/16/32 conversions without linking all of ICU.

Change callers who used to call SysUTF8ToWide/SysWideToUTF8 in base to using
these new functions. I will remove the Sys versions of these functions in a
later patch.

TEST=none
BUG=none
Review URL: http://codereview.chromium.org/243102

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@28219 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 61 insertions, 440 deletions

diff --git a/base/i18n/string_conversions.cc b/base/i18n/icu_string_conversions.cc
index 35c9d6d..225fe0b 100644
--- a/base/i18n/string_conversions.cc
+++ b/base/i18n/icu_string_conversions.cc
@@ -2,7 +2,7 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
-#include "base/i18n/string_conversions.h"
+#include "base/i18n/icu_string_conversions.h"
 
 #include <vector>
 
@@ -82,166 +82,6 @@ void ToUnicodeCallbackSubstitute(const void* context,
   // else ignore the reset, close and clone calls.
 }
 
-// ReadUnicodeCharacter --------------------------------------------------------
-
-// Reads a UTF-8 stream, placing the next code point into the given output
-// |*code_point|. |src| represents the entire string to read, and |*char_index|
-// is the character offset within the string to start reading at. |*char_index|
-// will be updated to index the last character read, such that incrementing it
-// (as in a for loop) will take the reader to the next character.
-//
-// Returns true on success. On false, |*code_point| will be invalid.
-bool ReadUnicodeCharacter(const char* src, int32 src_len,
-                          int32* char_index, uint32* code_point_out) {
-  // U8_NEXT expects to be able to use -1 to signal an error, so we must
-  // use a signed type for code_point.  But this function returns false
-  // on error anyway, so code_point_out is unsigned.
-  int32 code_point;
-  U8_NEXT(src, *char_index, src_len, code_point);
-  *code_point_out = static_cast<uint32>(code_point);
-
-  // The ICU macro above moves to the next char, we want to point to the last
-  // char consumed.
-  (*char_index)--;
-
-  // Validate the decoded value.
-  return IsValidCodepoint(code_point);
-}
-
-// Reads a UTF-16 character. The usage is the same as the 8-bit version above.
-bool ReadUnicodeCharacter(const char16* src, int32 src_len,
-                          int32* char_index, uint32* code_point) {
-  if (U16_IS_SURROGATE(src[*char_index])) {
-    if (!U16_IS_SURROGATE_LEAD(src[*char_index]) ||
-        *char_index + 1 >= src_len ||
-        !U16_IS_TRAIL(src[*char_index + 1])) {
-      // Invalid surrogate pair.
-      return false;
-    }
-
-    // Valid surrogate pair.
-    *code_point = U16_GET_SUPPLEMENTARY(src[*char_index],
-                                        src[*char_index + 1]);
-    (*char_index)++;
-  } else {
-    // Not a surrogate, just one 16-bit word.
-    *code_point = src[*char_index];
-  }
-
-  return IsValidCodepoint(*code_point);
-}
-
-#if defined(WCHAR_T_IS_UTF32)
-// Reads UTF-32 character. The usage is the same as the 8-bit version above.
-bool ReadUnicodeCharacter(const wchar_t* src, int32 src_len,
-                          int32* char_index, uint32* code_point) {
-  // Conversion is easy since the source is 32-bit.
-  *code_point = src[*char_index];
-
-  // Validate the value.
-  return IsValidCodepoint(*code_point);
-}
-#endif  // defined(WCHAR_T_IS_UTF32)
-
-// WriteUnicodeCharacter -------------------------------------------------------
-
-// Appends a UTF-8 character to the given 8-bit string.
-void WriteUnicodeCharacter(uint32 code_point, std::string* output) {
-  if (code_point <= 0x7f) {
-    // Fast path the common case of one byte.
-    output->push_back(code_point);
-    return;
-  }
-
-  // U8_APPEND_UNSAFE can append up to 4 bytes.
-  int32 char_offset = static_cast<int32>(output->length());
-  output->resize(char_offset + U8_MAX_LENGTH);
-
-  U8_APPEND_UNSAFE(&(*output)[0], char_offset, code_point);
-
-  // U8_APPEND_UNSAFE will advance our pointer past the inserted character, so
-  // it will represent the new length of the string.
-  output->resize(char_offset);
-}
-
-// Appends the given code point as a UTF-16 character to the STL string.
-void WriteUnicodeCharacter(uint32 code_point, string16* output) {
-  if (U16_LENGTH(code_point) == 1) {
-    // Thie code point is in the Basic Multilingual Plane (BMP).
-    output->push_back(static_cast<char16>(code_point));
-  } else {
-    // Non-BMP characters use a double-character encoding.
-    int32 char_offset = static_cast<int32>(output->length());
-    output->resize(char_offset + U16_MAX_LENGTH);
-    U16_APPEND_UNSAFE(&(*output)[0], char_offset, code_point);
-  }
-}
-
-#if defined(WCHAR_T_IS_UTF32)
-// Appends the given UTF-32 character to the given 32-bit string.
-inline void WriteUnicodeCharacter(uint32 code_point, std::wstring* output) {
-  // This is the easy case, just append the character.
-  output->push_back(code_point);
-}
-#endif  // defined(WCHAR_T_IS_UTF32)
-
-// Generalized Unicode converter -----------------------------------------------
-
-// Converts the given source Unicode character type to the given destination
-// Unicode character type as a STL string. The given input buffer and size
-// determine the source, and the given output STL string will be replaced by
-// the result.
-template<typename SRC_CHAR, typename DEST_STRING>
-bool ConvertUnicode(const SRC_CHAR* src, size_t src_len, DEST_STRING* output) {
-  output->clear();
-
-  // ICU requires 32-bit numbers.
-  bool success = true;
-  int32 src_len32 = static_cast<int32>(src_len);
-  for (int32 i = 0; i < src_len32; i++) {
-    uint32 code_point;
-    if (ReadUnicodeCharacter(src, src_len32, &i, &code_point)) {
-      WriteUnicodeCharacter(code_point, output);
-    } else {
-      // TODO(jungshik): consider adding 'Replacement character' (U+FFFD)
-      // in place of an invalid codepoint.
-      success = false;
-    }
-  }
-  return success;
-}
-
-
-// Guesses the length of the output in UTF-8 in bytes, and reserves that amount
-// of space in the given string. We also assume that the input character types
-// are unsigned, which will be true for UTF-16 and -32 on our systems. We assume
-// the string length is greater than zero.
-template<typename CHAR>
-void ReserveUTF8Output(const CHAR* src, size_t src_len, std::string* output) {
-  if (src[0] < 0x80) {
-    // Assume that the entire input will be ASCII.
-    output->reserve(src_len);
-  } else {
-    // Assume that the entire input is non-ASCII and will have 3 bytes per char.
-    output->reserve(src_len * 3);
-  }
-}
-
-// Guesses the size of the output buffer (containing either UTF-16 or -32 data)
-// given some UTF-8 input that will be converted to it. See ReserveUTF8Output.
-// We assume the source length is > 0.
-template<typename STRING>
-void ReserveUTF16Or32Output(const char* src, size_t src_len, STRING* output) {
-  if (static_cast<unsigned char>(src[0]) < 0x80) {
-    // Assume the input is all ASCII, which means 1:1 correspondence.
-    output->reserve(src_len);
-  } else {
-    // Otherwise assume that the UTF-8 sequences will have 2 bytes for each
-    // character.
-    output->reserve(src_len / 2);
-  }
-}
-
 bool ConvertFromUTF16(UConverter* converter, const UChar* uchar_src,
                       int uchar_len, OnStringUtilConversionError::Type on_error,
                       std::string* encoded) {
@@ -308,184 +148,6 @@ inline UConverterType utf32_platform_endian() {
 
 }  // namespace
 
-// UTF-8 <-> Wide --------------------------------------------------------------
-
-std::string WideToUTF8(const std::wstring& wide) {
-  std::string ret;
-  if (wide.empty())
-    return ret;
-
-  // Ignore the success flag of this call, it will do the best it can for
-  // invalid input, which is what we want here.
-  WideToUTF8(wide.data(), wide.length(), &ret);
-  return ret;
-}
-
-bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  ReserveUTF8Output(src, src_len, output);
-  return ConvertUnicode<wchar_t, std::string>(src, src_len, output);
-}
-
-std::wstring UTF8ToWide(const base::StringPiece& utf8) {
-  std::wstring ret;
-  if (utf8.empty())
-    return ret;
-
-  UTF8ToWide(utf8.data(), utf8.length(), &ret);
-  return ret;
-}
-
-bool UTF8ToWide(const char* src, size_t src_len, std::wstring* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  ReserveUTF16Or32Output(src, src_len, output);
-  return ConvertUnicode<char, std::wstring>(src, src_len, output);
-}
-
-// UTF-16 <-> Wide -------------------------------------------------------------
-
-#if defined(WCHAR_T_IS_UTF16)
-
-// When wide == UTF-16, then conversions are a NOP.
-string16 WideToUTF16(const std::wstring& wide) {
-  return wide;
-}
-
-bool WideToUTF16(const wchar_t* src, size_t src_len, string16* output) {
-  output->assign(src, src_len);
-  return true;
-}
-
-std::wstring UTF16ToWide(const string16& utf16) {
-  return utf16;
-}
-
-bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) {
-  output->assign(src, src_len);
-  return true;
-}
-
-#elif defined(WCHAR_T_IS_UTF32)
-
-string16 WideToUTF16(const std::wstring& wide) {
-  string16 ret;
-  if (wide.empty())
-    return ret;
-
-  WideToUTF16(wide.data(), wide.length(), &ret);
-  return ret;
-}
-
-bool WideToUTF16(const wchar_t* src, size_t src_len, string16* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  // Assume that normally we won't have any non-BMP characters so the counts
-  // will be the same.
-  output->reserve(src_len);
-  return ConvertUnicode<wchar_t, string16>(src, src_len, output);
-}
-
-std::wstring UTF16ToWide(const string16& utf16) {
-  std::wstring ret;
-  if (utf16.empty())
-    return ret;
-
-  UTF16ToWide(utf16.data(), utf16.length(), &ret);
-  return ret;
-}
-
-bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  // Assume that normally we won't have any non-BMP characters so the counts
-  // will be the same.
-  output->reserve(src_len);
-  return ConvertUnicode<char16, std::wstring>(src, src_len, output);
-}
-
-#endif  // defined(WCHAR_T_IS_UTF32)
-
-// UTF16 <-> UTF8 --------------------------------------------------------------
-
-#if defined(WCHAR_T_IS_UTF32)
-
-bool UTF8ToUTF16(const char* src, size_t src_len, string16* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  ReserveUTF16Or32Output(src, src_len, output);
-  return ConvertUnicode<char, string16>(src, src_len, output);
-}
-
-string16 UTF8ToUTF16(const std::string& utf8) {
-  string16 ret;
-  if (utf8.empty())
-    return ret;
-
-  // Ignore the success flag of this call, it will do the best it can for
-  // invalid input, which is what we want here.
-  UTF8ToUTF16(utf8.data(), utf8.length(), &ret);
-  return ret;
-}
-
-bool UTF16ToUTF8(const char16* src, size_t src_len, std::string* output) {
-  if (src_len == 0) {
-    output->clear();
-    return true;
-  }
-
-  ReserveUTF8Output(src, src_len, output);
-  return ConvertUnicode<char16, std::string>(src, src_len, output);
-}
-
-std::string UTF16ToUTF8(const string16& utf16) {
-  std::string ret;
-  if (utf16.empty())
-    return ret;
-
-  // Ignore the success flag of this call, it will do the best it can for
-  // invalid input, which is what we want here.
-  UTF16ToUTF8(utf16.data(), utf16.length(), &ret);
-  return ret;
-}
-
-#elif defined(WCHAR_T_IS_UTF16)
-// Easy case since we can use the "wide" versions we already wrote above.
-
-bool UTF8ToUTF16(const char* src, size_t src_len, string16* output) {
-  return UTF8ToWide(src, src_len, output);
-}
-
-string16 UTF8ToUTF16(const std::string& utf8) {
-  return UTF8ToWide(utf8);
-}
-
-bool UTF16ToUTF8(const char16* src, size_t src_len, std::string* output) {
-  return WideToUTF8(src, src_len, output);
-}
-
-std::string UTF16ToUTF8(const string16& utf16) {
-  return WideToUTF8(utf16);
-}
-
-#endif
-
 // Codepage <-> Wide/UTF-16  ---------------------------------------------------
 
 // Convert a wstring into the specified codepage_name.  If the codepage
diff --git a/base/i18n/icu_string_conversions.h b/base/i18n/icu_string_conversions.h
new file mode 100644
index 0000000..d849c71
--- /dev/null
+++ b/base/i18n/icu_string_conversions.h
@@ -0,0 +1,60 @@
+// Copyright (c) 2009 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef BASE_I18N_ICU_STRING_CONVERSIONS_H_
+#define BASE_I18N_ICU_STRING_CONVERSIONS_H_
+
+#include <string>
+
+#include "base/string16.h"
+#include "base/string_piece.h"
+
+// Defines the error handling modes of UTF16ToCodepage, CodepageToUTF16,
+// WideToCodepage and CodepageToWide.
+class OnStringUtilConversionError {
+ public:
+  enum Type {
+    // The function will return failure. The output buffer will be empty.
+    FAIL,
+
+    // The offending characters are skipped and the conversion will proceed as
+    // if they did not exist.
+    SKIP,
+
+    // When converting to Unicode, the offending byte sequences are substituted
+    // by Unicode replacement character (U+FFFD). When converting from Unicode,
+    // this is the same as SKIP.
+    SUBSTITUTE,
+  };
+
+ private:
+  OnStringUtilConversionError();
+};
+
+// Converts between UTF-16 strings and the encoding specified.  If the
+// encoding doesn't exist or the encoding fails (when on_error is FAIL),
+// returns false.
+bool UTF16ToCodepage(const string16& utf16,
+                     const char* codepage_name,
+                     OnStringUtilConversionError::Type on_error,
+                     std::string* encoded);
+
+bool CodepageToUTF16(const std::string& encoded,
+                     const char* codepage_name,
+                     OnStringUtilConversionError::Type on_error,
+                     string16* utf16);
+
+// Converts between wide strings and the encoding specified.  If the
+// encoding doesn't exist or the encoding fails (when on_error is FAIL),
+// returns false.
+bool WideToCodepage(const std::wstring& wide,
+                    const char* codepage_name,
+                    OnStringUtilConversionError::Type on_error,
+                    std::string* encoded);
+bool CodepageToWide(const std::string& encoded,
+                    const char* codepage_name,
+                    OnStringUtilConversionError::Type on_error,
+                    std::wstring* wide);
+
+#endif  // BASE_I18N_ICU_STRING_CONVERSIONS_H_
diff --git a/base/i18n/string_conversions.h b/base/i18n/string_conversions.h
deleted file mode 100644
index c055bb1..0000000
--- a/base/i18n/string_conversions.h
+++ /dev/null
@@ -1,101 +0,0 @@
-// Copyright (c) 2009 The Chromium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-#ifndef BASE_I18N_STRING_CONVERSIONS_H_
-#define BASE_I18N_STRING_CONVERSIONS_H_
-
-#include <string>
-
-#include "base/string16.h"
-#include "base/string_piece.h"
-
-// These convert between UTF-8, -16, and -32 strings. They are potentially slow,
-// so avoid unnecessary conversions. The low-level versions return a boolean
-// indicating whether the conversion was 100% valid. In this case, it will still
-// do the best it can and put the result in the output buffer. The versions that
-// return strings ignore this error and just return the best conversion
-// possible.
-//
-// Note that only the structural validity is checked and non-character
-// codepoints and unassigned are regarded as valid.
-// TODO(jungshik): Consider replacing an invalid input sequence with
-// the Unicode replacement character or adding |replacement_char| parameter.
-// Currently, it's skipped in the ouput, which could be problematic in
-// some situations.
-bool WideToUTF8(const wchar_t* src, size_t src_len, std::string* output);
-std::string WideToUTF8(const std::wstring& wide);
-bool UTF8ToWide(const char* src, size_t src_len, std::wstring* output);
-std::wstring UTF8ToWide(const base::StringPiece& utf8);
-
-bool WideToUTF16(const wchar_t* src, size_t src_len, string16* output);
-string16 WideToUTF16(const std::wstring& wide);
-bool UTF16ToWide(const char16* src, size_t src_len, std::wstring* output);
-std::wstring UTF16ToWide(const string16& utf16);
-
-bool UTF8ToUTF16(const char* src, size_t src_len, string16* output);
-string16 UTF8ToUTF16(const std::string& utf8);
-bool UTF16ToUTF8(const char16* src, size_t src_len, std::string* output);
-std::string UTF16ToUTF8(const string16& utf16);
-
-// We are trying to get rid of wstring as much as possible, but it's too big
-// a mess to do it all at once.  These conversions should be used when we
-// really should just be passing a string16 around, but we haven't finished
-// porting whatever module uses wstring and the conversion is being used as a
-// stopcock.  This makes it easy to grep for the ones that should be removed.
-#if defined(OS_WIN)
-# define WideToUTF16Hack
-# define UTF16ToWideHack
-#else
-# define WideToUTF16Hack WideToUTF16
-# define UTF16ToWideHack UTF16ToWide
-#endif
-
-// Defines the error handling modes of UTF16ToCodepage, CodepageToUTF16,
-// WideToCodepage and CodepageToWide.
-class OnStringUtilConversionError {
- public:
-  enum Type {
-    // The function will return failure. The output buffer will be empty.
-    FAIL,
-
-    // The offending characters are skipped and the conversion will proceed as
-    // if they did not exist.
-    SKIP,
-
-    // When converting to Unicode, the offending byte sequences are substituted
-    // by Unicode replacement character (U+FFFD). When converting from Unicode,
-    // this is the same as SKIP.
-    SUBSTITUTE,
-  };
-
- private:
-  OnStringUtilConversionError();
-};
-
-// Converts between UTF-16 strings and the encoding specified.  If the
-// encoding doesn't exist or the encoding fails (when on_error is FAIL),
-// returns false.
-bool UTF16ToCodepage(const string16& utf16,
-                     const char* codepage_name,
-                     OnStringUtilConversionError::Type on_error,
-                     std::string* encoded);
-
-bool CodepageToUTF16(const std::string& encoded,
-                     const char* codepage_name,
-                     OnStringUtilConversionError::Type on_error,
-                     string16* utf16);
-
-// Converts between wide strings and the encoding specified.  If the
-// encoding doesn't exist or the encoding fails (when on_error is FAIL),
-// returns false.
-bool WideToCodepage(const std::wstring& wide,
-                    const char* codepage_name,
-                    OnStringUtilConversionError::Type on_error,
-                    std::string* encoded);
-bool CodepageToWide(const std::string& encoded,
-                    const char* codepage_name,
-                    OnStringUtilConversionError::Type on_error,
-                    std::wstring* wide);
-
-#endif  // BASE_I18N_STRING_CONVERSIONS_H_