Make IsStringUTF8 reject (U+FDD0 .. U+FDEF) ...

base/utf_string_conversion_utils.h

 // 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_UTF_STRING_CONVERSION_UTILS_H_
 #define BASE_UTF_STRING_CONVERSION_UTILS_H_
 
 // This should only be used by the various UTF string conversion files.
 
 #include "base/string16.h"
 
 namespace base {
 
 inline bool IsValidCodepoint(uint32 code_point) {
   // Excludes the surrogate code points ([0xD800, 0xDFFF]) and
   // codepoints larger than 0x10FFFF (the highest codepoint allowed).
   // Non-characters and unassigned codepoints are allowed.
   return code_point < 0xD800u ||
          (code_point >= 0xE000u && code_point <= 0x10FFFFu);
 }
 
+inline bool IsValidCharacter(uint32 code_point) {
+  // Excludes non-characters (U+FDD0..U+FDEF, and all codepoints ending in
+  // 0xFFFE or 0xFFFF) from the set of valid code points.
+  return code_point < 0xD800u || (code_point >= 0xE000u &&
+      code_point < 0xFDD0u) || (code_point > 0xFDEFu &&
+      code_point <= 0x10FFFFu && (code_point & 0xFFFEu) != 0xFFFEu);
+}
+
 // 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,
@@ skipping 45 matching lines @@
 void PrepareForUTF8Output(const CHAR* src, size_t src_len, std::string* output);
 
 // Prepares an output buffer (containing either UTF-16 or -32 data) given some
 // UTF-8 input that will be converted to it.  See PrepareForUTF8Output().
 template<typename STRING>
 void PrepareForUTF16Or32Output(const char* src, size_t src_len, STRING* output);
 
 }  // namespace base
 
 #endif  // BASE_UTF_STRING_CONVERSION_UTILS_H_