Rewrite base::JSONReader to be 35-40% faster, depending on the input string.

base/string_util.cc

 // Copyright (c) 2012 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/string_util.h"
 
 #include "build/build_config.h"
 
 #include <ctype.h>
 #include <errno.h>
@@ skipping 457 matching lines @@
 }
 
 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;
+    if (!base::IsValidCharacter(code_point)) {
+      for (size_t i = 0; i < str.length(); ++i) {

[Mark Mentovai, 2012/05/08 20:19:41]

Dump the debugging code.

[Robert Sesek, 2012/05/15 16:57:51]

Done.

+        printf("0x%x ", str[i]);
+      }
+      printf("\n");
+      LOG(INFO) << "STRING = <<<" << str << ">>>";
+      LOG(INFO) << "**** char_index = " << char_index << " // " << 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)
@@ skipping 556 matching lines @@
 }
 
 }  // 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);
 }