base/json/string_escape.cc

   1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "base/json/string_escape.h"
   6
   7 #include <string>
   8
   9 #include "base/strings/string_util.h"
  10 #include "base/strings/stringprintf.h"
  11 #include "base/strings/utf_string_conversion_utils.h"
  12 #include "base/strings/utf_string_conversions.h"
  13 #include "base/third_party/icu/icu_utf.h"
  14
  15 namespace base {
  16
  17 namespace {
  18
  19 // Format string for printing a \uXXXX escape sequence.
  20 const char kU16EscapeFormat[] = "\\u%04X";
  21
  22 // The code point to output for an invalid input code unit.
  23 const uint32 kReplacementCodePoint = 0xFFFD;
  24
  25 // Used below in EscapeSpecialCodePoint().
  26 COMPILE_ASSERT('<' == 0x3C, less_than_sign_is_0x3c);
  27
  28 // Try to escape the |code_point| if it is a known special character. If
  29 // successful, returns true and appends the escape sequence to |dest|. This
  30 // isn't required by the spec, but it's more readable by humans.
  31 bool EscapeSpecialCodePoint(uint32 code_point, std::string* dest) {
  32   // WARNING: if you add a new case here, you need to update the reader as well.
  33   // Note: \v is in the reader, but not here since the JSON spec doesn't
  34   // allow it.
  35   switch (code_point) {
  36     case '\b':
  37       dest->append("\\b");
  38       break;
  39     case '\f':
  40       dest->append("\\f");
  41       break;
  42     case '\n':
  43       dest->append("\\n");
  44       break;
  45     case '\r':
  46       dest->append("\\r");
  47       break;
  48     case '\t':
  49       dest->append("\\t");
  50       break;
  51     case '\\':
  52       dest->append("\\\\");
  53       break;
  54     case '"':
  55       dest->append("\\\"");
  56       break;
  57     // Escape < to prevent script execution; escaping > is not necessary and
  58     // not doing so save a few bytes.
  59     case '<':
  60       dest->append("\\u003C");
  61       break;
  62     // Escape the "Line Separator" and "Paragraph Separator" characters, since
  63     // they should be treated like a new line \r or \n.
  64     case 0x2028:
  65       dest->append("\\u2028");
  66       break;
  67     case 0x2029:
  68       dest->append("\\u2029");
  69       break;
  70     default:
  71       return false;
  72   }
  73   return true;
  74 }
  75
  76 template <typename S>
  77 bool EscapeJSONStringImpl(const S& str, bool put_in_quotes, std::string* dest) {
  78   bool did_replacement = false;
  79
  80   if (put_in_quotes)
  81     dest->push_back('"');
  82
  83   // Casting is necessary because ICU uses int32. Try and do so safely.
  84   CHECK_LE(str.length(), static_cast<size_t>(kint32max));
  85   const int32 length = static_cast<int32>(str.length());
  86
  87   for (int32 i = 0; i < length; ++i) {
  88     uint32 code_point;
  89     if (!ReadUnicodeCharacter(str.data(), length, &i, &code_point)) {
  90       code_point = kReplacementCodePoint;
  91       did_replacement = true;
  92     }
  93
  94     if (EscapeSpecialCodePoint(code_point, dest))
  95       continue;
  96
  97     // Escape non-printing characters.
  98     if (code_point < 32)
  99       base::StringAppendF(dest, kU16EscapeFormat, code_point);
 100     else
 101       WriteUnicodeCharacter(code_point, dest);
 102   }
 103
 104   if (put_in_quotes)
 105     dest->push_back('"');
 106
 107   return !did_replacement;
 108 }
 109
 110 }  // namespace
 111
 112 bool EscapeJSONString(const StringPiece& str,
 113                       bool put_in_quotes,
 114                       std::string* dest) {
 115   return EscapeJSONStringImpl(str, put_in_quotes, dest);
 116 }
 117
 118 bool EscapeJSONString(const StringPiece16& str,
 119                       bool put_in_quotes,
 120                       std::string* dest) {
 121   return EscapeJSONStringImpl(str, put_in_quotes, dest);
 122 }
 123
 124 std::string GetQuotedJSONString(const StringPiece& str) {
 125   std::string dest;
 126   bool ok = EscapeJSONStringImpl(str, true, &dest);
 127   DCHECK(ok);
 128   return dest;
 129 }
 130
 131 std::string GetQuotedJSONString(const StringPiece16& str) {
 132   std::string dest;
 133   bool ok = EscapeJSONStringImpl(str, true, &dest);
 134   DCHECK(ok);
 135   return dest;
 136 }
 137
 138 std::string EscapeBytesAsInvalidJSONString(const StringPiece& str,
 139                                            bool put_in_quotes) {
 140   std::string dest;
 141
 142   if (put_in_quotes)
 143     dest.push_back('"');
 144
 145   for (StringPiece::const_iterator it = str.begin(); it != str.end(); ++it) {
 146     unsigned char c = *it;
 147     if (EscapeSpecialCodePoint(c, &dest))
 148       continue;
 149
 150     if (c < 32 || c > 126)
 151       base::StringAppendF(&dest, kU16EscapeFormat, c);
 152     else
 153       dest.push_back(*it);
 154   }
 155
 156   if (put_in_quotes)
 157     dest.push_back('"');
 158
 159   return dest;
 160 }
 161
 162 }  // namespace base