1 // Copyright (c) 2013 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 <stddef.h> 8 9 #include "base/macros.h" 10 #include "base/strings/string_util.h" 11 #include "base/strings/utf_string_conversions.h" 12 #include "testing/gtest/include/gtest/gtest.h" 13 14 namespace base { 15 16 TEST(JSONStringEscapeTest, EscapeUTF8) { 17 const struct { 18 const char* to_escape; 19 const char* escaped; 20 } cases[] = { 21 {"\b\001aZ\"\\wee", "\\b\\u0001aZ\\\"\\\\wee"}, 22 {"a\b\f\n\r\t\v\1\\.\"z", 23 "a\\b\\f\\n\\r\\t\\u000B\\u0001\\\\.\\\"z"}, 24 {"b\x0f\x7f\xf0\xff!", // \xf0\xff is not a valid UTF-8 unit. 25 "b\\u000F\x7F\xEF\xBF\xBD\xEF\xBF\xBD!"}, 26 {"c<>d", "c\\u003C>d"}, 27 {"Hello\xe2\x80\xa8world", "Hello\\u2028world"}, 28 {"\xe2\x80\xa9purple", "\\u2029purple"}, 29 }; 30 31 for (size_t i = 0; i < arraysize(cases); ++i) { 32 const char* in_ptr = cases[i].to_escape; 33 std::string in_str = in_ptr; 34 35 std::string out; 36 EscapeJSONString(in_ptr, false, &out); 37 EXPECT_EQ(std::string(cases[i].escaped), out); 38 EXPECT_TRUE(IsStringUTF8(out)); 39 40 out.erase(); 41 bool convert_ok = EscapeJSONString(in_str, false, &out); 42 EXPECT_EQ(std::string(cases[i].escaped), out); 43 EXPECT_TRUE(IsStringUTF8(out)); 44 45 if (convert_ok) { 46 std::string fooout = GetQuotedJSONString(in_str); 47 EXPECT_EQ("\"" + std::string(cases[i].escaped) + "\"", fooout); 48 EXPECT_TRUE(IsStringUTF8(out)); 49 } 50 } 51 52 std::string in = cases[0].to_escape; 53 std::string out; 54 EscapeJSONString(in, false, &out); 55 EXPECT_TRUE(IsStringUTF8(out)); 56 57 // test quoting 58 std::string out_quoted; 59 EscapeJSONString(in, true, &out_quoted); 60 EXPECT_EQ(out.length() + 2, out_quoted.length()); 61 EXPECT_EQ(out_quoted.find(out), 1U); 62 EXPECT_TRUE(IsStringUTF8(out_quoted)); 63 64 // now try with a NULL in the string 65 std::string null_prepend = "test"; 66 null_prepend.push_back(0); 67 in = null_prepend + in; 68 std::string expected = "test\\u0000"; 69 expected += cases[0].escaped; 70 out.clear(); 71 EscapeJSONString(in, false, &out); 72 EXPECT_EQ(expected, out); 73 EXPECT_TRUE(IsStringUTF8(out)); 74 } 75 76 TEST(JSONStringEscapeTest, EscapeUTF16) { 77 const struct { 78 const wchar_t* to_escape; 79 const char* escaped; 80 } cases[] = { 81 {L"b\uffb1\u00ff", "b\xEF\xBE\xB1\xC3\xBF"}, 82 {L"\b\001aZ\"\\wee", "\\b\\u0001aZ\\\"\\\\wee"}, 83 {L"a\b\f\n\r\t\v\1\\.\"z", 84 "a\\b\\f\\n\\r\\t\\u000B\\u0001\\\\.\\\"z"}, 85 {L"b\x0f\x7f\xf0\xff!", "b\\u000F\x7F\xC3\xB0\xC3\xBF!"}, 86 {L"c<>d", "c\\u003C>d"}, 87 {L"Hello\u2028world", "Hello\\u2028world"}, 88 {L"\u2029purple", "\\u2029purple"}, 89 }; 90 91 for (size_t i = 0; i < arraysize(cases); ++i) { 92 string16 in = WideToUTF16(cases[i].to_escape); 93 94 std::string out; 95 EscapeJSONString(in, false, &out); 96 EXPECT_EQ(std::string(cases[i].escaped), out); 97 EXPECT_TRUE(IsStringUTF8(out)); 98 99 out = GetQuotedJSONString(in); 100 EXPECT_EQ("\"" + std::string(cases[i].escaped) + "\"", out); 101 EXPECT_TRUE(IsStringUTF8(out)); 102 } 103 104 string16 in = WideToUTF16(cases[0].to_escape); 105 std::string out; 106 EscapeJSONString(in, false, &out); 107 EXPECT_TRUE(IsStringUTF8(out)); 108 109 // test quoting 110 std::string out_quoted; 111 EscapeJSONString(in, true, &out_quoted); 112 EXPECT_EQ(out.length() + 2, out_quoted.length()); 113 EXPECT_EQ(out_quoted.find(out), 1U); 114 EXPECT_TRUE(IsStringUTF8(out)); 115 116 // now try with a NULL in the string 117 string16 null_prepend = WideToUTF16(L"test"); 118 null_prepend.push_back(0); 119 in = null_prepend + in; 120 std::string expected = "test\\u0000"; 121 expected += cases[0].escaped; 122 out.clear(); 123 EscapeJSONString(in, false, &out); 124 EXPECT_EQ(expected, out); 125 EXPECT_TRUE(IsStringUTF8(out)); 126 } 127 128 TEST(JSONStringEscapeTest, EscapeUTF16OutsideBMP) { 129 { 130 // {a, U+10300, !}, SMP. 131 string16 test; 132 test.push_back('a'); 133 test.push_back(0xD800); 134 test.push_back(0xDF00); 135 test.push_back('!'); 136 std::string actual; 137 EXPECT_TRUE(EscapeJSONString(test, false, &actual)); 138 EXPECT_EQ("a\xF0\x90\x8C\x80!", actual); 139 } 140 { 141 // {U+20021, U+2002B}, SIP. 142 string16 test; 143 test.push_back(0xD840); 144 test.push_back(0xDC21); 145 test.push_back(0xD840); 146 test.push_back(0xDC2B); 147 std::string actual; 148 EXPECT_TRUE(EscapeJSONString(test, false, &actual)); 149 EXPECT_EQ("\xF0\xA0\x80\xA1\xF0\xA0\x80\xAB", actual); 150 } 151 { 152 // {?, U+D800, @}, lone surrogate. 153 string16 test; 154 test.push_back('?'); 155 test.push_back(0xD800); 156 test.push_back('@'); 157 std::string actual; 158 EXPECT_FALSE(EscapeJSONString(test, false, &actual)); 159 EXPECT_EQ("?\xEF\xBF\xBD@", actual); 160 } 161 } 162 163 TEST(JSONStringEscapeTest, EscapeBytes) { 164 const struct { 165 const char* to_escape; 166 const char* escaped; 167 } cases[] = { 168 {"b\x0f\x7f\xf0\xff!", "b\\u000F\\u007F\\u00F0\\u00FF!"}, 169 {"\xe5\xc4\x4f\x05\xb6\xfd", "\\u00E5\\u00C4O\\u0005\\u00B6\\u00FD"}, 170 }; 171 172 for (size_t i = 0; i < arraysize(cases); ++i) { 173 std::string in = std::string(cases[i].to_escape); 174 EXPECT_FALSE(IsStringUTF8(in)); 175 176 EXPECT_EQ(std::string(cases[i].escaped), 177 EscapeBytesAsInvalidJSONString(in, false)); 178 EXPECT_EQ("\"" + std::string(cases[i].escaped) + "\"", 179 EscapeBytesAsInvalidJSONString(in, true)); 180 } 181 182 const char kEmbedNull[] = { '\xab', '\x39', '\0', '\x9f', '\xab' }; 183 std::string in(kEmbedNull, arraysize(kEmbedNull)); 184 EXPECT_FALSE(IsStringUTF8(in)); 185 EXPECT_EQ(std::string("\\u00AB9\\u0000\\u009F\\u00AB"), 186 EscapeBytesAsInvalidJSONString(in, false)); 187 } 188 189 } // namespace base 190
