1 // Copyright 2014 The Chromium Authors
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
5 // All data that is passed through a WebSocket with type "Text" needs to be
6 // validated as UTF8. Since this is done on the IO thread, it needs to be
9 // We are only interested in the performance on valid UTF8. Invalid UTF8 will
10 // result in a connection failure, so is unlikely to become a source of
11 // performance issues.
13 #include "base/i18n/streaming_utf8_validator.h"
19 #include "base/functional/bind.h"
20 #include "base/functional/callback.h"
21 #include "base/strings/string_util.h"
22 #include "base/strings/stringprintf.h"
23 #include "base/test/perf_time_logger.h"
24 #include "testing/gtest/include/gtest/gtest.h"
29 // We want to test ranges of valid UTF-8 sequences. These ranges are inclusive.
30 // They are intended to be large enough that the validator needs to do
31 // meaningful work while being in some sense "realistic" (eg. control characters
33 const char kOneByteSeqRangeStart[] = " "; // U+0020
34 const char kOneByteSeqRangeEnd[] = "~"; // U+007E
36 const char kTwoByteSeqRangeStart[] = "\xc2\xa0"; // U+00A0 non-breaking space
37 const char kTwoByteSeqRangeEnd[] = "\xc9\x8f"; // U+024F small y with stroke
39 const char kThreeByteSeqRangeStart[] = "\xe3\x81\x82"; // U+3042 Hiragana "a"
40 const char kThreeByteSeqRangeEnd[] = "\xe9\xbf\x83"; // U+9FC3 "to blink"
42 const char kFourByteSeqRangeStart[] = "\xf0\xa0\x80\x8b"; // U+2000B
43 const char kFourByteSeqRangeEnd[] = "\xf0\xaa\x9a\xb2"; // U+2A6B2
45 // The different lengths of strings to test.
46 const size_t kTestLengths[] = {1, 32, 256, 32768, 1 << 20};
48 // Simplest possible byte-at-a-time validator, to provide a baseline
49 // for comparison. This is only tried on 1-byte UTF-8 sequences, as
50 // the results will not be meaningful with sequences containing
52 bool IsString7Bit(const std::string& s) {
60 // Assumes that |previous| is a valid UTF-8 sequence, and attempts to return
61 // the next one. Is just barely smart enough to iterate through the ranges
63 std::string NextUtf8Sequence(const std::string& previous) {
64 DCHECK(StreamingUtf8Validator::Validate(previous));
65 std::string next = previous;
66 for (int i = static_cast<int>(previous.length() - 1); i >= 0; --i) {
67 // All bytes in a UTF-8 sequence except the first one are
68 // constrained to the range 0x80 to 0xbf, inclusive. When we
69 // increment past 0xbf, we carry into the previous byte.
70 if (i > 0 && next[i] == '\xbf') {
77 DCHECK(StreamingUtf8Validator::Validate(next))
78 << "Result \"" << next << "\" failed validation";
82 typedef bool (*TestTargetType)(const std::string&);
84 // Run fuction |target| over |test_string| |times| times, and report the results
85 // using |description|.
86 bool RunTest(const std::string& description,
87 TestTargetType target,
88 const std::string& test_string,
90 base::PerfTimeLogger timer(description.c_str());
92 for (int i = 0; i < times; ++i) {
93 result = target(test_string) && result;
99 // Construct a string by repeating |input| enough times to equal or exceed
101 std::string ConstructRepeatedTestString(const std::string& input,
103 std::string output = input;
104 while (output.length() * 2 < length) {
107 if (output.length() < length) {
108 output += ConstructRepeatedTestString(input, length - output.length());
113 // Construct a string by expanding the range of UTF-8 sequences
114 // between |input_start| and |input_end|, inclusive, and then
115 // repeating the resulting string until it equals or exceeds |length|
116 // bytes. |input_start| and |input_end| must be valid UTF-8
118 std::string ConstructRangedTestString(const std::string& input_start,
119 const std::string& input_end,
121 std::string output = input_start;
122 std::string input = input_start;
123 while (output.length() < length && input != input_end) {
124 input = NextUtf8Sequence(input);
127 if (output.length() < length) {
128 output = ConstructRepeatedTestString(output, length);
133 struct TestFunctionDescription {
134 TestTargetType function;
135 const char* function_name;
138 bool IsStringUTF8(const std::string& str) {
139 return base::IsStringUTF8(base::StringPiece(str));
142 // IsString7Bit is intentionally placed last so it can be excluded easily.
143 const TestFunctionDescription kTestFunctions[] = {
144 {&StreamingUtf8Validator::Validate, "StreamingUtf8Validator"},
145 {&IsStringUTF8, "IsStringUTF8"}, {&IsString7Bit, "IsString7Bit"}};
147 // Construct a test string from |construct_test_string| for each of the lengths
148 // in |kTestLengths| in turn. For each string, run each test in |test_functions|
149 // for a number of iterations such that the total number of bytes validated
153 base::RepeatingCallback<std::string(size_t length)> construct_test_string,
154 const TestFunctionDescription* test_functions,
156 for (auto length : kTestLengths) {
157 const std::string test_string = construct_test_string.Run(length);
158 const int real_length = static_cast<int>(test_string.length());
159 const int times = (1 << 24) / real_length;
160 for (size_t test_index = 0; test_index < test_count; ++test_index) {
161 EXPECT_TRUE(RunTest(StringPrintfNonConstexpr(
162 format, test_functions[test_index].function_name,
164 test_functions[test_index].function, test_string,
170 TEST(StreamingUtf8ValidatorPerfTest, OneByteRepeated) {
172 "%s: bytes=1 repeated length=%d repeat=%d",
173 base::BindRepeating(ConstructRepeatedTestString, kOneByteSeqRangeStart),
177 TEST(StreamingUtf8ValidatorPerfTest, OneByteRange) {
178 RunSomeTests("%s: bytes=1 ranged length=%d repeat=%d",
179 base::BindRepeating(ConstructRangedTestString,
180 kOneByteSeqRangeStart, kOneByteSeqRangeEnd),
184 TEST(StreamingUtf8ValidatorPerfTest, TwoByteRepeated) {
186 "%s: bytes=2 repeated length=%d repeat=%d",
187 base::BindRepeating(ConstructRepeatedTestString, kTwoByteSeqRangeStart),
191 TEST(StreamingUtf8ValidatorPerfTest, TwoByteRange) {
192 RunSomeTests("%s: bytes=2 ranged length=%d repeat=%d",
193 base::BindRepeating(ConstructRangedTestString,
194 kTwoByteSeqRangeStart, kTwoByteSeqRangeEnd),
198 TEST(StreamingUtf8ValidatorPerfTest, ThreeByteRepeated) {
200 "%s: bytes=3 repeated length=%d repeat=%d",
201 base::BindRepeating(ConstructRepeatedTestString, kThreeByteSeqRangeStart),
205 TEST(StreamingUtf8ValidatorPerfTest, ThreeByteRange) {
207 "%s: bytes=3 ranged length=%d repeat=%d",
208 base::BindRepeating(ConstructRangedTestString, kThreeByteSeqRangeStart,
209 kThreeByteSeqRangeEnd),
213 TEST(StreamingUtf8ValidatorPerfTest, FourByteRepeated) {
215 "%s: bytes=4 repeated length=%d repeat=%d",
216 base::BindRepeating(ConstructRepeatedTestString, kFourByteSeqRangeStart),
220 TEST(StreamingUtf8ValidatorPerfTest, FourByteRange) {
222 "%s: bytes=4 ranged length=%d repeat=%d",
223 base::BindRepeating(ConstructRangedTestString, kFourByteSeqRangeStart,
224 kFourByteSeqRangeEnd),