| // Copyright 2016 the V8 project 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 <memory> |
| #include <string> |
| #include <vector> |
| |
| #include "src/unicode-decoder.h" |
| #include "src/unicode-inl.h" |
| #include "testing/gtest/include/gtest/gtest.h" |
| |
| namespace v8 { |
| namespace internal { |
| |
| namespace { |
| |
| using Utf8Decoder = unibrow::Utf8Decoder<512>; |
| |
| void Decode(Utf8Decoder* decoder, const std::string& str) { |
| // Put the string in its own buffer on the heap to make sure that |
| // AddressSanitizer's heap-buffer-overflow logic can see what's going on. |
| std::unique_ptr<char[]> buffer(new char[str.length()]); |
| memcpy(buffer.get(), str.data(), str.length()); |
| decoder->Reset(buffer.get(), str.length()); |
| } |
| |
| void DecodeNormally(const std::vector<byte>& bytes, |
| std::vector<unibrow::uchar>* output) { |
| size_t cursor = 0; |
| while (cursor < bytes.size()) { |
| output->push_back( |
| unibrow::Utf8::ValueOf(bytes.data() + cursor, bytes.size(), &cursor)); |
| } |
| } |
| |
| void DecodeIncrementally(const std::vector<byte>& bytes, |
| std::vector<unibrow::uchar>* output) { |
| unibrow::Utf8::Utf8IncrementalBuffer buffer = 0; |
| unibrow::Utf8::State state = unibrow::Utf8::State::kAccept; |
| for (size_t i = 0; i < bytes.size();) { |
| unibrow::uchar result = |
| unibrow::Utf8::ValueOfIncremental(bytes[i], &i, &state, &buffer); |
| if (result != unibrow::Utf8::kIncomplete) { |
| output->push_back(result); |
| } |
| } |
| unibrow::uchar result = unibrow::Utf8::ValueOfIncrementalFinish(&state); |
| if (result != unibrow::Utf8::kBufferEmpty) { |
| output->push_back(result); |
| } |
| } |
| |
| } // namespace |
| |
| TEST(UnicodeTest, ReadOffEndOfUtf8String) { |
| Utf8Decoder decoder; |
| |
| // Not enough continuation bytes before string ends. |
| Decode(&decoder, "\xE0"); |
| Decode(&decoder, "\xED"); |
| Decode(&decoder, "\xF0"); |
| Decode(&decoder, "\xF4"); |
| } |
| |
| TEST(UnicodeTest, IncrementalUTF8DecodingVsNonIncrementalUtf8Decoding) { |
| // Unfortunately, V8 has two UTF-8 decoders. This test checks that they |
| // produce the same result. This test was inspired by |
| // https://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt . |
| typedef struct { |
| std::vector<byte> bytes; |
| std::vector<unibrow::uchar> unicode_expected; |
| } TestCase; |
| |
| TestCase data[] = { |
| // Correct UTF-8 text. |
| {{0xCE, 0xBA, 0xE1, 0xBD, 0xB9, 0xCF, 0x83, 0xCE, 0xBC, 0xCE, 0xB5}, |
| {0x3BA, 0x1F79, 0x3C3, 0x3BC, 0x3B5}}, |
| |
| // First possible sequence of a certain length: |
| // 1 byte |
| {{0x00}, {0x0}}, |
| // 2 bytes |
| {{0xC2, 0x80}, {0x80}}, |
| // 3 bytes |
| {{0xE0, 0xA0, 0x80}, {0x800}}, |
| // 4 bytes |
| {{0xF0, 0x90, 0x80, 0x80}, {0x10000}}, |
| // 5 bytes (not supported) |
| {{0xF8, 0x88, 0x80, 0x80, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6 bytes (not supported) |
| {{0xFC, 0x84, 0x80, 0x80, 0x80, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Last possible sequence of certain length: |
| // 1 byte |
| {{0x7F}, {0x7F}}, |
| // 2 bytes |
| {{0xDF, 0xBF}, {0x7FF}}, |
| // 3 bytes |
| {{0xEF, 0xBF, 0xBF}, {0xFFFF}}, |
| // 4 bytes (this sequence is not a valid code point) |
| {{0xF7, 0xBF, 0xBF, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5 bytes (not supported) |
| {{0xFB, 0xBF, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6 bytes (not supported) |
| {{0xFD, 0xBF, 0xBF, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // Other boundary conditions: |
| {{0xED, 0x9F, 0xBF}, {0xD7FF}}, |
| {{0xEE, 0x80, 0x80}, {0xE000}}, |
| // U+fffd (invalid code point) |
| {{0xEF, 0xBF, 0xBD}, {0xFFFD}}, |
| // U+10ffff (last valid code point) |
| {{0xF4, 0x8F, 0xBF, 0xBF}, {0x10FFFF}}, |
| // First invalid (too large) code point |
| {{0xF4, 0x90, 0x80, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Malformed sequences: |
| // Unexpected continuation bytes: |
| // First continuation byte |
| {{0x80}, {0xFFFD}}, |
| // Last continuation byte |
| {{0xBF}, {0xFFFD}}, |
| // 2 continuation bytes |
| {{0x80, 0xBF}, {0xFFFD, 0xFFFD}}, |
| // 3 continuation bytes |
| {{0x80, 0xBF, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 4 continuation bytes |
| {{0x80, 0xBF, 0x80, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5 continuation bytes |
| {{0x80, 0xBF, 0x80, 0xBF, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6 continuation bytes |
| {{0x80, 0xBF, 0x80, 0xBF, 0x80, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 7 continuation bytes |
| {{0x80, 0xBF, 0x80, 0xBF, 0x80, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // Sequence of all 64 possible continuation bytes |
| {{0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, |
| 0x8B, 0x8C, 0x8D, 0x8E, 0x8F, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, |
| 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F, 0xA0, |
| 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xAB, |
| 0xAC, 0xAD, 0xAE, 0xAF, 0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, |
| 0xB7, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // Using each possible continuation byte in a two-byte sequence: |
| {{0xD0, 0x80, 0xD0, 0x81, 0xD0, 0x82, 0xD0, 0x83, 0xD0, 0x84, 0xD0, 0x85, |
| 0xD0, 0x86, 0xD0, 0x87, 0xD0, 0x88, 0xD0, 0x89, 0xD0, 0x8A, 0xD0, 0x8B, |
| 0xD0, 0x8C, 0xD0, 0x8D, 0xD0, 0x8E, 0xD0, 0x8F, 0xD0, 0x90, 0xD0, 0x91, |
| 0xD0, 0x92, 0xD0, 0x93, 0xD0, 0x94, 0xD0, 0x95, 0xD0, 0x96, 0xD0, 0x97, |
| 0xD0, 0x98, 0xD0, 0x99, 0xD0, 0x9A, 0xD0, 0x9B, 0xD0, 0x9C, 0xD0, 0x9D, |
| 0xD0, 0x9E, 0xD0, 0x9F, 0xD0, 0xA0, 0xD0, 0xA1, 0xD0, 0xA2, 0xD0, 0xA3, |
| 0xD0, 0xA4, 0xD0, 0xA5, 0xD0, 0xA6, 0xD0, 0xA7, 0xD0, 0xA8, 0xD0, 0xA9, |
| 0xD0, 0xAA, 0xD0, 0xAB, 0xD0, 0xAC, 0xD0, 0xAD, 0xD0, 0xAE, 0xD0, 0xAF, |
| 0xD0, 0xB0, 0xD0, 0xB1, 0xD0, 0xB2, 0xD0, 0xB3, 0xD0, 0xB4, 0xD0, 0xB5, |
| 0xD0, 0xB6, 0xD0, 0xB7, 0xD0, 0xB8, 0xD0, 0xB9, 0xD0, 0xBA, 0xD0, 0xBB, |
| 0xD0, 0xBC, 0xD0, 0xBD, 0xD0, 0xBE, 0xD0, 0xBF}, |
| {0x400, 0x401, 0x402, 0x403, 0x404, 0x405, 0x406, 0x407, 0x408, 0x409, |
| 0x40A, 0x40B, 0x40C, 0x40D, 0x40E, 0x40F, 0x410, 0x411, 0x412, 0x413, |
| 0x414, 0x415, 0x416, 0x417, 0x418, 0x419, 0x41A, 0x41B, 0x41C, 0x41D, |
| 0x41E, 0x41F, 0x420, 0x421, 0x422, 0x423, 0x424, 0x425, 0x426, 0x427, |
| 0x428, 0x429, 0x42A, 0x42B, 0x42C, 0x42D, 0x42E, 0x42F, 0x430, 0x431, |
| 0x432, 0x433, 0x434, 0x435, 0x436, 0x437, 0x438, 0x439, 0x43A, 0x43B, |
| 0x43C, 0x43D, 0x43E, 0x43F}}, |
| |
| // Lonely first bytes: |
| // All 32 first bytes of 32-byte sequences, each followed by a space |
| // (generates 32 invalid char + space sequences. |
| {{0xC0, 0x20, 0xC1, 0x20, 0xC2, 0x20, 0xC3, 0x20, 0xC4, 0x20, 0xC5, |
| 0x20, 0xC6, 0x20, 0xC7, 0x20, 0xC8, 0x20, 0xC9, 0x20, 0xCA, 0x20, |
| 0xCB, 0x20, 0xCC, 0x20, 0xCD, 0x20, 0xCE, 0x20, 0xCF, 0x20, 0xD0, |
| 0x20, 0xD1, 0x20, 0xD2, 0x20, 0xD3, 0x20, 0xD4, 0x20, 0xD5, 0x20, |
| 0xD6, 0x20, 0xD7, 0x20, 0xD8, 0x20, 0xD9, 0x20, 0xDA, 0x20, 0xDB, |
| 0x20, 0xDC, 0x20, 0xDD, 0x20, 0xDE, 0x20, 0xDF, 0x20}, |
| {0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20}}, |
| // All 16 first bytes of 3-byte sequences, each followed by a space |
| // (generates 16 invalid char + space sequences): |
| {{0xE0, 0x20, 0xE1, 0x20, 0xE2, 0x20, 0xE3, 0x20, 0xE4, 0x20, 0xE5, |
| 0x20, 0xE6, 0x20, 0xE7, 0x20, 0xE8, 0x20, 0xE9, 0x20, 0xEA, 0x20, |
| 0xEB, 0x20, 0xEC, 0x20, 0xED, 0x20, 0xEE, 0x20, 0xEF, 0x20}, |
| {0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20}}, |
| // All 8 first bytes of 4-byte sequences, each followed by a space |
| // (generates 8 invalid char + space sequences): |
| {{0xF0, 0x20, 0xF1, 0x20, 0xF2, 0x20, 0xF3, 0x20, 0xF4, 0x20, 0xF5, 0x20, |
| 0xF6, 0x20, 0xF7, 0x20}, |
| {0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, |
| 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20}}, |
| // All 4 first bytes of 5-byte sequences (not supported), each followed by |
| // a space (generates 4 invalid char + space sequences): |
| {{0xF8, 0x20, 0xF9, 0x20, 0xFA, 0x20, 0xFB, 0x20}, |
| {0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20, 0xFFFD, 0x20}}, |
| // All 2 first bytes of 6-byte sequences (not supported), each followed by |
| // a space (generates 2 invalid char + space sequences): |
| {{0xFC, 0x20, 0xFD, 0x20}, {0xFFFD, 0x20, 0xFFFD, 0x20}}, |
| |
| // Sequences with last continuation byte missing. Normally the whole |
| // incomplete sequence generates a single invalid character (exceptions |
| // explained below). |
| |
| // 2-byte sequences with last byte missing |
| {{0xC0}, {0xFFFD}}, |
| {{0xDF}, {0xFFFD}}, |
| // 3-byte sequences with last byte missing. |
| {{0xE8, 0x80}, {0xFFFD}}, |
| {{0xE0, 0xBF}, {0xFFFD}}, |
| {{0xEF, 0xBF}, {0xFFFD}}, |
| // Start of an overlong sequence. The first "maximal subpart" is the first |
| // byte; it creates an invalid character. Each following byte generates an |
| // invalid character too. |
| {{0xE0, 0x80}, {0xFFFD, 0xFFFD}}, |
| // 4-byte sequences with last byte missing |
| {{0xF1, 0x80, 0x80}, {0xFFFD}}, |
| {{0xF4, 0x8F, 0xBF}, {0xFFFD}}, |
| // Start of an overlong sequence. The first "maximal subpart" is the first |
| // byte; it creates an invalid character. Each following byte generates an |
| // invalid character too. |
| {{0xF0, 0x80, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5-byte sequences (not supported) with last byte missing |
| {{0xF8, 0x80, 0x80, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFB, 0xBF, 0xBF, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6-byte sequences (not supported) with last byte missing |
| {{0xFC, 0x80, 0x80, 0x80, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFD, 0xBF, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Concatenation of incomplete sequences: above incomplete sequences |
| // concatenated. |
| {{0xC0, 0xDF, 0xE8, 0x80, 0xE0, 0xBF, 0xEF, 0xBF, 0xE0, 0x80, |
| 0xF1, 0x80, 0x80, 0xF4, 0x8F, 0xBF, 0xF0, 0x80, 0x80, 0xF8, |
| 0x80, 0x80, 0x80, 0xFB, 0xBF, 0xBF, 0xBF, 0xFC, 0x80, 0x80, |
| 0x80, 0x80, 0xFD, 0xBF, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, |
| 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Incomplete sequence tests repeated with a space after the incomplete |
| // sequence. |
| |
| // 2-byte sequences with last byte missing |
| {{0xC0, 0x20}, {0xFFFD, 0x20}}, |
| {{0xDF, 0x20}, {0xFFFD, 0x20}}, |
| // 3-byte sequences with last byte missing |
| {{0xE8, 0x80, 0x20}, {0xFFFD, 0x20}}, |
| {{0xE0, 0xBF, 0x20}, {0xFFFD, 0x20}}, |
| {{0xEF, 0xBF, 0x20}, {0xFFFD, 0x20}}, |
| // Start of overlong 3-byte sequence with last byte missing |
| {{0xE0, 0x80, 0x20}, {0xFFFD, 0xFFFD, 0x20}}, |
| // 4-byte sequences with last byte missing |
| {{0xF1, 0x80, 0x80, 0x20}, {0xFFFD, 0x20}}, |
| {{0xF4, 0x8F, 0xBF, 0x20}, {0xFFFD, 0x20}}, |
| // Start of overlong 4-byte sequence with last byte missing |
| {{0xF0, 0x80, 0x80, 0x20}, {0xFFFD, 0xFFFD, 0xFFFD, 0x20}}, |
| // 5-byte sequences (not supported) with last byte missing |
| {{0xF8, 0x80, 0x80, 0x80, 0x20}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x20}}, |
| {{0xFB, 0xBF, 0xBF, 0xBF, 0x20}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x20}}, |
| // 6-byte sequences (not supported) with last byte missing |
| {{0xFC, 0x80, 0x80, 0x80, 0x80, 0x20}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x20}}, |
| {{0xFD, 0xBF, 0xBF, 0xBF, 0xBF, 0x20}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0x20}}, |
| |
| // Impossible bytes |
| {{0xFE}, {0xFFFD}}, |
| {{0xFF}, {0xFFFD}}, |
| {{0xFE, 0xFE, 0xFF, 0xFF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // Lead-byte-like bytes which aren't valid lead bytes. |
| {{0xC0}, {0xFFFD}}, |
| {{0xC0, 0xAA}, {0xFFFD, 0xFFFD}}, |
| {{0xC1}, {0xFFFD}}, |
| {{0xC1, 0xAA}, {0xFFFD, 0xFFFD}}, |
| {{0xF5}, {0xFFFD}}, |
| {{0xF5, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF6}, {0xFFFD}}, |
| {{0xF6, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF7}, {0xFFFD}}, |
| {{0xF7, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF8}, {0xFFFD}}, |
| {{0xF8, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF9}, {0xFFFD}}, |
| {{0xF9, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFA}, {0xFFFD}}, |
| {{0xFA, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFB}, {0xFFFD}}, |
| {{0xFB, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFC}, {0xFFFD}}, |
| {{0xFC, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFD}, {0xFFFD}}, |
| {{0xFD, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFE}, {0xFFFD}}, |
| {{0xFE, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xFF}, {0xFFFD}}, |
| {{0xFF, 0xAA, 0xAA, 0xAA}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Overlong sequences: |
| |
| // Overlong encodings for "/" |
| {{0xC0, 0xAF}, {0xFFFD, 0xFFFD}}, |
| {{0xE0, 0x80, 0xAF}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF0, 0x80, 0x80, 0xAF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5-byte sequence (not supported anyway) |
| {{0xF8, 0x80, 0x80, 0x80, 0xAF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6-byte sequence (not supported anyway) |
| {{0xFC, 0x80, 0x80, 0x80, 0x80, 0xAF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Maximum overlong sequences |
| {{0xC1, 0xBF}, {0xFFFD, 0xFFFD}}, |
| {{0xE0, 0x9F, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF0, 0x8F, 0xBF, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5-byte sequence (not supported anyway) |
| {{0xF8, 0x87, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6-byte sequence (not supported anyway) |
| {{0xFC, 0x83, 0xBF, 0xBF, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Overlong encodings for 0 |
| {{0xC0, 0x80}, {0xFFFD, 0xFFFD}}, |
| {{0xE0, 0x80, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xF0, 0x80, 0x80, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 5-byte sequence (not supported anyway) |
| {{0xF8, 0x80, 0x80, 0x80, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| // 6-byte sequence (not supported anyway) |
| {{0xFC, 0x80, 0x80, 0x80, 0x80, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Illegal code positions: |
| |
| // Single UTF-16 surrogates |
| {{0xED, 0xA0, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xA0, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAD, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAE, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAF, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xB0, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xBE, 0x80}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xBF, 0xBF}, {0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Paired surrogates |
| {{0xED, 0xA0, 0x80, 0xED, 0xB0, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xA0, 0x80, 0xED, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAD, 0xBF, 0xED, 0xB0, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAD, 0xBF, 0xED, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAE, 0x80, 0xED, 0xB0, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAE, 0x80, 0xED, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAF, 0xBF, 0xED, 0xB0, 0x80}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAF, 0xBF, 0xED, 0xBF, 0xBF}, |
| {0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD, 0xFFFD}}, |
| |
| // Surrogates with the last byte missing. |
| {{0xED, 0xA0}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xA0}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAD}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAE}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xAF}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xB0}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xBE}, {0xFFFD, 0xFFFD}}, |
| {{0xED, 0xBF}, {0xFFFD, 0xFFFD}}, |
| |
| // Other non-characters |
| {{0xEF, 0xBF, 0xBE}, {0xFFFE}}, |
| {{0xEF, 0xBF, 0xBF}, {0xFFFF}}, |
| {{0xEF, 0xB7, 0x90, 0xEF, 0xB7, 0x91, 0xEF, 0xB7, 0x92, 0xEF, 0xB7, 0x93, |
| 0xEF, 0xB7, 0x94, 0xEF, 0xB7, 0x95, 0xEF, 0xB7, 0x96, 0xEF, 0xB7, 0x97, |
| 0xEF, 0xB7, 0x98, 0xEF, 0xB7, 0x99, 0xEF, 0xB7, 0x9A, 0xEF, 0xB7, 0x9B, |
| 0xEF, 0xB7, 0x9C, 0xEF, 0xB7, 0x9D, 0xEF, 0xB7, 0x9E, 0xEF, 0xB7, 0x9F, |
| 0xEF, 0xB7, 0xA0, 0xEF, 0xB7, 0xA1, 0xEF, 0xB7, 0xA2, 0xEF, 0xB7, 0xA3, |
| 0xEF, 0xB7, 0xA4, 0xEF, 0xB7, 0xA5, 0xEF, 0xB7, 0xA6, 0xEF, 0xB7, 0xA7, |
| 0xEF, 0xB7, 0xA8, 0xEF, 0xB7, 0xA9, 0xEF, 0xB7, 0xAA, 0xEF, 0xB7, 0xAB, |
| 0xEF, 0xB7, 0xAC, 0xEF, 0xB7, 0xAD, 0xEF, 0xB7, 0xAE, 0xEF, 0xB7, 0xAF}, |
| {0xFDD0, 0xFDD1, 0xFDD2, 0xFDD3, 0xFDD4, 0xFDD5, 0xFDD6, 0xFDD7, |
| 0xFDD8, 0xFDD9, 0xFDDA, 0xFDDB, 0xFDDC, 0xFDDD, 0xFDDE, 0xFDDF, |
| 0xFDE0, 0xFDE1, 0xFDE2, 0xFDE3, 0xFDE4, 0xFDE5, 0xFDE6, 0xFDE7, |
| 0xFDE8, 0xFDE9, 0xFDEA, 0xFDEB, 0xFDEC, 0xFDED, 0xFDEE, 0xFDEF}}, |
| {{0xF0, 0x9F, 0xBF, 0xBE, 0xF0, 0x9F, 0xBF, 0xBF, 0xF0, 0xAF, 0xBF, |
| 0xBE, 0xF0, 0xAF, 0xBF, 0xBF, 0xF0, 0xBF, 0xBF, 0xBE, 0xF0, 0xBF, |
| 0xBF, 0xBF, 0xF1, 0x8F, 0xBF, 0xBE, 0xF1, 0x8F, 0xBF, 0xBF, 0xF1, |
| 0x9F, 0xBF, 0xBE, 0xF1, 0x9F, 0xBF, 0xBF, 0xF1, 0xAF, 0xBF, 0xBE, |
| 0xF1, 0xAF, 0xBF, 0xBF, 0xF1, 0xBF, 0xBF, 0xBE, 0xF1, 0xBF, 0xBF, |
| 0xBF, 0xF2, 0x8F, 0xBF, 0xBE, 0xF2, 0x8F, 0xBF, 0xBF}, |
| {0x1FFFE, 0x1FFFF, 0x2FFFE, 0x2FFFF, 0x3FFFE, 0x3FFFF, 0x4FFFE, 0x4FFFF, |
| 0x5FFFE, 0x5FFFF, 0x6FFFE, 0x6FFFF, 0x7FFFE, 0x7FFFF, 0x8FFFE, |
| 0x8FFFF}}, |
| }; |
| |
| for (auto test : data) { |
| // For figuring out which test fails: |
| fprintf(stderr, "test: "); |
| for (auto b : test.bytes) { |
| fprintf(stderr, "%x ", b); |
| } |
| fprintf(stderr, "\n"); |
| |
| std::vector<unibrow::uchar> output_normal; |
| DecodeNormally(test.bytes, &output_normal); |
| |
| CHECK_EQ(output_normal.size(), test.unicode_expected.size()); |
| for (size_t i = 0; i < output_normal.size(); ++i) { |
| CHECK_EQ(output_normal[i], test.unicode_expected[i]); |
| } |
| |
| std::vector<unibrow::uchar> output_incremental; |
| DecodeIncrementally(test.bytes, &output_incremental); |
| |
| CHECK_EQ(output_incremental.size(), test.unicode_expected.size()); |
| for (size_t i = 0; i < output_incremental.size(); ++i) { |
| CHECK_EQ(output_incremental[i], test.unicode_expected[i]); |
| } |
| } |
| } |
| |
| } // namespace internal |
| } // namespace v8 |