|
| 1 | +#include "../src/unicode.hpp" |
| 2 | +#include "../submodules/iutest/include/iutest.hpp" |
| 3 | +#include <sstream> |
| 4 | +#include <string_view> |
| 5 | + |
| 6 | +IUTEST(unicode, test_read_utf8_stream) { |
| 7 | + std::stringstream ss(u8"\1\x7F\u0080\uFFFF\U00010000\U0010FFFF"); |
| 8 | + IUTEST_ASSERT_EQ(U'\1', fafnir::read_utf8_stream(ss)); |
| 9 | + IUTEST_ASSERT_EQ(U'\x7F', fafnir::read_utf8_stream(ss)); |
| 10 | + IUTEST_ASSERT_EQ(U'\u0080', fafnir::read_utf8_stream(ss)); |
| 11 | + IUTEST_ASSERT_EQ(U'\uFFFF', fafnir::read_utf8_stream(ss)); |
| 12 | + IUTEST_ASSERT_EQ(U'\U00010000', fafnir::read_utf8_stream(ss)); |
| 13 | + IUTEST_ASSERT_EQ(U'\U0010FFFF', fafnir::read_utf8_stream(ss)); |
| 14 | +} |
| 15 | + |
| 16 | +IUTEST(unicode, test_read_bad_utf8_stream) { |
| 17 | + std::stringstream ss("\x80\xF0\x80\x80\x80\xED\x9F\xBF\xED\xA0\x80\xED\xBF\xBF\xEE\x80\x80"); |
| 18 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf8_stream(ss)); |
| 19 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf8_stream(ss)); |
| 20 | + IUTEST_ASSERT_EQ(U'\uD7FF', fafnir::read_utf8_stream(ss)); |
| 21 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf8_stream(ss)); |
| 22 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf8_stream(ss)); |
| 23 | + IUTEST_ASSERT_EQ(U'\uE000', fafnir::read_utf8_stream(ss)); |
| 24 | +} |
| 25 | + |
| 26 | +IUTEST(unicode, test_read_utf16_stream) { |
| 27 | + std::u16string_view str = u"\1\uFFFF\U00010000\U0010FFFF"; |
| 28 | + std::stringstream ss({ |
| 29 | + reinterpret_cast<const char*>(str.data()), |
| 30 | + reinterpret_cast<const char*>(str.data() + str.size()) |
| 31 | + }); |
| 32 | + IUTEST_ASSERT_EQ(U'\1', fafnir::read_utf16_stream(ss)); |
| 33 | + IUTEST_ASSERT_EQ(U'\uFFFF', fafnir::read_utf16_stream(ss)); |
| 34 | + IUTEST_ASSERT_EQ(U'\U00010000', fafnir::read_utf16_stream(ss)); |
| 35 | + IUTEST_ASSERT_EQ(U'\U0010FFFF', fafnir::read_utf16_stream(ss)); |
| 36 | +} |
| 37 | + |
| 38 | +IUTEST(unicode, test_read_bad_stream) { |
| 39 | + char str[] = "\xFF\xD7\x00\xD8\x01\x00\xFF\xDF\x00\xE0"; |
| 40 | + std::stringstream ss({str, str + sizeof(str) - 1}); |
| 41 | + IUTEST_ASSERT_EQ(U'\uD7FF', fafnir::read_utf16_stream(ss)); |
| 42 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf16_stream(ss)); |
| 43 | + IUTEST_ASSERT_EQ(U'\uFFFD', fafnir::read_utf16_stream(ss)); |
| 44 | + IUTEST_ASSERT_EQ(U'\uE000', fafnir::read_utf16_stream(ss)); |
| 45 | +} |
| 46 | + |
| 47 | +IUTEST(unicode, test_write_utf8_stream) { |
| 48 | + std::stringstream ss; |
| 49 | + fafnir::write_utf8_stream(ss, U'\1'); |
| 50 | + fafnir::write_utf8_stream(ss, U'\x7F'); |
| 51 | + fafnir::write_utf8_stream(ss, U'\u0080'); |
| 52 | + fafnir::write_utf8_stream(ss, U'\uFFFF'); |
| 53 | + fafnir::write_utf8_stream(ss, U'\U00010000'); |
| 54 | + fafnir::write_utf8_stream(ss, U'\U0010FFFF'); |
| 55 | + IUTEST_ASSERT_EQ(u8"\1\x7F\u0080\uFFFF\U00010000\U0010FFFF", ss.str()); |
| 56 | +} |
| 57 | + |
| 58 | +IUTEST(unicode, test_write_bad_utf8_stream) { |
| 59 | + std::stringstream ss; |
| 60 | + fafnir::write_utf8_stream(ss, U'\uD7FF'); |
| 61 | + fafnir::write_utf8_stream(ss, 0xD800); |
| 62 | + fafnir::write_utf8_stream(ss, 0xDFFF); |
| 63 | + fafnir::write_utf8_stream(ss, U'\uE000'); |
| 64 | + fafnir::write_utf8_stream(ss, U'\U0010FFFF'); |
| 65 | + fafnir::write_utf8_stream(ss, 0x00110000); |
| 66 | + IUTEST_ASSERT_EQ(u8"\uD7FF\uFFFD\uFFFD\uE000\U0010FFFF\uFFFD", ss.str()); |
| 67 | +} |
| 68 | + |
| 69 | +IUTEST(unicode, test_write_utf16_stream) { |
| 70 | + std::stringstream ss; |
| 71 | + fafnir::write_utf16_stream(ss, U'\1'); |
| 72 | + fafnir::write_utf16_stream(ss, U'\uFFFF'); |
| 73 | + fafnir::write_utf16_stream(ss, U'\U00010000'); |
| 74 | + fafnir::write_utf16_stream(ss, U'\U0010FFFF'); |
| 75 | + auto str = ss.str(); |
| 76 | + IUTEST_ASSERT_EQ( |
| 77 | + u"\1\uFFFF\U00010000\U0010FFFF", |
| 78 | + std::u16string( |
| 79 | + reinterpret_cast<const char16_t*>(str.data()), |
| 80 | + reinterpret_cast<const char16_t*>(str.data() + str.size()) |
| 81 | + ) |
| 82 | + ); |
| 83 | +} |
| 84 | + |
| 85 | +IUTEST(unicode, test_write_bad_utf16_stream) { |
| 86 | + std::stringstream ss; |
| 87 | + fafnir::write_utf16_stream(ss, U'\uD7FF'); |
| 88 | + fafnir::write_utf16_stream(ss, 0xD800); |
| 89 | + fafnir::write_utf16_stream(ss, 0xDFFF); |
| 90 | + fafnir::write_utf16_stream(ss, U'\uE000'); |
| 91 | + fafnir::write_utf16_stream(ss, U'\U0010FFFF'); |
| 92 | + fafnir::write_utf16_stream(ss, 0x00110000); |
| 93 | + auto str = ss.str(); |
| 94 | + IUTEST_ASSERT_EQ( |
| 95 | + u"\uD7FF\uFFFD\uFFFD\uE000\U0010FFFF\uFFFD", |
| 96 | + std::u16string( |
| 97 | + reinterpret_cast<const char16_t*>(str.data()), |
| 98 | + reinterpret_cast<const char16_t*>(str.data() + str.size()) |
| 99 | + ) |
| 100 | + ); |
| 101 | +} |
| 102 | + |
| 103 | +int main(int argc, char** argv) { |
| 104 | + IUTEST_INIT(&argc, argv); |
| 105 | + return IUTEST_RUN_ALL_TESTS(); |
| 106 | +} |
0 commit comments