| 1 | // |
| 2 | // TextBufferIteratorTest.cpp |
| 3 | // |
| 4 | // Copyright (c) 2010, Applied Informatics Software Engineering GmbH. |
| 5 | // and Contributors. |
| 6 | // |
| 7 | // SPDX-License-Identifier: BSL-1.0 |
| 8 | // |
| 9 | |
| 10 | |
| 11 | #include "TextBufferIteratorTest.h" |
| 12 | #include "Poco/CppUnit/TestCaller.h" |
| 13 | #include "Poco/CppUnit/TestSuite.h" |
| 14 | #include "Poco/TextBufferIterator.h" |
| 15 | #include "Poco/Latin1Encoding.h" |
| 16 | #include "Poco/UTF8Encoding.h" |
| 17 | #include "Poco/UTF16Encoding.h" |
| 18 | |
| 19 | |
| 20 | using Poco::TextBufferIterator; |
| 21 | using Poco::Latin1Encoding; |
| 22 | using Poco::UTF8Encoding; |
| 23 | using Poco::UTF16Encoding; |
| 24 | |
| 25 | |
| 26 | TextBufferIteratorTest::TextBufferIteratorTest(const std::string& rName): CppUnit::TestCase(rName) |
| 27 | { |
| 28 | } |
| 29 | |
| 30 | |
| 31 | TextBufferIteratorTest::~TextBufferIteratorTest() |
| 32 | { |
| 33 | } |
| 34 | |
| 35 | |
| 36 | void TextBufferIteratorTest::testEmptyLatin1() |
| 37 | { |
| 38 | Latin1Encoding encoding; |
| 39 | const char* text = "" ; |
| 40 | TextBufferIterator it(text, encoding); |
| 41 | TextBufferIterator end(it.end()); |
| 42 | |
| 43 | assertTrue (it == end); |
| 44 | } |
| 45 | |
| 46 | |
| 47 | void TextBufferIteratorTest::testOneLatin1() |
| 48 | { |
| 49 | Latin1Encoding encoding; |
| 50 | const char* text = "x" ; |
| 51 | TextBufferIterator it(text, encoding); |
| 52 | TextBufferIterator end(it.end()); |
| 53 | |
| 54 | assertTrue (it != end); |
| 55 | assertTrue (*it == 'x'); |
| 56 | ++it; |
| 57 | assertTrue (it == end); |
| 58 | } |
| 59 | |
| 60 | |
| 61 | void TextBufferIteratorTest::testLatin1() |
| 62 | { |
| 63 | Latin1Encoding encoding; |
| 64 | const char* text = "Latin1" ; |
| 65 | TextBufferIterator it(text, encoding); |
| 66 | TextBufferIterator end(it.end()); |
| 67 | |
| 68 | assertTrue (it != end); |
| 69 | assertTrue (*it++ == 'L'); |
| 70 | assertTrue (it != end); |
| 71 | assertTrue (*it++ == 'a'); |
| 72 | assertTrue (it != end); |
| 73 | assertTrue (*it++ == 't'); |
| 74 | assertTrue (it != end); |
| 75 | assertTrue (*it++ == 'i'); |
| 76 | assertTrue (it != end); |
| 77 | assertTrue (*it++ == 'n'); |
| 78 | assertTrue (it != end); |
| 79 | assertTrue (*it++ == '1'); |
| 80 | assertTrue (it == end); |
| 81 | |
| 82 | const char* empty = "" ; |
| 83 | it = TextBufferIterator(empty, encoding); |
| 84 | end = it.end(); |
| 85 | assertTrue (it == end); |
| 86 | } |
| 87 | |
| 88 | |
| 89 | void TextBufferIteratorTest::testEmptyUTF8() |
| 90 | { |
| 91 | UTF8Encoding encoding; |
| 92 | const char* text = "" ; |
| 93 | TextBufferIterator it(text, encoding); |
| 94 | TextBufferIterator end(text); |
| 95 | |
| 96 | assertTrue (it == end); |
| 97 | } |
| 98 | |
| 99 | |
| 100 | void TextBufferIteratorTest::testOneUTF8() |
| 101 | { |
| 102 | UTF8Encoding encoding; |
| 103 | |
| 104 | // 1 byte sequence |
| 105 | const char* text = "x" ; |
| 106 | TextBufferIterator it(text, encoding); |
| 107 | TextBufferIterator end(it.end()); |
| 108 | |
| 109 | assertTrue (it != end); |
| 110 | assertTrue (*it == 'x'); |
| 111 | ++it; |
| 112 | assertTrue (it == end); |
| 113 | |
| 114 | unsigned char data[Poco::TextEncoding::MAX_SEQUENCE_LENGTH]; |
| 115 | |
| 116 | // 2 byte sequence |
| 117 | int n = encoding.convert(0xab, data, sizeof(data)); |
| 118 | assertTrue (n == 2); |
| 119 | it = TextBufferIterator(reinterpret_cast<const char*>(data), n, encoding); |
| 120 | end = it.end(); |
| 121 | |
| 122 | assertTrue (it != end); |
| 123 | assertTrue (*it++ == 0xab); |
| 124 | assertTrue (it == end); |
| 125 | |
| 126 | // 3 byte sequence |
| 127 | n = encoding.convert(0xabcd, data, sizeof(data)); |
| 128 | assertTrue (n == 3); |
| 129 | it = TextBufferIterator(reinterpret_cast<const char*>(data), n, encoding); |
| 130 | end = it.end(); |
| 131 | |
| 132 | assertTrue (it != end); |
| 133 | assertTrue (*it++ == 0xabcd); |
| 134 | assertTrue (it == end); |
| 135 | |
| 136 | // 4 byte sequence |
| 137 | n = encoding.convert(0xabcde, data, sizeof(data)); |
| 138 | assertTrue (n == 4); |
| 139 | it = TextBufferIterator(reinterpret_cast<const char*>(data), n, encoding); |
| 140 | end = it.end(); |
| 141 | |
| 142 | assertTrue (it != end); |
| 143 | assertTrue (*it++ == 0xabcde); |
| 144 | assertTrue (it == end); |
| 145 | |
| 146 | // 5 byte sequence - not supported |
| 147 | n = encoding.convert(0xabcdef, data, sizeof(data)); |
| 148 | assertTrue (n == 0); |
| 149 | |
| 150 | // 6 byte sequence - not supported |
| 151 | n = encoding.convert(0xfabcdef, data, sizeof(data)); |
| 152 | assertTrue (n == 0); |
| 153 | } |
| 154 | |
| 155 | |
| 156 | void TextBufferIteratorTest::testUTF8() |
| 157 | { |
| 158 | UTF8Encoding encoding; |
| 159 | const unsigned char text[] = {0x20, 0xce, 0xba, 0xe1, 0xbd, 0xb9, 0xcf, 0x83, 0xce, 0xbc, 0xce, 0xb5, 0x20, 0x00}; |
| 160 | TextBufferIterator it(reinterpret_cast<const char*>(text), encoding); |
| 161 | TextBufferIterator end(it.end()); |
| 162 | |
| 163 | assertTrue (it != end); |
| 164 | assertTrue (*it++ == 0x0020); |
| 165 | assertTrue (it != end); |
| 166 | assertTrue (*it++ == 0x03ba); |
| 167 | assertTrue (it != end); |
| 168 | assertTrue (*it++ == 0x1f79); |
| 169 | assertTrue (it != end); |
| 170 | assertTrue (*it++ == 0x03c3); |
| 171 | assertTrue (it != end); |
| 172 | assertTrue (*it++ == 0x03bc); |
| 173 | assertTrue (it != end); |
| 174 | assertTrue (*it++ == 0x03b5); |
| 175 | assertTrue (it != end); |
| 176 | assertTrue (*it++ == 0x0020); |
| 177 | assertTrue (it == end); |
| 178 | } |
| 179 | |
| 180 | |
| 181 | void TextBufferIteratorTest::testUTF8Supplementary() |
| 182 | { |
| 183 | UTF8Encoding encoding; |
| 184 | const unsigned char text[] = {0x41, 0x42, 0xf0, 0x90, 0x82, 0xa4, 0xf0, 0xaf, 0xa6, 0xa0, 0xf0, 0xaf, 0xa8, 0x9d, 0x00}; |
| 185 | TextBufferIterator it(reinterpret_cast<const char*>(text), encoding); |
| 186 | TextBufferIterator end(it.end()); |
| 187 | |
| 188 | assertTrue (it != end); |
| 189 | assertTrue (*it++ == 0x0041); |
| 190 | assertTrue (it != end); |
| 191 | assertTrue (*it++ == 0x0042); |
| 192 | assertTrue (it != end); |
| 193 | assertTrue (*it++ == 0x100a4); |
| 194 | assertTrue (it != end); |
| 195 | assertTrue (*it++ == 0x2f9a0); |
| 196 | assertTrue (it != end); |
| 197 | assertTrue (*it++ == 0x2fa1d); |
| 198 | assertTrue (it == end); |
| 199 | } |
| 200 | |
| 201 | |
| 202 | void TextBufferIteratorTest::testUTF16Supplementary() |
| 203 | { |
| 204 | UTF16Encoding encoding; |
| 205 | const Poco::UInt16 text[] = { 0x0041, 0x0042, 0xD800, 0xDCA4, 0xD87E, 0xDDA0, 0xD87E, 0xDE1D}; |
| 206 | TextBufferIterator it(reinterpret_cast<const char*>(text), sizeof(text), encoding); |
| 207 | TextBufferIterator end(it.end()); |
| 208 | |
| 209 | assertTrue (it != end); |
| 210 | assertTrue (*it++ == 0x0041); |
| 211 | assertTrue (it != end); |
| 212 | assertTrue (*it++ == 0x0042); |
| 213 | assertTrue (it != end); |
| 214 | assertTrue (*it++ == 0x100a4); |
| 215 | assertTrue (it != end); |
| 216 | assertTrue (*it++ == 0x2f9a0); |
| 217 | assertTrue (it != end); |
| 218 | assertTrue (*it++ == 0x2fa1d); |
| 219 | assertTrue (it == end); |
| 220 | } |
| 221 | |
| 222 | |
| 223 | void TextBufferIteratorTest::testSwap() |
| 224 | { |
| 225 | Latin1Encoding encoding; |
| 226 | const char* text = "x" ; |
| 227 | TextBufferIterator it1(text, encoding); |
| 228 | TextBufferIterator it2(text, encoding); |
| 229 | TextBufferIterator end(it1.end()); |
| 230 | |
| 231 | assertTrue (it1 == it2); |
| 232 | it2.swap(end); |
| 233 | assertTrue (it1 != it2); |
| 234 | it2.swap(end); |
| 235 | assertTrue (it1 == it2); |
| 236 | } |
| 237 | |
| 238 | |
| 239 | void TextBufferIteratorTest::setUp() |
| 240 | { |
| 241 | } |
| 242 | |
| 243 | |
| 244 | void TextBufferIteratorTest::tearDown() |
| 245 | { |
| 246 | } |
| 247 | |
| 248 | |
| 249 | CppUnit::Test* TextBufferIteratorTest::suite() |
| 250 | { |
| 251 | CppUnit::TestSuite* pSuite = new CppUnit::TestSuite("TextBufferIteratorTest" ); |
| 252 | |
| 253 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testEmptyLatin1); |
| 254 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testOneLatin1); |
| 255 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testLatin1); |
| 256 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testEmptyUTF8); |
| 257 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testOneUTF8); |
| 258 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testUTF8); |
| 259 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testUTF8Supplementary); |
| 260 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testUTF16Supplementary); |
| 261 | CppUnit_addTest(pSuite, TextBufferIteratorTest, testSwap); |
| 262 | |
| 263 | return pSuite; |
| 264 | } |
| 265 | |