OLD | NEW |
(Empty) | |
| 1 // Copyright (c) 2012, the Dart project authors. Please see the AUTHORS file |
| 2 // for details. All rights reserved. Use of this source code is governed by a |
| 3 // BSD-style license that can be found in the LICENSE file. |
| 4 |
| 5 library utf.utf32_test; |
| 6 |
| 7 import 'package:expect/expect.dart'; |
| 8 import 'package:utf/utf.dart'; |
| 9 |
| 10 const String testKoreanCharSubset = """ |
| 11 가각갂갃간갅갆갇갈갉갊갋갌갍갎갏감갑값갓갔강갖갗갘같갚갛 |
| 12 개객갞갟갠갡갢갣갤갥갦갧갨갩갪갫갬갭갮갯갰갱갲갳갴갵갶갷 |
| 13 갸갹갺갻갼갽갾갿걀걁걂걃걄걅걆걇걈걉걊걋걌걍걎걏걐걑걒걓"""; |
| 14 |
| 15 const String testHanTwice = "二"; |
| 16 |
| 17 const List<int> testKoreanCharSubsetUtf32beBom = const<int>[ |
| 18 0x00, 0x00, 0xfe, 0xff, 0x00, 0x00, 0xac, 0x00, |
| 19 0x00, 0x00, 0xac, 0x01, 0x00, 0x00, 0xac, 0x02, |
| 20 0x00, 0x00, 0xac, 0x03, 0x00, 0x00, 0xac, 0x04, |
| 21 0x00, 0x00, 0xac, 0x05, 0x00, 0x00, 0xac, 0x06, |
| 22 0x00, 0x00, 0xac, 0x07, 0x00, 0x00, 0xac, 0x08, |
| 23 0x00, 0x00, 0xac, 0x09, 0x00, 0x00, 0xac, 0x0a, |
| 24 0x00, 0x00, 0xac, 0x0b, 0x00, 0x00, 0xac, 0x0c, |
| 25 0x00, 0x00, 0xac, 0x0d, 0x00, 0x00, 0xac, 0x0e, |
| 26 0x00, 0x00, 0xac, 0x0f, 0x00, 0x00, 0xac, 0x10, |
| 27 0x00, 0x00, 0xac, 0x11, 0x00, 0x00, 0xac, 0x12, |
| 28 0x00, 0x00, 0xac, 0x13, 0x00, 0x00, 0xac, 0x14, |
| 29 0x00, 0x00, 0xac, 0x15, 0x00, 0x00, 0xac, 0x16, |
| 30 0x00, 0x00, 0xac, 0x17, 0x00, 0x00, 0xac, 0x18, |
| 31 0x00, 0x00, 0xac, 0x19, 0x00, 0x00, 0xac, 0x1a, |
| 32 0x00, 0x00, 0xac, 0x1b, 0x00, 0x00, 0x00, 0x0a, |
| 33 0x00, 0x00, 0xac, 0x1c, 0x00, 0x00, 0xac, 0x1d, |
| 34 0x00, 0x00, 0xac, 0x1e, 0x00, 0x00, 0xac, 0x1f, |
| 35 0x00, 0x00, 0xac, 0x20, 0x00, 0x00, 0xac, 0x21, |
| 36 0x00, 0x00, 0xac, 0x22, 0x00, 0x00, 0xac, 0x23, |
| 37 0x00, 0x00, 0xac, 0x24, 0x00, 0x00, 0xac, 0x25, |
| 38 0x00, 0x00, 0xac, 0x26, 0x00, 0x00, 0xac, 0x27, |
| 39 0x00, 0x00, 0xac, 0x28, 0x00, 0x00, 0xac, 0x29, |
| 40 0x00, 0x00, 0xac, 0x2a, 0x00, 0x00, 0xac, 0x2b, |
| 41 0x00, 0x00, 0xac, 0x2c, 0x00, 0x00, 0xac, 0x2d, |
| 42 0x00, 0x00, 0xac, 0x2e, 0x00, 0x00, 0xac, 0x2f, |
| 43 0x00, 0x00, 0xac, 0x30, 0x00, 0x00, 0xac, 0x31, |
| 44 0x00, 0x00, 0xac, 0x32, 0x00, 0x00, 0xac, 0x33, |
| 45 0x00, 0x00, 0xac, 0x34, 0x00, 0x00, 0xac, 0x35, |
| 46 0x00, 0x00, 0xac, 0x36, 0x00, 0x00, 0xac, 0x37, |
| 47 0x00, 0x00, 0x00, 0x0a, 0x00, 0x00, 0xac, 0x38, |
| 48 0x00, 0x00, 0xac, 0x39, 0x00, 0x00, 0xac, 0x3a, |
| 49 0x00, 0x00, 0xac, 0x3b, 0x00, 0x00, 0xac, 0x3c, |
| 50 0x00, 0x00, 0xac, 0x3d, 0x00, 0x00, 0xac, 0x3e, |
| 51 0x00, 0x00, 0xac, 0x3f, 0x00, 0x00, 0xac, 0x40, |
| 52 0x00, 0x00, 0xac, 0x41, 0x00, 0x00, 0xac, 0x42, |
| 53 0x00, 0x00, 0xac, 0x43, 0x00, 0x00, 0xac, 0x44, |
| 54 0x00, 0x00, 0xac, 0x45, 0x00, 0x00, 0xac, 0x46, |
| 55 0x00, 0x00, 0xac, 0x47, 0x00, 0x00, 0xac, 0x48, |
| 56 0x00, 0x00, 0xac, 0x49, 0x00, 0x00, 0xac, 0x4a, |
| 57 0x00, 0x00, 0xac, 0x4b, 0x00, 0x00, 0xac, 0x4c, |
| 58 0x00, 0x00, 0xac, 0x4d, 0x00, 0x00, 0xac, 0x4e, |
| 59 0x00, 0x00, 0xac, 0x4f, 0x00, 0x00, 0xac, 0x50, |
| 60 0x00, 0x00, 0xac, 0x51, 0x00, 0x00, 0xac, 0x52, |
| 61 0x00, 0x00, 0xac, 0x53]; |
| 62 |
| 63 const List<int> testKoreanCharSubsetUtf32le = const<int>[ |
| 64 0x00, 0xac, 0x00, 0x00, 0x01, 0xac, 0x00, 0x00, |
| 65 0x02, 0xac, 0x00, 0x00, 0x03, 0xac, 0x00, 0x00, |
| 66 0x04, 0xac, 0x00, 0x00, 0x05, 0xac, 0x00, 0x00, |
| 67 0x06, 0xac, 0x00, 0x00, 0x07, 0xac, 0x00, 0x00, |
| 68 0x08, 0xac, 0x00, 0x00, 0x09, 0xac, 0x00, 0x00, |
| 69 0x0a, 0xac, 0x00, 0x00, 0x0b, 0xac, 0x00, 0x00, |
| 70 0x0c, 0xac, 0x00, 0x00, 0x0d, 0xac, 0x00, 0x00, |
| 71 0x0e, 0xac, 0x00, 0x00, 0x0f, 0xac, 0x00, 0x00, |
| 72 0x10, 0xac, 0x00, 0x00, 0x11, 0xac, 0x00, 0x00, |
| 73 0x12, 0xac, 0x00, 0x00, 0x13, 0xac, 0x00, 0x00, |
| 74 0x14, 0xac, 0x00, 0x00, 0x15, 0xac, 0x00, 0x00, |
| 75 0x16, 0xac, 0x00, 0x00, 0x17, 0xac, 0x00, 0x00, |
| 76 0x18, 0xac, 0x00, 0x00, 0x19, 0xac, 0x00, 0x00, |
| 77 0x1a, 0xac, 0x00, 0x00, 0x1b, 0xac, 0x00, 0x00, |
| 78 0x0a, 0x00, 0x00, 0x00, 0x1c, 0xac, 0x00, 0x00, |
| 79 0x1d, 0xac, 0x00, 0x00, 0x1e, 0xac, 0x00, 0x00, |
| 80 0x1f, 0xac, 0x00, 0x00, 0x20, 0xac, 0x00, 0x00, |
| 81 0x21, 0xac, 0x00, 0x00, 0x22, 0xac, 0x00, 0x00, |
| 82 0x23, 0xac, 0x00, 0x00, 0x24, 0xac, 0x00, 0x00, |
| 83 0x25, 0xac, 0x00, 0x00, 0x26, 0xac, 0x00, 0x00, |
| 84 0x27, 0xac, 0x00, 0x00, 0x28, 0xac, 0x00, 0x00, |
| 85 0x29, 0xac, 0x00, 0x00, 0x2a, 0xac, 0x00, 0x00, |
| 86 0x2b, 0xac, 0x00, 0x00, 0x2c, 0xac, 0x00, 0x00, |
| 87 0x2d, 0xac, 0x00, 0x00, 0x2e, 0xac, 0x00, 0x00, |
| 88 0x2f, 0xac, 0x00, 0x00, 0x30, 0xac, 0x00, 0x00, |
| 89 0x31, 0xac, 0x00, 0x00, 0x32, 0xac, 0x00, 0x00, |
| 90 0x33, 0xac, 0x00, 0x00, 0x34, 0xac, 0x00, 0x00, |
| 91 0x35, 0xac, 0x00, 0x00, 0x36, 0xac, 0x00, 0x00, |
| 92 0x37, 0xac, 0x00, 0x00, 0x0a, 0x00, 0x00, 0x00, |
| 93 0x38, 0xac, 0x00, 0x00, 0x39, 0xac, 0x00, 0x00, |
| 94 0x3a, 0xac, 0x00, 0x00, 0x3b, 0xac, 0x00, 0x00, |
| 95 0x3c, 0xac, 0x00, 0x00, 0x3d, 0xac, 0x00, 0x00, |
| 96 0x3e, 0xac, 0x00, 0x00, 0x3f, 0xac, 0x00, 0x00, |
| 97 0x40, 0xac, 0x00, 0x00, 0x41, 0xac, 0x00, 0x00, |
| 98 0x42, 0xac, 0x00, 0x00, 0x43, 0xac, 0x00, 0x00, |
| 99 0x44, 0xac, 0x00, 0x00, 0x45, 0xac, 0x00, 0x00, |
| 100 0x46, 0xac, 0x00, 0x00, 0x47, 0xac, 0x00, 0x00, |
| 101 0x48, 0xac, 0x00, 0x00, 0x49, 0xac, 0x00, 0x00, |
| 102 0x4a, 0xac, 0x00, 0x00, 0x4b, 0xac, 0x00, 0x00, |
| 103 0x4c, 0xac, 0x00, 0x00, 0x4d, 0xac, 0x00, 0x00, |
| 104 0x4e, 0xac, 0x00, 0x00, 0x4f, 0xac, 0x00, 0x00, |
| 105 0x50, 0xac, 0x00, 0x00, 0x51, 0xac, 0x00, 0x00, |
| 106 0x52, 0xac, 0x00, 0x00, 0x53, 0xac, 0x00, 0x00]; |
| 107 |
| 108 void main() { |
| 109 testUtf32BytesToString(); |
| 110 testEncodeToUtf32(); |
| 111 testIterableMethods(); |
| 112 } |
| 113 |
| 114 void testEncodeToUtf32() { |
| 115 Expect.listEquals([], encodeUtf32le(""), "no input"); // TODO(dcarlson) skip b
om on empty? |
| 116 Expect.listEquals(testKoreanCharSubsetUtf32beBom, |
| 117 encodeUtf32(testKoreanCharSubset), |
| 118 "encode UTF-32(BE by default) Korean"); |
| 119 Expect.listEquals(testKoreanCharSubsetUtf32le, |
| 120 encodeUtf32le(testKoreanCharSubset), |
| 121 "encode UTF-32(LE by default) Korean"); |
| 122 } |
| 123 |
| 124 void testUtf32BytesToString() { |
| 125 Expect.stringEquals("", decodeUtf32([]), "no input"); |
| 126 Expect.stringEquals("\ufffd", decodeUtf32([0]), "single byte"); |
| 127 Expect.stringEquals("\ufffd", decodeUtf32([0, 0, 0x4e]), |
| 128 "short a byte"); |
| 129 Expect.stringEquals("\u4e8c\ufffd", decodeUtf32([0, 0, 0x4e, 0x8c, 0]), |
| 130 "extra byte"); |
| 131 |
| 132 Expect.stringEquals(testHanTwice, decodeUtf32([0, 0, 0x4e, 0x8c]), |
| 133 "twice variation 1"); |
| 134 Expect.stringEquals(testHanTwice, |
| 135 decodeUtf32([0, 0, 0xfe, 0xff, 0, 0, 0x4e, 0x8c]), |
| 136 "twice variation 2"); |
| 137 Expect.stringEquals(testHanTwice, |
| 138 decodeUtf32([0xff, 0xfe, 0, 0, 0x8c, 0x4e, 0, 0]), |
| 139 "twice variation 3"); |
| 140 |
| 141 Expect.stringEquals(testHanTwice, decodeUtf32be([0, 0, 0x4e, 0x8c]), |
| 142 "twice variation 4"); |
| 143 Expect.stringEquals(testHanTwice, |
| 144 decodeUtf32be([0, 0, 0xfe, 0xff, 0, 0, 0x4e, 0x8c]), |
| 145 "twice variation 5"); |
| 146 |
| 147 Expect.stringEquals(testHanTwice, decodeUtf32le([0x8c, 0x4e, 0, 0]), |
| 148 "twice variation 6"); |
| 149 Expect.stringEquals(testHanTwice, |
| 150 decodeUtf32le([0xff, 0xfe, 0, 0, 0x8c, 0x4e, 0, 0]), |
| 151 "twice variation 7"); |
| 152 |
| 153 Expect.stringEquals(testKoreanCharSubset, |
| 154 decodeUtf32(testKoreanCharSubsetUtf32beBom), |
| 155 "UTF-32BE Korean"); |
| 156 } |
| 157 |
| 158 void testIterableMethods() { |
| 159 // empty input |
| 160 Expect.isFalse(decodeUtf32AsIterable([]).iterator.moveNext()); |
| 161 |
| 162 IterableUtf32Decoder koreanDecoder = |
| 163 decodeUtf32AsIterable(testKoreanCharSubsetUtf32beBom); |
| 164 // get the first character |
| 165 Expect.equals(testKoreanCharSubset.codeUnits[0], koreanDecoder.first); |
| 166 // get the whole translation using the Iterable interface |
| 167 Expect.stringEquals(testKoreanCharSubset, |
| 168 new String.fromCharCodes(new List<int>.from(koreanDecoder))); |
| 169 |
| 170 // specify types |
| 171 Expect.equals(44032, (new List<int> |
| 172 .from(decodeUtf32beAsIterable(testKoreanCharSubsetUtf32beBom)))[0]); |
| 173 Expect.equals(44032, (new List<int> |
| 174 .from(decodeUtf32leAsIterable(testKoreanCharSubsetUtf32le)))[0]); |
| 175 bool stripBom = false; |
| 176 Expect.equals(UNICODE_BOM, (new List<int> |
| 177 .from(decodeUtf32beAsIterable(testKoreanCharSubsetUtf32beBom, |
| 178 0, null, stripBom)))[0]); |
| 179 } |
OLD | NEW |