mcs/class/I18N/CJK/Test/I18N.CJK.Test.cs

   1 //
   2 // I18N.CJK.Test.cs
   3 //
   4 // Author:
   5 //      Atsushi Enomoto  <atsushi@ximian.com>
   6 //
   7 // Copyright (C) 2005 Novell, Inc.  http://www.novell.com
   8 //
   9
  10 using System;
  11 using System.IO;
  12 using System.Text;
  13 using NUnit.Framework;
  14
  15 namespace MonoTests.I18N.CJK
  16 {
  17         [TestFixture]
  18         public class TestCJK
  19         {
  20                 void AssertEncode (string utf8file, string decfile, int codepage)
  21                 {
  22                         string decoded = null;
  23                         byte [] encoded = null;
  24                         using (StreamReader sr = new StreamReader (utf8file,
  25                                 Encoding.UTF8)) {
  26                                 decoded = sr.ReadToEnd ();
  27                         }
  28                         using (FileStream fs = File.OpenRead (decfile)) {
  29                                 encoded = new byte [fs.Length];
  30                                 fs.Read (encoded, 0, (int) fs.Length);
  31                         }
  32                         Encoding enc = Encoding.GetEncoding (codepage);
  33                         byte [] actual;
  34
  35                         // simple string case
  36                         //Assert.AreEqual (encoded.Length,
  37                         //      enc.GetByteCount (decoded),
  38                         //      "GetByteCount(string)");
  39                         actual = enc.GetBytes (decoded);
  40                         Assert.AreEqual (encoded, actual,
  41                                 "GetBytes(string)");
  42
  43                         // simple char[] case
  44                         Assert.AreEqual (encoded.Length,
  45                                 enc.GetByteCount (decoded.ToCharArray (), 0, decoded.Length),
  46                                 "GetByteCount(char[], 0, len)");
  47                         actual = enc.GetBytes (decoded.ToCharArray (), 0, decoded.Length);
  48                         Assert.AreEqual (encoded, actual,
  49                                 "GetBytes(char[], 0, len)");
  50                 }
  51
  52                 void AssertDecode (string utf8file, string decfile, int codepage)
  53                 {
  54                         string decoded = null;
  55                         byte [] encoded = null;
  56                         using (StreamReader sr = new StreamReader (utf8file,
  57                                 Encoding.UTF8)) {
  58                                 decoded = sr.ReadToEnd ();
  59                         }
  60                         using (FileStream fs = File.OpenRead (decfile)) {
  61                                 encoded = new byte [fs.Length];
  62                                 fs.Read (encoded, 0, (int) fs.Length);
  63                         }
  64                         Encoding enc = Encoding.GetEncoding (codepage);
  65                         char [] actual;
  66
  67                         Assert.AreEqual (decoded.Length,
  68                                 enc.GetCharCount (encoded, 0, encoded.Length),
  69                                 "GetCharCount(byte[], 0, len)");
  70                         actual = enc.GetChars (encoded, 0, encoded.Length);
  71                         Assert.AreEqual (decoded.ToCharArray (), actual,
  72                                 "GetChars(byte[], 0, len)");
  73                 }
  74
  75                 #region Chinese
  76
  77                 // GB2312
  78
  79                 [Test]
  80                 public void CP936_Encode ()
  81                 {
  82                         AssertEncode ("Test/texts/chinese-utf8.txt", "Test/texts/chinese-936.txt", 936);
  83                 }
  84
  85                 [Test]
  86                 public void CP936_Decode ()
  87                 {
  88                         AssertDecode ("Test/texts/chinese-utf8.txt", "Test/texts/chinese-936.txt", 936);
  89                 }
  90
  91                 // BIG5
  92
  93                 [Test]
  94                 public void CP950_Encode ()
  95                 {
  96                         AssertEncode ("Test/texts/chinese2-utf8.txt", "Test/texts/chinese2-950.txt", 950);
  97                 }
  98
  99                 [Test]
 100                 public void CP950_Decode ()
 101                 {
 102                         AssertDecode ("Test/texts/chinese2-utf8.txt", "Test/texts/chinese2-950.txt", 950);
 103                 }
 104
 105                 // GB18030
 106
 107                 [Test]
 108                 public void CP54936_Encode ()
 109                 {
 110                         AssertEncode ("Test/texts/chinese-utf8.txt", "Test/texts/chinese-54936.txt", 54936);
 111                 }
 112
 113                 [Test]
 114                 public void CP54936_Decode ()
 115                 {
 116                         AssertDecode ("Test/texts/chinese-utf8.txt", "Test/texts/chinese-54936.txt", 54936);
 117                 }
 118
 119                 #endregion
 120
 121                 #region Japanese
 122
 123                 // Shift_JIS
 124
 125                 [Test]
 126                 public void CP932_Encode ()
 127                 {
 128                         AssertEncode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-932.txt", 932);
 129                 }
 130
 131                 [Test]
 132                 public void CP932_Decode ()
 133                 {
 134                         AssertDecode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-932.txt", 932);
 135                 }
 136
 137                 // EUC-JP
 138
 139                 [Test]
 140                 public void CP51932_Encode ()
 141                 {
 142                         AssertEncode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-51932.txt", 51932);
 143                 }
 144
 145                 [Test]
 146                 public void CP51932_Decode ()
 147                 {
 148                         AssertDecode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-51932.txt", 51932);
 149                 }
 150
 151                 // ISO-2022-JP
 152
 153                 [Test]
 154                 public void CP50220_Encode ()
 155                 {
 156                         AssertEncode ("Test/texts/japanese2-utf8.txt", "Test/texts/japanese2-50220.txt", 50220);
 157                 }
 158
 159                 [Test]
 160                 public void CP50220_Decode ()
 161                 {
 162                         AssertDecode ("Test/texts/japanese2-utf8.txt", "Test/texts/japanese2-50220.txt", 50220);
 163                 }
 164
 165                 [Test]
 166                 public void CP50221_Encode ()
 167                 {
 168                         AssertEncode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-50221.txt", 50221);
 169                 }
 170
 171                 [Test]
 172                 public void CP50221_Decode ()
 173                 {
 174                         AssertDecode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-50221.txt", 50221);
 175                 }
 176
 177                 [Test]
 178 #if !NET_2_0
 179                 [Category ("NotDotNet")] // MS is buggy here
 180 #endif
 181                 public void CP50222_Encode ()
 182                 {
 183                         AssertEncode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-50222.txt", 50222);
 184                 }
 185
 186                 [Test]
 187 #if !NET_2_0
 188                 [Category ("NotDotNet")] // MS is buggy here
 189 #endif
 190                 public void CP50222_Decode ()
 191                 {
 192                         AssertDecode ("Test/texts/japanese-utf8.txt", "Test/texts/japanese-50222.txt", 50222);
 193                 }
 194
 195                 [Test]
 196 #if !NET_2_0
 197                 [Category ("NotDotNet")] // MS bug
 198 #endif
 199                 public void Bug77723 ()
 200                 {
 201                         GetBytesAllSingleChars (51932);
 202                 }
 203
 204                 [Test]
 205                 public void Bug77724 ()
 206                 {
 207                         GetBytesAllSingleChars (932);
 208                 }
 209
 210                 [Test]
 211                 public void Bug77307 ()
 212                 {
 213                         GetBytesAllSingleChars (54936);
 214                 }
 215
 216                 void GetBytesAllSingleChars (int enc)
 217                 {
 218                         Encoding e = Encoding.GetEncoding (enc);
 219                         for (int i = 0; i < 0x10000; i++)
 220                                 e.GetBytes (new char [] { (char)i });
 221                 }
 222
 223                 void GetCharsAllBytePairs (int enc)
 224                 {
 225                         Encoding e = Encoding.GetEncoding (enc);
 226                         byte [] bytes = new byte [2];
 227                         for (int i0 = 0; i0 < 0x100; i0++) {
 228                                 bytes [0] = (byte) i0;
 229                                 for (int i1 = 0; i1 < 0x100; i1++) {
 230                                         bytes [1] = (byte) i1;
 231                                         e.GetChars (bytes);
 232                                 }
 233                         }
 234                 }
 235
 236                 [Test]
 237                 public void Bug77222 ()
 238                 {
 239                         GetCharsAllBytePairs (51932);
 240                 }
 241
 242                 [Test]
 243                 public void Bug77238 ()
 244                 {
 245                         GetCharsAllBytePairs (936);
 246                 }
 247
 248                 [Test]
 249                 public void Bug77306 ()
 250                 {
 251                         GetCharsAllBytePairs (54936);
 252                 }
 253
 254                 [Test]
 255                 public void Bug77298 ()
 256                 {
 257                         GetCharsAllBytePairs (949);
 258                 }
 259
 260                 [Test]
 261                 public void Bug77274 ()
 262                 {
 263                         GetCharsAllBytePairs (950);
 264                 }
 265
 266                 [Test]
 267 #if !NET_2_0
 268                 [Category ("NotDotNet")] // MS bug
 269 #endif
 270                 public void Encoder54936Refresh ()
 271                 {
 272                         Encoding e = Encoding.GetEncoding ("gb18030");
 273                         Encoder d = e.GetEncoder ();
 274                         byte [] bytes;
 275
 276                         bytes = new byte [4];
 277                         Assert.AreEqual (0, d.GetBytes (new char [] {'\uD800'}, 0, 1, bytes, 0, false), "#1");
 278                         Assert.AreEqual (new byte [] {00, 00, 00, 00},
 279                                 bytes, "#2");
 280
 281                         bytes = new byte [4];
 282                         Assert.AreEqual (4, d.GetBytes (new char [] {'\uDC00'}, 0, 1, bytes, 0, true), "#3");
 283                         Assert.AreEqual (new byte [] {0x90, 0x30, 0x81, 0x30},
 284                                 bytes, "#4");
 285
 286                         bytes = new byte [4];
 287                         Assert.AreEqual (1, d.GetBytes (new char [] {'\uD800'}, 0, 1, bytes, 0, true), "#5");
 288                         Assert.AreEqual (new byte [] {0x3F, 00, 00, 00},
 289                                 bytes, "#6");
 290                 }
 291
 292 #if NET_2_0
 293                 [Test]
 294                 public void Decoder932Refresh ()
 295                 {
 296                         Encoding e = Encoding.GetEncoding (932);
 297                         Decoder d = e.GetDecoder ();
 298                         char [] chars;
 299
 300                         chars = new char [1];
 301                         Assert.AreEqual (0, d.GetChars (new byte [] {0x81}, 0, 1, chars, 0, false), "#1");
 302                         Assert.AreEqual (new char [] {'\0'}, chars, "#2");
 303
 304                         chars = new char [1];
 305                         Assert.AreEqual (1, d.GetChars (new byte [] {0x81}, 0, 1, chars, 0, true), "#3");
 306                         Assert.AreEqual (new char [] {'\uFF1D'}, chars, "#4");
 307
 308                         chars = new char [1];
 309                         Assert.AreEqual (1, d.GetChars (new byte [] {0x81}, 0, 1, chars, 0, true), "#5");
 310                         Assert.AreEqual (new char [] {'\u30FB'}, chars, "#6");
 311                 }
 312
 313                 [Test]
 314                 public void Decoder51932Refresh ()
 315                 {
 316                         Encoding e = Encoding.GetEncoding (51932);
 317                         Decoder d = e.GetDecoder ();
 318                         char [] chars;
 319
 320                         // invalid one
 321                         chars = new char [1];
 322                         Assert.AreEqual (1, d.GetChars (new byte [] {0x81}, 0, 1, chars, 0, false), "#0.1");
 323                         Assert.AreEqual (new char [] {'\u30FB'}, chars, "#0.2");
 324
 325                         // incomplete
 326                         chars = new char [1];
 327                         Assert.AreEqual (0, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0, false), "#1");
 328                         Assert.AreEqual (new char [] {'\0'}, chars, "#2");
 329
 330                         // became complete
 331                         chars = new char [1];
 332                         Assert.AreEqual (1, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0, true), "#3");
 333                         Assert.AreEqual (new char [] {'\u3000'}, chars, "#4");
 334
 335                         // incomplete but refreshed
 336                         chars = new char [1];
 337                         Assert.AreEqual (1, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0, true), "#5");
 338                         Assert.AreEqual (new char [] {'\u30FB'}, chars, "#6");
 339                 }
 340 #endif
 341
 342
 343                 [Test]
 344                 public void Decoder51932NoRefresh ()
 345                 {
 346                         Encoding e = Encoding.GetEncoding (51932);
 347                         Decoder d = e.GetDecoder ();
 348                         char [] chars;
 349
 350                         // incomplete
 351                         chars = new char [1];
 352                         Assert.AreEqual (0, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0), "#1");
 353                         Assert.AreEqual (new char [] {'\0'}, chars, "#2");
 354
 355                         // became complete
 356                         chars = new char [1];
 357                         Assert.AreEqual (1, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0), "#3");
 358                         Assert.AreEqual (new char [] {'\u3000'}, chars, "#4");
 359
 360                         // incomplete but refreshed
 361                         chars = new char [1];
 362                         Assert.AreEqual (0, d.GetChars (new byte [] {0xA1}, 0, 1, chars, 0), "#5");
 363                         Assert.AreEqual (new char [] {'\0'}, chars, "#6");
 364                 }
 365                 #endregion
 366
 367                 #region Korean
 368
 369                 [Test]
 370                 public void CP949_Encode ()
 371                 {
 372                         AssertEncode ("Test/texts/korean-utf8.txt", "Test/texts/korean-949.txt", 949);
 373                 }
 374
 375                 [Test]
 376                 public void CP949_Decode ()
 377                 {
 378                         AssertDecode ("Test/texts/korean-utf8.txt", "Test/texts/korean-949.txt", 949);
 379                 }
 380
 381                 #endregion
 382         }
 383 }