class/corlib/System.Text/UnicodeEncoding.cs

   1 /*
   2  * UnicodeEncoding.cs - Implementation of the
   3  *              "System.Text.UnicodeEncoding" class.
   4  *
   5  * Copyright (c) 2001, 2002  Southern Storm Software, Pty Ltd
   6  * Copyright (C) 2003, 2004 Novell, Inc.
   7  * Copyright (C) 2006 Kornél Pál <http://www.kornelpal.hu/>
   8  *
   9  * Permission is hereby granted, free of charge, to any person obtaining
  10  * a copy of this software and associated documentation files (the "Software"),
  11  * to deal in the Software without restriction, including without limitation
  12  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
  13  * and/or sell copies of the Software, and to permit persons to whom the
  14  * Software is furnished to do so, subject to the following conditions:
  15  *
  16  * The above copyright notice and this permission notice shall be included
  17  * in all copies or substantial portions of the Software.
  18  *
  19  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  20  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  21  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  22  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
  23  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
  24  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
  25  * OTHER DEALINGS IN THE SOFTWARE.
  26  */
  27
  28 namespace System.Text
  29 {
  30
  31 using System;
  32 using System.Runtime.InteropServices;
  33
  34 [Serializable]
  35 [ComVisible (true)]
  36 [MonoLimitation ("Serialization format not compatible with .NET")]
  37 public class UnicodeEncoding : Encoding
  38 {
  39         // Magic numbers used by Windows for Unicode.
  40         internal const int UNICODE_CODE_PAGE     = 1200;
  41         internal const int BIG_UNICODE_CODE_PAGE = 1201;
  42
  43 #if !ECMA_COMPAT
  44         // Size of characters in this encoding.
  45         public const int CharSize = 2;
  46 #endif
  47
  48         // Internal state.
  49         private bool bigEndian;
  50         private bool byteOrderMark;
  51
  52         // Constructors.
  53         public UnicodeEncoding () : this (false, true)
  54         {
  55                 bigEndian = false;
  56                 byteOrderMark = true;
  57         }
  58         public UnicodeEncoding (bool bigEndian, bool byteOrderMark)
  59                 : this (bigEndian, byteOrderMark, false)
  60         {
  61         }
  62
  63         public UnicodeEncoding (bool bigEndian, bool byteOrderMark, bool throwOnInvalidBytes)
  64                 : base ((bigEndian ? BIG_UNICODE_CODE_PAGE : UNICODE_CODE_PAGE))
  65         {
  66                 if (throwOnInvalidBytes)
  67                         SetFallbackInternal (null, new DecoderExceptionFallback ());
  68                 else
  69                         SetFallbackInternal (null, new DecoderReplacementFallback ("\uFFFD"));
  70
  71                 this.bigEndian = bigEndian;
  72                 this.byteOrderMark = byteOrderMark;
  73
  74                 if (bigEndian){
  75                         body_name = "unicodeFFFE";
  76                         encoding_name = "Unicode (Big-Endian)";
  77                         header_name = "unicodeFFFE";
  78                         is_browser_save = false;
  79                         web_name = "unicodeFFFE";
  80                 } else {
  81                         body_name = "utf-16";
  82                         encoding_name = "Unicode";
  83                         header_name = "utf-16";
  84                         is_browser_save = true;
  85                         web_name = "utf-16";
  86                 }
  87
  88                 // Windows reports the same code page number for
  89                 // both the little-endian and big-endian forms.
  90                 windows_code_page = UNICODE_CODE_PAGE;
  91         }
  92
  93         // Get the number of bytes needed to encode a character buffer.
  94         public override int GetByteCount (char[] chars, int index, int count)
  95         {
  96                 if (chars == null) {
  97                         throw new ArgumentNullException ("chars");
  98                 }
  99                 if (index < 0 || index > chars.Length) {
 100                         throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
 101                 }
 102                 if (count < 0 || count > (chars.Length - index)) {
 103                         throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
 104                 }
 105                 return count * 2;
 106         }
 107
 108         public override int GetByteCount (String s)
 109         {
 110                 if (s == null) {
 111                         throw new ArgumentNullException ("s");
 112                 }
 113                 return s.Length * 2;
 114         }
 115
 116         [CLSCompliantAttribute (false)]
 117         [ComVisible (false)]
 118         public unsafe override int GetByteCount (char* chars, int count)
 119         {
 120                 if (chars == null)
 121                         throw new ArgumentNullException ("chars");
 122                 if (count < 0)
 123                         throw new ArgumentOutOfRangeException ("count");
 124
 125                 return count * 2;
 126         }
 127
 128         // Get the bytes that result from encoding a character buffer.
 129         public unsafe override int GetBytes (char [] chars, int charIndex, int charCount,
 130                                                                                 byte [] bytes, int byteIndex)
 131         {
 132                 if (chars == null) {
 133                         throw new ArgumentNullException ("chars");
 134                 }
 135                 if (bytes == null) {
 136                         throw new ArgumentNullException ("bytes");
 137                 }
 138                 if (charIndex < 0 || charIndex > chars.Length) {
 139                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_Array"));
 140                 }
 141                 if (charCount < 0 || charCount > (chars.Length - charIndex)) {
 142                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_Array"));
 143                 }
 144                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 145                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 146                 }
 147
 148                 if (charCount == 0)
 149                         return 0;
 150
 151                 int byteCount = bytes.Length - byteIndex;
 152                 if (bytes.Length == 0)
 153                         bytes = new byte [1];
 154
 155                 fixed (char* charPtr = chars)
 156                         fixed (byte* bytePtr = bytes)
 157                                 return GetBytesInternal (charPtr + charIndex, charCount, bytePtr + byteIndex, byteCount);
 158         }
 159
 160         public unsafe override int GetBytes (String s, int charIndex, int charCount,
 161                                                                                 byte [] bytes, int byteIndex)
 162         {
 163                 if (s == null) {
 164                         throw new ArgumentNullException ("s");
 165                 }
 166                 if (bytes == null) {
 167                         throw new ArgumentNullException ("bytes");
 168                 }
 169                 if (charIndex < 0 || charIndex > s.Length) {
 170                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_StringIndex"));
 171                 }
 172                 if (charCount < 0 || charCount > (s.Length - charIndex)) {
 173                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_StringRange"));
 174                 }
 175                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 176                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 177                 }
 178
 179                 // For consistency
 180                 if (charCount == 0)
 181                         return 0;
 182
 183                 int byteCount = bytes.Length - byteIndex;
 184                 if (bytes.Length == 0)
 185                         bytes = new byte [1];
 186
 187                 fixed (char* charPtr = s)
 188                         fixed (byte* bytePtr = bytes)
 189                                 return GetBytesInternal (charPtr + charIndex, charCount, bytePtr + byteIndex, byteCount);
 190         }
 191
 192         [CLSCompliantAttribute (false)]
 193         [ComVisible (false)]
 194         public unsafe override int GetBytes (char* chars, int charCount,
 195                                                                                 byte* bytes, int byteCount)
 196         {
 197                 if (bytes == null)
 198                         throw new ArgumentNullException ("bytes");
 199                 if (chars == null)
 200                         throw new ArgumentNullException ("chars");
 201                 if (charCount < 0)
 202                         throw new ArgumentOutOfRangeException ("charCount");
 203                 if (byteCount < 0)
 204                         throw new ArgumentOutOfRangeException ("byteCount");
 205
 206                 return GetBytesInternal (chars, charCount, bytes, byteCount);
 207         }
 208
 209         private unsafe int GetBytesInternal (char* chars, int charCount,
 210                                                                                 byte* bytes, int byteCount)
 211         {
 212                 int count = charCount * 2;
 213
 214                 if (byteCount < count)
 215                         throw new ArgumentException (_("Arg_InsufficientSpace"));
 216
 217                 CopyChars ((byte*) chars, bytes, count, bigEndian);
 218                 return count;
 219         }
 220
 221         // Get the number of characters needed to decode a byte buffer.
 222         public override int GetCharCount (byte[] bytes, int index, int count)
 223         {
 224                 if (bytes == null) {
 225                         throw new ArgumentNullException ("bytes");
 226                 }
 227                 if (index < 0 || index > bytes.Length) {
 228                         throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
 229                 }
 230                 if (count < 0 || count > (bytes.Length - index)) {
 231                         throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
 232                 }
 233                 return count / 2;
 234         }
 235
 236         [CLSCompliantAttribute (false)]
 237         [ComVisible (false)]
 238         public unsafe override int GetCharCount (byte* bytes, int count)
 239         {
 240                 if (bytes == null)
 241                         throw new ArgumentNullException ("bytes");
 242                 if (count < 0)
 243                         throw new ArgumentOutOfRangeException ("count");
 244
 245                 return count / 2;
 246         }
 247
 248         // Get the characters that result from decoding a byte buffer.
 249         public unsafe override int GetChars (byte [] bytes, int byteIndex, int byteCount,
 250                                                                                 char [] chars, int charIndex)
 251         {
 252                 if (bytes == null) {
 253                         throw new ArgumentNullException ("bytes");
 254                 }
 255                 if (chars == null) {
 256                         throw new ArgumentNullException ("chars");
 257                 }
 258                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 259                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 260                 }
 261                 if (byteCount < 0 || byteCount > (bytes.Length - byteIndex)) {
 262                         throw new ArgumentOutOfRangeException ("byteCount", _("ArgRange_Array"));
 263                 }
 264                 if (charIndex < 0 || charIndex > chars.Length) {
 265                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_Array"));
 266                 }
 267
 268                 if (byteCount == 0)
 269                         return 0;
 270
 271                 int charCount = chars.Length - charIndex;
 272                 if (chars.Length == 0)
 273                         chars = new char [1];
 274
 275                 fixed (byte* bytePtr = bytes)
 276                         fixed (char* charPtr = chars)
 277                                 return GetCharsInternal (bytePtr + byteIndex, byteCount, charPtr + charIndex, charCount);
 278 }
 279
 280         [CLSCompliantAttribute (false)]
 281         [ComVisible (false)]
 282         public unsafe override int GetChars (byte* bytes, int byteCount,
 283                                                                                 char* chars, int charCount)
 284         {
 285                 if (bytes == null)
 286                         throw new ArgumentNullException ("bytes");
 287                 if (chars == null)
 288                         throw new ArgumentNullException ("chars");
 289                 if (charCount < 0)
 290                         throw new ArgumentOutOfRangeException ("charCount");
 291                 if (byteCount < 0)
 292                         throw new ArgumentOutOfRangeException ("byteCount");
 293
 294                 return GetCharsInternal (bytes, byteCount, chars, charCount);
 295         }
 296
 297         // Decode a buffer of bytes into a string.
 298         [ComVisible (false)]
 299         public unsafe override String GetString (byte [] bytes, int index, int count)
 300         {
 301                 if (bytes == null)
 302                         throw new ArgumentNullException ("bytes");
 303                 if (index < 0 || index > bytes.Length)
 304                         throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
 305                 if (count < 0 || count > (bytes.Length - index))
 306                         throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
 307
 308                 if (count == 0)
 309                         return string.Empty;
 310
 311                 // GetCharCountInternal
 312                 int charCount = count / 2;
 313                 string s = string.InternalAllocateStr (charCount);
 314
 315                 fixed (byte* bytePtr = bytes)
 316                         fixed (char* charPtr = s)
 317                                 GetCharsInternal (bytePtr + index, count, charPtr, charCount);
 318
 319                 return s;
 320         }
 321
 322         private unsafe int GetCharsInternal (byte* bytes, int byteCount,
 323                                                                                 char* chars, int charCount)
 324         {
 325                 int count = byteCount / 2;
 326
 327                 // Validate that we have sufficient space in "chars".
 328                 if (charCount < count)
 329                         throw new ArgumentException (_("Arg_InsufficientSpace"));
 330
 331                 CopyChars (bytes, (byte*) chars, byteCount, bigEndian);
 332                 return count;
 333         }
 334
 335         [ComVisible (false)]
 336         public override Encoder GetEncoder ()
 337         {
 338                 return(base.GetEncoder ());
 339         }
 340
 341         // Get the maximum number of bytes needed to encode a
 342         // specified number of characters.
 343         public override int GetMaxByteCount (int charCount)
 344         {
 345                 if (charCount < 0) {
 346                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_NonNegative"));
 347                 }
 348                 return charCount * 2;
 349         }
 350
 351         // Get the maximum number of characters needed to decode a
 352         // specified number of bytes.
 353         public override int GetMaxCharCount (int byteCount)
 354         {
 355                 if (byteCount < 0) {
 356                         throw new ArgumentOutOfRangeException
 357                                 ("byteCount", _("ArgRange_NonNegative"));
 358                 }
 359                 return byteCount / 2;
 360         }
 361
 362         // Get a Unicode-specific decoder that is attached to this instance.
 363         public override Decoder GetDecoder ()
 364         {
 365                 return new UnicodeDecoder (bigEndian);
 366         }
 367
 368         // Get the Unicode preamble.
 369         public override byte[] GetPreamble ()
 370         {
 371                 if (byteOrderMark) {
 372                         byte[] preamble = new byte[2];
 373                         if (bigEndian) {
 374                                 preamble[0] = (byte)0xFE;
 375                                 preamble[1] = (byte)0xFF;
 376                         } else {
 377                                 preamble[0] = (byte)0xFF;
 378                                 preamble[1] = (byte)0xFE;
 379                         }
 380                         return preamble;
 381                 } else {
 382                         return new byte [0];
 383                 }
 384         }
 385
 386         // Determine if this object is equal to another.
 387         public override bool Equals (Object value)
 388         {
 389                 UnicodeEncoding enc = (value as UnicodeEncoding);
 390                 if (enc != null) {
 391                         return (codePage == enc.codePage &&
 392                                         bigEndian == enc.bigEndian &&
 393                                         byteOrderMark == enc.byteOrderMark);
 394                 } else {
 395                         return false;
 396                 }
 397         }
 398
 399         // Get the hash code for this object.
 400         public override int GetHashCode ()
 401         {
 402                 return base.GetHashCode ();
 403         }
 404
 405         private unsafe static void CopyChars (byte* src, byte* dest, int count, bool bigEndian)
 406         {
 407                 if (BitConverter.IsLittleEndian != bigEndian) {
 408                         string.memcpy (dest, src, count & unchecked ((int) 0xFFFFFFFE));
 409                         return;
 410                 }
 411
 412                 switch (count) {
 413                 case 0:
 414                         return;
 415                 case 1:
 416                         return;
 417                 case 2:
 418                         goto Count2;
 419                 case 3:
 420                         goto Count2;
 421                 case 4:
 422                         goto Count4;
 423                 case 5:
 424                         goto Count4;
 425                 case 6:
 426                         goto Count4;
 427                 case 7:
 428                         goto Count4;
 429                 case 8:
 430                         goto Count8;
 431                 case 9:
 432                         goto Count8;
 433                 case 10:
 434                         goto Count8;
 435                 case 11:
 436                         goto Count8;
 437                 case 12:
 438                         goto Count8;
 439                 case 13:
 440                         goto Count8;
 441                 case 14:
 442                         goto Count8;
 443                 case 15:
 444                         goto Count8;
 445                 }
 446
 447                 do {
 448                         dest [0] = src [1];
 449                         dest [1] = src [0];
 450                         dest [2] = src [3];
 451                         dest [3] = src [2];
 452                         dest [4] = src [5];
 453                         dest [5] = src [4];
 454                         dest [6] = src [7];
 455                         dest [7] = src [6];
 456                         dest [8] = src [9];
 457                         dest [9] = src [8];
 458                         dest [10] = src [11];
 459                         dest [11] = src [10];
 460                         dest [12] = src [13];
 461                         dest [13] = src [12];
 462                         dest [14] = src [15];
 463                         dest [15] = src [14];
 464                         dest += 16;
 465                         src += 16;
 466                         count -= 16;
 467                 } while ((count & unchecked ((int) 0xFFFFFFF0)) != 0);
 468
 469                 switch (count) {
 470                 case 0:
 471                         return;
 472                 case 1:
 473                         return;
 474                 case 2:
 475                         goto Count2;
 476                 case 3:
 477                         goto Count2;
 478                 case 4:
 479                         goto Count4;
 480                 case 5:
 481                         goto Count4;
 482                 case 6:
 483                         goto Count4;
 484                 case 7:
 485                         goto Count4;
 486                 }
 487
 488                 Count8:;
 489                 dest [0] = src [1];
 490                 dest [1] = src [0];
 491                 dest [2] = src [3];
 492                 dest [3] = src [2];
 493                 dest [4] = src [5];
 494                 dest [5] = src [4];
 495                 dest [6] = src [7];
 496                 dest [7] = src [6];
 497                 dest += 8;
 498                 src += 8;
 499
 500                 if ((count & 4) == 0)
 501                         goto TestCount2;
 502                 Count4:;
 503                 dest [0] = src [1];
 504                 dest [1] = src [0];
 505                 dest [2] = src [3];
 506                 dest [3] = src [2];
 507                 dest += 4;
 508                 src += 4;
 509
 510                 TestCount2:;
 511                 if ((count & 2) == 0)
 512                         return;
 513                 Count2:;
 514                 dest [0] = src [1];
 515                 dest [1] = src [0];
 516         }
 517
 518         // Unicode decoder implementation.
 519         private sealed class UnicodeDecoder : Decoder
 520         {
 521                 private bool bigEndian;
 522                 private int leftOverByte;
 523
 524                 // Constructor.
 525                 public UnicodeDecoder (bool bigEndian)
 526                 {
 527                         this.bigEndian = bigEndian;
 528                         leftOverByte = -1;
 529                 }
 530
 531                 // Override inherited methods.
 532                 public override int GetCharCount (byte[] bytes, int index, int count)
 533                 {
 534                         if (bytes == null) {
 535                                 throw new ArgumentNullException ("bytes");
 536                         }
 537                         if (index < 0 || index > bytes.Length) {
 538                                 throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
 539                         }
 540                         if (count < 0 || count > (bytes.Length - index)) {
 541                                 throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
 542                         }
 543                         if (leftOverByte != -1) {
 544                                 return (count + 1) / 2;
 545                         } else {
 546                                 return count / 2;
 547                         }
 548                 }
 549
 550                 public unsafe override int GetChars (byte [] bytes, int byteIndex,
 551                                                                                         int byteCount, char [] chars,
 552                                                                                         int charIndex)
 553                 {
 554                         if (bytes == null) {
 555                                 throw new ArgumentNullException ("bytes");
 556                         }
 557                         if (chars == null) {
 558                                 throw new ArgumentNullException ("chars");
 559                         }
 560                         if (byteIndex < 0 || byteIndex > bytes.Length) {
 561                                 throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 562                         }
 563                         if (byteCount < 0 || byteCount > (bytes.Length - byteIndex)) {
 564                                 throw new ArgumentOutOfRangeException ("byteCount", _("ArgRange_Array"));
 565                         }
 566                         if (charIndex < 0 || charIndex > chars.Length) {
 567                                 throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_Array"));
 568                         }
 569
 570                         if (byteCount == 0)
 571                                 return 0;
 572
 573                         int leftOver = leftOverByte;
 574                         int count;
 575
 576                         if (leftOver != -1)
 577                                 count = (byteCount + 1) / 2;
 578                         else
 579                                 count = byteCount / 2;
 580
 581                         if (chars.Length - charIndex < count)
 582                                 throw new ArgumentException (_("Arg_InsufficientSpace"));
 583
 584                         if (leftOver != -1) {
 585                                 if (bigEndian)
 586                                         chars [charIndex] = unchecked ((char) ((leftOver << 8) | (int) bytes [byteIndex]));
 587                                 else
 588                                         chars [charIndex] = unchecked ((char) (((int) bytes [byteIndex] << 8) | leftOver));
 589                                 charIndex++;
 590                                 byteIndex++;
 591                                 byteCount--;
 592                         }
 593
 594                         if ((byteCount & unchecked ((int) 0xFFFFFFFE)) != 0)
 595                                 fixed (byte* bytePtr = bytes)
 596                                         fixed (char* charPtr = chars)
 597                                                 CopyChars (bytePtr + byteIndex, (byte*) (charPtr + charIndex), byteCount, bigEndian);
 598
 599                         if ((byteCount & 1) == 0)
 600                                 leftOverByte = -1;
 601                         else
 602                                 leftOverByte = bytes [byteCount + byteIndex - 1];
 603
 604                         return count;
 605                 }
 606
 607         } // class UnicodeDecoder
 608
 609 }; // class UnicodeEncoding
 610
 611 }; // namespace System.Text