mcs/class/referencesource/System.Runtime.Serialization/System/Xml/EncodingStreamWrapper.cs

   1 //-----------------------------------------------------------------------------
   2 // Copyright (c) Microsoft Corporation.  All rights reserved.
   3 //-----------------------------------------------------------------------------
   4 namespace System.Xml
   5 {
   6     using System;
   7     using System.IO;
   8     using System.Text;
   9     using System.Runtime.Serialization;
  10
  11     // This wrapper does not support seek.
  12     // Constructors consume/emit byte order mark.
  13     // Supports: UTF-8, Unicode, BigEndianUnicode
  14     // ASSUMPTION ([....]): This class will only be used for EITHER reading OR writing.  It can be done, it would just mean more buffers.
  15     // ASSUMPTION ([....]): The byte buffer is large enough to hold the declaration
  16     // ASSUMPTION ([....]): The buffer manipulation methods (FillBuffer/Compare/etc.) will only be used to parse the declaration
  17     //                      during construction.
  18     class EncodingStreamWrapper : Stream
  19     {
  20         enum SupportedEncoding { UTF8, UTF16LE, UTF16BE, None }
  21         static readonly UTF8Encoding SafeUTF8 = new UTF8Encoding(false, false);
  22         static readonly UnicodeEncoding SafeUTF16 = new UnicodeEncoding(false, false, false);
  23         static readonly UnicodeEncoding SafeBEUTF16 = new UnicodeEncoding(true, false, false);
  24         static readonly UTF8Encoding ValidatingUTF8 = new UTF8Encoding(false, true);
  25         static readonly UnicodeEncoding ValidatingUTF16 = new UnicodeEncoding(false, false, true);
  26         static readonly UnicodeEncoding ValidatingBEUTF16 = new UnicodeEncoding(true, false, true);
  27         const int BufferLength = 128;
  28
  29         // UTF-8 is fastpath, so that's how these are stored
  30         // Compare methods adapt to unicodes.
  31         static readonly byte[] encodingAttr = new byte[] { (byte)'e', (byte)'n', (byte)'c', (byte)'o', (byte)'d', (byte)'i', (byte)'n', (byte)'g' };
  32         static readonly byte[] encodingUTF8 = new byte[] { (byte)'u', (byte)'t', (byte)'f', (byte)'-', (byte)'8' };
  33         static readonly byte[] encodingUnicode = new byte[] { (byte)'u', (byte)'t', (byte)'f', (byte)'-', (byte)'1', (byte)'6' };
  34         static readonly byte[] encodingUnicodeLE = new byte[] { (byte)'u', (byte)'t', (byte)'f', (byte)'-', (byte)'1', (byte)'6', (byte)'l', (byte)'e' };
  35         static readonly byte[] encodingUnicodeBE = new byte[] { (byte)'u', (byte)'t', (byte)'f', (byte)'-', (byte)'1', (byte)'6', (byte)'b', (byte)'e' };
  36
  37         SupportedEncoding encodingCode;
  38         Encoding encoding;
  39         Encoder enc;
  40         Decoder dec;
  41         bool isReading;
  42
  43         Stream stream;
  44         char[] chars;
  45         byte[] bytes;
  46         int byteOffset;
  47         int byteCount;
  48
  49         byte[] byteBuffer = new byte[1];
  50
  51         // Reading constructor
  52         public EncodingStreamWrapper(Stream stream, Encoding encoding)
  53         {
  54             try
  55             {
  56                 this.isReading = true;
  57                 this.stream = new BufferedStream(stream);
  58
  59                 // Decode the expected encoding
  60                 SupportedEncoding expectedEnc = GetSupportedEncoding(encoding);
  61
  62                 // Get the byte order mark so we can determine the encoding
  63                 // May want to try to delay allocating everything until we know the BOM
  64                 SupportedEncoding declEnc = ReadBOMEncoding(encoding == null);
  65
  66                 // Check that the expected encoding matches the decl encoding.
  67                 if (expectedEnc != SupportedEncoding.None && expectedEnc != declEnc)
  68                     ThrowExpectedEncodingMismatch(expectedEnc, declEnc);
  69
  70                 // Fastpath: UTF-8 BOM
  71                 if (declEnc == SupportedEncoding.UTF8)
  72                 {
  73                     // Fastpath: UTF-8 BOM, No declaration
  74                     FillBuffer(2);
  75                     if (bytes[byteOffset + 1] != '?' || bytes[byteOffset] != '<')
  76                     {
  77                         return;
  78                     }
  79
  80                     FillBuffer(BufferLength);
  81                     CheckUTF8DeclarationEncoding(bytes, byteOffset, byteCount, declEnc, expectedEnc);
  82                 }
  83                 else
  84                 {
  85                     // Convert to UTF-8
  86                     EnsureBuffers();
  87                     FillBuffer((BufferLength - 1) * 2);
  88                     SetReadDocumentEncoding(declEnc);
  89                     CleanupCharBreak();
  90                     int count = this.encoding.GetChars(bytes, byteOffset, byteCount, chars, 0);
  91                     byteOffset = 0;
  92                     byteCount = ValidatingUTF8.GetBytes(chars, 0, count, bytes, 0);
  93
  94                     // Check for declaration
  95                     if (bytes[1] == '?' && bytes[0] == '<')
  96                     {
  97                         CheckUTF8DeclarationEncoding(bytes, 0, byteCount, declEnc, expectedEnc);
  98                     }
  99                     else
 100                     {
 101                         // Declaration required if no out-of-band encoding
 102                         if (expectedEnc == SupportedEncoding.None)
 103                             throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclarationRequired)));
 104                     }
 105                 }
 106             }
 107             catch (DecoderFallbackException ex)
 108             {
 109                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlInvalidBytes), ex));
 110             }
 111         }
 112
 113         void SetReadDocumentEncoding(SupportedEncoding e)
 114         {
 115             EnsureBuffers();
 116             this.encodingCode = e;
 117             this.encoding = GetEncoding(e);
 118         }
 119
 120         static Encoding GetEncoding(SupportedEncoding e)
 121         {
 122             switch (e)
 123             {
 124                 case SupportedEncoding.UTF8:
 125                     return ValidatingUTF8;
 126
 127                 case SupportedEncoding.UTF16LE:
 128                     return ValidatingUTF16;
 129
 130                 case SupportedEncoding.UTF16BE:
 131                     return ValidatingBEUTF16;
 132
 133                 default:
 134                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlEncodingNotSupported)));
 135             }
 136         }
 137
 138         static Encoding GetSafeEncoding(SupportedEncoding e)
 139         {
 140             switch (e)
 141             {
 142                 case SupportedEncoding.UTF8:
 143                     return SafeUTF8;
 144
 145                 case SupportedEncoding.UTF16LE:
 146                     return SafeUTF16;
 147
 148                 case SupportedEncoding.UTF16BE:
 149                     return SafeBEUTF16;
 150
 151                 default:
 152                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlEncodingNotSupported)));
 153             }
 154         }
 155
 156         static string GetEncodingName(SupportedEncoding enc)
 157         {
 158             switch (enc)
 159             {
 160                 case SupportedEncoding.UTF8:
 161                     return "utf-8";
 162
 163                 case SupportedEncoding.UTF16LE:
 164                     return "utf-16LE";
 165
 166                 case SupportedEncoding.UTF16BE:
 167                     return "utf-16BE";
 168
 169                 default:
 170                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlEncodingNotSupported)));
 171             }
 172         }
 173
 174         static SupportedEncoding GetSupportedEncoding(Encoding encoding)
 175         {
 176             if (encoding == null)
 177                 return SupportedEncoding.None;
 178             else if (encoding.WebName == ValidatingUTF8.WebName)
 179                 return SupportedEncoding.UTF8;
 180             else if (encoding.WebName == ValidatingUTF16.WebName)
 181                 return SupportedEncoding.UTF16LE;
 182             else if (encoding.WebName == ValidatingBEUTF16.WebName)
 183                 return SupportedEncoding.UTF16BE;
 184             else
 185                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlEncodingNotSupported)));
 186         }
 187
 188         // Writing constructor
 189         public EncodingStreamWrapper(Stream stream, Encoding encoding, bool emitBOM)
 190         {
 191             this.isReading = false;
 192             this.encoding = encoding;
 193             this.stream = new BufferedStream(stream);
 194
 195             // Set the encoding code
 196             this.encodingCode = GetSupportedEncoding(encoding);
 197
 198             if (encodingCode != SupportedEncoding.UTF8)
 199             {
 200                 EnsureBuffers();
 201                 dec = ValidatingUTF8.GetDecoder();
 202                 enc = this.encoding.GetEncoder();
 203
 204                 // Emit BOM
 205                 if (emitBOM)
 206                 {
 207                     byte[] bom = this.encoding.GetPreamble();
 208                     if (bom.Length > 0)
 209                         this.stream.Write(bom, 0, bom.Length);
 210                 }
 211             }
 212         }
 213
 214         SupportedEncoding ReadBOMEncoding(bool notOutOfBand)
 215         {
 216             int b1 = this.stream.ReadByte();
 217             int b2 = this.stream.ReadByte();
 218             int b3 = this.stream.ReadByte();
 219             int b4 = this.stream.ReadByte();
 220
 221             // Premature end of stream
 222             if (b4 == -1)
 223                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.UnexpectedEndOfFile)));
 224
 225             int preserve;
 226             SupportedEncoding e = ReadBOMEncoding((byte)b1, (byte)b2, (byte)b3, (byte)b4, notOutOfBand, out preserve);
 227
 228             EnsureByteBuffer();
 229             switch (preserve)
 230             {
 231                 case 1:
 232                     bytes[0] = (byte)b4;
 233                     break;
 234
 235                 case 2:
 236                     bytes[0] = (byte)b3;
 237                     bytes[1] = (byte)b4;
 238                     break;
 239
 240                 case 4:
 241                     bytes[0] = (byte)b1;
 242                     bytes[1] = (byte)b2;
 243                     bytes[2] = (byte)b3;
 244                     bytes[3] = (byte)b4;
 245                     break;
 246             }
 247             byteCount = preserve;
 248
 249             return e;
 250         }
 251
 252         static SupportedEncoding ReadBOMEncoding(byte b1, byte b2, byte b3, byte b4, bool notOutOfBand, out int preserve)
 253         {
 254             SupportedEncoding e = SupportedEncoding.UTF8; // Default
 255
 256             preserve = 0;
 257             if (b1 == '<' && b2 != 0x00) // UTF-8, no BOM
 258             {
 259                 e = SupportedEncoding.UTF8;
 260                 preserve = 4;
 261             }
 262             else if (b1 == 0xFF && b2 == 0xFE) // UTF-16 little endian
 263             {
 264                 e = SupportedEncoding.UTF16LE;
 265                 preserve = 2;
 266             }
 267             else if (b1 == 0xFE && b2 == 0xFF) // UTF-16 big endian
 268             {
 269                 e = SupportedEncoding.UTF16BE;
 270                 preserve = 2;
 271             }
 272             else if (b1 == 0x00 && b2 == '<') // UTF-16 big endian, no BOM
 273             {
 274                 e = SupportedEncoding.UTF16BE;
 275
 276                 if (notOutOfBand && (b3 != 0x00 || b4 != '?'))
 277                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclMissing)));
 278                 preserve = 4;
 279             }
 280             else if (b1 == '<' && b2 == 0x00) // UTF-16 little endian, no BOM
 281             {
 282                 e = SupportedEncoding.UTF16LE;
 283
 284                 if (notOutOfBand && (b3 != '?' || b4 != 0x00))
 285                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclMissing)));
 286                 preserve = 4;
 287             }
 288             else if (b1 == 0xEF && b2 == 0xBB) // UTF8 with BOM
 289             {
 290                 // Encoding error
 291                 if (notOutOfBand && b3 != 0xBF)
 292                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlBadBOM)));
 293                 preserve = 1;
 294             }
 295             else  // Assume UTF8
 296             {
 297                 preserve = 4;
 298             }
 299
 300             return e;
 301         }
 302
 303         void FillBuffer(int count)
 304         {
 305             count -= byteCount;
 306             while (count > 0)
 307             {
 308                 int read = stream.Read(bytes, byteOffset + byteCount, count);
 309                 if (read == 0)
 310                     break;
 311
 312                 byteCount += read;
 313                 count -= read;
 314             }
 315         }
 316
 317         void EnsureBuffers()
 318         {
 319             EnsureByteBuffer();
 320             if (chars == null)
 321                 chars = new char[BufferLength];
 322         }
 323
 324         void EnsureByteBuffer()
 325         {
 326             if (bytes != null)
 327                 return;
 328
 329             bytes = new byte[BufferLength * 4];
 330             byteOffset = 0;
 331             byteCount = 0;
 332         }
 333
 334         static void CheckUTF8DeclarationEncoding(byte[] buffer, int offset, int count, SupportedEncoding e, SupportedEncoding expectedEnc)
 335         {
 336             byte quot = 0;
 337             int encEq = -1;
 338             int max = offset + Math.Min(count, BufferLength);
 339
 340             // Encoding should be second "=", abort at first "?"
 341             int i = 0;
 342             int eq = 0;
 343             for (i = offset + 2; i < max; i++)  // Skip the "<?" so we don't get caught by the first "?"
 344             {
 345                 if (quot != 0)
 346                 {
 347                     if (buffer[i] == quot)
 348                     {
 349                         quot = 0;
 350                     }
 351                     continue;
 352                 }
 353
 354                 if (buffer[i] == (byte)'\'' || buffer[i] == (byte)'"')
 355                 {
 356                     quot = buffer[i];
 357                 }
 358                 else if (buffer[i] == (byte)'=')
 359                 {
 360                     if (eq == 1)
 361                     {
 362                         encEq = i;
 363                         break;
 364                     }
 365                     eq++;
 366                 }
 367                 else if (buffer[i] == (byte)'?')  // Not legal character in a decl before second "="
 368                 {
 369                     break;
 370                 }
 371             }
 372
 373             // No encoding found
 374             if (encEq == -1)
 375             {
 376                 if (e != SupportedEncoding.UTF8 && expectedEnc == SupportedEncoding.None)
 377                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclarationRequired)));
 378                 return;
 379             }
 380
 381             if (encEq < 28) // Earliest second "=" can appear
 382                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlMalformedDecl)));
 383
 384             // Back off whitespace
 385             for (i = encEq - 1; IsWhitespace(buffer[i]); i--);
 386
 387             // Check for encoding attribute
 388             if (!Compare(encodingAttr, buffer, i - encodingAttr.Length + 1))
 389             {
 390                 if (e != SupportedEncoding.UTF8 && expectedEnc == SupportedEncoding.None)
 391                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclarationRequired)));
 392                 return;
 393             }
 394
 395             // Move ahead of whitespace
 396             for (i = encEq + 1; i < max && IsWhitespace(buffer[i]); i++);
 397
 398             // Find the quotes
 399             if (buffer[i] != '\'' && buffer[i] != '"')
 400                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlMalformedDecl)));
 401             quot = buffer[i];
 402
 403             int q = i;
 404             for (i = q + 1; buffer[i] != quot && i < max; ++i);
 405
 406             if (buffer[i] != quot)
 407                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlMalformedDecl)));
 408
 409             int encStart = q + 1;
 410             int encCount = i - encStart;
 411
 412             // lookup the encoding
 413             SupportedEncoding declEnc = e;
 414             if (encCount == encodingUTF8.Length && CompareCaseInsensitive(encodingUTF8, buffer, encStart))
 415             {
 416                 declEnc = SupportedEncoding.UTF8;
 417             }
 418             else if (encCount == encodingUnicodeLE.Length && CompareCaseInsensitive(encodingUnicodeLE, buffer, encStart))
 419             {
 420                 declEnc = SupportedEncoding.UTF16LE;
 421             }
 422             else if (encCount == encodingUnicodeBE.Length && CompareCaseInsensitive(encodingUnicodeBE, buffer, encStart))
 423             {
 424                 declEnc = SupportedEncoding.UTF16BE;
 425             }
 426             else if (encCount == encodingUnicode.Length && CompareCaseInsensitive(encodingUnicode, buffer, encStart))
 427             {
 428                 if (e == SupportedEncoding.UTF8)
 429                     ThrowEncodingMismatch(SafeUTF8.GetString(buffer, encStart, encCount), SafeUTF8.GetString(encodingUTF8, 0, encodingUTF8.Length));
 430             }
 431             else
 432             {
 433                 ThrowEncodingMismatch(SafeUTF8.GetString(buffer, encStart, encCount), e);
 434             }
 435
 436             if (e != declEnc)
 437                 ThrowEncodingMismatch(SafeUTF8.GetString(buffer, encStart, encCount), e);
 438         }
 439
 440         static bool CompareCaseInsensitive(byte[] key, byte[] buffer, int offset)
 441         {
 442             for (int i = 0; i < key.Length; i++)
 443             {
 444                 if (key[i] == buffer[offset + i])
 445                     continue;
 446
 447                 if (key[i] != Char.ToLower((char)buffer[offset + i], System.Globalization.CultureInfo.InvariantCulture))
 448                     return false;
 449             }
 450             return true;
 451         }
 452
 453         static bool Compare(byte[] key, byte[] buffer, int offset)
 454         {
 455             for (int i = 0; i < key.Length; i++)
 456             {
 457                 if (key[i] != buffer[offset + i])
 458                     return false;
 459             }
 460             return true;
 461         }
 462
 463         static bool IsWhitespace(byte ch)
 464         {
 465             return ch == (byte)' ' || ch == (byte)'\n' || ch == (byte)'\t' || ch == (byte)'\r';
 466         }
 467
 468         internal static ArraySegment<byte> ProcessBuffer(byte[] buffer, int offset, int count, Encoding encoding)
 469         {
 470             if (count < 4)
 471                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.UnexpectedEndOfFile)));
 472
 473             try
 474             {
 475                 int preserve;
 476                 ArraySegment<byte> seg;
 477
 478                 SupportedEncoding expectedEnc = GetSupportedEncoding(encoding);
 479                 SupportedEncoding declEnc = ReadBOMEncoding(buffer[offset], buffer[offset + 1], buffer[offset + 2], buffer[offset + 3], encoding == null, out preserve);
 480                 if (expectedEnc != SupportedEncoding.None && expectedEnc != declEnc)
 481                     ThrowExpectedEncodingMismatch(expectedEnc, declEnc);
 482
 483                 offset += 4 - preserve;
 484                 count -= 4 - preserve;
 485
 486                 // Fastpath: UTF-8
 487                 char[] chars;
 488                 byte[] bytes;
 489                 Encoding localEnc;
 490                 if (declEnc == SupportedEncoding.UTF8)
 491                 {
 492                     // Fastpath: No declaration
 493                     if (buffer[offset + 1] != '?' || buffer[offset] != '<')
 494                     {
 495                         seg = new ArraySegment<byte>(buffer, offset, count);
 496                         return seg;
 497                     }
 498
 499                     CheckUTF8DeclarationEncoding(buffer, offset, count, declEnc, expectedEnc);
 500                     seg = new ArraySegment<byte>(buffer, offset, count);
 501                     return seg;
 502                 }
 503
 504                 // Convert to UTF-8
 505                 localEnc = GetSafeEncoding(declEnc);
 506                 int inputCount = Math.Min(count, BufferLength * 2);
 507                 chars = new char[localEnc.GetMaxCharCount(inputCount)];
 508                 int ccount = localEnc.GetChars(buffer, offset, inputCount, chars, 0);
 509                 bytes = new byte[ValidatingUTF8.GetMaxByteCount(ccount)];
 510                 int bcount = ValidatingUTF8.GetBytes(chars, 0, ccount, bytes, 0);
 511
 512                 // Check for declaration
 513                 if (bytes[1] == '?' && bytes[0] == '<')
 514                 {
 515                     CheckUTF8DeclarationEncoding(bytes, 0, bcount, declEnc, expectedEnc);
 516                 }
 517                 else
 518                 {
 519                     // Declaration required if no out-of-band encoding
 520                     if (expectedEnc == SupportedEncoding.None)
 521                         throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlDeclarationRequired)));
 522                 }
 523
 524                 seg = new ArraySegment<byte>(ValidatingUTF8.GetBytes(GetEncoding(declEnc).GetChars(buffer, offset, count)));
 525                 return seg;
 526             }
 527             catch (DecoderFallbackException e)
 528             {
 529                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlInvalidBytes), e));
 530             }
 531         }
 532
 533         static void ThrowExpectedEncodingMismatch(SupportedEncoding expEnc, SupportedEncoding actualEnc)
 534         {
 535             throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlExpectedEncoding, GetEncodingName(expEnc), GetEncodingName(actualEnc))));
 536         }
 537
 538         static void ThrowEncodingMismatch(string declEnc, SupportedEncoding enc)
 539         {
 540             ThrowEncodingMismatch(declEnc, GetEncodingName(enc));
 541         }
 542
 543         static void ThrowEncodingMismatch(string declEnc, string docEnc)
 544         {
 545             throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlEncodingMismatch, declEnc, docEnc)));
 546         }
 547
 548         // This stream wrapper does not support duplex
 549         public override bool CanRead
 550         {
 551             get
 552             {
 553                 if (!isReading)
 554                     return false;
 555
 556                 return this.stream.CanRead;
 557             }
 558         }
 559
 560         // The encoding conversion and buffering breaks seeking.
 561         public override bool CanSeek
 562         {
 563             get
 564             {
 565                 return false;
 566             }
 567         }
 568
 569         // This stream wrapper does not support duplex
 570         public override bool CanWrite
 571         {
 572             get
 573             {
 574                 if (isReading)
 575                     return false;
 576
 577                 return this.stream.CanWrite;
 578             }
 579         }
 580
 581
 582         // The encoding conversion and buffering breaks seeking.
 583         public override long Position
 584         {
 585             get
 586             {
 587 #pragma warning suppress 56503 // The contract for non seekable stream is to throw exception
 588                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new NotSupportedException());
 589             }
 590             set
 591             {
 592                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new NotSupportedException());
 593             }
 594         }
 595
 596         public override void Close()
 597         {
 598             Flush();
 599             base.Close();
 600             this.stream.Close();
 601         }
 602
 603         public override void Flush()
 604         {
 605             this.stream.Flush();
 606         }
 607
 608         public override int ReadByte()
 609         {
 610             if (byteCount == 0 && encodingCode == SupportedEncoding.UTF8)
 611                 return this.stream.ReadByte();
 612             if (Read(byteBuffer, 0, 1) == 0)
 613                 return -1;
 614             return byteBuffer[0];
 615         }
 616
 617         public override int Read(byte[] buffer, int offset, int count)
 618         {
 619             try
 620             {
 621                 if (byteCount == 0)
 622                 {
 623                     if (encodingCode == SupportedEncoding.UTF8)
 624                         return this.stream.Read(buffer, offset, count);
 625
 626                     // No more bytes than can be turned into characters
 627                     byteOffset = 0;
 628                     byteCount = this.stream.Read(bytes, byteCount, (chars.Length - 1) * 2);
 629
 630                     // Check for end of stream
 631                     if (byteCount == 0)
 632                         return 0;
 633
 634                     // Fix up incomplete chars
 635                     CleanupCharBreak();
 636
 637                     // Change encoding
 638                     int charCount = this.encoding.GetChars(bytes, 0, byteCount, chars, 0);
 639                     byteCount = Encoding.UTF8.GetBytes(chars, 0, charCount, bytes, 0);
 640                 }
 641
 642                 // Give them bytes
 643                 if (byteCount < count)
 644                     count = byteCount;
 645                 Buffer.BlockCopy(bytes, byteOffset, buffer, offset, count);
 646                 byteOffset += count;
 647                 byteCount -= count;
 648                 return count;
 649             }
 650             catch (DecoderFallbackException ex)
 651             {
 652                 throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.XmlInvalidBytes), ex));
 653             }
 654         }
 655
 656         void CleanupCharBreak()
 657         {
 658             int max = byteOffset + byteCount;
 659
 660             // Read on 2 byte boundaries
 661             if ((byteCount % 2) != 0)
 662             {
 663                 int b = this.stream.ReadByte();
 664                 if (b < 0)
 665                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.UnexpectedEndOfFile)));
 666
 667                 bytes[max++] = (byte)b;
 668                 byteCount++;
 669             }
 670
 671             // Don't cut off a surrogate character
 672             int w;
 673             if (encodingCode == SupportedEncoding.UTF16LE)
 674             {
 675                 w = bytes[max - 2] + (bytes[max - 1] << 8);
 676             }
 677             else
 678             {
 679                 w = bytes[max - 1] + (bytes[max - 2] << 8);
 680             }
 681             if ((w & 0xDC00) != 0xDC00 && w >= 0xD800 && w <= 0xDBFF)  // First 16-bit number of surrogate pair
 682             {
 683                 int b1 = this.stream.ReadByte();
 684                 int b2 = this.stream.ReadByte();
 685                 if (b2 < 0)
 686                     throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new XmlException(SR.GetString(SR.UnexpectedEndOfFile)));
 687                 bytes[max++] = (byte)b1;
 688                 bytes[max++] = (byte)b2;
 689                 byteCount += 2;
 690             }
 691         }
 692
 693         public override long Seek(long offset, SeekOrigin origin)
 694         {
 695             throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new NotSupportedException());
 696         }
 697
 698         public override void WriteByte(byte b)
 699         {
 700             if (encodingCode == SupportedEncoding.UTF8)
 701             {
 702                 this.stream.WriteByte(b);
 703                 return;
 704             }
 705             byteBuffer[0] = b;
 706             Write(byteBuffer, 0, 1);
 707         }
 708
 709         public override void Write(byte[] buffer, int offset, int count)
 710         {
 711             // Optimize UTF-8 case
 712             if (encodingCode == SupportedEncoding.UTF8)
 713             {
 714                 this.stream.Write(buffer, offset, count);
 715                 return;
 716             }
 717
 718             while (count > 0)
 719             {
 720                 int size = chars.Length < count ? chars.Length : count;
 721                 int charCount = dec.GetChars(buffer, offset, size, chars, 0, false);
 722                 byteCount = enc.GetBytes(chars, 0, charCount, bytes, 0, false);
 723                 this.stream.Write(bytes, 0, byteCount);
 724                 offset += size;
 725                 count -= size;
 726             }
 727         }
 728
 729         // Delegate properties
 730         public override bool CanTimeout { get { return this.stream.CanTimeout; } }
 731         public override long Length { get { return this.stream.Length; } }
 732         public override int ReadTimeout
 733         {
 734             get { return this.stream.ReadTimeout; }
 735             set { this.stream.ReadTimeout = value; }
 736         }
 737         public override int WriteTimeout
 738         {
 739             get { return this.stream.WriteTimeout; }
 740             set { this.stream.WriteTimeout = value; }
 741         }
 742
 743         // Delegate methods
 744         public override void SetLength(long value)
 745         {
 746             throw DiagnosticUtility.ExceptionUtility.ThrowHelperError(new NotSupportedException());
 747         }
 748     }
 749
 750     // Add format exceptions
 751     // Do we need to modify the stream position/Seek to account for the buffer?
 752     // ASSUMPTION ([....]): This class will only be used for EITHER reading OR writing.
 753 #if NO
 754     class UTF16Stream : Stream
 755     {
 756         const int BufferLength = 128;
 757
 758         Stream stream;
 759         bool bigEndian;
 760         byte[] streamBuffer;
 761         int streamOffset;
 762         int streamMax;
 763         byte[] trailBytes = new byte[4];
 764         int trailCount;
 765
 766         public UTF16Stream(Stream stream, bool bigEndian)
 767         {
 768             this.stream = stream;
 769             this.bigEndian = bigEndian;
 770             this.streamBuffer = byte[BufferLength];
 771         }
 772
 773         public override void Close()
 774         {
 775             Flush();
 776             base.Close();
 777             this.stream.Close();
 778         }
 779
 780         public override void Flush()
 781         {
 782             this.stream.Flush();
 783         }
 784
 785         public override int Read(byte[] buffer, int offset, int count)
 786         {
 787             // Validate args
 788
 789             // Read what we can if we aren't sure we have enough for a single character
 790             if (this.streamMax < 4)
 791                 this.streamMax += this.stream.Read(this.streamBuffer, streamOffset, streamBuffer.Length - this.streamMax);
 792
 793             int totalWritten = 0;
 794             while (streamOffset < streamMax && count > 0)
 795             {
 796                 int ch;
 797                 int read;
 798
 799                 read = ReadUTF16Char(out ch, streamBuffer, streamOffset, streamBuffer.Length - streamMax);
 800                 if (read == 0)
 801                     break;
 802
 803                 int written = WriteUTF8Char(ch, buffer, offset, count);
 804                 if (written == 0)
 805                     break;
 806
 807                 totalWritten += written;
 808                 streamOffset += read;
 809                 offset += written;
 810                 count -= written;
 811             }
 812
 813             // Shift down the leftover data
 814             if (this.streamOffset > 0 && this.streamOffset < this.streamMax)
 815             {
 816                 Buffer.BlockCopy(this.streamBuffer, this.streamOffset, this.streamBuffer, 0, this.streamMax - this.streamOffset);
 817                 this.streamMax -= this.streamOffset;
 818                 this.streamOffset = 0;
 819             }
 820
 821             return totalWritten;
 822         }
 823
 824         int ReadUTF8Char(out int ch, byte[] buffer, int offset, int count)
 825         {
 826             ch = -1;
 827             if (buffer[offset] < 0x80)
 828             {
 829                 ch = buffer[offset];
 830                 return 1;
 831             }
 832
 833             int mask = buffer[offset] & 0xF0;
 834             byte b1, b2, b3, b4;
 835             if (mask == 0xC0)
 836             {
 837                 if (count < 2)
 838                     return 0;
 839
 840                 b1 = buffer[offset + 0];
 841                 b2 = buffer[offset + 1];
 842
 843                 ch = ((b1 & 0x1F) << 6) + (b2 & 0x3F);
 844
 845                 return 2;
 846             }
 847             else if (mask == 0xE0)
 848             {
 849                 if (count < 3)
 850                     return 0;
 851
 852                 b1 = buffer[offset + 0];
 853                 b2 = buffer[offset + 1];
 854                 b3 = buffer[offset + 2];
 855
 856                 ch = ((((b1 & 0x0F) << 6) + (b2 & 0x3F)) << 6) + (b3 & 0x3F);
 857
 858                 return 3;
 859             }
 860             else if (mask == 0xF0)
 861             {
 862                 if (count < 4)
 863                     return 0;
 864
 865                 b1 = buffer[offset + 0];
 866                 b2 = buffer[offset + 1];
 867                 b3 = buffer[offset + 2];
 868                 b4 = buffer[offset + 3];
 869
 870                 ch = ((((((b1 & 0x0F) << 6) + (b2 & 0x3F)) << 6) + (b3 & 0x3F)) << 6) + (b4 & 0x3F);
 871
 872                 return 4;
 873             }
 874
 875             // Invalid
 876             return 0;
 877         }
 878
 879         int ReadUTF16Char(out int ch, byte[] buffer, int offset, int count)
 880         {
 881             ch = -1;
 882
 883             if (count < 2)
 884                 return 0;
 885
 886             int w1 = ReadEndian(buffer, offset);
 887
 888             if (w1 < 0xD800 || w1 > 0xDFFF)
 889             {
 890                 ch = w1;
 891                 return 2;
 892             }
 893
 894             if (count < 4)
 895                 return 0;
 896
 897             int w2 = ReadEndian(buffer, offset + 2);
 898
 899             ch = ((w1 & 0x03FF) << 10) + (w2 & 0x03FF);
 900             return 4;
 901         }
 902
 903         int ReadEndian(byte[] buffer, int offset)
 904         {
 905             if (bigEndian)
 906             {
 907                 return (buffer[offset + 0] << 8) + buffer[offset + 1];
 908             }
 909             else
 910             {
 911                 return (buffer[offset + 1] << 8) + buffer[offset + 0];
 912             }
 913         }
 914
 915         int WriteUTF8Char(int ch, byte[] buffer, int offset, int count)
 916         {
 917             if (ch < 0x80)
 918             {
 919                 buffer[offset] = (byte)ch;
 920                 return 1;
 921             }
 922             else if (ch < 0x800)
 923             {
 924                 if (count < 2)
 925                     return 0;
 926
 927                 buffer[offset + 1] = 0x80 | (ch & 0x3F);
 928                 ch >>= 6;
 929                 buffer[offset + 0] = 0xC0 | ch;
 930                 return 2
 931             }
 932             else if (ch < 0x10000)
 933             {
 934                 if (count < 3)
 935                     return 0;
 936
 937                 buffer[offset + 2] = 0x80 | (ch & 0x3F);
 938                 ch >>= 6;
 939                 buffer[offset + 1] = 0x80 | (ch & 0x3F);
 940                 ch >>= 6;
 941                 buffer[offset + 0] = 0xE0 | ch;
 942                 return 3;
 943             }
 944             else if (ch <= 0x110000)
 945             {
 946                 if (count < 4)
 947                     return 0;
 948                 buffer[offset + 3] = 0x80 | (ch & 0x3F);
 949                 ch >>= 6;
 950                 buffer[offset + 2] = 0x80 | (ch & 0x3F);
 951                 ch >>= 6;
 952                 buffer[offset + 1] = 0x80 | (ch & 0x3F);
 953                 ch >>= 6;
 954                 buffer[offset + 0] = 0xF0 | ch;
 955                 return 4;
 956             }
 957
 958             // Invalid?
 959             return 0;
 960         }
 961
 962         int WriteUTF16Char(int ch, byte[] buffer, int offset, int count)
 963         {
 964             if (ch < 0x10000)
 965             {
 966                 if (count < 2)
 967                     return 0;
 968
 969                 WriteEndian(ch, buffer, offset);
 970                 return 2;
 971             }
 972
 973             if (count < 4)
 974                 return 0;
 975
 976             ch -= 0x10000;
 977             int w2 = 0xDC00 | (ch & 0x03FF);
 978             int w1 = 0xD800 | ch >> 10;
 979             WriteEndian(w1, buffer, offset);
 980             WriteEndian(w2, buffer, offset + 2);
 981             return 4;
 982         }
 983
 984         void WriteEndian(int ch, byte[] buffer, int offset)
 985         {
 986             if (bigEndian)
 987             {
 988                 buffer[offset + 1] = (byte)ch;
 989                 buffer[offset + 0] = ch >> 8;
 990             }
 991             else
 992             {
 993                 buffer[offset + 0] = (byte)ch;
 994                 buffer[offset + 1] = ch >> 8;
 995             }
 996         }
 997
 998         public override void Write(byte[] buffer, int offset, int count)
 999         {
1000             // Validate args
1001
1002             // Write the trail bytes
1003             if (trailCount > 0)
1004             {
1005                 int free = 4-trailCount;
1006                 int total = (count < free ? count : free) + trialCount;
1007                 Buffer.BlockCopy(buffer, offset, trailBytes, trailCount, total);
1008
1009                 int c;
1010                 int r = ReadUTF8Char(out c, trailBuffer, 0, total);
1011                 if (r == 0 && count < free)
1012                 {
1013                     trailCount = total;
1014                     return;
1015                 }
1016
1017                 int diff = r - trailCount;
1018                 offset += diff;
1019                 count -= diff;
1020                 streamOffset = WriteUTF16Char(c, streamBuffer, 0, streamBuffer.Length - streamOffset);
1021             }
1022
1023             while (count > 0)
1024             {
1025                 if (streamBuffer.Length - streamOffset < 4)
1026                 {
1027                     this.stream.Write(streamBuffer, 0, streamOffset);
1028                     streamOffset = 0;
1029                 }
1030
1031                 int ch;
1032                 int read = ReadUTF8Char(out ch, buffer, offset, count);
1033                 if (read == 0)
1034                     break;
1035
1036                 int written = WriteUTF16Char(ch, streamBuffer, streamOffset, streamBuffer.Length - streamOffset);
1037                 if (written == 0)
1038                     break;
1039
1040                 streamOffset += written;
1041                 offset += read;
1042                 count -= read;
1043             }
1044
1045             if (streamOffset > 0)
1046             {
1047                 this.stream.Write(streamBuffer, 0, streamOffset);
1048                 streamOffset = 0;
1049             }
1050
1051             // Save trailing bytes
1052             if (count > 0)
1053             {
1054                 Buffer.BlockCopy(buffer, offset, trailBytes, 0, count);
1055                 trailCount = count;
1056             }
1057         }
1058
1059         // Delegate properties
1060         public override bool CanRead { get { return this.stream.CanRead; } }
1061         public override bool CanSeek { get { return this.stream.CanSeek; } }
1062         public override bool CanTimeout { get { return this.stream.CanTimeout; } }
1063         public override bool CanWrite { get { return this.stream.CanWrite; } }
1064         public override long Length { get { return this.stream.Length; } }
1065         public override long Position
1066         {
1067             get { return this.stream.Position; }
1068             set { this.stream.Position = value; }
1069         }
1070         public override int ReadTimeout
1071         {
1072             get { return this.stream.ReadTimeout; }
1073             set { this.stream.ReadTimeout = value; }
1074         }
1075         public override int WriteTimeout
1076         {
1077             get { return this.stream.WriteTimeout; }
1078             set { this.stream.WriteTimeout = value; }
1079         }
1080
1081         // Delegate methods
1082         public override long Seek(long offset, SeekOrigin origin)
1083         {
1084             return this.stream.Seek(offset, origin);
1085         }
1086
1087         public override void SetLength(long value)
1088         {
1089             this.stream.SetLength(value);
1090         }
1091     }
1092 #endif
1093 }