mcs/class/Mono.Posix/Mono.Unix/UnixEncoding.cs

   1 /*
   2  * Mono.Unix/UnixEncoding.cs
   3  *
   4  * Authors:
   5  *   Jonathan Pryor (jonpryor@vt.edu)
   6  *
   7  * Copyright (c) 2001, 2002  Southern Storm Software, Pty Ltd
   8  * Copyright (C) 2004 Novell, Inc (http://www.novell.com)
   9  * Copyright (C) 2005 Jonathan Pryor
  10  *
  11  * Permission is hereby granted, free of charge, to any person obtaining
  12  * a copy of this software and associated documentation files (the "Software"),
  13  * to deal in the Software without restriction, including without limitation
  14  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
  15  * and/or sell copies of the Software, and to permit persons to whom the
  16  * Software is furnished to do so, subject to the following conditions:
  17  *
  18  * The above copyright notice and this permission notice shall be included
  19  * in all copies or substantial portions of the Software.
  20  *
  21  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  22  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  23  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  24  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
  25  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
  26  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
  27  * OTHER DEALINGS IN THE SOFTWARE.
  28  */
  29
  30 namespace Mono.Unix
  31 {
  32
  33 using System;
  34 using System.Text;
  35
  36 [Serializable]
  37 public class UnixEncoding : Encoding
  38 {
  39         public static readonly Encoding Instance = new UnixEncoding ();
  40
  41         public static readonly char EscapeByte = '\u0000';
  42
  43         // Constructors.
  44         public UnixEncoding ()
  45         {
  46         }
  47
  48         // Internal version of "GetByteCount" which can handle a rolling
  49         // state between multiple calls to this method.
  50         private static int InternalGetByteCount (char[] chars, int index, int count, uint leftOver, bool flush)
  51         {
  52                 // Validate the parameters.
  53                 if (chars == null) {
  54                         throw new ArgumentNullException ("chars");
  55                 }
  56                 if (index < 0 || index > chars.Length) {
  57                         throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
  58                 }
  59                 if (count < 0 || count > (chars.Length - index)) {
  60                         throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
  61                 }
  62
  63                 // Determine the lengths of all characters.
  64                 char ch;
  65                 int length = 0;
  66                 uint pair = leftOver;
  67                 while (count > 0) {
  68                         ch = chars[index];
  69                         if (pair == 0) {
  70                                 if (ch == EscapeByte && count > 1) {
  71                                         ++length;
  72                                         ++index;
  73                                         --count;
  74                                 } else if (ch < '\u0080') {
  75                                         ++length;
  76                                 } else if (ch < '\u0800') {
  77                                         length += 2;
  78                                 } else if (ch >= '\uD800' && ch <= '\uDBFF') {
  79                                         // This is the start of a surrogate pair.
  80                                         pair = (uint)ch;
  81                                 } else {
  82                                         length += 3;
  83                                 }
  84                         } else if (ch >= '\uDC00' && ch <= '\uDFFF') {
  85                                 // We have a surrogate pair.
  86                                 length += 4;
  87                                 pair = 0;
  88                         } else {
  89                                 // We have a surrogate start followed by a
  90                                 // regular character.  Technically, this is
  91                                 // invalid, but we have to do something.
  92                                 // We write out the surrogate start and then
  93                                 // re-visit the current character again.
  94                                 length += 3;
  95                                 pair = 0;
  96                                 continue;
  97                         }
  98                         ++index;
  99                         --count;
 100                 }
 101                 if (flush && pair != 0) {
 102                         // Flush the left-over surrogate pair start.
 103                         length += 3;
 104                 }
 105
 106                 // Return the final length to the caller.
 107                 return length;
 108         }
 109
 110         // Get the number of bytes needed to encode a character buffer.
 111         public override int GetByteCount (char[] chars, int index, int count)
 112         {
 113                 return InternalGetByteCount (chars, index, count, 0, true);
 114         }
 115
 116         // Convenience wrappers for "GetByteCount".
 117         public override int GetByteCount (String s)
 118         {
 119                 // Validate the parameters.
 120                 if (s == null) {
 121                         throw new ArgumentNullException ("s");
 122                 }
 123
 124                 // Determine the lengths of all characters.
 125                 char ch;
 126                 int index = 0;
 127                 int count = s.Length;
 128                 int length = 0;
 129                 uint pair;
 130                 while (count > 0) {
 131                         ch = s[index++];
 132                         if (ch == EscapeByte && count > 1) {
 133                                 ++length;
 134                                 ++index;
 135                                 --count;
 136                         } else if (ch < '\u0080') {
 137                                 ++length;
 138                         } else if (ch < '\u0800') {
 139                                 length += 2;
 140                         } else if (ch >= '\uD800' && ch <= '\uDBFF' && count > 1) {
 141                                 // This may be the start of a surrogate pair.
 142                                 pair = (uint)(s[index]);
 143                                 if (pair >= (uint)0xDC00 && pair <= (uint)0xDFFF) {
 144                                         length += 4;
 145                                         ++index;
 146                                         --count;
 147                                 } else {
 148                                         length += 3;
 149                                 }
 150                         } else {
 151                                 length += 3;
 152                         }
 153                         --count;
 154                 }
 155
 156                 // Return the final length to the caller.
 157                 return length;
 158         }
 159
 160         // Internal version of "GetBytes" which can handle a rolling
 161         // state between multiple calls to this method.
 162         private static int InternalGetBytes (char[] chars, int charIndex,
 163                                              int charCount, byte[] bytes,
 164                                              int byteIndex, ref uint leftOver,
 165                                              bool flush)
 166         {
 167                 // Validate the parameters.
 168                 if (chars == null) {
 169                         throw new ArgumentNullException ("chars");
 170                 }
 171                 if (bytes == null) {
 172                         throw new ArgumentNullException ("bytes");
 173                 }
 174                 if (charIndex < 0 || charIndex > chars.Length) {
 175                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_Array"));
 176                 }
 177                 if (charCount < 0 || charCount > (chars.Length - charIndex)) {
 178                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_Array"));
 179                 }
 180                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 181                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 182                 }
 183
 184                 // Convert the characters into bytes.
 185                 char ch;
 186                 int length = bytes.Length;
 187                 uint pair;
 188                 uint left = leftOver;
 189                 int posn = byteIndex;
 190                 while (charCount > 0) {
 191                         // Fetch the next UTF-16 character pair value.
 192                         ch = chars[charIndex++];
 193                         --charCount;
 194                         if (left == 0) {
 195                                 if (ch >= '\uD800' && ch <= '\uDBFF') {
 196                                         // This is the start of a surrogate pair.
 197                                         left = (uint)ch;
 198                                         continue;
 199                                 } else if (ch == EscapeByte) {
 200                                         if (posn >= length) {
 201                                                 throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 202                                         }
 203                                         if (--charCount >= 0) {
 204                                                 bytes[posn++] = (byte) chars [charIndex++];
 205                                         }
 206                                         continue;
 207                                 } else {
 208                                         // This is a regular character.
 209                                         pair = (uint)ch;
 210                                 }
 211                         } else if (ch >= '\uDC00' && ch <= '\uDFFF') {
 212                                 // We have a surrogate pair.
 213                                 pair = ((left - (uint)0xD800) << 10) +
 214                                            (((uint)ch) - (uint)0xDC00) +
 215                                            (uint)0x10000;
 216                                 left = 0;
 217                         } else {
 218                                 // We have a surrogate start followed by a
 219                                 // regular character.  Technically, this is
 220                                 // invalid, but we have to do something.
 221                                 // We write out the surrogate start and then
 222                                 // re-visit the current character again.
 223                                 pair = (uint)left;
 224                                 left = 0;
 225                                 --charIndex;
 226                                 ++charCount;
 227                         }
 228
 229                         // Encode the character pair value.
 230                         if (pair < (uint)0x0080) {
 231                                 if (posn >= length) {
 232                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 233                                 }
 234                                 bytes[posn++] = (byte)pair;
 235                         } else if (pair < (uint)0x0800) {
 236                                 if ((posn + 2) > length) {
 237                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 238                                 }
 239                                 bytes[posn++] = (byte)(0xC0 | (pair >> 6));
 240                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 241                         } else if (pair < (uint)0x10000) {
 242                                 if ((posn + 3) > length) {
 243                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 244                                 }
 245                                 bytes[posn++] = (byte)(0xE0 | (pair >> 12));
 246                                 bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
 247                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 248                         } else {
 249                                 if ((posn + 4) > length) {
 250                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 251                                 }
 252                                 bytes[posn++] = (byte)(0xF0 | (pair >> 18));
 253                                 bytes[posn++] = (byte)(0x80 | ((pair >> 12) & 0x3F));
 254                                 bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
 255                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 256                         }
 257                 }
 258                 if (flush && left != 0) {
 259                         // Flush the left-over surrogate pair start.
 260                         if ((posn + 3) > length) {
 261                                 throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 262                         }
 263                         bytes[posn++] = (byte)(0xE0 | (left >> 12));
 264                         bytes[posn++] = (byte)(0x80 | ((left >> 6) & 0x3F));
 265                         bytes[posn++] = (byte)(0x80 | (left & 0x3F));
 266                         left = 0;
 267                 }
 268                 leftOver = left;
 269
 270                 // Return the final count to the caller.
 271                 return posn - byteIndex;
 272         }
 273
 274         // Get the bytes that result from encoding a character buffer.
 275         public override int GetBytes (char[] chars, int charIndex, int charCount,
 276                                                                  byte[] bytes, int byteIndex)
 277         {
 278                 uint leftOver = 0;
 279                 return InternalGetBytes (chars, charIndex, charCount, bytes, byteIndex, ref leftOver, true);
 280         }
 281
 282         // Convenience wrappers for "GetBytes".
 283         public override int GetBytes (String s, int charIndex, int charCount,
 284                                                                  byte[] bytes, int byteIndex)
 285         {
 286                 // Validate the parameters.
 287                 if (s == null) {
 288                         throw new ArgumentNullException ("s");
 289                 }
 290                 if (bytes == null) {
 291                         throw new ArgumentNullException ("bytes");
 292                 }
 293                 if (charIndex < 0 || charIndex > s.Length) {
 294                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_StringIndex"));
 295                 }
 296                 if (charCount < 0 || charCount > (s.Length - charIndex)) {
 297                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_StringRange"));
 298                 }
 299                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 300                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 301                 }
 302
 303                 unsafe {
 304                         fixed (char* p = s) {
 305                                 fixed (byte* b = bytes) {
 306                                         return GetBytes (p + charIndex, charCount, b + byteIndex, bytes.Length - byteIndex);
 307                                 }
 308                         }
 309                 }
 310         }
 311
 312         public unsafe override int GetBytes(char* chars, int charCount, byte* bytes, int byteCount)
 313         {
 314                 if (bytes == null || chars == null)
 315                         throw new ArgumentNullException (bytes == null ? "bytes" : "chars");
 316
 317                 if (charCount < 0 || byteCount < 0)
 318                         throw new ArgumentOutOfRangeException (charCount < 0 ? "charCount" : "byteCount");
 319
 320                 // Convert the characters into bytes.
 321                 char ch;
 322                 int length = byteCount;
 323                 uint pair;
 324                 int posn = 0;
 325                 int charIndex = 0;
 326                 while (charCount > 0) {
 327                         // Fetch the next UTF-16 character pair value.
 328                         ch = chars [charIndex++];
 329                         if (ch >= '\uD800' && ch <= '\uDBFF' && charCount > 1) {
 330                                 // This may be the start of a surrogate pair.
 331                                 pair = (uint)(chars[charIndex]);
 332                                 if (pair >= (uint)0xDC00 && pair <= (uint)0xDFFF) {
 333                                         pair = (pair - (uint)0xDC00) +
 334                                                    ((((uint)ch) - (uint)0xD800) << 10) +
 335                                                    (uint)0x10000;
 336                                         ++charIndex;
 337                                         --charCount;
 338                                 } else {
 339                                         pair = (uint)ch;
 340                                 }
 341                         } else if (ch == EscapeByte && charCount > 1) {
 342                                 if (posn >= length) {
 343                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 344                                 }
 345                                 charCount -= 2;
 346                                 if (charCount >= 0) {
 347                                         bytes[posn++] = (byte)chars [charIndex++];
 348                                 }
 349                                 continue;
 350                         } else {
 351                                 pair = (uint)ch;
 352                         }
 353                         --charCount;
 354
 355                         // Encode the character pair value.
 356                         if (pair < (uint)0x0080) {
 357                                 if (posn >= length) {
 358                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 359                                 }
 360                                 bytes[posn++] = (byte)pair;
 361                         } else if (pair < (uint)0x0800) {
 362                                 if ((posn + 2) > length) {
 363                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 364                                 }
 365                                 bytes[posn++] = (byte)(0xC0 | (pair >> 6));
 366                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 367                         } else if (pair < (uint)0x10000) {
 368                                 if ((posn + 3) > length) {
 369                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 370                                 }
 371                                 bytes[posn++] = (byte)(0xE0 | (pair >> 12));
 372                                 bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
 373                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 374                         } else {
 375                                 if ((posn + 4) > length) {
 376                                         throw new ArgumentException (_("Arg_InsufficientSpace"), "bytes");
 377                                 }
 378                                 bytes[posn++] = (byte)(0xF0 | (pair >> 18));
 379                                 bytes[posn++] = (byte)(0x80 | ((pair >> 12) & 0x3F));
 380                                 bytes[posn++] = (byte)(0x80 | ((pair >> 6) & 0x3F));
 381                                 bytes[posn++] = (byte)(0x80 | (pair & 0x3F));
 382                         }
 383                 }
 384
 385                 // Return the final count to the caller.
 386                 return posn;
 387         }
 388
 389         // Internal version of "GetCharCount" which can handle a rolling
 390         // state between multiple calls to this method.
 391         private static int InternalGetCharCount (byte[] bytes, int index, int count,
 392                                                                                    uint leftOverBits,
 393                                                                                    uint leftOverCount,
 394                                                                                    bool throwOnInvalid, bool flush)
 395         {
 396                 // Validate the parameters.
 397                 if (bytes == null) {
 398                         throw new ArgumentNullException ("bytes");
 399                 }
 400                 if (index < 0 || index > bytes.Length) {
 401                         throw new ArgumentOutOfRangeException ("index", _("ArgRange_Array"));
 402                 }
 403                 if (count < 0 || count > (bytes.Length - index)) {
 404                         throw new ArgumentOutOfRangeException ("count", _("ArgRange_Array"));
 405                 }
 406
 407                 // Determine the number of characters that we have.
 408                 int next_raw = 0;
 409                 uint ch;
 410                 int length = 0;
 411                 uint leftBits = leftOverBits;
 412                 uint leftSoFar = (leftOverCount & (uint)0x0F);
 413                 uint leftSize = ((leftOverCount >> 4) & (uint)0x0F);
 414                 while (count > 0) {
 415                         ch = (uint)(bytes [index++]);
 416                         ++next_raw;
 417                         --count;
 418                         if (leftSize == 0) {
 419                                 // Process a UTF-8 start character.
 420                                 if (ch < (uint)0x0080) {
 421                                         // Single-byte UTF-8 character.
 422                                         ++length;
 423                                         next_raw = 0;
 424                                 } else if ((ch & (uint)0xE0) == (uint)0xC0) {
 425                                         // Double-byte UTF-8 character.
 426                                         leftBits = (ch & (uint)0x1F);
 427                                         leftSoFar = 1;
 428                                         leftSize = 2;
 429                                 } else if ((ch & (uint)0xF0) == (uint)0xE0) {
 430                                         // Three-byte UTF-8 character.
 431                                         leftBits = (ch & (uint)0x0F);
 432                                         leftSoFar = 1;
 433                                         leftSize = 3;
 434                                 } else if ((ch & (uint)0xF8) == (uint)0xF0) {
 435                                         // Four-byte UTF-8 character.
 436                                         leftBits = (ch & (uint)0x07);
 437                                         leftSoFar = 1;
 438                                         leftSize = 4;
 439                                 } else if ((ch & (uint)0xFC) == (uint)0xF8) {
 440                                         // Five-byte UTF-8 character.
 441                                         leftBits = (ch & (uint)0x03);
 442                                         leftSoFar = 1;
 443                                         leftSize = 5;
 444                                 } else if ((ch & (uint)0xFE) == (uint)0xFC) {
 445                                         // Six-byte UTF-8 character.
 446                                         leftBits = (ch & (uint)0x03);
 447                                         leftSoFar = 1;
 448                                         leftSize = 6;
 449                                 } else {
 450                                         // Invalid UTF-8 start character.
 451                                         if (throwOnInvalid) {
 452                                                 // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 453                                         }
 454                                         length += next_raw*2;
 455                                         next_raw = 0;
 456                                 }
 457                         } else {
 458                                 // Process an extra byte in a multi-byte sequence.
 459                                 if ((ch & (uint)0xC0) == (uint)0x80) {
 460                                         leftBits = ((leftBits << 6) | (ch & (uint)0x3F));
 461                                         if (++leftSoFar >= leftSize) {
 462                                                 // We have a complete character now.
 463                                                 if (leftBits < (uint)0x10000) {
 464                                                         // is it an overlong ?
 465                                                         bool overlong = false;
 466                                                         switch (leftSize) {
 467                                                         case 2:
 468                                                                 overlong = (leftBits <= 0x7F);
 469                                                                 break;
 470                                                         case 3:
 471                                                                 overlong = (leftBits <= 0x07FF);
 472                                                                 break;
 473                                                         case 4:
 474                                                                 overlong = (leftBits <= 0xFFFF);
 475                                                                 break;
 476                                                         case 5:
 477                                                                 overlong = (leftBits <= 0x1FFFFF);
 478                                                                 break;
 479                                                         case 6:
 480                                                                 overlong = (leftBits <= 0x03FFFFFF);
 481                                                                 break;
 482                                                         }
 483                                                         if (overlong) {
 484                                                                 // if (throwOnInvalid)
 485                                                                 //      throw new ArgumentException (_("Overlong"), leftBits.ToString ());
 486                                                                 length += next_raw*2;
 487                                                         }
 488                                                         else
 489                                                                 ++length;
 490                                                 } else if (leftBits < (uint)0x110000) {
 491                                                         length += 2;
 492                                                 } else if (throwOnInvalid) {
 493                                                         // ???
 494                                                         // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 495                                                         length += next_raw*2;
 496                                                 }
 497                                                 leftSize = 0;
 498                                                 next_raw = 0;
 499                                         }
 500                                 } else {
 501                                         // Invalid UTF-8 sequence: clear and restart.
 502                                         if (throwOnInvalid) {
 503                                                 // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 504                                         }
 505                                         // don't escape the current byte, process it normally
 506                                         if (ch < (uint)0x0080) {
 507                                                 --index;
 508                                                 ++count;
 509                                                 --next_raw;
 510                                         }
 511                                         length += next_raw*2;
 512                                         leftSize = 0;
 513                                         next_raw = 0;
 514                                 }
 515                         }
 516                 }
 517                 if (flush && leftSize != 0 && throwOnInvalid) {
 518                         // We had left-over bytes that didn't make up
 519                         // a complete UTF-8 character sequence.
 520                         // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 521                         length += next_raw * 2;
 522                 }
 523
 524                 // Return the final length to the caller.
 525                 return length;
 526         }
 527
 528         // Get the number of characters needed to decode a byte buffer.
 529         public override int GetCharCount (byte[] bytes, int index, int count)
 530         {
 531                 return InternalGetCharCount (bytes, index, count, 0, 0, true, true);
 532         }
 533
 534         // Get the characters that result from decoding a byte buffer.
 535         private static int InternalGetChars (byte[] bytes, int byteIndex,
 536                                                                            int byteCount, char[] chars,
 537                                                                            int charIndex, ref uint leftOverBits,
 538                                                                            ref uint leftOverCount,
 539                                                                            bool throwOnInvalid, bool flush)
 540         {
 541                 // Validate the parameters.
 542                 if (bytes == null) {
 543                         throw new ArgumentNullException ("bytes");
 544                 }
 545                 if (chars == null) {
 546                         throw new ArgumentNullException ("chars");
 547                 }
 548                 if (byteIndex < 0 || byteIndex > bytes.Length) {
 549                         throw new ArgumentOutOfRangeException ("byteIndex", _("ArgRange_Array"));
 550                 }
 551                 if (byteCount < 0 || byteCount > (bytes.Length - byteIndex)) {
 552                         throw new ArgumentOutOfRangeException ("byteCount", _("ArgRange_Array"));
 553                 }
 554                 if (charIndex < 0 || charIndex > chars.Length) {
 555                         throw new ArgumentOutOfRangeException ("charIndex", _("ArgRange_Array"));
 556                 }
 557
 558                 if (charIndex == chars.Length)
 559                         return 0;
 560
 561                 // Convert the bytes into the output buffer.
 562                 byte[] raw = new byte[6];
 563                 int next_raw = 0;
 564                 uint ch;
 565                 int length = chars.Length;
 566                 int posn = charIndex;
 567                 uint leftBits = leftOverBits;
 568                 uint leftSoFar = (leftOverCount & (uint)0x0F);
 569                 uint leftSize = ((leftOverCount >> 4) & (uint)0x0F);
 570                 while (byteCount > 0) {
 571                         // Fetch the next character from the byte buffer.
 572                         ch = (uint)(bytes[byteIndex++]);
 573                         raw [next_raw++] = (byte) ch;
 574                         --byteCount;
 575                         if (leftSize == 0) {
 576                                 // Process a UTF-8 start character.
 577                                 if (ch < (uint)0x0080) {
 578                                         // Single-byte UTF-8 character.
 579                                         if (posn >= length) {
 580                                                 throw new ArgumentException (_("Arg_InsufficientSpace"), "chars");
 581                                         }
 582                                         next_raw = 0;
 583                                         chars[posn++] = (char)ch;
 584                                 } else if ((ch & (uint)0xE0) == (uint)0xC0) {
 585                                         // Double-byte UTF-8 character.
 586                                         leftBits = (ch & (uint)0x1F);
 587                                         leftSoFar = 1;
 588                                         leftSize = 2;
 589                                 } else if ((ch & (uint)0xF0) == (uint)0xE0) {
 590                                         // Three-byte UTF-8 character.
 591                                         leftBits = (ch & (uint)0x0F);
 592                                         leftSoFar = 1;
 593                                         leftSize = 3;
 594                                 } else if ((ch & (uint)0xF8) == (uint)0xF0) {
 595                                         // Four-byte UTF-8 character.
 596                                         leftBits = (ch & (uint)0x07);
 597                                         leftSoFar = 1;
 598                                         leftSize = 4;
 599                                 } else if ((ch & (uint)0xFC) == (uint)0xF8) {
 600                                         // Five-byte UTF-8 character.
 601                                         leftBits = (ch & (uint)0x03);
 602                                         leftSoFar = 1;
 603                                         leftSize = 5;
 604                                 } else if ((ch & (uint)0xFE) == (uint)0xFC) {
 605                                         // Six-byte UTF-8 character.
 606                                         leftBits = (ch & (uint)0x03);
 607                                         leftSoFar = 1;
 608                                         leftSize = 6;
 609                                 } else {
 610                                         // Invalid UTF-8 start character.
 611                                         if (throwOnInvalid) {
 612                                                 // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 613                                         }
 614                                         next_raw = 0;
 615                                         chars[posn++] = EscapeByte;
 616                                         chars[posn++] = (char) ch;
 617                                 }
 618                         } else {
 619                                 // Process an extra byte in a multi-byte sequence.
 620                                 if ((ch & (uint)0xC0) == (uint)0x80) {
 621                                         leftBits = ((leftBits << 6) | (ch & (uint)0x3F));
 622                                         if (++leftSoFar >= leftSize) {
 623                                                 // We have a complete character now.
 624                                                 if (leftBits < (uint)0x10000) {
 625                                                         // is it an overlong ?
 626                                                         bool overlong = false;
 627                                                         switch (leftSize) {
 628                                                         case 2:
 629                                                                 overlong = (leftBits <= 0x7F);
 630                                                                 break;
 631                                                         case 3:
 632                                                                 overlong = (leftBits <= 0x07FF);
 633                                                                 break;
 634                                                         case 4:
 635                                                                 overlong = (leftBits <= 0xFFFF);
 636                                                                 break;
 637                                                         case 5:
 638                                                                 overlong = (leftBits <= 0x1FFFFF);
 639                                                                 break;
 640                                                         case 6:
 641                                                                 overlong = (leftBits <= 0x03FFFFFF);
 642                                                                 break;
 643                                                         }
 644                                                         if (overlong) {
 645                                                                 // if (throwOnInvalid)
 646                                                                 //      throw new ArgumentException (_("Overlong"), leftBits.ToString ());
 647                                                                 CopyRaw (raw, ref next_raw, chars, ref posn, length);
 648                                                         }
 649                                                         else {
 650                                                                 if (posn >= length) {
 651                                                                         throw new ArgumentException
 652                                                                                 (_("Arg_InsufficientSpace"), "chars");
 653                                                                 }
 654                                                                 chars[posn++] = (char)leftBits;
 655                                                         }
 656                                                 } else if (leftBits < (uint)0x110000) {
 657                                                         if ((posn + 2) > length) {
 658                                                                 throw new ArgumentException
 659                                                                         (_("Arg_InsufficientSpace"), "chars");
 660                                                         }
 661                                                         leftBits -= (uint)0x10000;
 662                                                         chars[posn++] = (char)((leftBits >> 10) +
 663                                                                                                    (uint)0xD800);
 664                                                         chars[posn++] =
 665                                                                 (char)((leftBits & (uint)0x3FF) + (uint)0xDC00);
 666                                                 } else if (throwOnInvalid) {
 667                                                         // ???
 668                                                         // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 669                                                         CopyRaw (raw, ref next_raw, chars, ref posn, length);
 670                                                 }
 671                                                 leftSize = 0;
 672                                                 next_raw = 0;
 673                                         }
 674                                 } else {
 675                                         // Invalid UTF-8 sequence: clear and restart.
 676                                         if (throwOnInvalid) {
 677                                                 // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 678                                         }
 679                                         // don't escape the current byte, process it normally
 680                                         if (ch < (uint)0x0080) {
 681                                                 --byteIndex;
 682                                                 ++byteCount;
 683                                                 --next_raw;
 684                                         }
 685                                         CopyRaw (raw, ref next_raw, chars, ref posn, length);
 686                                         leftSize = 0;
 687                                         next_raw = 0;
 688                                 }
 689                         }
 690                 }
 691                 if (flush && leftSize != 0 && throwOnInvalid) {
 692                         // We had left-over bytes that didn't make up
 693                         // a complete UTF-8 character sequence.
 694                         // throw new ArgumentException (_("Arg_InvalidUTF8"), "bytes");
 695                         CopyRaw (raw, ref next_raw, chars, ref posn, length);
 696                 }
 697                 leftOverBits = leftBits;
 698                 leftOverCount = (leftSoFar | (leftSize << 4));
 699
 700                 // Return the final length to the caller.
 701                 return posn - charIndex;
 702         }
 703
 704         private static void CopyRaw (byte[] raw, ref int next_raw, char[] chars, ref int posn, int length)
 705         {
 706                 if (posn+(next_raw*2) > length)
 707                         throw new ArgumentException (_("Arg_InsufficientSpace"), "chars");
 708
 709                 for (int i = 0; i < next_raw; ++i) {
 710                         chars[posn++] = EscapeByte;
 711                         chars[posn++] = (char) raw [i];
 712                 }
 713
 714                 next_raw = 0;
 715         }
 716
 717         // Get the characters that result from decoding a byte buffer.
 718         public override int GetChars (byte[] bytes, int byteIndex, int byteCount,
 719                                                                  char[] chars, int charIndex)
 720         {
 721                 uint leftOverBits = 0;
 722                 uint leftOverCount = 0;
 723                 return InternalGetChars (bytes, byteIndex, byteCount, chars,
 724                                 charIndex, ref leftOverBits, ref leftOverCount, true, true);
 725         }
 726
 727         // Get the maximum number of bytes needed to encode a
 728         // specified number of characters.
 729         public override int GetMaxByteCount (int charCount)
 730         {
 731                 if (charCount < 0) {
 732                         throw new ArgumentOutOfRangeException ("charCount", _("ArgRange_NonNegative"));
 733                 }
 734                 return charCount * 4;
 735         }
 736
 737         // Get the maximum number of characters needed to decode a
 738         // specified number of bytes.
 739         public override int GetMaxCharCount (int byteCount)
 740         {
 741                 if (byteCount < 0) {
 742                         throw new ArgumentOutOfRangeException ("byteCount", _("ArgRange_NonNegative"));
 743                 }
 744                 return byteCount;
 745         }
 746
 747         // Get a Unix-specific decoder that is attached to this instance.
 748         public override Decoder GetDecoder ()
 749         {
 750                 return new UnixDecoder ();
 751         }
 752
 753         // Get a Unix-specific encoder that is attached to this instance.
 754         public override Encoder GetEncoder ()
 755         {
 756                 return new UnixEncoder ();
 757         }
 758
 759         // Get the Unix preamble.
 760         public override byte[] GetPreamble ()
 761         {
 762                 return new byte [0];
 763         }
 764
 765         // Determine if this object is equal to another.
 766         public override bool Equals (Object value)
 767         {
 768                 UnixEncoding enc = (value as UnixEncoding);
 769                 if (enc != null) {
 770                         return true;
 771                 }
 772                 else {
 773                         return false;
 774                 }
 775         }
 776
 777         // Get the hash code for this object.
 778         public override int GetHashCode ()
 779         {
 780                 return base.GetHashCode ();
 781         }
 782
 783         public override byte [] GetBytes (String s)
 784         {
 785                 if (s == null)
 786                         throw new ArgumentNullException ("s");
 787
 788                 int length = GetByteCount (s);
 789                 byte [] bytes = new byte [length];
 790                 GetBytes (s, 0, s.Length, bytes, 0);
 791                 return bytes;
 792         }
 793
 794         // Unix decoder implementation.
 795         [Serializable]
 796         private class UnixDecoder : Decoder
 797         {
 798                 private uint leftOverBits;
 799                 private uint leftOverCount;
 800
 801                 // Constructor.
 802                 public UnixDecoder ()
 803                 {
 804                         leftOverBits = 0;
 805                         leftOverCount = 0;
 806                 }
 807
 808                 // Override inherited methods.
 809                 public override int GetCharCount (byte[] bytes, int index, int count)
 810                 {
 811                         return InternalGetCharCount (bytes, index, count,
 812                                         leftOverBits, leftOverCount, true, false);
 813                 }
 814                 public override int GetChars (byte[] bytes, int byteIndex,
 815                                                  int byteCount, char[] chars, int charIndex)
 816                 {
 817                         return InternalGetChars (bytes, byteIndex, byteCount,
 818                                 chars, charIndex, ref leftOverBits, ref leftOverCount, true, false);
 819                 }
 820
 821         }
 822
 823         // Unix encoder implementation.
 824         [Serializable]
 825         private class UnixEncoder : Encoder
 826         {
 827                 private uint leftOver;
 828
 829                 // Constructor.
 830                 public UnixEncoder ()
 831                 {
 832                         leftOver = 0;
 833                 }
 834
 835                 // Override inherited methods.
 836                 public override int GetByteCount (char[] chars, int index,
 837                                          int count, bool flush)
 838                 {
 839                         return InternalGetByteCount (chars, index, count, leftOver, flush);
 840                 }
 841                 public override int GetBytes (char[] chars, int charIndex,
 842                                          int charCount, byte[] bytes, int byteCount, bool flush)
 843                 {
 844                         int result;
 845                         result = InternalGetBytes (chars, charIndex, charCount, bytes, byteCount, ref leftOver, flush);
 846                         return result;
 847                 }
 848         }
 849
 850         private static string _ (string arg)
 851         {
 852                 return arg;
 853         }
 854 }
 855 }
 856