mcs/class/corlib/System.IO/StreamReader.cs

   1 //
   2 // System.IO.StreamReader.cs
   3 //
   4 // Author:
   5 //   Dietmar Maurer (dietmar@ximian.com)
   6 //   Miguel de Icaza (miguel@ximian.com)
   7 //   Marek Safar (marek.safar@gmail.com)
   8 //
   9 // (C) Ximian, Inc.  http://www.ximian.com
  10 // Copyright (C) 2004 Novell (http://www.novell.com)
  11 //
  12
  13 //
  14 // Copyright (C) 2004 Novell, Inc (http://www.novell.com)
  15 //
  16 // Permission is hereby granted, free of charge, to any person obtaining
  17 // a copy of this software and associated documentation files (the
  18 // "Software"), to deal in the Software without restriction, including
  19 // without limitation the rights to use, copy, modify, merge, publish,
  20 // distribute, sublicense, and/or sell copies of the Software, and to
  21 // permit persons to whom the Software is furnished to do so, subject to
  22 // the following conditions:
  23 //
  24 // The above copyright notice and this permission notice shall be
  25 // included in all copies or substantial portions of the Software.
  26 //
  27 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  28 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  29 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
  30 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
  31 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
  32 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  33 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  34 //
  35
  36 using System;
  37 using System.Text;
  38 using System.Runtime.InteropServices;
  39
  40 namespace System.IO {
  41         [Serializable]
  42         [ComVisible (true)]
  43         public class StreamReader : TextReader {
  44
  45                 const int DefaultBufferSize = 1024;
  46                 const int DefaultFileBufferSize = 4096;
  47                 const int MinimumBufferSize = 128;
  48
  49                 //
  50                 // The input buffer
  51                 //
  52                 byte [] input_buffer;
  53
  54                 // Input buffer ready for recycling
  55                 static byte [] input_buffer_recycle;
  56                 static object input_buffer_recycle_lock = new object ();
  57
  58                 //
  59                 // The decoded buffer from the above input buffer
  60                 //
  61                 char [] decoded_buffer;
  62                 static char[] decoded_buffer_recycle;
  63
  64                 //
  65                 // Decoded bytes in decoded_buffer.
  66                 //
  67                 int decoded_count;
  68
  69                 //
  70                 // Current position in the decoded_buffer
  71                 //
  72                 int pos;
  73
  74                 //
  75                 // The buffer size that we are using
  76                 //
  77                 int buffer_size;
  78
  79                 int do_checks;
  80
  81                 Encoding encoding;
  82                 Decoder decoder;
  83
  84                 Stream base_stream;
  85                 bool mayBlock;
  86                 StringBuilder line_builder;
  87
  88                 private class NullStreamReader : StreamReader {
  89                         public override int Peek ()
  90                         {
  91                                 return -1;
  92                         }
  93
  94                         public override int Read ()
  95                         {
  96                                 return -1;
  97                         }
  98
  99                         public override int Read ([In, Out] char[] buffer, int index, int count)
 100                         {
 101                                 return 0;
 102                         }
 103
 104                         public override string ReadLine ()
 105                         {
 106                                 return null;
 107                         }
 108
 109                         public override string ReadToEnd ()
 110                         {
 111                                 return String.Empty;
 112                         }
 113
 114                         public override Stream BaseStream
 115                         {
 116                                 get { return Stream.Null; }
 117                         }
 118
 119                         public override Encoding CurrentEncoding
 120                         {
 121                                 get { return Encoding.Unicode; }
 122                         }
 123                 }
 124
 125                 public new static readonly StreamReader Null =  new NullStreamReader ();
 126
 127                 internal StreamReader() {}
 128
 129                 public StreamReader(Stream stream)
 130                         : this (stream, Encoding.UTF8Unmarked, true, DefaultBufferSize) { }
 131
 132                 public StreamReader(Stream stream, bool detectEncodingFromByteOrderMarks)
 133                         : this (stream, Encoding.UTF8Unmarked, detectEncodingFromByteOrderMarks, DefaultBufferSize) { }
 134
 135                 public StreamReader(Stream stream, Encoding encoding)
 136                         : this (stream, encoding, true, DefaultBufferSize) { }
 137
 138                 public StreamReader(Stream stream, Encoding encoding, bool detectEncodingFromByteOrderMarks)
 139                         : this (stream, encoding, detectEncodingFromByteOrderMarks, DefaultBufferSize) { }
 140
 141                 public StreamReader(Stream stream, Encoding encoding, bool detectEncodingFromByteOrderMarks, int bufferSize)
 142                 {
 143                         Initialize (stream, encoding, detectEncodingFromByteOrderMarks, bufferSize);
 144                 }
 145
 146                 public StreamReader(string path)
 147                         : this (path, Encoding.UTF8Unmarked, true, DefaultFileBufferSize) { }
 148
 149                 public StreamReader(string path, bool detectEncodingFromByteOrderMarks)
 150                         : this (path, Encoding.UTF8Unmarked, detectEncodingFromByteOrderMarks, DefaultFileBufferSize) { }
 151
 152                 public StreamReader(string path, Encoding encoding)
 153                         : this (path, encoding, true, DefaultFileBufferSize) { }
 154
 155                 public StreamReader(string path, Encoding encoding, bool detectEncodingFromByteOrderMarks)
 156                         : this (path, encoding, detectEncodingFromByteOrderMarks, DefaultFileBufferSize) { }
 157
 158                 public StreamReader(string path, Encoding encoding, bool detectEncodingFromByteOrderMarks, int bufferSize)
 159                 {
 160                         if (null == path)
 161                                 throw new ArgumentNullException("path");
 162                         if (String.Empty == path)
 163                                 throw new ArgumentException("Empty path not allowed");
 164                         if (path.IndexOfAny (Path.InvalidPathChars) != -1)
 165                                 throw new ArgumentException("path contains invalid characters");
 166                         if (null == encoding)
 167                                 throw new ArgumentNullException ("encoding");
 168                         if (bufferSize <= 0)
 169                                 throw new ArgumentOutOfRangeException ("bufferSize", "The minimum size of the buffer must be positive");
 170
 171                         Stream stream = (Stream) File.OpenRead (path);
 172                         Initialize (stream, encoding, detectEncodingFromByteOrderMarks, bufferSize);
 173                 }
 174
 175                 internal void Initialize (Stream stream, Encoding encoding, bool detectEncodingFromByteOrderMarks, int bufferSize)
 176                 {
 177                         if (null == stream)
 178                                 throw new ArgumentNullException ("stream");
 179                         if (null == encoding)
 180                                 throw new ArgumentNullException ("encoding");
 181                         if (!stream.CanRead)
 182                                 throw new ArgumentException ("Cannot read stream");
 183                         if (bufferSize <= 0)
 184                                 throw new ArgumentOutOfRangeException ("bufferSize", "The minimum size of the buffer must be positive");
 185
 186                         if (bufferSize < MinimumBufferSize)
 187                                 bufferSize = MinimumBufferSize;
 188
 189                         // since GetChars() might add flushed character, it
 190                         // should have additional char buffer for extra 1
 191                         // (probably 1 is ok, but might be insufficient. I'm not sure)
 192                         var decoded_buffer_size = encoding.GetMaxCharCount (bufferSize) + 1;
 193
 194                         //
 195                         // Instead of allocating a new default buffer use the
 196                         // last one if there is any available
 197                         //
 198                         if (bufferSize <= DefaultBufferSize && input_buffer_recycle != null) {
 199                                 lock (input_buffer_recycle_lock) {
 200                                         if (input_buffer_recycle != null) {
 201                                                 input_buffer = input_buffer_recycle;
 202                                                 input_buffer_recycle = null;
 203                                         }
 204
 205                                         if (decoded_buffer_recycle != null && decoded_buffer_size <= decoded_buffer_recycle.Length) {
 206                                                 decoded_buffer = decoded_buffer_recycle;
 207                                                 decoded_buffer_recycle = null;
 208                                         }
 209                                 }
 210                         }
 211
 212                         if (input_buffer == null)
 213                                 input_buffer = new byte [bufferSize];
 214                         else
 215                                 Array.Clear (input_buffer, 0, bufferSize);
 216
 217                         if (decoded_buffer == null)
 218                                 decoded_buffer = new char [decoded_buffer_size];
 219                         else
 220                                 Array.Clear (decoded_buffer, 0, decoded_buffer_size);
 221
 222                         base_stream = stream;
 223                         this.buffer_size = bufferSize;
 224                         this.encoding = encoding;
 225                         decoder = encoding.GetDecoder ();
 226
 227                         byte [] preamble = encoding.GetPreamble ();
 228                         do_checks = detectEncodingFromByteOrderMarks ? 1 : 0;
 229                         do_checks += (preamble.Length == 0) ? 0 : 2;
 230
 231                         decoded_count = 0;
 232                         pos = 0;
 233                 }
 234
 235                 public virtual Stream BaseStream
 236                 {
 237                         get {
 238                                 return base_stream;
 239                         }
 240                 }
 241
 242                 public virtual Encoding CurrentEncoding
 243                 {
 244                         get {
 245                                 if (encoding == null)
 246                                         throw new Exception ();
 247                                 return encoding;
 248                         }
 249                 }
 250
 251                 public bool EndOfStream {
 252                         get { return Peek () < 0; }
 253                 }
 254
 255                 public override void Close ()
 256                 {
 257                         Dispose (true);
 258                 }
 259
 260                 protected override void Dispose (bool disposing)
 261                 {
 262                         if (disposing && base_stream != null)
 263                                 base_stream.Close ();
 264
 265                         if (input_buffer != null && input_buffer.Length == DefaultBufferSize && input_buffer_recycle == null) {
 266                                 lock (input_buffer_recycle_lock) {
 267                                         if (input_buffer_recycle == null) {
 268                                                 input_buffer_recycle = input_buffer;
 269                                         }
 270
 271                                         if (decoded_buffer_recycle == null) {
 272                                                 decoded_buffer_recycle = decoded_buffer;
 273                                         }
 274                                 }
 275                         }
 276
 277                         input_buffer = null;
 278                         decoded_buffer = null;
 279                         encoding = null;
 280                         decoder = null;
 281                         base_stream = null;
 282                         base.Dispose (disposing);
 283                 }
 284
 285                 //
 286                 // Provides auto-detection of the encoding, as well as skipping over
 287                 // byte marks at the beginning of a stream.
 288                 //
 289                 int DoChecks (int count)
 290                 {
 291                         if ((do_checks & 2) == 2){
 292                                 byte [] preamble = encoding.GetPreamble ();
 293                                 int c = preamble.Length;
 294                                 if (count >= c){
 295                                         int i;
 296
 297                                         for (i = 0; i < c; i++)
 298                                                 if (input_buffer [i] != preamble [i])
 299                                                         break;
 300
 301                                         if (i == c)
 302                                                 return i;
 303                                 }
 304                         }
 305
 306                         if ((do_checks & 1) == 1){
 307                                 if (count < 2)
 308                                         return 0;
 309
 310                                 if (input_buffer [0] == 0xfe && input_buffer [1] == 0xff){
 311                                         this.encoding = Encoding.BigEndianUnicode;
 312                                         return 2;
 313                                 }
 314                                 if (input_buffer [0] == 0xff && input_buffer [1] == 0xfe && count < 4) {
 315                                         // If we don't have enough bytes we can't check for UTF32, so use Unicode
 316                                         this.encoding = Encoding.Unicode;
 317                                         return 2;
 318                                 }
 319
 320                                 if (count < 3)
 321                                         return 0;
 322
 323                                 if (input_buffer [0] == 0xef && input_buffer [1] == 0xbb && input_buffer [2] == 0xbf){
 324                                         this.encoding = Encoding.UTF8Unmarked;
 325                                         return 3;
 326                                 }
 327
 328                                 if (count < 4) {
 329                                         if (input_buffer [0] == 0xff && input_buffer [1] == 0xfe && input_buffer [2] != 0) {
 330                                                 this.encoding = Encoding.Unicode;
 331                                                 return 2;
 332                                         }
 333                                         return 0;
 334                                 }
 335
 336                                 if (input_buffer [0] == 0 && input_buffer [1] == 0
 337                                         && input_buffer [2] == 0xfe && input_buffer [3] == 0xff)
 338                                 {
 339                                         this.encoding = Encoding.BigEndianUTF32;
 340                                         return 4;
 341                                 }
 342
 343                                 if (input_buffer [0] == 0xff && input_buffer [1] == 0xfe) {
 344                                         if (input_buffer [2] == 0 && input_buffer[3] == 0) {
 345                                                 this.encoding = Encoding.UTF32;
 346                                                 return 4;
 347                                         }
 348
 349                                         this.encoding = Encoding.Unicode;
 350                                         return 2;
 351                                 }
 352                         }
 353
 354                         return 0;
 355                 }
 356
 357                 public void DiscardBufferedData ()
 358                 {
 359                         pos = decoded_count = 0;
 360                         mayBlock = false;
 361                         // Discard internal state of the decoder too.
 362                         decoder = encoding.GetDecoder ();
 363                 }
 364
 365                 // the buffer is empty, fill it again
 366                 private int ReadBuffer ()
 367                 {
 368                         pos = 0;
 369                         int cbEncoded = 0;
 370
 371                         // keep looping until the decoder gives us some chars
 372                         decoded_count = 0;
 373                         int parse_start = 0;
 374                         do
 375                         {
 376                                 cbEncoded = base_stream.Read (input_buffer, 0, buffer_size);
 377
 378                                 if (cbEncoded <= 0)
 379                                         return 0;
 380
 381                                 mayBlock = (cbEncoded < buffer_size);
 382                                 if (do_checks > 0){
 383                                         Encoding old = encoding;
 384                                         parse_start = DoChecks (cbEncoded);
 385                                         if (old != encoding){
 386                                                 int old_decoded_size = old.GetMaxCharCount (buffer_size) + 1;
 387                                                 int new_decoded_size = encoding.GetMaxCharCount (buffer_size) + 1;
 388                                                 if (old_decoded_size != new_decoded_size)
 389                                                         decoded_buffer = new char [new_decoded_size];
 390                                                 decoder = encoding.GetDecoder ();
 391                                         }
 392                                         do_checks = 0;
 393                                         cbEncoded -= parse_start;
 394                                 }
 395
 396                                 decoded_count += decoder.GetChars (input_buffer, parse_start, cbEncoded, decoded_buffer, 0);
 397                                 parse_start = 0;
 398                         } while (decoded_count == 0);
 399
 400                         return decoded_count;
 401                 }
 402
 403                 //
 404                 // Peek can block:
 405                 // http://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?FeedbackID=96484
 406                 //
 407                 public override int Peek ()
 408                 {
 409                         if (base_stream == null)
 410                                 throw new ObjectDisposedException ("StreamReader", "Cannot read from a closed StreamReader");
 411                         if (pos >= decoded_count && ReadBuffer () == 0)
 412                                 return -1;
 413
 414                         return decoded_buffer [pos];
 415                 }
 416
 417                 //
 418                 // Used internally by our console, as it previously depended on Peek() being a
 419                 // routine that would not block.
 420                 //
 421                 internal bool DataAvailable ()
 422                 {
 423                         return pos < decoded_count;
 424                 }
 425
 426                 public override int Read ()
 427                 {
 428                         if (base_stream == null)
 429                                 throw new ObjectDisposedException ("StreamReader", "Cannot read from a closed StreamReader");
 430                         if (pos >= decoded_count && ReadBuffer () == 0)
 431                                 return -1;
 432
 433                         return decoded_buffer [pos++];
 434                 }
 435
 436                 public override int Read ([In, Out] char[] buffer, int index, int count)
 437                 {
 438                         if (base_stream == null)
 439                                 throw new ObjectDisposedException ("StreamReader", "Cannot read from a closed StreamReader");
 440                         if (buffer == null)
 441                                 throw new ArgumentNullException ("buffer");
 442                         if (index < 0)
 443                                 throw new ArgumentOutOfRangeException ("index", "< 0");
 444                         if (count < 0)
 445                                 throw new ArgumentOutOfRangeException ("count", "< 0");
 446                         // re-ordered to avoid possible integer overflow
 447                         if (index > buffer.Length - count)
 448                                 throw new ArgumentException ("index + count > buffer.Length");
 449
 450                         int chars_read = 0;
 451                         while (count > 0)
 452                         {
 453                                 if (pos >= decoded_count && ReadBuffer () == 0)
 454                                         return chars_read > 0 ? chars_read : 0;
 455
 456                                 int cch = Math.Min (decoded_count - pos, count);
 457                                 Array.Copy (decoded_buffer, pos, buffer, index, cch);
 458                                 pos += cch;
 459                                 index += cch;
 460                                 count -= cch;
 461                                 chars_read += cch;
 462                                 if (mayBlock)
 463                                         break;
 464                         }
 465                         return chars_read;
 466                 }
 467
 468                 bool foundCR;
 469                 int FindNextEOL ()
 470                 {
 471                         char c = '\0';
 472                         for (; pos < decoded_count; pos++) {
 473                                 c = decoded_buffer [pos];
 474                                 if (c == '\n') {
 475                                         pos++;
 476                                         int res = (foundCR) ? (pos - 2) : (pos - 1);
 477                                         if (res < 0)
 478                                                 res = 0; // if a new buffer starts with a \n and there was a \r at
 479                                                         // the end of the previous one, we get here.
 480                                         foundCR = false;
 481                                         return res;
 482                                 } else if (foundCR) {
 483                                         foundCR = false;
 484                                         if (pos == 0)
 485                                                 return -2; // Need to flush the current buffered line.
 486                                                            // This is a \r at the end of the previous decoded buffer that
 487                                                            // is not followed by a \n in the current decoded buffer.
 488                                         return pos - 1;
 489                                 }
 490
 491                                 foundCR = (c == '\r');
 492                         }
 493
 494                         return -1;
 495                 }
 496
 497                 public override string ReadLine()
 498                 {
 499                         if (base_stream == null)
 500                                 throw new ObjectDisposedException ("StreamReader", "Cannot read from a closed StreamReader");
 501
 502                         if (pos >= decoded_count && ReadBuffer () == 0)
 503                                 return null;
 504
 505                         int begin = pos;
 506                         int end = FindNextEOL ();
 507                         if (end < decoded_count && end >= begin)
 508                                 return new string (decoded_buffer, begin, end - begin);
 509                         else if (end == -2)
 510                                 return line_builder.ToString (0, line_builder.Length);
 511
 512                         if (line_builder == null)
 513                                 line_builder = new StringBuilder ();
 514                         else
 515                                 line_builder.Length = 0;
 516
 517                         while (true) {
 518                                 if (foundCR) // don't include the trailing CR if present
 519                                         decoded_count--;
 520
 521                                 line_builder.Append (decoded_buffer, begin, decoded_count - begin);
 522                                 if (ReadBuffer () == 0) {
 523                                         if (line_builder.Capacity > 32768) {
 524                                                 StringBuilder sb = line_builder;
 525                                                 line_builder = null;
 526                                                 return sb.ToString (0, sb.Length);
 527                                         }
 528                                         return line_builder.ToString (0, line_builder.Length);
 529                                 }
 530
 531                                 begin = pos;
 532                                 end = FindNextEOL ();
 533                                 if (end < decoded_count && end >= begin) {
 534                                         line_builder.Append (decoded_buffer, begin, end - begin);
 535                                         if (line_builder.Capacity > 32768) {
 536                                                 StringBuilder sb = line_builder;
 537                                                 line_builder = null;
 538                                                 return sb.ToString (0, sb.Length);
 539                                         }
 540                                         return line_builder.ToString (0, line_builder.Length);
 541                                 } else if (end == -2)
 542                                         return line_builder.ToString (0, line_builder.Length);
 543                         }
 544                 }
 545
 546                 public override string ReadToEnd()
 547                 {
 548                         if (base_stream == null)
 549                                 throw new ObjectDisposedException ("StreamReader", "Cannot read from a closed StreamReader");
 550
 551                         StringBuilder text = new StringBuilder ();
 552
 553                         int size = decoded_buffer.Length;
 554                         char [] buffer = new char [size];
 555                         int len;
 556
 557                         while ((len = Read (buffer, 0, size)) > 0)
 558                                 text.Append (buffer, 0, len);
 559
 560                         return text.ToString ();
 561                 }
 562         }
 563 }