GNU Classpath (0.97.2) | |
Frames | No Frames |
1: /* BufferedReader.java 2: Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005 3: Free Software Foundation, Inc. 4: 5: This file is part of GNU Classpath. 6: 7: GNU Classpath is free software; you can redistribute it and/or modify 8: it under the terms of the GNU General Public License as published by 9: the Free Software Foundation; either version 2, or (at your option) 10: any later version. 11: 12: GNU Classpath is distributed in the hope that it will be useful, but 13: WITHOUT ANY WARRANTY; without even the implied warranty of 14: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 15: General Public License for more details. 16: 17: You should have received a copy of the GNU General Public License 18: along with GNU Classpath; see the file COPYING. If not, write to the 19: Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 20: 02110-1301 USA. 21: 22: Linking this library statically or dynamically with other modules is 23: making a combined work based on this library. Thus, the terms and 24: conditions of the GNU General Public License cover the whole 25: combination. 26: 27: As a special exception, the copyright holders of this library give you 28: permission to link this library with independent modules to produce an 29: executable, regardless of the license terms of these independent 30: modules, and to copy and distribute the resulting executable under 31: terms of your choice, provided that you also meet, for each linked 32: independent module, the terms and conditions of the license of that 33: module. An independent module is a module which is not derived from 34: or based on this library. If you modify this library, you may extend 35: this exception to your version of the library, but you are not 36: obligated to do so. If you do not wish to do so, delete this 37: exception statement from your version. */ 38: 39: 40: package java.io; 41: 42: /* Written using "Java Class Libraries", 2nd edition, plus online 43: * API docs for JDK 1.2 beta from http://www.javasoft.com. 44: * Status: Believed complete and correct. 45: */ 46: 47: /** 48: * This subclass of <code>FilterReader</code> buffers input from an 49: * underlying implementation to provide a possibly more efficient read 50: * mechanism. It maintains the buffer and buffer state in instance 51: * variables that are available to subclasses. The default buffer size 52: * of 8192 chars can be overridden by the creator of the stream. 53: * <p> 54: * This class also implements mark/reset functionality. It is capable 55: * of remembering any number of input chars, to the limits of 56: * system memory or the size of <code>Integer.MAX_VALUE</code> 57: * 58: * @author Per Bothner (bothner@cygnus.com) 59: * @author Aaron M. Renn (arenn@urbanophile.com) 60: */ 61: public class BufferedReader extends Reader 62: { 63: Reader in; 64: char[] buffer; 65: /* Index of current read position. Must be >= 0 and <= limit. */ 66: /* There is a special case where pos may be equal to limit+1; this 67: * is used as an indicator that a readLine was done with a '\r' was 68: * the very last char in the buffer. Since we don't want to read-ahead 69: * and potentially block, we set pos this way to indicate the situation 70: * and deal with it later. Doing it this way rather than having a 71: * separate boolean field to indicate the condition has the advantage 72: * that it is self-clearing on things like mark/reset. 73: */ 74: int pos; 75: /* Limit of valid data in buffer. Must be >= pos and <= buffer.length. */ 76: /* This can be < pos in the one special case described above. */ 77: int limit; 78: 79: /* The value -1 means there is no mark, or the mark has been invalidated. 80: Otherwise, markPos is the index in the buffer of the marked position. 81: Must be >= 0 and <= pos. 82: Note we do not explicitly store the read-limit. 83: The implicit read-limit is (buffer.length - markPos), which is 84: guaranteed to be >= the read-limit requested in the call to mark. */ 85: int markPos = -1; 86: 87: // The JCL book specifies the default buffer size as 8K characters. 88: // This is package-private because it is used by LineNumberReader. 89: static final int DEFAULT_BUFFER_SIZE = 8192; 90: 91: /** 92: * Create a new <code>BufferedReader</code> that will read from the 93: * specified subordinate stream with a default buffer size of 8192 chars. 94: * 95: * @param in The subordinate stream to read from 96: */ 97: public BufferedReader(Reader in) 98: { 99: this(in, DEFAULT_BUFFER_SIZE); 100: } 101: 102: /** 103: * Create a new <code>BufferedReader</code> that will read from the 104: * specified subordinate stream with a buffer size that is specified by the 105: * caller. 106: * 107: * @param in The subordinate stream to read from 108: * @param size The buffer size to use 109: * 110: * @exception IllegalArgumentException if size <= 0 111: */ 112: public BufferedReader(Reader in, int size) 113: { 114: super(in.lock); 115: if (size <= 0) 116: throw new IllegalArgumentException("Illegal buffer size: " + size); 117: this.in = in; 118: buffer = new char[size]; 119: } 120: 121: /** 122: * This method closes the underlying stream and frees any associated 123: * resources. 124: * 125: * @exception IOException If an error occurs 126: */ 127: public void close() throws IOException 128: { 129: synchronized (lock) 130: { 131: if (in != null) 132: in.close(); 133: in = null; 134: buffer = null; 135: } 136: } 137: 138: /** 139: * Returns <code>true</code> to indicate that this class supports mark/reset 140: * functionality. 141: * 142: * @return <code>true</code> 143: */ 144: public boolean markSupported() 145: { 146: return true; 147: } 148: 149: /** 150: * Mark a position in the input to which the stream can be 151: * "reset" by calling the <code>reset()</code> method. The parameter 152: * <code>readLimit</code> is the number of chars that can be read from the 153: * stream after setting the mark before the mark becomes invalid. For 154: * example, if <code>mark()</code> is called with a read limit of 10, then 155: * when 11 chars of data are read from the stream before the 156: * <code>reset()</code> method is called, then the mark is invalid and the 157: * stream object instance is not required to remember the mark. 158: * <p> 159: * Note that the number of chars that can be remembered by this method 160: * can be greater than the size of the internal read buffer. It is also 161: * not dependent on the subordinate stream supporting mark/reset 162: * functionality. 163: * 164: * @param readLimit The number of chars that can be read before the mark 165: * becomes invalid 166: * 167: * @exception IOException If an error occurs 168: * @exception IllegalArgumentException if readLimit is negative. 169: */ 170: public void mark(int readLimit) throws IOException 171: { 172: if (readLimit < 0) 173: throw new IllegalArgumentException("Read-ahead limit is negative"); 174: 175: synchronized (lock) 176: { 177: checkStatus(); 178: // In this method we need to be aware of the special case where 179: // pos + 1 == limit. This indicates that a '\r' was the last char 180: // in the buffer during a readLine. We'll want to maintain that 181: // condition after we shift things around and if a larger buffer is 182: // needed to track readLimit, we'll have to make it one element 183: // larger to ensure we don't invalidate the mark too early, if the 184: // char following the '\r' is NOT a '\n'. This is ok because, per 185: // the spec, we are not required to invalidate when passing readLimit. 186: // 187: // Note that if 'pos > limit', then doing 'limit -= pos' will cause 188: // limit to be negative. This is the only way limit will be < 0. 189: 190: if (pos + readLimit > limit) 191: { 192: char[] old_buffer = buffer; 193: int extraBuffSpace = 0; 194: if (pos > limit) 195: extraBuffSpace = 1; 196: if (readLimit + extraBuffSpace > limit) 197: buffer = new char[readLimit + extraBuffSpace]; 198: limit -= pos; 199: if (limit >= 0) 200: { 201: System.arraycopy(old_buffer, pos, buffer, 0, limit); 202: pos = 0; 203: } 204: } 205: 206: if (limit < 0) 207: { 208: // Maintain the relationship of 'pos > limit'. 209: pos = 1; 210: limit = markPos = 0; 211: } 212: else 213: markPos = pos; 214: // Now pos + readLimit <= buffer.length. thus if we need to read 215: // beyond buffer.length, then we are allowed to invalidate markPos. 216: } 217: } 218: 219: /** 220: * Reset the stream to the point where the <code>mark()</code> method 221: * was called. Any chars that were read after the mark point was set will 222: * be re-read during subsequent reads. 223: * <p> 224: * This method will throw an IOException if the number of chars read from 225: * the stream since the call to <code>mark()</code> exceeds the mark limit 226: * passed when establishing the mark. 227: * 228: * @exception IOException If an error occurs; 229: */ 230: public void reset() throws IOException 231: { 232: synchronized (lock) 233: { 234: checkStatus(); 235: if (markPos < 0) 236: throw new IOException("mark never set or invalidated"); 237: 238: // Need to handle the extremely unlikely case where a readLine was 239: // done with a '\r' as the last char in the buffer; which was then 240: // immediately followed by a mark and a reset with NO intervening 241: // read of any sort. In that case, setting pos to markPos would 242: // lose that info and a subsequent read would thus not skip a '\n' 243: // (if one exists). The value of limit in this rare case is zero. 244: // We can assume that if limit is zero for other reasons, then 245: // pos is already set to zero and doesn't need to be readjusted. 246: if (limit > 0) 247: pos = markPos; 248: } 249: } 250: 251: /** 252: * This method determines whether or not a stream is ready to be read. If 253: * this method returns <code>false</code> then this stream could (but is 254: * not guaranteed to) block on the next read attempt. 255: * 256: * @return <code>true</code> if this stream is ready to be read, 257: * <code>false</code> otherwise 258: * 259: * @exception IOException If an error occurs 260: */ 261: public boolean ready() throws IOException 262: { 263: synchronized (lock) 264: { 265: checkStatus(); 266: return pos < limit || in.ready(); 267: } 268: } 269: 270: /** 271: * This method read chars from a stream and stores them into a caller 272: * supplied buffer. It starts storing the data at index 273: * <code>offset</code> into 274: * the buffer and attempts to read <code>len</code> chars. This method can 275: * return before reading the number of chars requested. The actual number 276: * of chars read is returned as an int. A -1 is returned to indicate the 277: * end of the stream. 278: * <p> 279: * This method will block until some data can be read. 280: * 281: * @param buf The array into which the chars read should be stored 282: * @param offset The offset into the array to start storing chars 283: * @param count The requested number of chars to read 284: * 285: * @return The actual number of chars read, or -1 if end of stream. 286: * 287: * @exception IOException If an error occurs. 288: * @exception IndexOutOfBoundsException If offset and count are not 289: * valid regarding buf. 290: */ 291: public int read(char[] buf, int offset, int count) throws IOException 292: { 293: if (offset < 0 || offset + count > buf.length || count < 0) 294: throw new IndexOutOfBoundsException(); 295: 296: synchronized (lock) 297: { 298: checkStatus(); 299: // Once again, we need to handle the special case of a readLine 300: // that has a '\r' at the end of the buffer. In this case, we'll 301: // need to skip a '\n' if it is the next char to be read. 302: // This special case is indicated by 'pos > limit'. 303: boolean retAtEndOfBuffer = false; 304: 305: int avail = limit - pos; 306: if (count > avail) 307: { 308: if (avail > 0) 309: count = avail; 310: else // pos >= limit 311: { 312: if (limit == buffer.length) 313: markPos = -1; // read too far - invalidate the mark. 314: if (pos > limit) 315: { 316: // Set a boolean and make pos == limit to simplify things. 317: retAtEndOfBuffer = true; 318: --pos; 319: } 320: if (markPos < 0) 321: { 322: // Optimization: can read directly into buf. 323: if (count >= buffer.length && !retAtEndOfBuffer) 324: return in.read(buf, offset, count); 325: pos = limit = 0; 326: } 327: avail = in.read(buffer, limit, buffer.length - limit); 328: if (retAtEndOfBuffer && avail > 0 && buffer[limit] == '\n') 329: { 330: --avail; 331: limit++; 332: } 333: if (avail < count) 334: { 335: if (avail <= 0) 336: return avail; 337: count = avail; 338: } 339: limit += avail; 340: } 341: } 342: System.arraycopy(buffer, pos, buf, offset, count); 343: pos += count; 344: return count; 345: } 346: } 347: 348: /* Read more data into the buffer. Update pos and limit appropriately. 349: Assumes pos==limit initially. May invalidate the mark if read too much. 350: Return number of chars read (never 0), or -1 on eof. */ 351: private int fill() throws IOException 352: { 353: checkStatus(); 354: // Handle the special case of a readLine that has a '\r' at the end of 355: // the buffer. In this case, we'll need to skip a '\n' if it is the 356: // next char to be read. This special case is indicated by 'pos > limit'. 357: boolean retAtEndOfBuffer = false; 358: if (pos > limit) 359: { 360: retAtEndOfBuffer = true; 361: --pos; 362: } 363: 364: if (markPos >= 0 && limit == buffer.length) 365: markPos = -1; 366: if (markPos < 0) 367: pos = limit = 0; 368: int count = in.read(buffer, limit, buffer.length - limit); 369: if (count > 0) 370: limit += count; 371: 372: if (retAtEndOfBuffer && buffer[pos] == '\n') 373: { 374: --count; 375: // If the mark was set to the location of the \n, then we 376: // must change it to fully pretend that the \n does not 377: // exist. 378: if (markPos == pos) 379: ++markPos; 380: ++pos; 381: } 382: 383: return count; 384: } 385: 386: public int read() throws IOException 387: { 388: synchronized (lock) 389: { 390: checkStatus(); 391: if (pos >= limit && fill () <= 0) 392: return -1; 393: return buffer[pos++]; 394: } 395: } 396: 397: /* Return the end of the line starting at this.pos and ending at limit. 398: * The index returns is *before* any line terminators, or limit 399: * if no line terminators were found. 400: */ 401: private int lineEnd(int limit) 402: { 403: int i = pos; 404: for (; i < limit; i++) 405: { 406: char ch = buffer[i]; 407: if (ch == '\n' || ch == '\r') 408: break; 409: } 410: return i; 411: } 412: 413: /** 414: * This method reads a single line of text from the input stream, returning 415: * it as a <code>String</code>. A line is terminated by "\n", a "\r", or 416: * an "\r\n" sequence. The system dependent line separator is not used. 417: * The line termination characters are not returned in the resulting 418: * <code>String</code>. 419: * 420: * @return The line of text read, or <code>null</code> if end of stream. 421: * 422: * @exception IOException If an error occurs 423: */ 424: public String readLine() throws IOException 425: { 426: checkStatus(); 427: // Handle the special case where a previous readLine (with no intervening 428: // reads/skips) had a '\r' at the end of the buffer. 429: // In this case, we'll need to skip a '\n' if it's the next char to be read. 430: // This special case is indicated by 'pos > limit'. 431: if (pos > limit) 432: { 433: int ch = read(); 434: if (ch < 0) 435: return null; 436: if (ch != '\n') 437: --pos; 438: } 439: int i = lineEnd(limit); 440: if (i < limit) 441: { 442: String str = String.valueOf(buffer, pos, i - pos); 443: pos = i + 1; 444: // If the last char in the buffer is a '\r', we must remember 445: // to check if the next char to be read after the buffer is refilled 446: // is a '\n'. If so, skip it. To indicate this condition, we set pos 447: // to be limit + 1, which normally is never possible. 448: if (buffer[i] == '\r') 449: if (pos == limit || buffer[pos] == '\n') 450: pos++; 451: return str; 452: } 453: StringBuilder sbuf = new StringBuilder(200); 454: sbuf.append(buffer, pos, i - pos); 455: pos = i; 456: // We only want to return null when no characters were read before 457: // EOF. So we must keep track of this separately. Otherwise we 458: // would treat an empty `sbuf' as an EOF condition, which is wrong 459: // when there is just a newline. 460: boolean eof = false; 461: for (;;) 462: { 463: // readLine should block. So we must not return until a -1 is reached. 464: if (pos >= limit) 465: { 466: // here count == 0 isn't sufficient to give a failure. 467: int count = fill(); 468: if (count < 0) 469: { 470: eof = true; 471: break; 472: } 473: continue; 474: } 475: int ch = buffer[pos++]; 476: if (ch == '\n' || ch == '\r') 477: { 478: // Check here if a '\r' was the last char in the buffer; if so, 479: // mark it as in the comment above to indicate future reads 480: // should skip a newline that is the next char read after 481: // refilling the buffer. 482: if (ch == '\r') 483: if (pos == limit || buffer[pos] == '\n') 484: pos++; 485: break; 486: } 487: i = lineEnd(limit); 488: sbuf.append(buffer, pos - 1, i - (pos - 1)); 489: pos = i; 490: } 491: return (sbuf.length() == 0 && eof) ? null : sbuf.toString(); 492: } 493: 494: /** 495: * This method skips the specified number of chars in the stream. It 496: * returns the actual number of chars skipped, which may be less than the 497: * requested amount. 498: * <p> 499: * This method first discards chars in the buffer, then calls the 500: * <code>skip</code> method on the underlying stream to skip the 501: * remaining chars. 502: * 503: * @param count The requested number of chars to skip 504: * 505: * @return The actual number of chars skipped. 506: * 507: * @exception IOException If an error occurs. 508: * @exception IllegalArgumentException If count is negative. 509: */ 510: public long skip(long count) throws IOException 511: { 512: synchronized (lock) 513: { 514: checkStatus(); 515: if (count < 0) 516: throw new IllegalArgumentException("skip value is negative"); 517: if (count == 0) 518: return 0; 519: // Yet again, we need to handle the special case of a readLine 520: // that has a '\r' at the end of the buffer. In this case, we need 521: // to ignore a '\n' if it is the next char to be read. 522: // This special case is indicated by 'pos > limit' (i.e. avail < 0). 523: // To simplify things, if we're dealing with the special case for 524: // readLine, just read the next char (since the fill method will 525: // skip the '\n' for us). By doing this, we'll have to back up pos. 526: // That's easier than trying to keep track of whether we've skipped 527: // one element or not. 528: if (pos > limit) 529: { 530: if (read() < 0) 531: return 0; 532: else 533: --pos; 534: } 535: 536: int avail = limit - pos; 537: 538: if (count < avail) 539: { 540: pos += count; 541: return count; 542: } 543: 544: pos = limit; 545: long todo = count - avail; 546: if (todo > buffer.length) 547: { 548: markPos = -1; 549: todo -= in.skip(todo); 550: } 551: else 552: { 553: while (todo > 0) 554: { 555: avail = fill(); 556: if (avail <= 0) 557: break; 558: if (avail > todo) 559: avail = (int) todo; 560: pos += avail; 561: todo -= avail; 562: } 563: } 564: return count - todo; 565: } 566: } 567: 568: private void checkStatus() throws IOException 569: { 570: if (in == null) 571: throw new IOException("Stream closed"); 572: } 573: }
GNU Classpath (0.97.2) |