View Javadoc
1   /*
2    *  Licensed to the Apache Software Foundation (ASF) under one or more
3    *  contributor license agreements.  See the NOTICE file distributed with
4    *  this work for additional information regarding copyright ownership.
5    *  The ASF licenses this file to You under the Apache License, Version 2.0
6    *  (the "License"); you may not use this file except in compliance with
7    *  the License.  You may obtain a copy of the License at
8    *
9    *     http://www.apache.org/licenses/LICENSE-2.0
10   *
11   *  Unless required by applicable law or agreed to in writing, software
12   *  distributed under the License is distributed on an "AS IS" BASIS,
13   *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   *  See the License for the specific language governing permissions and
15   *  limitations under the License.
16   */
17  
18  package org.apache.commons.io.input;
19  
20  import java.io.BufferedInputStream;
21  import java.io.IOException;
22  import java.io.InputStream;
23  
24  import org.apache.commons.io.IOUtils;
25  import org.apache.commons.io.build.AbstractStreamBuilder;
26  
27  /**
28   * An unsynchronized version of {@link BufferedInputStream}, not thread-safe.
29   * <p>
30   * Wraps an existing {@link InputStream} and <em>buffers</em> the input. Expensive interaction with the underlying input stream is minimized, since most
31   * (smaller) requests can be satisfied by accessing the buffer alone. The drawback is that some extra space is required to hold the buffer and that copying
32   * takes place when filling that buffer, but this is usually outweighed by the performance benefits.
33   * </p>
34   * <p>
35   * To build an instance, use {@link Builder}.
36   * </p>
37   * <p>
38   * A typical application pattern for the class looks like this:
39   * </p>
40   *
41   * <pre>
42   * UnsynchronizedBufferedInputStream s = new UnsynchronizedBufferedInputStream.Builder().
43   *   .setInputStream(new FileInputStream(&quot;file.java&quot;))
44   *   .setBufferSize(8192)
45   *   .get();
46   * </pre>
47   * <p>
48   * Provenance: Apache Harmony and modified.
49   * </p>
50   *
51   * @see Builder
52   * @see BufferedInputStream
53   * @since 2.12.0
54   */
55  //@NotThreadSafe
56  public final class UnsynchronizedBufferedInputStream extends UnsynchronizedFilterInputStream {
57  
58      // @formatter:off
59      /**
60       * Builds a new {@link UnsynchronizedBufferedInputStream}.
61       *
62       * <p>
63       * Using File IO:
64       * </p>
65       * <pre>{@code
66       * UnsynchronizedBufferedInputStream s = UnsynchronizedBufferedInputStream.builder()
67       *   .setFile(file)
68       *   .setBufferSize(8192)
69       *   .get();}
70       * </pre>
71       * <p>
72       * Using NIO Path:
73       * </p>
74       * <pre>{@code
75       * UnsynchronizedBufferedInputStream s = UnsynchronizedBufferedInputStream.builder()
76       *   .setPath(path)
77       *   .setBufferSize(8192)
78       *   .get();}
79       * </pre>
80       *
81       * @see #get()
82       */
83      // @formatter:on
84      public static class Builder extends AbstractStreamBuilder<UnsynchronizedBufferedInputStream, Builder> {
85  
86          /**
87           * Constructs a builder of {@link UnsynchronizedBufferedInputStream}.
88           */
89          public Builder() {
90              // empty
91          }
92  
93          /**
94           * Builds a new {@link UnsynchronizedBufferedInputStream}.
95           * <p>
96           * You must set an aspect that supports {@link #getInputStream()} on this builder, otherwise, this method throws an exception.
97           * </p>
98           * <p>
99           * This builder uses the following aspects:
100          * </p>
101          * <ul>
102          * <li>{@link #getInputStream()}</li>
103          * <li>{@link #getBufferSize()}</li>
104          * </ul>
105          *
106          * @return a new instance.
107          * @throws IllegalStateException         if the {@code origin} is {@code null}.
108          * @throws UnsupportedOperationException if the origin cannot be converted to an {@link InputStream}.
109          * @throws IOException                   if an I/O error occurs converting to an {@link InputStream} using {@link #getInputStream()}.
110          * @see #getInputStream()
111          * @see #getBufferSize()
112          * @see #getUnchecked()
113          */
114         @Override
115         public UnsynchronizedBufferedInputStream get() throws IOException {
116             return new UnsynchronizedBufferedInputStream(getInputStream(), getBufferSize());
117         }
118 
119     }
120 
121     /**
122      * The buffer containing the current bytes read from the target InputStream.
123      */
124     protected volatile byte[] buffer;
125 
126     /**
127      * The total number of bytes inside the byte array {@code buffer}.
128      */
129     protected int count;
130 
131     /**
132      * The current limit, which when passed, invalidates the current mark.
133      */
134     protected int markLimit;
135 
136     /**
137      * The currently marked position. -1 indicates no mark has been set or the mark has been invalidated.
138      */
139     protected int markPos = IOUtils.EOF;
140 
141     /**
142      * The current position within the byte array {@code buffer}.
143      */
144     protected int pos;
145 
146     /**
147      * Constructs a new {@code BufferedInputStream} on the {@link InputStream} {@code in}. The buffer size is specified by the parameter {@code size} and all
148      * reads are now filtered through this stream.
149      *
150      * @param in   the input stream the buffer reads from.
151      * @param size the size of buffer to allocate.
152      * @throws IllegalArgumentException if {@code size < 0}.
153      */
154     private UnsynchronizedBufferedInputStream(final InputStream in, final int size) {
155         super(in);
156         if (size <= 0) {
157             throw new IllegalArgumentException("Size must be > 0");
158         }
159         buffer = new byte[size];
160     }
161 
162     /**
163      * Returns the number of bytes that are available before this stream will block. This method returns the number of bytes available in the buffer plus those
164      * available in the source stream.
165      *
166      * @return the number of bytes available before blocking.
167      * @throws IOException if this stream is closed.
168      */
169     @Override
170     public int available() throws IOException {
171         final InputStream localIn = inputStream; // 'in' could be invalidated by close()
172         if (buffer == null || localIn == null) {
173             throw new IOException("Stream is closed");
174         }
175         return count - pos + localIn.available();
176     }
177 
178     /**
179      * Closes this stream. The source stream is closed and any resources associated with it are released.
180      *
181      * @throws IOException if an error occurs while closing this stream.
182      */
183     @Override
184     public void close() throws IOException {
185         buffer = null;
186         final InputStream localIn = inputStream;
187         inputStream = null;
188         if (localIn != null) {
189             localIn.close();
190         }
191     }
192 
193     private int fillBuffer(final InputStream localIn, byte[] localBuf) throws IOException {
194         if (markPos == IOUtils.EOF || pos - markPos >= markLimit) {
195             /* Mark position not set or exceeded readLimit */
196             final int result = localIn.read(localBuf);
197             if (result > 0) {
198                 markPos = IOUtils.EOF;
199                 pos = 0;
200                 count = result;
201             }
202             return result;
203         }
204         if (markPos == 0 && markLimit > localBuf.length) {
205             /* Increase buffer size to accommodate the readLimit */
206             int newLength = localBuf.length * 2;
207             if (newLength > markLimit) {
208                 newLength = markLimit;
209             }
210             final byte[] newbuf = new byte[newLength];
211             System.arraycopy(localBuf, 0, newbuf, 0, localBuf.length);
212             // Reassign buffer, which will invalidate any local references
213             // FIXME: what if buffer was null?
214             localBuf = buffer = newbuf;
215         } else if (markPos > 0) {
216             System.arraycopy(localBuf, markPos, localBuf, 0, localBuf.length - markPos);
217         }
218         // Set the new position and mark position
219         pos -= markPos;
220         count = markPos = 0;
221         final int bytesread = localIn.read(localBuf, pos, localBuf.length - pos);
222         count = bytesread <= 0 ? pos : pos + bytesread;
223         return bytesread;
224     }
225 
226     byte[] getBuffer() {
227         return buffer;
228     }
229 
230     /**
231      * Sets a mark position in this stream. The parameter {@code readLimit} indicates how many bytes can be read before a mark is invalidated. Calling
232      * {@code reset()} will reposition the stream back to the marked position if {@code readLimit} has not been surpassed. The underlying buffer may be
233      * increased in size to allow {@code readLimit} number of bytes to be supported.
234      *
235      * @param readLimit the number of bytes that can be read before the mark is invalidated.
236      * @see #reset()
237      */
238     @Override
239     public void mark(final int readLimit) {
240         markLimit = readLimit;
241         markPos = pos;
242     }
243 
244     /**
245      * Indicates whether {@code BufferedInputStream} supports the {@code mark()} and {@code reset()} methods.
246      *
247      * @return {@code true} for BufferedInputStreams.
248      * @see #mark(int)
249      * @see #reset()
250      */
251     @Override
252     public boolean markSupported() {
253         return true;
254     }
255 
256     /**
257      * Reads a single byte from this stream and returns it as an integer in the range from 0 to 255. Returns -1 if the end of the source string has been
258      * reached. If the internal buffer does not contain any available bytes then it is filled from the source stream and the first byte is returned.
259      *
260      * @return the byte read or -1 if the end of the source stream has been reached.
261      * @throws IOException if this stream is closed or another IOException occurs.
262      */
263     @Override
264     public int read() throws IOException {
265         // Use local refs since buf and in may be invalidated by an
266         // unsynchronized close()
267         byte[] localBuf = buffer;
268         final InputStream localIn = inputStream;
269         if (localBuf == null || localIn == null) {
270             throw new IOException("Stream is closed");
271         }
272 
273         /* Are there buffered bytes available? */
274         if (pos >= count && fillBuffer(localIn, localBuf) == IOUtils.EOF) {
275             return IOUtils.EOF; /* no, fill buffer */
276         }
277         // localBuf may have been invalidated by fillbuf
278         if (localBuf != buffer) {
279             localBuf = buffer;
280             if (localBuf == null) {
281                 throw new IOException("Stream is closed");
282             }
283         }
284 
285         /* Did filling the buffer fail with -1 (EOF)? */
286         if (count - pos > 0) {
287             return localBuf[pos++] & 0xFF;
288         }
289         return IOUtils.EOF;
290     }
291 
292     /**
293      * Reads at most {@code length} bytes from this stream and stores them in byte array {@code buffer} starting at offset {@code offset}. Returns the number of
294      * bytes actually read or -1 if no bytes were read and the end of the stream was encountered. If all the buffered bytes have been used, a mark has not been
295      * set and the requested number of bytes is larger than the receiver's buffer size, this implementation bypasses the buffer and simply places the results
296      * directly into {@code buffer}.
297      *
298      * @param dest the byte array in which to store the bytes read.
299      * @param offset the initial position in {@code buffer} to store the bytes read from this stream.
300      * @param length the maximum number of bytes to store in {@code buffer}.
301      * @return the number of bytes actually read or -1 if end of stream.
302      * @throws IndexOutOfBoundsException if {@code offset < 0} or {@code length < 0}, or if {@code offset + length} is greater than the size of {@code buffer}.
303      * @throws IOException               if the stream is already closed or another IOException occurs.
304      */
305     @Override
306     public int read(final byte[] dest, int offset, final int length) throws IOException {
307         // Use local ref since buf may be invalidated by an unsynchronized
308         // close()
309         byte[] localBuf = buffer;
310         if (localBuf == null) {
311             throw new IOException("Stream is closed");
312         }
313         // avoid int overflow
314         if (offset > dest.length - length || offset < 0 || length < 0) {
315             throw new IndexOutOfBoundsException();
316         }
317         if (length == 0) {
318             return 0;
319         }
320         final InputStream localIn = inputStream;
321         if (localIn == null) {
322             throw new IOException("Stream is closed");
323         }
324 
325         int required;
326         if (pos < count) {
327             /* There are bytes available in the buffer. */
328             final int copylength = count - pos >= length ? length : count - pos;
329             System.arraycopy(localBuf, pos, dest, offset, copylength);
330             pos += copylength;
331             if (copylength == length || localIn.available() == 0) {
332                 return copylength;
333             }
334             offset += copylength;
335             required = length - copylength;
336         } else {
337             required = length;
338         }
339 
340         while (true) {
341             final int read;
342             /*
343              * If we're not marked and the required size is greater than the buffer, simply read the bytes directly bypassing the buffer.
344              */
345             if (markPos == IOUtils.EOF && required >= localBuf.length) {
346                 read = localIn.read(dest, offset, required);
347                 if (read == IOUtils.EOF) {
348                     return required == length ? IOUtils.EOF : length - required;
349                 }
350             } else {
351                 if (fillBuffer(localIn, localBuf) == IOUtils.EOF) {
352                     return required == length ? IOUtils.EOF : length - required;
353                 }
354                 // localBuf may have been invalidated by fillBuffer()
355                 if (localBuf != buffer) {
356                     localBuf = buffer;
357                     if (localBuf == null) {
358                         throw new IOException("Stream is closed");
359                     }
360                 }
361 
362                 read = count - pos >= required ? required : count - pos;
363                 System.arraycopy(localBuf, pos, dest, offset, read);
364                 pos += read;
365             }
366             required -= read;
367             if (required == 0) {
368                 return length;
369             }
370             if (localIn.available() == 0) {
371                 return length - required;
372             }
373             offset += read;
374         }
375     }
376 
377     /**
378      * Resets this stream to the last marked location.
379      *
380      * @throws IOException if this stream is closed, no mark has been set or the mark is no longer valid because more than {@code readLimit} bytes have been
381      *                     read since setting the mark.
382      * @see #mark(int)
383      */
384     @Override
385     public void reset() throws IOException {
386         if (buffer == null) {
387             throw new IOException("Stream is closed");
388         }
389         if (IOUtils.EOF == markPos) {
390             throw new IOException("Mark has been invalidated");
391         }
392         pos = markPos;
393     }
394 
395     /**
396      * Skips {@code amount} number of bytes in this stream. Subsequent {@code read()}'s will not return these bytes unless {@code reset()} is used.
397      *
398      * @param amount the number of bytes to skip. {@code skip} does nothing and returns 0 if {@code amount} is less than zero.
399      * @return the number of bytes actually skipped.
400      * @throws IOException if this stream is closed or another IOException occurs.
401      */
402     @Override
403     public long skip(final long amount) throws IOException {
404         // Use local refs since buf and in may be invalidated by an
405         // unsynchronized close()
406         final byte[] localBuf = buffer;
407         final InputStream localIn = inputStream;
408         if (localBuf == null) {
409             throw new IOException("Stream is closed");
410         }
411         if (amount < 1) {
412             return 0;
413         }
414         if (localIn == null) {
415             throw new IOException("Stream is closed");
416         }
417 
418         if (count - pos >= amount) {
419             // (int count - int pos) here is always an int so amount is also in the int range if the above test is true.
420             // We can safely cast to int and avoid static analysis warnings.
421             pos += (int) amount;
422             return amount;
423         }
424         int read = count - pos;
425         pos = count;
426 
427         if (markPos != IOUtils.EOF && amount <= markLimit) {
428             if (fillBuffer(localIn, localBuf) == IOUtils.EOF) {
429                 return read;
430             }
431             if (count - pos >= amount - read) {
432                 // (int count - int pos) here is always an int so (amount - read) is also in the int range if the above test is true.
433                 // We can safely cast to int and avoid static analysis warnings.
434                 pos += (int) amount - read;
435                 return amount;
436             }
437             // Couldn't get all the bytes, skip what we read
438             read += count - pos;
439             pos = count;
440             return read;
441         }
442         return read + localIn.skip(amount - read);
443     }
444 }