annotate src/org/tmatesoft/hg/internal/InflaterDataAccess.java @ 575:8bf184c9d733

Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
author Artem Tikhomirov <tikhomirov.artem@gmail.com>
date Tue, 16 Apr 2013 16:59:59 +0200
parents 2a08466838d3
children 3c4db86e8c1f
rev   line source
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
1 /*
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
2 * Copyright (c) 2011-2012 TMate Software Ltd
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
3 *
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
4 * This program is free software; you can redistribute it and/or modify
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
5 * it under the terms of the GNU General Public License as published by
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
6 * the Free Software Foundation; version 2 of the License.
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
7 *
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
8 * This program is distributed in the hope that it will be useful,
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
11 * GNU General Public License for more details.
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
12 *
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
13 * For information on how to redistribute this software under
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
14 * the terms of a license other than GNU General Public License
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
15 * contact TMate Software at support@hg4j.com
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
16 */
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
17 package org.tmatesoft.hg.internal;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
18
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
19 import java.io.EOFException;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
20 import java.io.IOException;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
21 import java.util.zip.DataFormatException;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
22 import java.util.zip.Inflater;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
23 import java.util.zip.ZipException;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
24
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
25 /**
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
26 * DataAccess counterpart for InflaterInputStream.
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
27 * XXX is it really needed to be subclass of FilterDataAccess?
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
28 *
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
29 * @author Artem Tikhomirov
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
30 * @author TMate Software Ltd.
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
31 */
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
32 public class InflaterDataAccess extends FilterDataAccess {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
33
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
34 private final Inflater inflater;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
35 private final byte[] buffer;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
36 private final byte[] singleByte = new byte[1];
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
37 private int decompressedPos = 0;
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
38 private int decompressedLength;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
39
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
40 public InflaterDataAccess(DataAccess dataAccess, long offset, int compressedLength) {
263
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
41 this(dataAccess, offset, compressedLength, -1, new Inflater(), new byte[512]);
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
42 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
43
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
44 public InflaterDataAccess(DataAccess dataAccess, long offset, int compressedLength, int actualLength) {
263
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
45 this(dataAccess, offset, compressedLength, actualLength, new Inflater(), new byte[512]);
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
46 }
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
47
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
48 public InflaterDataAccess(DataAccess dataAccess, long offset, int compressedLength, int actualLength, Inflater inflater, byte[] buf) {
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
49 super(dataAccess, offset, compressedLength);
263
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
50 if (inflater == null || buf == null) {
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
51 throw new IllegalArgumentException();
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
52 }
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
53 this.inflater = inflater;
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
54 this.decompressedLength = actualLength;
263
31f67be94e71 RevlogStream - reduce number of object instances, reuse when possible
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 158
diff changeset
55 buffer = buf;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
56 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
57
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
58 @Override
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
59 public InflaterDataAccess reset() throws IOException {
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
60 super.reset();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
61 inflater.reset();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
62 decompressedPos = 0;
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
63 return this;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
64 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
65
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
66 @Override
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
67 protected int available() throws IOException {
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
68 return length() - decompressedPos;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
69 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
70
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
71 @Override
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
72 public boolean isEmpty() throws IOException {
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
73 // can't use super.available() <= 0 because even when 0 < super.count < 6(?)
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
74 // decompressedPos might be already == length()
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
75 return available() <= 0;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
76 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
77
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
78 @Override
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
79 public int length() throws IOException {
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
80 if (decompressedLength != -1) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
81 return decompressedLength;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
82 }
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
83 decompressedLength = 0; // guard to avoid endless loop in case length() would get invoked from below.
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
84 int c = 0;
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
85 int oldPos = decompressedPos;
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
86 byte[] dummy = new byte[buffer.length];
575
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
87 try {
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
88 int toRead = -1;
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
89 do {
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
90 while (!inflater.needsInput()) {
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
91 c += inflater.inflate(dummy, 0, dummy.length);
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
92 }
575
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
93 if (inflater.needsInput() && (toRead = super.available()) > 0) {
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
94 // fill:
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
95 if (toRead > buffer.length) {
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
96 toRead = buffer.length;
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
97 }
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
98 super.readBytes(buffer, 0, toRead);
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
99 inflater.setInput(buffer, 0, toRead);
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
100 }
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
101 } while(toRead > 0);
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
102 } catch (DataFormatException ex) {
8bf184c9d733 Issue 43: poor performance with InflaterDataAccess. Phase 1: test existing code, fix defects found
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 441
diff changeset
103 throw new IOException(ex.toString());
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
104 }
421
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
105 decompressedLength = c + oldPos;
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
106 reset();
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
107 seek(oldPos);
fdd7d756dea0 Allow IOException from DataAccess methods for subclasses with non-trivial implementations, to avoid exception dumps when inapropriate
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 420
diff changeset
108 return decompressedLength;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
109 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
110
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
111 @Override
158
b413b16d10a5 Integer offsets and file length explictly, rather than casts throughout code. Inflater may benefit from total length hint, but shall calculate it by its own if needed
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
112 public void seek(int localOffset) throws IOException {
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
113 if (localOffset < 0 /* || localOffset >= length() */) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
114 throw new IllegalArgumentException();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
115 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
116 if (localOffset >= decompressedPos) {
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
117 skip(localOffset - decompressedPos);
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
118 } else {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
119 reset();
420
6c22bdc0bdfd Respect long offsets in revlogs
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 399
diff changeset
120 skip(localOffset);
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
121 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
122 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
123
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
124 @Override
391
856517285256 Provide more details about exception
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 263
diff changeset
125 public void skip(final int bytesToSkip) throws IOException {
856517285256 Provide more details about exception
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 263
diff changeset
126 int bytes = bytesToSkip;
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
127 if (bytes < 0) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
128 bytes += decompressedPos;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
129 if (bytes < 0) {
391
856517285256 Provide more details about exception
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 263
diff changeset
130 throw new IOException(String.format("Underflow. Rewind past start of the slice. To skip:%d, decPos:%d, decLen:%d. Left:%d", bytesToSkip, decompressedPos, decompressedLength, bytes));
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
131 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
132 reset();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
133 // fall-through
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
134 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
135 while (!isEmpty() && bytes > 0) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
136 readByte();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
137 bytes--;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
138 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
139 if (bytes != 0) {
391
856517285256 Provide more details about exception
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 263
diff changeset
140 throw new IOException(String.format("Underflow. Rewind past end of the slice. To skip:%d, decPos:%d, decLen:%d. Left:%d", bytesToSkip, decompressedPos, decompressedLength, bytes));
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
141 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
142 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
143
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
144 @Override
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
145 public byte readByte() throws IOException {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
146 readBytes(singleByte, 0, 1);
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
147 return singleByte[0];
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
148 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
149
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
150 @Override
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
151 public void readBytes(byte[] b, int off, int len) throws IOException {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
152 try {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
153 int n;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
154 while (len > 0) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
155 while ((n = inflater.inflate(b, off, len)) == 0) {
399
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
156 // XXX few last bytes (checksum?) may be ignored by inflater, thus inflate may return 0 in
157
d5268ca7715b Merged branch wrap-data-access into default for resource-friendly data access. Updated API to promote that friendliness to clients (channels, not byte[]). More exceptions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 51
diff changeset
157 // perfectly legal conditions (when all data already expanded, but there are still some bytes
399
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
158 // in the input stream)
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
159 int toRead = -1;
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
160 if (inflater.needsInput() && (toRead = super.available()) > 0) {
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
161 // fill:
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
162 if (toRead > buffer.length) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
163 toRead = buffer.length;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
164 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
165 super.readBytes(buffer, 0, toRead);
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
166 inflater.setInput(buffer, 0, toRead);
399
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
167 } else {
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
168 // prevent hang up in this cycle if no more data is available, see Issue 25
fdc1db8f7f61 Issue 25: Underflow in InflaterDataAccess; test and fix for hang up when reading past end of compressed data (or zero-length data)
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 391
diff changeset
169 throw new EOFException(String.format("No more compressed data is available to satisfy request for %d bytes. [finished:%b, needDict:%b, needInp:%b, available:%d", len, inflater.finished(), inflater.needsDictionary(), inflater.needsInput(), toRead));
51
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
170 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
171 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
172 off += n;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
173 len -= n;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
174 decompressedPos += n;
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
175 if (len == 0) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
176 return; // filled
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
177 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
178 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
179 } catch (DataFormatException e) {
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
180 String s = e.getMessage();
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
181 throw new ZipException(s != null ? s : "Invalid ZLIB data format");
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
182 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
183 }
9429c7bd1920 Try DataAccess to reach revision data instead of plain byte arrays
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
184 }