annotate src/org/tmatesoft/hg/repo/HgManifest.java @ 304:85b8efde5586

Use memory-friendly set implementation to canonicalize filenames and nodeids
author Artem Tikhomirov <tikhomirov.artem@gmail.com>
date Wed, 21 Sep 2011 18:26:16 +0200
parents 650b45d290b1
children f9f3e9b67ccc
rev   line source
13
df8c67f3006a Basic manifest parsing to analyze what's in there
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 2
diff changeset
1 /*
74
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
2 * Copyright (c) 2010-2011 TMate Software Ltd
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
3 *
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
4 * This program is free software; you can redistribute it and/or modify
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
5 * it under the terms of the GNU General Public License as published by
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
6 * the Free Software Foundation; version 2 of the License.
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
7 *
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
8 * This program is distributed in the hope that it will be useful,
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
11 * GNU General Public License for more details.
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
12 *
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
13 * For information on how to redistribute this software under
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
14 * the terms of a license other than GNU General Public License
102
a3a2e5deb320 Updated contact address to support@hg4j.com
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 77
diff changeset
15 * contact TMate Software at support@hg4j.com
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
16 */
74
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
17 package org.tmatesoft.hg.repo;
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
18
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
19 import static org.tmatesoft.hg.repo.HgRepository.TIP;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
20
232
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
21 import java.io.ByteArrayOutputStream;
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: 51 102
diff changeset
22 import java.io.IOException;
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
23 import java.util.ArrayList;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
24 import java.util.Arrays;
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: 51 102
diff changeset
25
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: 51 102
diff changeset
26 import org.tmatesoft.hg.core.HgBadStateException;
74
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
27 import org.tmatesoft.hg.core.Nodeid;
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: 51 102
diff changeset
28 import org.tmatesoft.hg.internal.DataAccess;
266
0a2f445de774 Improve manifest parsing: reduce number of arrays instantiated for Nodeid
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 264
diff changeset
29 import org.tmatesoft.hg.internal.DigestHelper;
232
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
30 import org.tmatesoft.hg.internal.Experimental;
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
31 import org.tmatesoft.hg.internal.Lifecycle;
304
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
32 import org.tmatesoft.hg.internal.Pool2;
77
c677e1593919 Moved RevlogStream implementation into .internal
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 74
diff changeset
33 import org.tmatesoft.hg.internal.RevlogStream;
232
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
34 import org.tmatesoft.hg.util.Path;
74
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
35
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
36
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
37 /**
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
38 *
74
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
39 * @author Artem Tikhomirov
6f1b88693d48 Complete refactoring to org.tmatesoft
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 28
diff changeset
40 * @author TMate Software Ltd.
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
41 */
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
42 public class HgManifest extends Revlog {
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
43 private RevisionMapper revisionMap;
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
44
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
45 public enum Flags {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
46 Exec, Link;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
47
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
48 static Flags parse(String flags) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
49 if ("x".equalsIgnoreCase(flags)) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
50 return Exec;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
51 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
52 if ("l".equalsIgnoreCase(flags)) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
53 return Link;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
54 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
55 if (flags == null) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
56 return null;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
57 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
58 throw new IllegalStateException(flags);
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
59 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
60
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
61 static Flags parse(byte[] data, int start, int length) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
62 if (length == 0) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
63 return null;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
64 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
65 if (length == 1) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
66 if (data[start] == 'x') {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
67 return Exec;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
68 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
69 if (data[start] == 'l') {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
70 return Link;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
71 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
72 // FALL THROUGH
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
73 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
74 throw new IllegalStateException(new String(data, start, length));
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
75 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
76
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
77 String nativeString() {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
78 if (this == Exec) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
79 return "x";
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
80 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
81 if (this == Link) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
82 return "l";
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
83 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
84 throw new IllegalStateException(toString());
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
85 }
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
86 }
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
87
13
df8c67f3006a Basic manifest parsing to analyze what's in there
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 2
diff changeset
88 /*package-local*/ HgManifest(HgRepository hgRepo, RevlogStream content) {
21
e929cecae4e1 Refactor to move revlog content to base class
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 20
diff changeset
89 super(hgRepo, content);
13
df8c67f3006a Basic manifest parsing to analyze what's in there
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 2
diff changeset
90 }
df8c67f3006a Basic manifest parsing to analyze what's in there
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 2
diff changeset
91
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
92 /**
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
93 *
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
94 * @param start changelog (not manifest!) revision to begin with
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
95 * @param end changelog (not manifest!) revision to stop, inclusive.
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
96 * @param inspector can't be <code>null</code>
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
97 */
19
40532cdc92fc Inspector (visitor) for manifest
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 16
diff changeset
98 public void walk(int start, int end, final Inspector inspector) {
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
99 if (inspector == null) {
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
100 throw new IllegalArgumentException();
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
101 }
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
102 int start0 = fromChangelog(start);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
103 int end0 = fromChangelog(end);
300
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
104 if (end0 < start0) {
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
105 // there are tool-constructed repositories that got order of changeset revisions completely different from that of manifest
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
106 int x = end0;
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
107 end0 = start0;
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
108 start0 = x;
650b45d290b1 Share range check code
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 285
diff changeset
109 }
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
110 content.iterate(start0, end0, true, new ManifestParser(inspector));
19
40532cdc92fc Inspector (visitor) for manifest
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 16
diff changeset
111 }
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
112
254
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
113 /**
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
114 * "Sparse" iteration of the manifest
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
115 *
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
116 * @param inspector
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
117 * @param localRevisions local changeset revisions to visit
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
118 */
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
119 public void walk(final Inspector inspector, int... localRevisions) {
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
120 if (inspector == null || localRevisions == null) {
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
121 throw new IllegalArgumentException();
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
122 }
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
123 int[] manifestLocalRevs = new int[localRevisions.length];
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
124 boolean needsSort = false;
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
125 for (int i = 0; i < localRevisions.length; i++) {
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
126 final int manifestLocalRev = fromChangelog(localRevisions[i]);
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
127 manifestLocalRevs[i] = manifestLocalRev;
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
128 if (i > 0 && manifestLocalRevs[i-1] > manifestLocalRev) {
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
129 needsSort = true;
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
130 }
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
131 }
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
132 if (needsSort) {
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
133 Arrays.sort(manifestLocalRevs);
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
134 }
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
135 content.iterate(manifestLocalRevs, true, new ManifestParser(inspector));
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
136 }
a620f0663a37 Collect tags for a file - improve performance of 'sparse' manifest reads
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 232
diff changeset
137
232
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
138 // manifest revision number that corresponds to the given changeset
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
139 /*package-local*/ int fromChangelog(int revisionNumber) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
140 if (HgInternals.wrongLocalRevision(revisionNumber)) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
141 throw new IllegalArgumentException(String.valueOf(revisionNumber));
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
142 }
257
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
143 if (revisionNumber == HgRepository.WORKING_COPY || revisionNumber == HgRepository.BAD_REVISION) {
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
144 throw new IllegalArgumentException("Can't use constants like WORKING_COPY or BAD_REVISION");
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
145 }
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
146 // revisionNumber == TIP is processed by RevisionMapper
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
147 if (revisionMap == null) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
148 revisionMap = new RevisionMapper(getRepo());
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
149 content.iterate(0, TIP, false, revisionMap);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
150 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
151 return revisionMap.at(revisionNumber);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
152 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
153
232
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
154 /**
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
155 * Extracts file revision as it was known at the time of given changeset.
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
156 *
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
157 * @param revisionNumber local changeset index
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
158 * @param file path to file in question
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
159 * @return file revision or <code>null</code> if manifest at specified revision doesn't list such file
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
160 */
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
161 @Experimental(reason="Perhaps, HgDataFile shall own this method")
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
162 public Nodeid getFileRevision(int revisionNumber, final Path file) {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
163 int rev = fromChangelog(revisionNumber);
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
164 final Nodeid[] rv = new Nodeid[] { null };
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
165 content.iterate(rev, rev, true, new RevlogStream.Inspector() {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
166
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
167 public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess data) {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
168 ByteArrayOutputStream bos = new ByteArrayOutputStream();
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
169 try {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
170 byte b;
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
171 while (!data.isEmpty() && (b = data.readByte()) != '\n') {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
172 if (b != 0) {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
173 bos.write(b);
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
174 } else {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
175 String fname = new String(bos.toByteArray());
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
176 bos.reset();
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
177 if (file.toString().equals(fname)) {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
178 byte[] nid = new byte[40];
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
179 data.readBytes(nid, 0, 40);
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
180 rv[0] = Nodeid.fromAscii(nid, 0, 40);
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
181 break;
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
182 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
183 // else skip to the end of line
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
184 while (!data.isEmpty() && (b = data.readByte()) != '\n')
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
185 ;
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
186 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
187 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
188 } catch (IOException ex) {
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
189 throw new HgBadStateException(ex);
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
190 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
191 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
192 });
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
193 return rv[0];
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
194 }
b7347daa50e3 Allow to cat a file with changeset revision
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 222
diff changeset
195
19
40532cdc92fc Inspector (visitor) for manifest
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 16
diff changeset
196 public interface Inspector {
222
373e07cd3991 Due to discrepancy in manifest and changelor indexes, wrong manifest revisions got cached and compared
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 220
diff changeset
197 boolean begin(int mainfestRevision, Nodeid nid, int changelogRevision);
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
198 /**
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
199 * @deprecated switch to {@link Inspector2#next(Nodeid, Path, Flags)}
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
200 */
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
201 @Deprecated
19
40532cdc92fc Inspector (visitor) for manifest
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 16
diff changeset
202 boolean next(Nodeid nid, String fname, String flags);
222
373e07cd3991 Due to discrepancy in manifest and changelor indexes, wrong manifest revisions got cached and compared
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 220
diff changeset
203 boolean end(int manifestRevision);
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
204 }
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
205
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
206 @Experimental(reason="Explore Path alternative for filenames and enum for flags")
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
207 public interface Inspector2 extends Inspector {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
208 boolean next(Nodeid nid, Path fname, Flags flags);
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
209 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
210
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
211 /**
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
212 * When Pool uses Strings directly,
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
213 * ManifestParser creates new String instance with new char[] value, and does byte->char conversion.
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
214 * For cpython repo, walk(0..10k), there are over 16 million filenames, of them only 3020 unique.
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
215 * This means there are 15.9 million useless char[] instances and byte->char conversions
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
216 *
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
217 * When String (Path) is wrapped into {@link PathProxy}, there's extra overhead of byte[] representation
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
218 * of the String, but these are only for unique Strings (Paths) (3020 in the example above). Besides, I save
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
219 * useless char[] and byte->char conversions.
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
220 */
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
221 private static class PathProxy {
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
222 private byte[] data;
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
223 private int start;
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
224 private final int hash, length;
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
225 private Path result;
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
226
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
227 public PathProxy(byte[] data, int start, int length) {
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
228 this.data = data;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
229 this.start = start;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
230 this.length = length;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
231
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
232 // copy from String.hashCode(). In fact, not necessarily match result of String(data).hashCode
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
233 // just need some nice algorithm here
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
234 int h = 0;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
235 byte[] d = data;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
236 for (int i = 0, off = start, len = length; i < len; i++) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
237 h = 31 * h + d[off++];
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
238 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
239 hash = h;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
240 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
241
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
242 @Override
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
243 public boolean equals(Object obj) {
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
244 if (false == obj instanceof PathProxy) {
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
245 return false;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
246 }
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
247 PathProxy o = (PathProxy) obj;
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
248 if (o.result != null && result != null) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
249 return result.equals(o.result);
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
250 }
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
251 if (o.length != length || o.hash != hash) {
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
252 return false;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
253 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
254 for (int i = 0, x = o.start, y = start; i < length; i++) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
255 if (o.data[x++] != data[y++]) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
256 return false;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
257 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
258 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
259 return true;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
260 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
261 @Override
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
262 public int hashCode() {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
263 return hash;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
264 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
265
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
266 public Path freeze() {
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
267 if (result == null) {
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
268 result = Path.create(new String(data, start, length));
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
269 // release reference to bigger data array, make a copy of relevant part only
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
270 // use original bytes, not those from String above to avoid cache misses due to different encodings
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
271 byte[] d = new byte[length];
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
272 System.arraycopy(data, start, d, 0, length);
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
273 data = d;
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
274 start = 0;
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
275 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
276 return result;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
277 }
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
278 }
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
279
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
280 private static class ManifestParser implements RevlogStream.Inspector {
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
281 private boolean gtg = true; // good to go
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
282 private final Inspector inspector;
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
283 private final Inspector2 inspector2;
304
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
284 private Pool2<Nodeid> nodeidPool, thisRevPool;
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
285 private final Pool2<PathProxy> fnamePool;
266
0a2f445de774 Improve manifest parsing: reduce number of arrays instantiated for Nodeid
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 264
diff changeset
286 private byte[] nodeidLookupBuffer = new byte[20]; // get reassigned each time new Nodeid is added to pool
262
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
287
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
288 public ManifestParser(Inspector delegate) {
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
289 assert delegate != null;
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
290 inspector = delegate;
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
291 inspector2 = delegate instanceof Inspector2 ? (Inspector2) delegate : null;
304
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
292 nodeidPool = new Pool2<Nodeid>();
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
293 fnamePool = new Pool2<PathProxy>();
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
294 thisRevPool = new Pool2<Nodeid>();
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
295 }
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
296
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
297 public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess da) {
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
298 if (!gtg) {
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
299 return;
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
300 }
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
301 try {
222
373e07cd3991 Due to discrepancy in manifest and changelor indexes, wrong manifest revisions got cached and compared
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 220
diff changeset
302 gtg = gtg && inspector.begin(revisionNumber, new Nodeid(nodeid, true), linkRevision);
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
303 Path fname = null;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
304 Flags flags = null;
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
305 Nodeid nid = null;
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
306 int i;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
307 byte[] data = da.byteArray();
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
308 for (i = 0; gtg && i < actualLen; i++) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
309 int x = i;
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
310 for( ; data[i] != '\n' && i < actualLen; i++) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
311 if (fname == null && data[i] == 0) {
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
312 PathProxy px = fnamePool.unify(new PathProxy(data, x, i - x));
268
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
313 // if (cached = fnamePool.unify(px))== px then cacheMiss, else cacheHit
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
314 // cpython 0..10k: hits: 15 989 152, misses: 3020
c5980f287cc4 Use StringProxy when parsing manifest to minimize number of useless conversions and array instances
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 267
diff changeset
315 fname = px.freeze();
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
316 x = i+1;
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
317 }
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
318 }
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
319 if (i < actualLen) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
320 assert data[i] == '\n';
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
321 int nodeidLen = i - x < 40 ? i-x : 40; // if > 40, there are flags
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
322 DigestHelper.ascii2bin(data, x, nodeidLen, nodeidLookupBuffer); // ignore return value as it's unlikely to have NULL in manifest
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
323 nid = new Nodeid(nodeidLookupBuffer, false); // this Nodeid is for pool lookup only, mock object
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
324 Nodeid cached = nodeidPool.unify(nid);
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
325 if (cached == nid) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
326 // buffer now belongs to the cached nodeid
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
327 nodeidLookupBuffer = new byte[20];
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
328 } else {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
329 nid = cached; // use existing version, discard the lookup object
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
330 } // for cpython 0..10k, cache hits are 15 973 301, vs 18871 misses.
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
331 thisRevPool.record(nid); // memorize revision for the next iteration.
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
332 if (nodeidLen + x < i) {
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
333 // 'x' and 'l' for executable bits and symlinks?
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
334 // hg --debug manifest shows 644 for each regular file in my repo
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
335 // for cpython 0..10k, there are 4361062 flag checks, and there's only 1 unique flag
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
336 flags = Flags.parse(data, x + nodeidLen, i-x-nodeidLen);
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
337 } else {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
338 flags = null;
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
339 }
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
340 if (inspector2 == null) {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
341 String flagString = flags == null ? null : flags.nativeString();
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
342 gtg = inspector.next(nid, fname.toString(), flagString);
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
343 } else {
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
344 gtg = inspector2.next(nid, fname, flags);
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
345 }
267
ec921ef0628e Revert manifest parser changes - no single string and back to HashMap Pool
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 266
diff changeset
346 }
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
347 nid = null;
285
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
348 fname = null;
6dbbc53fc46d Use Path instead of plain String for manifest file names
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 268
diff changeset
349 flags = null;
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
350 }
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
351 gtg = gtg && inspector.end(revisionNumber);
262
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
352 //
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
353 // keep only actual file revisions, found at this version
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
354 // (next manifest is likely to refer to most of them, although in specific cases
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
355 // like commit in another branch a lot may be useless)
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
356 nodeidPool.clear();
304
85b8efde5586 Use memory-friendly set implementation to canonicalize filenames and nodeids
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 300
diff changeset
357 Pool2<Nodeid> t = nodeidPool;
262
3dcd3dd90c77 Improve manifest parsing: decode bytes to chars once, minimize arraycopy on String instantiation, keep set of file revisions from previous manifest only
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 257
diff changeset
358 nodeidPool = thisRevPool;
264
6bb5e7ed051a Optimize memory usage (reduce number of objects instantiated) when pooling file names and nodeids during manifest parsing
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 262
diff changeset
359 thisRevPool = t;
196
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
360 } catch (IOException ex) {
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
361 throw new HgBadStateException(ex);
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
362 }
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
363 }
e2115da4cf6a Pool objects to avoid memory polution with duplicates
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 157
diff changeset
364 }
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
365
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
366 private static class RevisionMapper implements RevlogStream.Inspector, Lifecycle {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
367
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
368 private final int changelogRevisions;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
369 private int[] changelog2manifest;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
370 private final HgRepository repo;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
371
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
372 public RevisionMapper(HgRepository hgRepo) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
373 repo = hgRepo;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
374 changelogRevisions = repo.getChangelog().getRevisionCount();
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
375 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
376
257
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
377 // respects TIP
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
378 public int at(int revisionNumber) {
257
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
379 if (revisionNumber == TIP) {
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
380 revisionNumber = changelogRevisions - 1;
f39fb6b3cc76 Handle revision constants correctly
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 254
diff changeset
381 }
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
382 if (changelog2manifest != null) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
383 return changelog2manifest[revisionNumber];
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
384 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
385 return revisionNumber;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
386 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
387
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
388 public void next(int revisionNumber, int actualLen, int baseRevision, int linkRevision, int parent1Revision, int parent2Revision, byte[] nodeid, DataAccess data) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
389 if (changelog2manifest != null) {
220
8de327242aa0 Basic information about branches
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 218
diff changeset
390 // next assertion is not an error, rather assumption check, which is too development-related to be explicit exception -
8de327242aa0 Basic information about branches
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 218
diff changeset
391 // I just wonder if there are manifests that have two entries pointing to single changeset. It seems unrealistic, though -
8de327242aa0 Basic information about branches
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 218
diff changeset
392 // changeset records one and only one manifest nodeid
8de327242aa0 Basic information about branches
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 218
diff changeset
393 assert changelog2manifest[linkRevision] == -1 : String.format("revision:%d, link:%d, already linked to revision:%d", revisionNumber, linkRevision, changelog2manifest[linkRevision]);
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
394 changelog2manifest[linkRevision] = revisionNumber;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
395 } else {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
396 if (revisionNumber != linkRevision) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
397 changelog2manifest = new int[changelogRevisions];
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
398 Arrays.fill(changelog2manifest, -1);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
399 for (int i = 0; i < revisionNumber; changelog2manifest[i] = i, i++)
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
400 ;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
401 changelog2manifest[linkRevision] = revisionNumber;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
402 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
403 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
404 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
405
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
406 public void start(int count, Callback callback, Object token) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
407 if (count != changelogRevisions) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
408 assert count < changelogRevisions; // no idea what to do if manifest has more revisions than changelog
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
409 // the way how manifest may contain more revisions than changelog, as I can imagine, is a result of
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
410 // some kind of an import tool (e.g. from SVN or CVS), that creates manifest and changelog independently.
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
411 // Note, it's pure guess, I didn't see such repository yet (although the way manifest revisions
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
412 // in cpython repo are numbered makes me think aforementioned way)
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
413 changelog2manifest = new int[changelogRevisions];
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
414 Arrays.fill(changelog2manifest, -1);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
415 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
416 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
417
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
418 public void finish(Object token) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
419 if (changelog2manifest == null) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
420 return;
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
421 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
422 // I assume there'd be not too many revisions we don't know manifest of
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
423 ArrayList<Integer> undefinedChangelogRevision = new ArrayList<Integer>();
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
424 for (int i = 0; i < changelog2manifest.length; i++) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
425 if (changelog2manifest[i] == -1) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
426 undefinedChangelogRevision.add(i);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
427 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
428 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
429 for (int u : undefinedChangelogRevision) {
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
430 Nodeid manifest = repo.getChangelog().range(u, u).get(0).manifest();
220
8de327242aa0 Basic information about branches
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 218
diff changeset
431 // FIXME calculate those missing effectively (e.g. cache and sort nodeids to speed lookup
218
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
432 // right away in the #next (may refactor ParentWalker's sequential and sorted into dedicated helper and reuse here)
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
433 changelog2manifest[u] = repo.getManifest().getLocalRevision(manifest);
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
434 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
435 }
047b1dec7a04 Issue 7: Correctly handle manifest and changelog with different number of (or non-matching) revisions
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents: 196
diff changeset
436 }
2
08db726a0fb7 Shaping out low-level Hg structures
Artem Tikhomirov <tikhomirov.artem@gmail.com>
parents:
diff changeset
437 }