Mercurial > hg4j

comparison hg4j/src/main/java/org/tmatesoft/hg/internal/KeywordFilter.java @ 213:6ec4af642ba8 gradle

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Project uses Gradle for build - actual changes
author Alexander Kitaev <kitaev@gmail.com>
date Tue, 10 May 2011 10:52:53 +0200
parents
children
comparison
equal deleted inserted replaced
212:edb2e2829352 213:6ec4af642ba8
1 /*
2 * Copyright (c) 2011 TMate Software Ltd
3 *
4 * This program is free software; you can redistribute it and/or modify
5 * it under the terms of the GNU General Public License as published by
6 * the Free Software Foundation; version 2 of the License.
7 *
8 * This program is distributed in the hope that it will be useful,
9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11 * GNU General Public License for more details.
12 *
13 * For information on how to redistribute this software under
14 * the terms of a license other than GNU General Public License
15 * contact TMate Software at support@hg4j.com
16 */
17 package org.tmatesoft.hg.internal;
18
19 import java.nio.ByteBuffer;
20 import java.util.ArrayList;
21 import java.util.Map;
22 import java.util.TreeMap;
23
24 import org.tmatesoft.hg.repo.HgChangelog.RawChangeset;
25 import org.tmatesoft.hg.repo.HgRepository;
26 import org.tmatesoft.hg.util.Path;
27
28 /**
29 *
30 * @author Artem Tikhomirov
31 * @author TMate Software Ltd.
32 */
33 public class KeywordFilter implements Filter {
34 // present implementation is stateless, however, filter use pattern shall not assume that. In fact, Factory may us that
35 private final HgRepository repo;
36 private final boolean isExpanding;
37 private final TreeMap<String,String> keywords;
38 private final int minBufferLen;
39 private final Path path;
40 private RawChangeset latestFileCset;
41
42 /**
43 *
44 * @param hgRepo
45 * @param path
46 * @param expand <code>true</code> to expand keywords, <code>false</code> to shrink
47 */
48 private KeywordFilter(HgRepository hgRepo, Path p, boolean expand) {
49 repo = hgRepo;
50 path = p;
51 isExpanding = expand;
52 keywords = new TreeMap<String,String>();
53 keywords.put("Id", "Id");
54 keywords.put("Revision", "Revision");
55 keywords.put("Author", "Author");
56 keywords.put("Date", "Date");
57 keywords.put("LastChangedRevision", "LastChangedRevision");
58 keywords.put("LastChangedBy", "LastChangedBy");
59 keywords.put("LastChangedDate", "LastChangedDate");
60 keywords.put("Source", "Source");
61 keywords.put("Header", "Header");
62
63 int l = 0;
64 for (String s : keywords.keySet()) {
65 if (s.length() > l) {
66 l = s.length();
67 }
68 }
69 // FIXME later may implement #filter() not to read full kw value (just "$kw:"). However, limit of maxLen + 2 would keep valid.
70 // for buffers less then minBufferLen, there are chances #filter() implementation would never end
71 // (i.e. for input "$LongestKey"$
72 minBufferLen = l + 2 + (isExpanding ? 0 : 120 /*any reasonable constant for max possible kw value length*/);
73 }
74
75 /**
76 * @param src buffer ready to be read
77 * @return buffer ready to be read and original buffer's position modified to reflect consumed bytes. IOW, if source buffer
78 * on return has remaining bytes, they are assumed not-read (not processed) and next chunk passed to filter is supposed to
79 * start with them
80 */
81 public ByteBuffer filter(ByteBuffer src) {
82 if (src.capacity() < minBufferLen) {
83 throw new IllegalStateException(String.format("Need buffer of at least %d bytes to ensure filter won't hang", minBufferLen));
84 }
85 ByteBuffer rv = null;
86 int keywordStart = -1;
87 int x = src.position();
88 int copyFrom = x; // needs to be updated each time we copy a slice, but not each time we modify source index (x)
89 while (x < src.limit()) {
90 if (keywordStart == -1) {
91 int i = indexOf(src, '$', x, false);
92 if (i == -1) {
93 if (rv == null) {
94 return src;
95 } else {
96 copySlice(src, copyFrom, src.limit(), rv);
97 rv.flip();
98 src.position(src.limit());
99 return rv;
100 }
101 }
102 keywordStart = i;
103 // fall-through
104 }
105 if (keywordStart >= 0) {
106 int i = indexOf(src, '$', keywordStart+1, true);
107 if (i == -1) {
108 // end of buffer reached
109 if (rv == null) {
110 if (keywordStart == x) {
111 // FIXME in fact, x might be equal to keywordStart and to src.position() here ('$' is first character in the buffer,
112 // and there are no other '$' not eols till the end of the buffer). This would lead to deadlock (filter won't consume any
113 // bytes). To prevent this, either shall copy bytes [keywordStart..buffer.limit()) to local buffer and use it on the next invocation,
114 // or add lookup of the keywords right after first '$' is found (do not wait for closing '$'). For now, large enough src buffer would be sufficient
115 // not to run into such situation
116 throw new IllegalStateException("Try src buffer of a greater size");
117 }
118 rv = ByteBuffer.allocate(keywordStart - copyFrom);
119 }
120 // copy all from source till latest possible kw start
121 copySlice(src, copyFrom, keywordStart, rv);
122 rv.flip();
123 // and tell caller we've consumed only to the potential kw start
124 src.position(keywordStart);
125 return rv;
126 } else if (src.get(i) == '$') {
127 // end of keyword, or start of a new one.
128 String keyword;
129 if ((keyword = matchKeyword(src, keywordStart, i)) != null) {
130 if (rv == null) {
131 // src.remaining(), not .capacity because src is not read, and remaining represents
132 // actual bytes count, while capacity - potential.
133 // Factor of 4 is pure guess and a HACK, need to be fixed with re-expanding buffer on demand
134 rv = ByteBuffer.allocate(isExpanding ? src.remaining() * 4 : src.remaining());
135 }
136 copySlice(src, copyFrom, keywordStart+1, rv);
137 rv.put(keyword.getBytes());
138 if (isExpanding) {
139 rv.put((byte) ':');
140 rv.put((byte) ' ');
141 expandKeywordValue(keyword, rv);
142 rv.put((byte) ' ');
143 }
144 rv.put((byte) '$');
145 keywordStart = -1;
146 x = i+1;
147 copyFrom = x;
148 continue;
149 } else {
150 if (rv != null) {
151 // we've already did some substitution, thus need to copy bytes we've scanned.
152 copySlice(src, x, i, rv);
153 copyFrom = i;
154 } // no else in attempt to avoid rv creation if no real kw would be found
155 keywordStart = i;
156 x = i; // '$' at i wasn't consumed, hence x points to i, not i+1. This is to avoid problems with case: "sdfsd $ asdfs $Id$ sdf"
157 continue;
158 }
159 } else {
160 assert src.get(i) == '\n' || src.get(i) == '\r';
161 // line break
162 if (rv != null) {
163 copySlice(src, x, i+1, rv);
164 copyFrom = i+1;
165 }
166 x = i+1;
167 keywordStart = -1; // Wasn't keyword, really
168 continue; // try once again
169 }
170 }
171 }
172 if (keywordStart != -1) {
173 if (rv == null) {
174 // no expansion happened yet, and we have potential kw start
175 rv = ByteBuffer.allocate(keywordStart - src.position());
176 copySlice(src, src.position(), keywordStart, rv);
177 }
178 src.position(keywordStart);
179 }
180 if (rv != null) {
181 rv.flip();
182 return rv;
183 }
184 return src;
185 }
186
187 /**
188 * @param keyword
189 * @param rv
190 */
191 private void expandKeywordValue(String keyword, ByteBuffer rv) {
192 if ("Id".equals(keyword)) {
193 rv.put(identityString().getBytes());
194 } else if ("Revision".equals(keyword)) {
195 rv.put(revision().getBytes());
196 } else if ("Author".equals(keyword)) {
197 rv.put(username().getBytes());
198 } else if ("Date".equals(keyword)) {
199 rv.put(date().getBytes());
200 } else {
201 throw new IllegalStateException(String.format("Keyword %s is not yet supported", keyword));
202 }
203 }
204
205 private String matchKeyword(ByteBuffer src, int kwStart, int kwEnd) {
206 assert kwEnd - kwStart - 1 > 0;
207 assert src.get(kwStart) == src.get(kwEnd) && src.get(kwEnd) == '$';
208 char[] chars = new char[kwEnd - kwStart - 1];
209 int i;
210 for (i = 0; i < chars.length; i++) {
211 char c = (char) src.get(kwStart + 1 + i);
212 if (c == ':') {
213 break;
214 }
215 chars[i] = c;
216 }
217 String kw = new String(chars, 0, i);
218 // XXX may use subMap to look up keywords based on few available characters (not waiting till closing $)
219 // System.out.println(keywords.subMap("I", "J"));
220 // System.out.println(keywords.subMap("A", "B"));
221 // System.out.println(keywords.subMap("Au", "B"));
222 return keywords.get(kw);
223 }
224
225 // copies part of the src buffer, [from..to). doesn't modify src position
226 static void copySlice(ByteBuffer src, int from, int to, ByteBuffer dst) {
227 if (to > src.limit()) {
228 throw new IllegalArgumentException("Bad right boundary");
229 }
230 if (dst.remaining() < to - from) {
231 throw new IllegalArgumentException("Not enough room in the destination buffer");
232 }
233 for (int i = from; i < to; i++) {
234 dst.put(src.get(i));
235 }
236 }
237
238 private static int indexOf(ByteBuffer b, char ch, int from, boolean newlineBreaks) {
239 for (int i = from; i < b.limit(); i++) {
240 byte c = b.get(i);
241 if (ch == c) {
242 return i;
243 }
244 if (newlineBreaks && (c == '\n' || c == '\r')) {
245 return i;
246 }
247 }
248 return -1;
249 }
250
251 private String identityString() {
252 return String.format("%s,v %s %s %s", path, revision(), date(), username());
253 }
254
255 private String revision() {
256 // FIXME add cset's nodeid into Changeset class
257 int csetRev = repo.getFileNode(path).getChangesetLocalRevision(HgRepository.TIP);
258 return repo.getChangelog().getRevision(csetRev).shortNotation();
259 }
260
261 private String username() {
262 return getChangeset().user();
263 }
264
265 private String date() {
266 return String.format("%tY/%<tm/%<td %<tH:%<tM:%<tS", getChangeset().date());
267 }
268
269 private RawChangeset getChangeset() {
270 if (latestFileCset == null) {
271 // XXX consider use of ChangelogHelper
272 int csetRev = repo.getFileNode(path).getChangesetLocalRevision(HgRepository.TIP);
273 latestFileCset = repo.getChangelog().range(csetRev, csetRev).get(0);
274 }
275 return latestFileCset;
276 }
277
278 public static class Factory implements Filter.Factory {
279
280 private HgRepository repo;
281 private Path.Matcher matcher;
282
283 public void initialize(HgRepository hgRepo, ConfigFile cfg) {
284 repo = hgRepo;
285 ArrayList<String> patterns = new ArrayList<String>();
286 for (Map.Entry<String,String> e : cfg.getSection("keyword").entrySet()) {
287 if (!"ignore".equalsIgnoreCase(e.getValue())) {
288 patterns.add(e.getKey());
289 }
290 }
291 matcher = new PathGlobMatcher(patterns.toArray(new String[patterns.size()]));
292 // TODO read and respect keyword patterns from [keywordmaps]
293 }
294
295 public Filter create(Path path, Options opts) {
296 if (matcher.accept(path)) {
297 return new KeywordFilter(repo, path, opts.getDirection() == Filter.Direction.FromRepo);
298 }
299 return null;
300 }
301 }
302
303 //
304 // public static void main(String[] args) throws Exception {
305 // FileInputStream fis = new FileInputStream(new File("/temp/kwoutput.txt"));
306 // FileOutputStream fos = new FileOutputStream(new File("/temp/kwoutput2.txt"));
307 // ByteBuffer b = ByteBuffer.allocate(256);
308 // KeywordFilter kwFilter = new KeywordFilter(false);
309 // while (fis.getChannel().read(b) != -1) {
310 // b.flip(); // get ready to be read
311 // ByteBuffer f = kwFilter.filter(b);
312 // fos.getChannel().write(f); // XXX in fact, f may not be fully consumed
313 // if (b.hasRemaining()) {
314 // b.compact();
315 // } else {
316 // b.clear();
317 // }
318 // }
319 // fis.close();
320 // fos.flush();
321 // fos.close();
322 // }
323 }