Mercurial Hosting > luan
annotate src/goodjava/lucene/logging/LoggingIndexWriter.java @ 1518:f989778ca9e1
stringify, json_string
| author | Franklin Schmidt <fschmidt@gmail.com> |
|---|---|
| date | Fri, 19 Jun 2020 17:50:25 -0600 |
| parents | 31b543826ca9 |
| children | 3bd4d7963456 |
| rev | line source |
|---|---|
| 1461 | 1 package goodjava.lucene.logging; |
| 2 | |
| 1465 | 3 import java.io.File; |
| 4 import java.io.RandomAccessFile; | |
| 5 import java.io.ByteArrayOutputStream; | |
| 6 import java.io.DataOutputStream; | |
| 7 import java.io.DataInputStream; | |
| 8 import java.io.FileInputStream; | |
| 1461 | 9 import java.io.IOException; |
| 10 import java.util.Map; | |
| 1465 | 11 import java.util.Set; |
| 12 import java.util.HashSet; | |
| 13 import java.util.List; | |
| 14 import java.util.ArrayList; | |
| 15 import java.util.Random; | |
| 16 import org.apache.lucene.document.Document; | |
| 17 import org.apache.lucene.index.DirectoryReader; | |
| 18 import org.apache.lucene.index.IndexReader; | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
19 import org.apache.lucene.index.Term; |
| 1465 | 20 import org.apache.lucene.search.IndexSearcher; |
| 1461 | 21 import org.apache.lucene.search.Query; |
| 1465 | 22 import org.apache.lucene.search.MatchAllDocsQuery; |
| 23 import org.apache.lucene.search.TopDocs; | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
24 import org.apache.lucene.search.PrefixQuery; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
25 import org.apache.lucene.search.SortField; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
26 import org.apache.lucene.search.Sort; |
| 1465 | 27 import org.apache.lucene.store.Directory; |
| 28 import org.apache.lucene.store.FSDirectory; | |
| 1473 | 29 import goodjava.io.IoUtils; |
| 1461 | 30 import goodjava.lucene.api.GoodIndexWriter; |
| 1465 | 31 import goodjava.lucene.api.LuceneIndexWriter; |
| 32 import goodjava.lucene.api.GoodCollector; | |
| 33 import goodjava.lucene.api.LuceneUtils; | |
| 34 import goodjava.logging.Logger; | |
| 35 import goodjava.logging.LoggerFactory; | |
| 1461 | 36 |
| 37 | |
| 1488 | 38 public class LoggingIndexWriter implements GoodIndexWriter { |
| 1465 | 39 private static final Logger logger = LoggerFactory.getLogger(LoggingIndexWriter.class); |
| 40 private static final int version = 1; | |
| 1461 | 41 private static final int OP_DELETE_ALL = 1; |
| 42 private static final int OP_DELETE_DOCUMENTS = 2; | |
| 43 private static final int OP_ADD_DOCUMENT = 3; | |
| 44 private static final int OP_UPDATE_DOCUMENT = 4; | |
| 1465 | 45 private static final Random rnd = new Random(); |
| 1461 | 46 |
| 1465 | 47 public final LuceneIndexWriter indexWriter; |
| 48 private final File logDir; | |
| 1488 | 49 protected final List<LogFile> logs = new ArrayList<LogFile>(); |
| 1486 | 50 private LogOutputStream log; |
| 1465 | 51 private final File index; |
| 52 private boolean isMerging = false; | |
| 1461 | 53 |
| 1465 | 54 public LoggingIndexWriter(LuceneIndexWriter indexWriter,File logDir) throws IOException { |
| 1461 | 55 this.indexWriter = indexWriter; |
| 1465 | 56 this.logDir = logDir; |
| 1501 | 57 IoUtils.mkdirs(logDir); |
| 1465 | 58 if( !logDir.isDirectory() ) |
| 59 throw new RuntimeException(); | |
| 60 index = new File(logDir,"index"); | |
| 61 if( index.exists() ) { | |
| 62 DataInputStream dis = new DataInputStream(new FileInputStream(index)); | |
| 63 try { | |
| 64 if( dis.readInt() == version ) { | |
| 65 final int n = dis.readInt(); | |
| 66 for( int i=0; i<n; i++ ) { | |
| 67 File file = new File( logDir, dis.readUTF() ); | |
| 1486 | 68 logs.add( new LogFile(file) ); |
| 1465 | 69 } |
| 70 deleteUnusedFiles(); | |
| 1486 | 71 setLog(); |
| 1465 | 72 return; |
| 73 } | |
| 74 } finally { | |
| 75 dis.close(); | |
| 76 } | |
| 77 } | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
78 newLogs(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
79 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
80 |
| 1486 | 81 private void setLog() throws IOException { |
| 82 if( log != null ) | |
| 83 log.close(); | |
| 84 log = logs.get(logs.size()-1).output(); | |
| 85 } | |
| 86 | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
87 public synchronized boolean isMerging() { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
88 return isMerging; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
89 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
90 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
91 private synchronized void isNotMerging() { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
92 isMerging = false; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
93 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
94 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
95 public synchronized void newLogs() throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
96 if( isMerging ) |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
97 throw new RuntimeException("merging"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
98 logger.info("building new logs"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
99 logs.clear(); |
| 1465 | 100 for( int i=0; i<2; i++ ) { |
| 101 logs.add( newLogFile() ); | |
| 102 } | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
103 logLucene( System.currentTimeMillis(), logs.get(0), indexWriter ); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
104 writeIndex(); |
| 1486 | 105 setLog(); |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
106 logger.info("done building new logs"); |
| 1461 | 107 } |
| 108 | |
| 1486 | 109 private static void logLucene(long time,LogFile logLucene,LuceneIndexWriter indexWriter) throws IOException { |
| 110 LogOutputStream log = logLucene.output(); | |
| 1465 | 111 IndexReader reader = indexWriter.openReader(); |
| 112 final IndexSearcher searcher = new IndexSearcher(reader); | |
| 113 Query query = new MatchAllDocsQuery(); | |
| 114 searcher.search( query, new GoodCollector(){ | |
| 115 public void collectDoc(int iDoc) throws IOException { | |
| 116 Document doc = searcher.doc(iDoc); | |
| 117 Map<String,Object> storedFields = LuceneUtils.toMap(doc); | |
| 118 log.writeLong(time); | |
| 119 log.writeByte(OP_ADD_DOCUMENT); | |
| 120 log.writeMap(storedFields); | |
| 121 } | |
| 122 }); | |
| 123 reader.close(); | |
| 124 log.commit(); | |
| 1486 | 125 log.close(); |
| 1465 | 126 } |
| 127 | |
| 128 private LogFile newLogFile() throws IOException { | |
| 129 File file; | |
| 130 do { | |
| 131 file = new File(logDir,"_"+rnd.nextInt(100)+".log"); | |
| 132 } while( file.exists() ); | |
| 1486 | 133 return new LogFile(file); |
| 1461 | 134 } |
| 135 | |
| 1473 | 136 private void deleteUnusedFiles() throws IOException { |
| 1499 | 137 deleteUnusedFiles(logs,index); |
| 138 } | |
| 139 | |
| 140 private static void deleteUnusedFiles(List<LogFile> logs,File index) throws IOException { | |
| 1465 | 141 Set<String> used = new HashSet<String>(); |
| 142 used.add( index.getName() ); | |
| 143 for( LogFile lf : logs ) { | |
| 144 used.add( lf.file.getName() ); | |
| 145 } | |
| 1499 | 146 for( File f : index.getParentFile().listFiles() ) { |
| 1465 | 147 if( !used.contains(f.getName()) ) { |
| 1475 | 148 IoUtils.deleteRecursively(f); |
| 1465 | 149 } |
| 150 } | |
| 1461 | 151 } |
| 152 | |
| 1465 | 153 private void writeIndex() throws IOException { |
| 1499 | 154 writeIndex(logs,index); |
| 155 } | |
| 156 | |
| 157 public static void writeIndex(List<LogFile> logs,File index) throws IOException { | |
| 1465 | 158 ByteArrayOutputStream baos = new ByteArrayOutputStream(); |
| 159 DataOutputStream dos = new DataOutputStream(baos); | |
| 160 dos.writeInt(version); | |
| 161 dos.writeInt(logs.size()); | |
| 162 for( LogFile lf : logs ) { | |
| 163 String fileName = lf.file.getName(); | |
| 164 dos.writeUTF(fileName); | |
| 165 } | |
| 166 dos.close(); | |
| 167 RandomAccessFile raf = new RandomAccessFile( index, "rwd" ); | |
| 168 raf.write( baos.toByteArray() ); | |
| 169 raf.close(); | |
| 1499 | 170 deleteUnusedFiles(logs,index); |
| 171 //logger.info("writeIndex "+logs.toString()); | |
| 1461 | 172 } |
| 173 | |
| 1465 | 174 private void mergeLogs() throws IOException { |
| 1512 | 175 //logger.info("merge"); |
| 1465 | 176 LogFile first = logs.get(0); |
| 177 LogFile second = logs.get(1); | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
178 long lastTime = second.file.lastModified(); |
| 1465 | 179 File dirFile = new File(logDir,"merge"); |
| 180 if( dirFile.exists() ) | |
| 181 throw new RuntimeException(); | |
| 182 Directory dir = FSDirectory.open(dirFile); | |
| 183 LuceneIndexWriter mergeWriter = new LuceneIndexWriter( indexWriter.luceneVersion, dir, indexWriter.goodConfig ); | |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
184 playLog( first.input(), mergeWriter ); |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
185 playLog( second.input(), mergeWriter ); |
| 1465 | 186 mergeWriter.commit(); |
| 187 LogFile merge = newLogFile(); | |
| 188 logLucene( lastTime, merge, mergeWriter ); | |
| 189 mergeWriter.close(); | |
| 190 synchronized(this) { | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
191 //check(); |
| 1465 | 192 logs.remove(0); |
| 193 logs.set(0,merge); | |
| 194 writeIndex(); | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
195 //check(null); |
| 1465 | 196 } |
| 1461 | 197 } |
| 1465 | 198 private final Runnable mergeLogs = new Runnable() { public void run() { |
| 199 try { | |
| 200 mergeLogs(); | |
| 201 } catch(IOException e) { | |
| 202 throw new RuntimeException(e); | |
| 203 } finally { | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
204 isNotMerging(); |
| 1465 | 205 } |
| 206 } }; | |
| 1461 | 207 |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
208 private static class DocIter { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
209 final IndexReader reader; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
210 final TopDocs td; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
211 final int n; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
212 int i = 0; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
213 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
214 DocIter(IndexReader reader,Query query,Sort sort) throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
215 this.reader = reader; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
216 IndexSearcher searcher = new IndexSearcher(reader); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
217 this.td = searcher.search(query,10000000,sort); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
218 this.n = td.scoreDocs.length; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
219 if( td.totalHits != n ) |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
220 throw new RuntimeException(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
221 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
222 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
223 Document next() throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
224 return i < n ? reader.document(td.scoreDocs[i++].doc) : null; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
225 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
226 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
227 |
| 1487 | 228 private volatile boolean isChecking = false; |
| 229 | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
230 public void check(SortField sortField) throws IOException { |
| 1487 | 231 if( isChecking ) |
| 232 throw new RuntimeException("another check is running"); | |
|
1508
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
233 isChecking = true; |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
234 try { |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
235 doCheck(sortField); |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
236 } finally { |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
237 isChecking = false; |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
238 } |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
239 } |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
240 |
|
86c5e7000ecf
lucene.backup checksum
Franklin Schmidt <fschmidt@gmail.com>
parents:
1504
diff
changeset
|
241 protected void doCheck(SortField sortField) throws IOException { |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
242 IndexReader indexReader; |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
243 List<LogInputStream> logReaders; |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
244 synchronized(this) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
245 indexReader = indexWriter.openReader(); |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
246 logReaders = logReaders(logs); |
| 1465 | 247 } |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
248 try { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
249 logger.info("check start"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
250 indexWriter.check(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
251 File dirFile = new File(logDir,"check"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
252 IoUtils.deleteRecursively(dirFile); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
253 Directory dir = FSDirectory.open(dirFile); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
254 LuceneIndexWriter checkWriter = new LuceneIndexWriter( indexWriter.luceneVersion, dir, indexWriter.goodConfig ); |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
255 playLogs(logReaders,checkWriter); |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
256 logger.info("check lucene"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
257 IndexReader checkReader = checkWriter.openReader(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
258 if( sortField == null ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
259 int nCheck = checkReader.numDocs(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
260 int nOrig = indexReader.numDocs(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
261 if( nCheck != nOrig ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
262 logger.error("numDocs mismatch: lucene="+nOrig+" logs="+nCheck); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
263 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
264 logger.info("numDocs="+nOrig); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
265 if( hash(indexReader) != hash(checkReader) ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
266 logger.error("hash mismatch"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
267 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
268 } else { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
269 Sort sort = new Sort(sortField); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
270 String sortFieldName = sortField.getField(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
271 Query query = new PrefixQuery(new Term(sortFieldName)); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
272 DocIter origIter = new DocIter(indexReader,query,sort); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
273 DocIter checkIter = new DocIter(checkReader,query,sort); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
274 Map<String,Object> origFields = LuceneUtils.toMap(origIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
275 Map<String,Object> checkFields = LuceneUtils.toMap(checkIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
276 while( origFields!=null && checkFields!=null ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
277 Comparable origFld = (Comparable)origFields.get(sortFieldName); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
278 Comparable checkFld = (Comparable)checkFields.get(sortFieldName); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
279 int cmp = origFld.compareTo(checkFld); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
280 if( cmp==0 ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
281 if( !origFields.equals(checkFields) ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
282 logger.error(sortFieldName+" "+origFld+" not equal"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
283 logger.error("lucene = "+origFields); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
284 logger.error("logs = "+checkFields); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
285 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
286 origFields = LuceneUtils.toMap(origIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
287 checkFields = LuceneUtils.toMap(checkIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
288 } else if( cmp < 0 ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
289 logger.error(sortFieldName+" "+origFld+" found in lucene but not logs"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
290 origFields = LuceneUtils.toMap(origIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
291 } else { // > |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
292 logger.error(sortFieldName+" "+checkFld+" found in logs but not lucene"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
293 checkFields = LuceneUtils.toMap(checkIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
294 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
295 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
296 while( origFields!=null ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
297 Comparable origFld = (Comparable)origFields.get(sortFieldName); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
298 logger.error(sortFieldName+" "+origFld+" found in lucene but not logs"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
299 origFields = LuceneUtils.toMap(origIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
300 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
301 while( checkFields!=null ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
302 Comparable checkFld = (Comparable)checkFields.get(sortFieldName); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
303 logger.error(sortFieldName+" "+checkFld+" found in logs but not lucene"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
304 checkFields = LuceneUtils.toMap(checkIter.next()); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
305 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
306 //logger.info("check done"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
307 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
308 checkReader.close(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
309 checkWriter.close(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
310 IoUtils.deleteRecursively(dirFile); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
311 logger.info("check done"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
312 } finally { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
313 indexReader.close(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
314 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
315 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
316 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
317 private static abstract class HashCollector extends GoodCollector { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
318 int total = 0; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
319 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
320 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
321 private static int hash(IndexReader reader) throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
322 final IndexSearcher searcher = new IndexSearcher(reader); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
323 Query query = new MatchAllDocsQuery(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
324 HashCollector col = new HashCollector() { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
325 public void collectDoc(int iDoc) throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
326 Document doc = searcher.doc(iDoc); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
327 Map<String,Object> storedFields = LuceneUtils.toMap(doc); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
328 total += storedFields.hashCode(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
329 } |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
330 }; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
331 searcher.search(query,col); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
332 return col.total; |
| 1461 | 333 } |
| 334 | |
| 1465 | 335 public synchronized void close() throws IOException { |
| 336 indexWriter.close(); | |
| 337 log.commit(); | |
| 1486 | 338 log.close(); |
| 1465 | 339 } |
| 340 | |
| 341 public synchronized void commit() throws IOException { | |
| 342 indexWriter.commit(); | |
| 343 log.commit(); | |
| 344 if( isMerging ) | |
| 345 return; | |
| 1486 | 346 if( log.logFile.end() > logs.get(0).end() ) { |
| 1465 | 347 logs.add( newLogFile() ); |
| 348 writeIndex(); | |
| 1486 | 349 setLog(); |
| 1465 | 350 } |
| 351 if( logs.size() > 3 ) { | |
| 352 isMerging = true; | |
| 1504 | 353 new Thread(mergeLogs).start(); |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
354 // mergeLogs.run(); |
| 1465 | 355 } |
| 1461 | 356 } |
| 357 | |
| 1465 | 358 public synchronized void rollback() throws IOException { |
| 359 indexWriter.rollback(); | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
360 log.rollback(); |
| 1465 | 361 } |
| 362 | |
| 363 public synchronized void deleteAll() throws IOException { | |
| 364 indexWriter.deleteAll(); | |
| 1486 | 365 writeOp(OP_DELETE_ALL); |
| 1461 | 366 } |
| 367 | |
| 1465 | 368 public synchronized void deleteDocuments(Query query) throws IOException { |
| 369 indexWriter.deleteDocuments(query); | |
| 1486 | 370 writeOp(OP_DELETE_DOCUMENTS); |
| 1465 | 371 log.writeQuery(query); |
| 372 } | |
| 373 | |
| 374 public synchronized void addDocument(Map<String,Object> storedFields) throws IOException { | |
| 375 indexWriter.addDocument(storedFields); | |
| 1486 | 376 writeOp(OP_ADD_DOCUMENT); |
| 1465 | 377 log.writeMap(storedFields); |
| 378 } | |
| 379 | |
| 380 public synchronized void updateDocument(String keyFieldName,Map<String,Object> storedFields) throws IOException { | |
| 381 indexWriter.updateDocument(keyFieldName,storedFields); | |
| 1486 | 382 writeOp(OP_UPDATE_DOCUMENT); |
| 1465 | 383 log.writeUTF(keyFieldName); |
| 384 log.writeMap(storedFields); | |
| 385 } | |
| 386 | |
| 387 public synchronized void reindexDocuments(String keyFieldName,Query query) throws IOException { | |
| 1461 | 388 indexWriter.reindexDocuments(keyFieldName,query); |
| 389 } | |
| 390 | |
| 1486 | 391 private void writeOp(int op) throws IOException { |
| 1465 | 392 log.writeLong(System.currentTimeMillis()); |
| 393 log.writeByte(op); | |
| 394 } | |
| 395 | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
396 public synchronized void playLogs() throws IOException { |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
397 playLogs( logReaders(logs), indexWriter ); |
| 1465 | 398 } |
| 399 | |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
400 private static List<LogInputStream> logReaders(List<LogFile> logs) throws IOException { |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
401 List<LogInputStream> logReaders = new ArrayList<LogInputStream>(); |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
402 for( LogFile log : logs ) { |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
403 logReaders.add( log.input() ); |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
404 } |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
405 return logReaders; |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
406 } |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
407 |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
408 private static void playLogs(List<LogInputStream> logReaders,LuceneIndexWriter indexWriter) throws IOException { |
| 1465 | 409 if( numDocs(indexWriter) != 0 ) |
| 410 throw new RuntimeException ("not empty"); | |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
411 for( LogInputStream reader : logReaders ) { |
|
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
412 playLog(reader,indexWriter); |
| 1465 | 413 } |
| 414 indexWriter.commit(); | |
| 415 } | |
| 416 | |
| 417 private static int numDocs(LuceneIndexWriter indexWriter) throws IOException { | |
| 418 IndexReader reader = indexWriter.openReader(); | |
| 419 int n = reader.numDocs(); | |
| 420 reader.close(); | |
| 421 return n; | |
| 422 } | |
| 423 | |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
424 private static void playLog(LogInputStream in,LuceneIndexWriter indexWriter) throws IOException { |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
425 while( in.available() > 0 ) { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
426 playOp(in,indexWriter); |
| 1465 | 427 } |
|
1484
1fa6e8ec2d53
lucene.logging cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
428 in.close(); |
| 1465 | 429 } |
| 430 | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
431 private static void playOp(LogInputStream in,LuceneIndexWriter indexWriter) throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
432 in.readLong(); // time |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
433 int op = in.readByte(); |
| 1461 | 434 switch(op) { |
| 435 case OP_DELETE_ALL: | |
| 436 indexWriter.deleteAll(); | |
| 437 return; | |
| 438 case OP_DELETE_DOCUMENTS: | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
439 indexWriter.deleteDocuments( in.readQuery() ); |
| 1461 | 440 return; |
| 441 case OP_ADD_DOCUMENT: | |
| 1465 | 442 { |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
443 Map storedFields = in.readMap(); |
| 1465 | 444 indexWriter.addDocument(storedFields); |
| 445 return; | |
| 446 } | |
| 1461 | 447 case OP_UPDATE_DOCUMENT: |
| 1465 | 448 { |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
449 String keyFieldName = in.readUTF(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1475
diff
changeset
|
450 Map storedFields = in.readMap(); |
| 1465 | 451 indexWriter.updateDocument(keyFieldName,storedFields); |
| 452 return; | |
| 453 } | |
| 1461 | 454 default: |
| 455 throw new RuntimeException("invalid op "+op); | |
| 456 } | |
| 457 } | |
| 458 | |
| 1465 | 459 private static void dump(LuceneIndexWriter indexWriter) throws IOException { |
| 460 IndexReader reader = indexWriter.openReader(); | |
| 461 IndexSearcher searcher = new IndexSearcher(reader); | |
| 462 Query query = new MatchAllDocsQuery(); | |
| 463 TopDocs td = searcher.search(query,100); | |
| 464 System.out.println("totalHits = "+td.totalHits); | |
| 465 for( int i=0; i<td.scoreDocs.length; i++ ) { | |
| 466 Document doc = searcher.doc(td.scoreDocs[i].doc); | |
| 467 System.out.println(LuceneUtils.toMap(doc)); | |
| 1461 | 468 } |
| 1465 | 469 System.out.println(); |
| 470 reader.close(); | |
| 1461 | 471 } |
| 1465 | 472 |
| 1461 | 473 } |