lucene-java-3.4.0/lucene/contrib/benchmark/src/java/org/apache/lucene/benchmark/byTask/tasks/CreateIndexTask.java

   1 package org.apache.lucene.benchmark.byTask.tasks;
   2
   3 /**
   4  * Licensed to the Apache Software Foundation (ASF) under one or more
   5  * contributor license agreements.  See the NOTICE file distributed with
   6  * this work for additional information regarding copyright ownership.
   7  * The ASF licenses this file to You under the Apache License, Version 2.0
   8  * (the "License"); you may not use this file except in compliance with
   9  * the License.  You may obtain a copy of the License at
  10  *
  11  *     http://www.apache.org/licenses/LICENSE-2.0
  12  *
  13  * Unless required by applicable law or agreed to in writing, software
  14  * distributed under the License is distributed on an "AS IS" BASIS,
  15  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  16  * See the License for the specific language governing permissions and
  17  * limitations under the License.
  18  */
  19
  20 import org.apache.lucene.benchmark.byTask.PerfRunData;
  21 import org.apache.lucene.benchmark.byTask.utils.Config;
  22 import org.apache.lucene.index.CorruptIndexException;
  23 import org.apache.lucene.index.IndexCommit;
  24 import org.apache.lucene.index.IndexDeletionPolicy;
  25 import org.apache.lucene.index.IndexWriter;
  26 import org.apache.lucene.index.IndexWriterConfig;
  27 import org.apache.lucene.index.LogMergePolicy;
  28 import org.apache.lucene.index.MergeScheduler;
  29 import org.apache.lucene.index.ConcurrentMergeScheduler;
  30 import org.apache.lucene.index.MergePolicy;
  31 import org.apache.lucene.index.NoDeletionPolicy;
  32 import org.apache.lucene.index.NoMergePolicy;
  33 import org.apache.lucene.index.NoMergeScheduler;
  34 import org.apache.lucene.index.IndexWriterConfig.OpenMode;
  35 import org.apache.lucene.index.TieredMergePolicy;
  36 import org.apache.lucene.store.LockObtainFailedException;
  37 import org.apache.lucene.util.Version;
  38
  39 import java.io.BufferedOutputStream;
  40 import java.io.File;
  41 import java.io.FileOutputStream;
  42 import java.io.IOException;
  43 import java.io.PrintStream;
  44
  45 /**
  46  * Create an index. <br>
  47  * Other side effects: index writer object in perfRunData is set. <br>
  48  * Relevant properties: <code>merge.factor (default 10),
  49  * max.buffered (default no flush), max.field.length (default
  50  * 10,000 tokens), max.field.length, compound (default true), ram.flush.mb [default 0],
  51  * merge.policy (default org.apache.lucene.index.LogByteSizeMergePolicy),
  52  * merge.scheduler (default
  53  * org.apache.lucene.index.ConcurrentMergeScheduler),
  54  * concurrent.merge.scheduler.max.thread.count and
  55  * concurrent.merge.scheduler.max.merge.count (defaults per
  56  * ConcurrentMergeScheduler) </code>.
  57  * <p>
  58  * This task also supports a "writer.info.stream" property with the following
  59  * values:
  60  * <ul>
  61  * <li>SystemOut - sets {@link IndexWriter#setInfoStream(java.io.PrintStream)}
  62  * to {@link System#out}.
  63  * <li>SystemErr - sets {@link IndexWriter#setInfoStream(java.io.PrintStream)}
  64  * to {@link System#err}.
  65  * <li>&lt;file_name&gt; - attempts to create a file given that name and sets
  66  * {@link IndexWriter#setInfoStream(java.io.PrintStream)} to that file. If this
  67  * denotes an invalid file name, or some error occurs, an exception will be
  68  * thrown.
  69  * </ul>
  70  */
  71 public class CreateIndexTask extends PerfTask {
  72
  73   public CreateIndexTask(PerfRunData runData) {
  74     super(runData);
  75   }
  76
  77
  78
  79   public static IndexDeletionPolicy getIndexDeletionPolicy(Config config) {
  80     String deletionPolicyName = config.get("deletion.policy", "org.apache.lucene.index.KeepOnlyLastCommitDeletionPolicy");
  81     if (deletionPolicyName.equals(NoDeletionPolicy.class.getName())) {
  82       return NoDeletionPolicy.INSTANCE;
  83     } else {
  84       try {
  85         return Class.forName(deletionPolicyName).asSubclass(IndexDeletionPolicy.class).newInstance();
  86       } catch (Exception e) {
  87         throw new RuntimeException("unable to instantiate class '" + deletionPolicyName + "' as IndexDeletionPolicy", e);
  88       }
  89     }
  90   }
  91
  92   @Override
  93   public int doLogic() throws IOException {
  94     PerfRunData runData = getRunData();
  95     Config config = runData.getConfig();
  96     runData.setIndexWriter(configureWriter(config, runData, OpenMode.CREATE, null));
  97     return 1;
  98   }
  99
 100   public static IndexWriterConfig createWriterConfig(Config config, PerfRunData runData, OpenMode mode, IndexCommit commit) {
 101     Version version = Version.valueOf(config.get("writer.version", Version.LUCENE_31.toString()));
 102     IndexWriterConfig iwConf = new IndexWriterConfig(version, runData.getAnalyzer());
 103     iwConf.setOpenMode(mode);
 104     IndexDeletionPolicy indexDeletionPolicy = getIndexDeletionPolicy(config);
 105     iwConf.setIndexDeletionPolicy(indexDeletionPolicy);
 106     if(commit != null)
 107       iwConf.setIndexCommit(commit);
 108
 109
 110     final String mergeScheduler = config.get("merge.scheduler",
 111                                              "org.apache.lucene.index.ConcurrentMergeScheduler");
 112     if (mergeScheduler.equals(NoMergeScheduler.class.getName())) {
 113       iwConf.setMergeScheduler(NoMergeScheduler.INSTANCE);
 114     } else {
 115       try {
 116         iwConf.setMergeScheduler(Class.forName(mergeScheduler).asSubclass(MergeScheduler.class).newInstance());
 117       } catch (Exception e) {
 118         throw new RuntimeException("unable to instantiate class '" + mergeScheduler + "' as merge scheduler", e);
 119       }
 120
 121       if (mergeScheduler.equals("org.apache.lucene.index.ConcurrentMergeScheduler")) {
 122         ConcurrentMergeScheduler cms = (ConcurrentMergeScheduler) iwConf.getMergeScheduler();
 123         int v = config.get("concurrent.merge.scheduler.max.thread.count", -1);
 124         if (v != -1) {
 125           cms.setMaxThreadCount(v);
 126         }
 127         v = config.get("concurrent.merge.scheduler.max.merge.count", -1);
 128         if (v != -1) {
 129           cms.setMaxMergeCount(v);
 130         }
 131       }
 132     }
 133
 134     final String mergePolicy = config.get("merge.policy",
 135                                           "org.apache.lucene.index.LogByteSizeMergePolicy");
 136     boolean isCompound = config.get("compound", true);
 137     if (mergePolicy.equals(NoMergePolicy.class.getName())) {
 138       iwConf.setMergePolicy(isCompound ? NoMergePolicy.COMPOUND_FILES : NoMergePolicy.NO_COMPOUND_FILES);
 139     } else {
 140       try {
 141         iwConf.setMergePolicy(Class.forName(mergePolicy).asSubclass(MergePolicy.class).newInstance());
 142       } catch (Exception e) {
 143         throw new RuntimeException("unable to instantiate class '" + mergePolicy + "' as merge policy", e);
 144       }
 145       if(iwConf.getMergePolicy() instanceof LogMergePolicy) {
 146         LogMergePolicy logMergePolicy = (LogMergePolicy) iwConf.getMergePolicy();
 147         logMergePolicy.setUseCompoundFile(isCompound);
 148         logMergePolicy.setMergeFactor(config.get("merge.factor",OpenIndexTask.DEFAULT_MERGE_PFACTOR));
 149       } else if(iwConf.getMergePolicy() instanceof TieredMergePolicy) {
 150         TieredMergePolicy tieredMergePolicy = (TieredMergePolicy) iwConf.getMergePolicy();
 151         tieredMergePolicy.setUseCompoundFile(isCompound);
 152       }
 153     }
 154     final double ramBuffer = config.get("ram.flush.mb",OpenIndexTask.DEFAULT_RAM_FLUSH_MB);
 155     final int maxBuffered = config.get("max.buffered",OpenIndexTask.DEFAULT_MAX_BUFFERED);
 156     if (maxBuffered == IndexWriterConfig.DISABLE_AUTO_FLUSH) {
 157       iwConf.setRAMBufferSizeMB(ramBuffer);
 158       iwConf.setMaxBufferedDocs(maxBuffered);
 159     } else {
 160       iwConf.setMaxBufferedDocs(maxBuffered);
 161       iwConf.setRAMBufferSizeMB(ramBuffer);
 162     }
 163
 164     return iwConf;
 165   }
 166
 167   public static IndexWriter configureWriter(Config config, PerfRunData runData, OpenMode mode, IndexCommit commit) throws CorruptIndexException, LockObtainFailedException, IOException {
 168     IndexWriter writer = new IndexWriter(runData.getDirectory(), createWriterConfig(config, runData, mode, commit));
 169     writer.setMaxFieldLength(config.get("max.field.length", OpenIndexTask.DEFAULT_MAX_FIELD_LENGTH));
 170
 171     String infoStreamVal = config.get("writer.info.stream", null);
 172     if (infoStreamVal != null) {
 173       if (infoStreamVal.equals("SystemOut")) {
 174         writer.setInfoStream(System.out);
 175       } else if (infoStreamVal.equals("SystemErr")) {
 176         writer.setInfoStream(System.err);
 177       } else {
 178         File f = new File(infoStreamVal).getAbsoluteFile();
 179         writer.setInfoStream(new PrintStream(new BufferedOutputStream(new FileOutputStream(f))));
 180       }
 181     }
 182     return writer;
 183   }
 184 }