public void testUpdateSameDoc() throws Exception {
    final Directory dir = newDirectory();

    final LineFileDocs docs = new LineFileDocs(random());
    for (int r = 0; r < 3; r++) {
      final IndexWriter w =
          new IndexWriter(
              dir, newIndexWriterConfig(new MockAnalyzer(random())).setMaxBufferedDocs(2));
      final int numUpdates = atLeast(20);
      int numThreads = TestUtil.nextInt(random(), 2, 6);
      IndexingThread[] threads = new IndexingThread[numThreads];
      for (int i = 0; i < numThreads; i++) {
        threads[i] = new IndexingThread(docs, w, numUpdates);
        threads[i].start();
      }

      for (int i = 0; i < numThreads; i++) {
        threads[i].join();
      }

      w.close();
    }

    IndexReader open = DirectoryReader.open(dir);
    assertEquals(1, open.numDocs());
    open.close();
    docs.close();
    dir.close();
  }
  private void doTest(Random random, PrintWriter out, boolean useCompoundFiles, int MAX_DOCS)
      throws Exception {
    Directory directory = newDirectory();
    Analyzer analyzer = new MockAnalyzer(random);
    IndexWriterConfig conf = newIndexWriterConfig(analyzer);
    final MergePolicy mp = conf.getMergePolicy();
    mp.setNoCFSRatio(useCompoundFiles ? 1.0 : 0.0);
    IndexWriter writer = new IndexWriter(directory, conf);
    if (VERBOSE) {
      System.out.println("TEST: now build index MAX_DOCS=" + MAX_DOCS);
    }

    for (int j = 0; j < MAX_DOCS; j++) {
      Document d = new Document();
      d.add(newTextField(PRIORITY_FIELD, HIGH_PRIORITY, Field.Store.YES));
      d.add(newTextField(ID_FIELD, Integer.toString(j), Field.Store.YES));
      writer.addDocument(d);
    }
    writer.close();

    // try a search without OR
    IndexReader reader = DirectoryReader.open(directory);
    IndexSearcher searcher = newSearcher(reader);

    Query query = new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY));
    out.println("Query: " + query.toString(PRIORITY_FIELD));
    if (VERBOSE) {
      System.out.println("TEST: search query=" + query);
    }

    final Sort sort = new Sort(SortField.FIELD_SCORE, new SortField(ID_FIELD, SortField.Type.INT));

    ScoreDoc[] hits = searcher.search(query, null, MAX_DOCS, sort).scoreDocs;
    printHits(out, hits, searcher);
    checkHits(hits, MAX_DOCS, searcher);

    // try a new search with OR
    searcher = newSearcher(reader);
    hits = null;

    BooleanQuery booleanQuery = new BooleanQuery();
    booleanQuery.add(
        new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY)), BooleanClause.Occur.SHOULD);
    booleanQuery.add(
        new TermQuery(new Term(PRIORITY_FIELD, MED_PRIORITY)), BooleanClause.Occur.SHOULD);
    out.println("Query: " + booleanQuery.toString(PRIORITY_FIELD));

    hits = searcher.search(booleanQuery, null, MAX_DOCS, sort).scoreDocs;
    printHits(out, hits, searcher);
    checkHits(hits, MAX_DOCS, searcher);

    reader.close();
    directory.close();
  }
 @Override
 public void run() {
   try {
     DirectoryReader open = null;
     for (int i = 0; i < num; i++) {
       Document doc = new Document(); // docs.nextDoc();
       BytesRef br = new BytesRef("test");
       doc.add(newStringField("id", br, Field.Store.NO));
       writer.updateDocument(new Term("id", br), doc);
       if (random().nextInt(3) == 0) {
         if (open == null) {
           open = DirectoryReader.open(writer, true);
         }
         DirectoryReader reader = DirectoryReader.openIfChanged(open);
         if (reader != null) {
           open.close();
           open = reader;
         }
         assertEquals(
             "iter: "
                 + i
                 + " numDocs: "
                 + open.numDocs()
                 + " del: "
                 + open.numDeletedDocs()
                 + " max: "
                 + open.maxDoc(),
             1,
             open.numDocs());
       }
     }
     if (open != null) {
       open.close();
     }
   } catch (Exception e) {
     throw new RuntimeException(e);
   }
 }
  public static void main(String[] args) throws IOException {

    IndexReader reader = null;

    /*
     *  Opening the index first simplifies the processing of the
     *  rest of the command line arguments.
     */
    for (int i = 0; i < args.length; i++) {
      if (("-index".equals(args[i])) && ((i + 1) < args.length)) {
        reader = DirectoryReader.open(FSDirectory.open(new File(args[i + 1])));

        if (reader == null) {
          System.err.println("Error:  Can't open index " + args[i + 1]);
          System.exit(1);
        }
        ;

        break;
      }
      ;
    }
    ;

    if (reader == null) {
      System.err.println(usage);
      System.exit(1);
    }
    ;

    /*
     *  Process the command line arguments sequentially.
     */
    for (int i = 0; i < args.length; i++) {

      if ("-index".equals(args[i])) {

        /*
         *  Handled in the previous loop, so just skip the argument.
         */
        i++;

      } else if ("-list-edocid".equals(args[i])) {

        System.out.println("-list-edocid:");

        if ((i + 1) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        Document d = reader.document(Integer.parseInt(args[i + 1]));

        System.out.println(
            "Internal docid --> External docid: " + args[i + 1] + " --> " + d.get("externalId"));

        i += 1;
      } else if ("-list-docids".equals(args[i])) {

        System.out.println("-list-docids:");

        for (int j = 0; j < reader.numDocs(); j++) {
          Document d = reader.document(j);
          System.out.println("Internal --> external docid: " + j + " --> " + d.get("externalId"));
        }
        ;

      } else if ("-list-fields".equals(args[i])) {

        Fields fields = MultiFields.getFields(reader);

        System.out.print("\nNumber of fields:  ");

        if (fields == null) System.out.println("0");
        else {
          System.out.println(fields.size());

          Iterator<String> is = fields.iterator();

          while (is.hasNext()) {
            System.out.println("\t" + is.next());
          }
          ;
        }
        ;

      } else if ("-list-postings".equals(args[i])) {

        if ((i + 2) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        listPostings(reader, args[i + 1], args[i + 2], Integer.MAX_VALUE);
        i += 2;

      } else if ("-list-postings-sample".equals(args[i])) {

        if ((i + 2) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        listPostings(reader, args[i + 1], args[i + 2], 5);
        i += 2;

      } else if ("-list-stats".equals(args[i])) {

        System.out.println("Corpus statistics:");
        System.out.println("\tnumdocs\t\t" + reader.numDocs());
        System.out.println(
            "\turl:\t"
                + "\tnumdocs="
                + reader.getDocCount("url")
                + "\tsumTotalTF="
                + reader.getSumTotalTermFreq("url")
                + "\tavglen="
                + reader.getSumTotalTermFreq("url") / (float) reader.getDocCount("url"));

        System.out.println(
            "\tkeywords:"
                + "\tnumdocs="
                + reader.getDocCount("keywords")
                + "\tsumTotalTF="
                + reader.getSumTotalTermFreq("keywords")
                + "\tavglen="
                + reader.getSumTotalTermFreq("keywords") / (float) reader.getDocCount("keywords"));

        System.out.println(
            "\ttitle:\t"
                + "\tnumdocs="
                + reader.getDocCount("title")
                + "\tsumTotalTF="
                + reader.getSumTotalTermFreq("title")
                + "\tavglen="
                + reader.getSumTotalTermFreq("title") / (float) reader.getDocCount("title"));

        System.out.println(
            "\tbody:\t"
                + "\tnumdocs="
                + reader.getDocCount("body")
                + "\tsumTotalTF="
                + reader.getSumTotalTermFreq("body")
                + "\tavglen="
                + reader.getSumTotalTermFreq("body") / (float) reader.getDocCount("body"));

        System.out.println(
            "\tinlink:\t"
                + "\tnumdocs="
                + reader.getDocCount("inlink")
                + "\tsumTotalTF="
                + reader.getSumTotalTermFreq("inlink")
                + "\tavglen="
                + reader.getSumTotalTermFreq("inlink") / (float) reader.getDocCount("inlink"));

      } else if ("-list-terms".equals(args[i])) {

        if ((i + 1) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        listTermDictionary(reader, args[i + 1]);
        i += 1;

      } else if ("-list-termvector".equals(args[i])) {

        if ((i + 1) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        listTermVectors(reader, args[i + 1]);
        i += 1;

      } else if ("-list-termvector-field".equals(args[i])) {

        if ((i + 2) >= args.length) {
          System.out.println(usage);
          break;
        }
        ;

        listTermVectorField(reader, args[i + 1], args[i + 2]);
        i += 2;

      } else System.err.println("\nWarning:  Unknown argument " + args[i] + " ignored.");
    }
    ;

    /*
     *  Close the index and exit gracefully.
     */
    reader.close();
  }
Beispiel #5
0
  @Test
  public void testMixedChecksums() throws IOException {
    final ShardId shardId = new ShardId(new Index("index"), 1);
    DirectoryService directoryService = new LuceneManagedDirectoryService(random());
    Store store =
        new Store(
            shardId,
            ImmutableSettings.EMPTY,
            directoryService,
            randomDistributor(directoryService),
            new DummyShardLock(shardId));
    // this time random codec....
    IndexWriter writer =
        new IndexWriter(
            store.directory(),
            newIndexWriterConfig(random(), new MockAnalyzer(random()))
                .setCodec(actualDefaultCodec()));
    int docs = 1 + random().nextInt(100);

    for (int i = 0; i < docs; i++) {
      Document doc = new Document();
      doc.add(
          new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new TextField(
              "body",
              TestUtil.randomRealisticUnicodeString(random()),
              random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new SortedDocValuesField(
              "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
      writer.addDocument(doc);
    }
    if (random().nextBoolean()) {
      for (int i = 0; i < docs; i++) {
        if (random().nextBoolean()) {
          Document doc = new Document();
          doc.add(
              new TextField(
                  "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          doc.add(
              new TextField(
                  "body",
                  TestUtil.randomRealisticUnicodeString(random()),
                  random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          writer.updateDocument(new Term("id", "" + i), doc);
        }
      }
    }
    if (random().nextBoolean()) {
      DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
    }
    Store.MetadataSnapshot metadata;
    // check before we committed
    try {
      store.getMetadata();
      fail("no index present - expected exception");
    } catch (IndexNotFoundException ex) {
      // expected
    }
    assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed
    writer.commit();
    writer.close();
    Store.LegacyChecksums checksums = new Store.LegacyChecksums();
    metadata = store.getMetadata();
    assertThat(metadata.asMap().isEmpty(), is(false));
    for (StoreFileMetaData meta : metadata) {
      try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
        if (meta.checksum() == null) {
          String checksum = null;
          try {
            CodecUtil.retrieveChecksum(input);
            fail("expected a corrupt index - posting format has not checksums");
          } catch (CorruptIndexException
              | IndexFormatTooOldException
              | IndexFormatTooNewException ex) {
            try (ChecksumIndexInput checksumIndexInput =
                store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) {
              checksumIndexInput.seek(meta.length());
              checksum = Store.digestToString(checksumIndexInput.getChecksum());
            }
            // fine - it's a postings format without checksums
            checksums.add(new StoreFileMetaData(meta.name(), meta.length(), checksum, null));
          }
        } else {
          String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
          assertThat(
              "File: " + meta.name() + " has a different checksum",
              meta.checksum(),
              equalTo(checksum));
          assertThat(meta.hasLegacyChecksum(), equalTo(false));
          assertThat(meta.writtenBy(), equalTo(Version.LATEST));
        }
      }
    }
    assertConsistent(store, metadata);
    checksums.write(store);
    metadata = store.getMetadata();
    assertThat(metadata.asMap().isEmpty(), is(false));
    for (StoreFileMetaData meta : metadata) {
      assertThat(
          "file: " + meta.name() + " has a null checksum", meta.checksum(), not(nullValue()));
      if (meta.hasLegacyChecksum()) {
        try (ChecksumIndexInput checksumIndexInput =
            store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) {
          checksumIndexInput.seek(meta.length());
          assertThat(
              meta.checksum(), equalTo(Store.digestToString(checksumIndexInput.getChecksum())));
        }
      } else {
        try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
          String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
          assertThat(
              "File: " + meta.name() + " has a different checksum",
              meta.checksum(),
              equalTo(checksum));
          assertThat(meta.hasLegacyChecksum(), equalTo(false));
          assertThat(meta.writtenBy(), equalTo(Version.LATEST));
        }
      }
    }
    assertConsistent(store, metadata);
    TestUtil.checkIndex(store.directory());
    assertDeleteContent(store, directoryService);
    IOUtils.close(store);
  }
Beispiel #6
0
  @Test
  public void testNewChecksums() throws IOException {
    final ShardId shardId = new ShardId(new Index("index"), 1);
    DirectoryService directoryService = new LuceneManagedDirectoryService(random());
    Store store =
        new Store(
            shardId,
            ImmutableSettings.EMPTY,
            directoryService,
            randomDistributor(directoryService),
            new DummyShardLock(shardId));
    // set default codec - all segments need checksums
    IndexWriter writer =
        new IndexWriter(
            store.directory(),
            newIndexWriterConfig(random(), new MockAnalyzer(random()))
                .setCodec(actualDefaultCodec()));
    int docs = 1 + random().nextInt(100);

    for (int i = 0; i < docs; i++) {
      Document doc = new Document();
      doc.add(
          new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new TextField(
              "body",
              TestUtil.randomRealisticUnicodeString(random()),
              random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new SortedDocValuesField(
              "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
      writer.addDocument(doc);
    }
    if (random().nextBoolean()) {
      for (int i = 0; i < docs; i++) {
        if (random().nextBoolean()) {
          Document doc = new Document();
          doc.add(
              new TextField(
                  "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          doc.add(
              new TextField(
                  "body",
                  TestUtil.randomRealisticUnicodeString(random()),
                  random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          writer.updateDocument(new Term("id", "" + i), doc);
        }
      }
    }
    if (random().nextBoolean()) {
      DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
    }
    Store.MetadataSnapshot metadata;
    // check before we committed
    try {
      store.getMetadata();
      fail("no index present - expected exception");
    } catch (IndexNotFoundException ex) {
      // expected
    }
    assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed
    writer.commit();
    writer.close();
    metadata = store.getMetadata();
    assertThat(metadata.asMap().isEmpty(), is(false));
    for (StoreFileMetaData meta : metadata) {
      try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
        String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
        assertThat(
            "File: " + meta.name() + " has a different checksum",
            meta.checksum(),
            equalTo(checksum));
        assertThat(meta.hasLegacyChecksum(), equalTo(false));
        assertThat(meta.writtenBy(), equalTo(Version.LATEST));
        if (meta.name().endsWith(".si") || meta.name().startsWith("segments_")) {
          assertThat(meta.hash().length, greaterThan(0));
        }
      }
    }
    assertConsistent(store, metadata);

    TestUtil.checkIndex(store.directory());
    assertDeleteContent(store, directoryService);
    IOUtils.close(store);
  }
Beispiel #7
0
  // IF THIS TEST FAILS ON UPGRADE GO LOOK AT THE
  // OldSIMockingCodec!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
  @Test
  public void testWriteLegacyChecksums() throws IOException {
    final ShardId shardId = new ShardId(new Index("index"), 1);
    DirectoryService directoryService = new LuceneManagedDirectoryService(random());
    Store store =
        new Store(
            shardId,
            ImmutableSettings.EMPTY,
            directoryService,
            randomDistributor(directoryService),
            new DummyShardLock(shardId));
    // set default codec - all segments need checksums
    final boolean usesOldCodec = randomBoolean();
    IndexWriter writer =
        new IndexWriter(
            store.directory(),
            newIndexWriterConfig(random(), new MockAnalyzer(random()))
                .setCodec(usesOldCodec ? new OldSIMockingCodec() : actualDefaultCodec()));
    int docs = 1 + random().nextInt(100);

    for (int i = 0; i < docs; i++) {
      Document doc = new Document();
      doc.add(
          new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new TextField(
              "body",
              TestUtil.randomRealisticUnicodeString(random()),
              random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
      doc.add(
          new SortedDocValuesField(
              "dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
      writer.addDocument(doc);
    }
    if (random().nextBoolean()) {
      for (int i = 0; i < docs; i++) {
        if (random().nextBoolean()) {
          Document doc = new Document();
          doc.add(
              new TextField(
                  "id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          doc.add(
              new TextField(
                  "body",
                  TestUtil.randomRealisticUnicodeString(random()),
                  random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
          writer.updateDocument(new Term("id", "" + i), doc);
        }
      }
    }
    if (random().nextBoolean()) {
      DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
    }
    Store.MetadataSnapshot metadata;
    // check before we committed
    try {
      store.getMetadata();
      fail("no index present - expected exception");
    } catch (IndexNotFoundException ex) {
      // expected
    }
    assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed

    writer.close();
    Store.LegacyChecksums checksums = new Store.LegacyChecksums();
    Map<String, StoreFileMetaData> legacyMeta = new HashMap<>();
    for (String file : store.directory().listAll()) {
      if (file.equals("write.lock") || file.equals(IndexFileNames.OLD_SEGMENTS_GEN)) {
        continue;
      }
      BytesRef hash = new BytesRef();
      if (file.startsWith("segments")) {
        hash = Store.MetadataSnapshot.hashFile(store.directory(), file);
      }
      StoreFileMetaData storeFileMetaData =
          new StoreFileMetaData(
              file, store.directory().fileLength(file), file + "checksum", null, hash);
      legacyMeta.put(file, storeFileMetaData);
      checksums.add(storeFileMetaData);
    }
    checksums.write(store);

    metadata = store.getMetadata();
    Map<String, StoreFileMetaData> stringStoreFileMetaDataMap = metadata.asMap();
    assertThat(legacyMeta.size(), equalTo(stringStoreFileMetaDataMap.size()));
    if (usesOldCodec) {
      for (StoreFileMetaData meta : legacyMeta.values()) {
        assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name()));
        assertEquals(meta.name() + "checksum", meta.checksum());
        assertTrue(
            meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()),
            stringStoreFileMetaDataMap.get(meta.name()).isSame(meta));
      }
    } else {

      // even if we have a legacy checksum - if we use a new codec we should reuse
      for (StoreFileMetaData meta : legacyMeta.values()) {
        assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name()));
        assertFalse(
            meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()),
            stringStoreFileMetaDataMap.get(meta.name()).isSame(meta));
        StoreFileMetaData storeFileMetaData = metadata.get(meta.name());
        try (IndexInput input =
            store.openVerifyingInput(meta.name(), IOContext.DEFAULT, storeFileMetaData)) {
          assertTrue(storeFileMetaData.toString(), input instanceof Store.VerifyingIndexInput);
          input.seek(meta.length());
          Store.verify(input);
        }
      }
    }
    assertDeleteContent(store, directoryService);
    IOUtils.close(store);
  }