Examples with DirectoryReader org.apache.lucene.index.DirectoryReader used on opensource projects

Example 66 with DirectoryReader

use of org.apache.lucene.index.DirectoryReader in project lucene-solr by apache.

the class TestSuggestField method testExtremeDeduplication.

public void testExtremeDeduplication() throws Exception {
    Analyzer analyzer = new MockAnalyzer(random());
    RandomIndexWriter iw = new RandomIndexWriter(random(), dir, iwcWithSuggestField(analyzer, "suggest_field"));
    final int num = atLeast(5000);
    int bestWeight = Integer.MIN_VALUE;
    for (int i = 0; i < num; i++) {
        Document document = new Document();
        int weight = TestUtil.nextInt(random(), 10, 100);
        bestWeight = Math.max(weight, bestWeight);
        document.add(new SuggestField("suggest_field", "abc", weight));
        iw.addDocument(document);
        if (rarely()) {
            iw.commit();
        }
    }
    Document document = new Document();
    document.add(new SuggestField("suggest_field", "abd", 7));
    iw.addDocument(document);
    if (random().nextBoolean()) {
        iw.forceMerge(1);
    }
    DirectoryReader reader = iw.getReader();
    Entry[] expectedEntries = new Entry[2];
    expectedEntries[0] = new Entry("abc", bestWeight);
    expectedEntries[1] = new Entry("abd", 7);
    SuggestIndexSearcher suggestIndexSearcher = new SuggestIndexSearcher(reader);
    PrefixCompletionQuery query = new PrefixCompletionQuery(analyzer, new Term("suggest_field", "a"));
    TopSuggestDocsCollector collector = new TopSuggestDocsCollector(2, true);
    suggestIndexSearcher.suggest(query, collector);
    TopSuggestDocs lookupDocs = collector.get();
    assertSuggestions(lookupDocs, expectedEntries);
    reader.close();
    iw.close();
}

Example 67 with DirectoryReader

use of org.apache.lucene.index.DirectoryReader in project lucene-solr by apache.

the class TestSuggestField method testSuggestOnAllDeletedDocuments.

@Test
public void testSuggestOnAllDeletedDocuments() throws Exception {
    Analyzer analyzer = new MockAnalyzer(random());
    // using IndexWriter instead of RandomIndexWriter
    IndexWriter iw = new IndexWriter(dir, iwcWithSuggestField(analyzer, "suggest_field"));
    int num = Math.min(1000, atLeast(10));
    for (int i = 0; i < num; i++) {
        Document document = new Document();
        document.add(new SuggestField("suggest_field", "abc_" + i, i));
        document.add(newStringField("delete", "delete", Field.Store.NO));
        iw.addDocument(document);
        if (usually()) {
            iw.commit();
        }
    }
    iw.deleteDocuments(new Term("delete", "delete"));
    DirectoryReader reader = DirectoryReader.open(iw);
    SuggestIndexSearcher indexSearcher = new SuggestIndexSearcher(reader);
    PrefixCompletionQuery query = new PrefixCompletionQuery(analyzer, new Term("suggest_field", "abc_"));
    TopSuggestDocs suggest = indexSearcher.suggest(query, num, false);
    assertThat(suggest.totalHits, equalTo(0));
    reader.close();
    iw.close();
}

Example 68 with DirectoryReader

use of org.apache.lucene.index.DirectoryReader in project lucene-solr by apache.

the class TestUtil method addIndexesSlowly.

public static void addIndexesSlowly(IndexWriter writer, DirectoryReader... readers) throws IOException {
    List<CodecReader> leaves = new ArrayList<>();
    for (DirectoryReader reader : readers) {
        for (LeafReaderContext context : reader.leaves()) {
            leaves.add(SlowCodecReaderWrapper.wrap(context.reader()));
        }
    }
    writer.addIndexes(leaves.toArray(new CodecReader[leaves.size()]));
}

Example 69 with DirectoryReader

use of org.apache.lucene.index.DirectoryReader in project lucene-solr by apache.

the class TestCompressingTermVectorsFormat method testChunkCleanup.

/**
   * writes some tiny segments with incomplete compressed blocks,
   * and ensures merge recompresses them.
   */
public void testChunkCleanup() throws IOException {
    Directory dir = newDirectory();
    IndexWriterConfig iwConf = newIndexWriterConfig(new MockAnalyzer(random()));
    iwConf.setMergePolicy(NoMergePolicy.INSTANCE);
    // we have to enforce certain things like maxDocsPerChunk to cause dirty chunks to be created
    // by this test.
    iwConf.setCodec(CompressingCodec.randomInstance(random(), 4 * 1024, 100, false, 8));
    IndexWriter iw = new IndexWriter(dir, iwConf);
    DirectoryReader ir = DirectoryReader.open(iw);
    for (int i = 0; i < 5; i++) {
        Document doc = new Document();
        FieldType ft = new FieldType(TextField.TYPE_NOT_STORED);
        ft.setStoreTermVectors(true);
        doc.add(new Field("text", "not very long at all", ft));
        iw.addDocument(doc);
        // force flush
        DirectoryReader ir2 = DirectoryReader.openIfChanged(ir);
        assertNotNull(ir2);
        ir.close();
        ir = ir2;
        // examine dirty counts:
        for (LeafReaderContext leaf : ir2.leaves()) {
            CodecReader sr = (CodecReader) leaf.reader();
            CompressingTermVectorsReader reader = (CompressingTermVectorsReader) sr.getTermVectorsReader();
            assertEquals(1, reader.getNumChunks());
            assertEquals(1, reader.getNumDirtyChunks());
        }
    }
    iw.getConfig().setMergePolicy(newLogMergePolicy());
    iw.forceMerge(1);
    DirectoryReader ir2 = DirectoryReader.openIfChanged(ir);
    assertNotNull(ir2);
    ir.close();
    ir = ir2;
    CodecReader sr = (CodecReader) getOnlyLeafReader(ir);
    CompressingTermVectorsReader reader = (CompressingTermVectorsReader) sr.getTermVectorsReader();
    // we could get lucky, and have zero, but typically one.
    assertTrue(reader.getNumDirtyChunks() <= 1);
    ir.close();
    iw.close();
    dir.close();
}

Example 70 with DirectoryReader

use of org.apache.lucene.index.DirectoryReader in project lucene-solr by apache.

the class DocSetUtil method createDocSet.

public static DocSet createDocSet(SolrIndexSearcher searcher, Term term) throws IOException {
    // raw reader to avoid extra wrapping overhead
    DirectoryReader reader = searcher.getRawReader();
    int maxDoc = searcher.getIndexReader().maxDoc();
    int smallSetSize = smallSetSize(maxDoc);
    String field = term.field();
    BytesRef termVal = term.bytes();
    int maxCount = 0;
    int firstReader = -1;
    List<LeafReaderContext> leaves = reader.leaves();
    // use array for slightly higher scanning cost, but fewer memory allocations
    PostingsEnum[] postList = new PostingsEnum[leaves.size()];
    for (LeafReaderContext ctx : leaves) {
        assert leaves.get(ctx.ord) == ctx;
        LeafReader r = ctx.reader();
        Fields f = r.fields();
        Terms t = f.terms(field);
        // field is missing
        if (t == null)
            continue;
        TermsEnum te = t.iterator();
        if (te.seekExact(termVal)) {
            maxCount += te.docFreq();
            postList[ctx.ord] = te.postings(null, PostingsEnum.NONE);
            if (firstReader < 0)
                firstReader = ctx.ord;
        }
    }
    DocSet answer = null;
    if (maxCount == 0) {
        answer = DocSet.EMPTY;
    } else if (maxCount <= smallSetSize) {
        answer = createSmallSet(leaves, postList, maxCount, firstReader);
    } else {
        answer = createBigSet(leaves, postList, maxDoc, firstReader);
    }
    return DocSetUtil.getDocSet(answer, searcher);
}