Examples with SortedSetDocValuesField org.apache.lucene.document.SortedSetDocValuesField used on opensource projects

Example 21 with SortedSetDocValuesField

use of org.apache.lucene.document.SortedSetDocValuesField in project lucene-solr by apache.

the class TestIndexSorting method testRandom2.

public void testRandom2() throws Exception {
    int numDocs = atLeast(100);
    FieldType POSITIONS_TYPE = new FieldType(TextField.TYPE_NOT_STORED);
    POSITIONS_TYPE.setIndexOptions(IndexOptions.DOCS_AND_FREQS_AND_POSITIONS_AND_OFFSETS);
    POSITIONS_TYPE.freeze();
    FieldType TERM_VECTORS_TYPE = new FieldType(TextField.TYPE_NOT_STORED);
    TERM_VECTORS_TYPE.setStoreTermVectors(true);
    TERM_VECTORS_TYPE.freeze();
    Analyzer a = new Analyzer() {

        @Override
        protected TokenStreamComponents createComponents(String fieldName) {
            Tokenizer tokenizer = new MockTokenizer();
            return new TokenStreamComponents(tokenizer, tokenizer);
        }
    };
    List<Document> docs = new ArrayList<>();
    for (int i = 0; i < numDocs; i++) {
        int id = i * 10;
        Document doc = new Document();
        doc.add(new StringField("id", Integer.toString(id), Store.YES));
        doc.add(new StringField("docs", "#all#", Store.NO));
        PositionsTokenStream positions = new PositionsTokenStream();
        positions.setId(id);
        doc.add(new Field("positions", positions, POSITIONS_TYPE));
        doc.add(new NumericDocValuesField("numeric", id));
        String value = IntStream.range(0, id).mapToObj(k -> Integer.toString(id)).collect(Collectors.joining(" "));
        TextField norms = new TextField("norms", value, Store.NO);
        doc.add(norms);
        doc.add(new BinaryDocValuesField("binary", new BytesRef(Integer.toString(id))));
        doc.add(new SortedDocValuesField("sorted", new BytesRef(Integer.toString(id))));
        doc.add(new SortedSetDocValuesField("multi_valued_string", new BytesRef(Integer.toString(id))));
        doc.add(new SortedSetDocValuesField("multi_valued_string", new BytesRef(Integer.toString(id + 1))));
        doc.add(new SortedNumericDocValuesField("multi_valued_numeric", id));
        doc.add(new SortedNumericDocValuesField("multi_valued_numeric", id + 1));
        doc.add(new Field("term_vectors", Integer.toString(id), TERM_VECTORS_TYPE));
        byte[] bytes = new byte[4];
        NumericUtils.intToSortableBytes(id, bytes, 0);
        doc.add(new BinaryPoint("points", bytes));
        docs.add(doc);
    }
    // Must use the same seed for both RandomIndexWriters so they behave identically
    long seed = random().nextLong();
    // We add document alread in ID order for the first writer:
    Directory dir1 = newFSDirectory(createTempDir());
    Random random1 = new Random(seed);
    IndexWriterConfig iwc1 = newIndexWriterConfig(random1, a);
    // for testing norms field
    iwc1.setSimilarity(new NormsSimilarity(iwc1.getSimilarity()));
    // preserve docIDs
    iwc1.setMergePolicy(newLogMergePolicy());
    if (VERBOSE) {
        System.out.println("TEST: now index pre-sorted");
    }
    RandomIndexWriter w1 = new RandomIndexWriter(random1, dir1, iwc1);
    for (Document doc : docs) {
        ((PositionsTokenStream) ((Field) doc.getField("positions")).tokenStreamValue()).setId(Integer.parseInt(doc.get("id")));
        w1.addDocument(doc);
    }
    // We shuffle documents, but set index sort, for the second writer:
    Directory dir2 = newFSDirectory(createTempDir());
    Random random2 = new Random(seed);
    IndexWriterConfig iwc2 = newIndexWriterConfig(random2, a);
    // for testing norms field
    iwc2.setSimilarity(new NormsSimilarity(iwc2.getSimilarity()));
    Sort sort = new Sort(new SortField("numeric", SortField.Type.INT));
    iwc2.setIndexSort(sort);
    Collections.shuffle(docs, random());
    if (VERBOSE) {
        System.out.println("TEST: now index with index-time sorting");
    }
    RandomIndexWriter w2 = new RandomIndexWriter(random2, dir2, iwc2);
    int count = 0;
    int commitAtCount = TestUtil.nextInt(random(), 1, numDocs - 1);
    for (Document doc : docs) {
        ((PositionsTokenStream) ((Field) doc.getField("positions")).tokenStreamValue()).setId(Integer.parseInt(doc.get("id")));
        if (count++ == commitAtCount) {
            // Ensure forceMerge really does merge
            w2.commit();
        }
        w2.addDocument(doc);
    }
    if (VERBOSE) {
        System.out.println("TEST: now force merge");
    }
    w2.forceMerge(1);
    DirectoryReader r1 = w1.getReader();
    DirectoryReader r2 = w2.getReader();
    if (VERBOSE) {
        System.out.println("TEST: now compare r1=" + r1 + " r2=" + r2);
    }
    assertEquals(sort, getOnlyLeafReader(r2).getMetaData().getSort());
    assertReaderEquals("left: sorted by hand; right: sorted by Lucene", r1, r2);
    IOUtils.close(w1, w2, r1, r2, dir1, dir2);
}

Example 22 with SortedSetDocValuesField

use of org.apache.lucene.document.SortedSetDocValuesField in project lucene-solr by apache.

the class TestIndexSorting method testRandom3.

// pits index time sorting against query time sorting
public void testRandom3() throws Exception {
    int numDocs;
    if (TEST_NIGHTLY) {
        numDocs = atLeast(100000);
    } else {
        numDocs = atLeast(1000);
    }
    List<RandomDoc> docs = new ArrayList<>();
    Sort sort = randomSort();
    if (VERBOSE) {
        System.out.println("TEST: numDocs=" + numDocs + " use sort=" + sort);
    }
    // no index sorting, all search-time sorting:
    Directory dir1 = newFSDirectory(createTempDir());
    IndexWriterConfig iwc1 = newIndexWriterConfig(new MockAnalyzer(random()));
    IndexWriter w1 = new IndexWriter(dir1, iwc1);
    // use index sorting:
    Directory dir2 = newFSDirectory(createTempDir());
    IndexWriterConfig iwc2 = newIndexWriterConfig(new MockAnalyzer(random()));
    iwc2.setIndexSort(sort);
    IndexWriter w2 = new IndexWriter(dir2, iwc2);
    Set<Integer> toDelete = new HashSet<>();
    double deleteChance = random().nextDouble();
    for (int id = 0; id < numDocs; id++) {
        RandomDoc docValues = new RandomDoc(id);
        docs.add(docValues);
        if (VERBOSE) {
            System.out.println("TEST: doc id=" + id);
            System.out.println("  int=" + docValues.intValue);
            System.out.println("  long=" + docValues.longValue);
            System.out.println("  float=" + docValues.floatValue);
            System.out.println("  double=" + docValues.doubleValue);
            System.out.println("  bytes=" + new BytesRef(docValues.bytesValue));
        }
        Document doc = new Document();
        doc.add(new StringField("id", Integer.toString(id), Field.Store.YES));
        doc.add(new NumericDocValuesField("id", id));
        doc.add(new NumericDocValuesField("int", docValues.intValue));
        doc.add(new NumericDocValuesField("long", docValues.longValue));
        doc.add(new DoubleDocValuesField("double", docValues.doubleValue));
        doc.add(new FloatDocValuesField("float", docValues.floatValue));
        doc.add(new SortedDocValuesField("bytes", new BytesRef(docValues.bytesValue)));
        for (int value : docValues.intValues) {
            doc.add(new SortedNumericDocValuesField("multi_valued_int", value));
        }
        for (long value : docValues.longValues) {
            doc.add(new SortedNumericDocValuesField("multi_valued_long", value));
        }
        for (float value : docValues.floatValues) {
            doc.add(new SortedNumericDocValuesField("multi_valued_float", NumericUtils.floatToSortableInt(value)));
        }
        for (double value : docValues.doubleValues) {
            doc.add(new SortedNumericDocValuesField("multi_valued_double", NumericUtils.doubleToSortableLong(value)));
        }
        for (byte[] value : docValues.bytesValues) {
            doc.add(new SortedSetDocValuesField("multi_valued_bytes", new BytesRef(value)));
        }
        w1.addDocument(doc);
        w2.addDocument(doc);
        if (random().nextDouble() < deleteChance) {
            toDelete.add(id);
        }
    }
    for (int id : toDelete) {
        w1.deleteDocuments(new Term("id", Integer.toString(id)));
        w2.deleteDocuments(new Term("id", Integer.toString(id)));
    }
    DirectoryReader r1 = DirectoryReader.open(w1);
    IndexSearcher s1 = newSearcher(r1);
    if (random().nextBoolean()) {
        int maxSegmentCount = TestUtil.nextInt(random(), 1, 5);
        if (VERBOSE) {
            System.out.println("TEST: now forceMerge(" + maxSegmentCount + ")");
        }
        w2.forceMerge(maxSegmentCount);
    }
    DirectoryReader r2 = DirectoryReader.open(w2);
    IndexSearcher s2 = newSearcher(r2);
    for (int iter = 0; iter < 100; iter++) {
        int numHits = TestUtil.nextInt(random(), 1, numDocs);
        if (VERBOSE) {
            System.out.println("TEST: iter=" + iter + " numHits=" + numHits);
        }
        TopFieldCollector c1 = TopFieldCollector.create(sort, numHits, true, true, true);
        s1.search(new MatchAllDocsQuery(), c1);
        TopDocs hits1 = c1.topDocs();
        TopFieldCollector c2 = TopFieldCollector.create(sort, numHits, true, true, true);
        EarlyTerminatingSortingCollector c3 = new EarlyTerminatingSortingCollector(c2, sort, numHits);
        s2.search(new MatchAllDocsQuery(), c3);
        TopDocs hits2 = c2.topDocs();
        if (VERBOSE) {
            System.out.println("  topDocs query-time sort: totalHits=" + hits1.totalHits);
            for (ScoreDoc scoreDoc : hits1.scoreDocs) {
                System.out.println("    " + scoreDoc.doc);
            }
            System.out.println("  topDocs index-time sort: totalHits=" + hits2.totalHits);
            for (ScoreDoc scoreDoc : hits2.scoreDocs) {
                System.out.println("    " + scoreDoc.doc);
            }
        }
        assertTrue(hits2.totalHits <= hits1.totalHits);
        assertEquals(hits2.scoreDocs.length, hits1.scoreDocs.length);
        for (int i = 0; i < hits2.scoreDocs.length; i++) {
            ScoreDoc hit1 = hits1.scoreDocs[i];
            ScoreDoc hit2 = hits2.scoreDocs[i];
            assertEquals(r1.document(hit1.doc).get("id"), r2.document(hit2.doc).get("id"));
            assertEquals(((FieldDoc) hit1).fields, ((FieldDoc) hit2).fields);
        }
    }
    IOUtils.close(r1, r2, w1, w2, dir1, dir2);
}

Example 23 with SortedSetDocValuesField

use of org.apache.lucene.document.SortedSetDocValuesField in project lucene-solr by apache.

the class TestIndexSorting method testBasicMultiValuedString.

public void testBasicMultiValuedString() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortedSetSortField("foo", false));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new NumericDocValuesField("id", 3));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("zzz")));
    w.addDocument(doc);
    // so we get more than one segment, so that forceMerge actually does merge, since we only get a sorted segment by merging:
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("id", 1));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("aaa")));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("zzz")));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("bcg")));
    w.addDocument(doc);
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("id", 2));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("mmm")));
    doc.add(new SortedSetDocValuesField("foo", new BytesRef("pppp")));
    w.addDocument(doc);
    w.forceMerge(1);
    DirectoryReader r = DirectoryReader.open(w);
    LeafReader leaf = getOnlyLeafReader(r);
    assertEquals(3, leaf.maxDoc());
    NumericDocValues values = leaf.getNumericDocValues("id");
    assertEquals(0, values.nextDoc());
    assertEquals(1l, values.longValue());
    assertEquals(1, values.nextDoc());
    assertEquals(2l, values.longValue());
    assertEquals(2, values.nextDoc());
    assertEquals(3l, values.longValue());
    r.close();
    w.close();
    dir.close();
}

Example 24 with SortedSetDocValuesField

use of org.apache.lucene.document.SortedSetDocValuesField in project lucene-solr by apache.

the class TestSortedSetSelector method testMaxMissingFirst.

public void testMaxMissingFirst() throws Exception {
    Directory dir = newDirectory();
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir);
    Document doc = new Document();
    doc.add(newStringField("id", "1", Field.Store.YES));
    writer.addDocument(doc);
    doc = new Document();
    doc.add(new SortedSetDocValuesField("value", new BytesRef("foo")));
    doc.add(new SortedSetDocValuesField("value", new BytesRef("bar")));
    doc.add(newStringField("id", "2", Field.Store.YES));
    writer.addDocument(doc);
    doc = new Document();
    doc.add(new SortedSetDocValuesField("value", new BytesRef("baz")));
    doc.add(newStringField("id", "3", Field.Store.YES));
    writer.addDocument(doc);
    IndexReader ir = writer.getReader();
    writer.close();
    // slow wrapper does not support random access ordinals (there is no need for that!)
    IndexSearcher searcher = newSearcher(ir, false);
    SortField sortField = new SortedSetSortField("value", false, SortedSetSelector.Type.MAX);
    sortField.setMissingValue(SortField.STRING_FIRST);
    Sort sort = new Sort(sortField);
    TopDocs td = searcher.search(new MatchAllDocsQuery(), 10, sort);
    assertEquals(3, td.totalHits);
    // null comes first
    assertEquals("1", searcher.doc(td.scoreDocs[0].doc).get("id"));
    // 'baz' comes before 'foo'
    assertEquals("3", searcher.doc(td.scoreDocs[1].doc).get("id"));
    assertEquals("2", searcher.doc(td.scoreDocs[2].doc).get("id"));
    ir.close();
    dir.close();
}

Example 25 with SortedSetDocValuesField

use of org.apache.lucene.document.SortedSetDocValuesField in project lucene-solr by apache.

the class TestSortedSetSelector method testMiddleMinMissingFirst.

public void testMiddleMinMissingFirst() throws Exception {
    Directory dir = newDirectory();
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir);
    Document doc = new Document();
    doc.add(newStringField("id", "3", Field.Store.YES));
    writer.addDocument(doc);
    doc = new Document();
    doc.add(new SortedSetDocValuesField("value", new BytesRef("c")));
    doc.add(newStringField("id", "2", Field.Store.YES));
    writer.addDocument(doc);
    doc = new Document();
    doc.add(new SortedSetDocValuesField("value", new BytesRef("a")));
    doc.add(new SortedSetDocValuesField("value", new BytesRef("b")));
    doc.add(new SortedSetDocValuesField("value", new BytesRef("c")));
    doc.add(new SortedSetDocValuesField("value", new BytesRef("d")));
    doc.add(newStringField("id", "1", Field.Store.YES));
    writer.addDocument(doc);
    IndexReader ir = writer.getReader();
    writer.close();
    // slow wrapper does not support random access ordinals (there is no need for that!)
    IndexSearcher searcher = newSearcher(ir, false);
    SortField sortField = new SortedSetSortField("value", false, SortedSetSelector.Type.MIDDLE_MIN);
    sortField.setMissingValue(SortField.STRING_FIRST);
    Sort sort = new Sort(sortField);
    TopDocs td = searcher.search(new MatchAllDocsQuery(), 10, sort);
    assertEquals(3, td.totalHits);
    // null comes first
    assertEquals("3", searcher.doc(td.scoreDocs[0].doc).get("id"));
    // 'b' comes before 'c'
    assertEquals("1", searcher.doc(td.scoreDocs[1].doc).get("id"));
    assertEquals("2", searcher.doc(td.scoreDocs[2].doc).get("id"));
    ir.close();
    dir.close();
}