Examples with SortField org.apache.lucene.search.SortField used on opensource projects

Example 66 with SortField

use of org.apache.lucene.search.SortField 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 67 with SortField

use of org.apache.lucene.search.SortField in project lucene-solr by apache.

the class TestIndexSorting method testMissingIntLast.

public void testMissingIntLast() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    SortField sortField = new SortField("foo", SortField.Type.INT);
    sortField.setMissingValue(Integer.valueOf(Integer.MAX_VALUE));
    Sort indexSort = new Sort(sortField);
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new NumericDocValuesField("foo", 18));
    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();
    // missing
    w.addDocument(new Document());
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("foo", 7));
    w.addDocument(doc);
    w.forceMerge(1);
    DirectoryReader r = DirectoryReader.open(w);
    LeafReader leaf = getOnlyLeafReader(r);
    assertEquals(3, leaf.maxDoc());
    NumericDocValues values = leaf.getNumericDocValues("foo");
    assertEquals(0, values.nextDoc());
    assertEquals(7, values.longValue());
    assertEquals(1, values.nextDoc());
    assertEquals(18, values.longValue());
    assertEquals(NO_MORE_DOCS, values.nextDoc());
    r.close();
    w.close();
    dir.close();
}

Example 68 with SortField

use of org.apache.lucene.search.SortField in project lucene-solr by apache.

the class TestIndexSorting method testMissingMultiValuedLongFirst.

public void testMissingMultiValuedLongFirst() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    SortField sortField = new SortedNumericSortField("foo", SortField.Type.LONG);
    sortField.setMissingValue(Long.valueOf(Long.MIN_VALUE));
    Sort indexSort = new Sort(sortField);
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new NumericDocValuesField("id", 3));
    doc.add(new SortedNumericDocValuesField("foo", 18));
    doc.add(new SortedNumericDocValuesField("foo", 27));
    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();
    // missing
    doc = new Document();
    doc.add(new NumericDocValuesField("id", 1));
    w.addDocument(doc);
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("id", 2));
    doc.add(new SortedNumericDocValuesField("foo", 7));
    doc.add(new SortedNumericDocValuesField("foo", 24));
    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(1, values.longValue());
    assertEquals(1, values.nextDoc());
    assertEquals(2, values.longValue());
    assertEquals(2, values.nextDoc());
    assertEquals(3, values.longValue());
    r.close();
    w.close();
    dir.close();
}

Example 69 with SortField

use of org.apache.lucene.search.SortField in project lucene-solr by apache.

the class TestIndexSorting method testBasicInt.

public void testBasicInt() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortField("foo", SortField.Type.INT));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new NumericDocValuesField("foo", 18));
    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("foo", -1));
    w.addDocument(doc);
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("foo", 7));
    w.addDocument(doc);
    w.forceMerge(1);
    DirectoryReader r = DirectoryReader.open(w);
    LeafReader leaf = getOnlyLeafReader(r);
    assertEquals(3, leaf.maxDoc());
    NumericDocValues values = leaf.getNumericDocValues("foo");
    assertEquals(0, values.nextDoc());
    assertEquals(-1, values.longValue());
    assertEquals(1, values.nextDoc());
    assertEquals(7, values.longValue());
    assertEquals(2, values.nextDoc());
    assertEquals(18, values.longValue());
    r.close();
    w.close();
    dir.close();
}

Example 70 with SortField

use of org.apache.lucene.search.SortField in project lucene-solr by apache.

the class TestIndexSorting method testIllegalChangeSort.

// you can't change the index sort on an existing index:
public void testIllegalChangeSort() throws Exception {
    final Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    iwc.setIndexSort(new Sort(new SortField("foo", SortField.Type.LONG)));
    IndexWriter w = new IndexWriter(dir, iwc);
    w.addDocument(new Document());
    DirectoryReader.open(w).close();
    w.addDocument(new Document());
    w.forceMerge(1);
    w.close();
    final IndexWriterConfig iwc2 = new IndexWriterConfig(new MockAnalyzer(random()));
    iwc2.setIndexSort(new Sort(new SortField("bar", SortField.Type.LONG)));
    IllegalArgumentException e = expectThrows(IllegalArgumentException.class, () -> {
        new IndexWriter(dir, iwc2);
    });
    String message = e.getMessage();
    assertTrue(message.contains("cannot change previous indexSort=<long: \"foo\">"));
    assertTrue(message.contains("to new indexSort=<long: \"bar\">"));
    dir.close();
}