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

Example 76 with NumericDocValuesField

use of org.apache.lucene.document.NumericDocValuesField 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 77 with NumericDocValuesField

use of org.apache.lucene.document.NumericDocValuesField 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 78 with NumericDocValuesField

use of org.apache.lucene.document.NumericDocValuesField 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 79 with NumericDocValuesField

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

the class TestIndexSorting method testBasicMultiValuedDouble.

public void testBasicMultiValuedDouble() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortedNumericSortField("foo", SortField.Type.DOUBLE));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new NumericDocValuesField("id", 3));
    doc.add(new SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(7.54)));
    doc.add(new SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(27.0)));
    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 SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(-1.0)));
    doc.add(new SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(0.0)));
    w.addDocument(doc);
    w.commit();
    doc = new Document();
    doc.add(new NumericDocValuesField("id", 2));
    doc.add(new SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(7.0)));
    doc.add(new SortedNumericDocValuesField("foo", NumericUtils.doubleToSortableLong(7.67)));
    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 80 with NumericDocValuesField

use of org.apache.lucene.document.NumericDocValuesField 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();
}