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

Example 66 with Sort

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

the class TestIndexSorting method randomSort.

private static Sort randomSort() {
    // at least 2
    int numFields = TestUtil.nextInt(random(), 2, 4);
    SortField[] sortFields = new SortField[numFields];
    for (int i = 0; i < numFields - 1; i++) {
        SortField sortField = randomIndexSortField();
        sortFields[i] = sortField;
    }
    // tie-break by id:
    sortFields[numFields - 1] = new SortField("id", SortField.Type.INT);
    return new Sort(sortFields);
}

Example 67 with Sort

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

the class TestIndexSorting method testRandom1.

public void testRandom1() throws IOException {
    boolean withDeletes = random().nextBoolean();
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortField("foo", SortField.Type.LONG));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    final int numDocs = atLeast(1000);
    final FixedBitSet deleted = new FixedBitSet(numDocs);
    for (int i = 0; i < numDocs; ++i) {
        Document doc = new Document();
        doc.add(new NumericDocValuesField("foo", random().nextInt(20)));
        doc.add(new StringField("id", Integer.toString(i), Store.YES));
        doc.add(new NumericDocValuesField("id", i));
        w.addDocument(doc);
        if (random().nextInt(5) == 0) {
            w.getReader().close();
        } else if (random().nextInt(30) == 0) {
            w.forceMerge(2);
        } else if (random().nextInt(4) == 0) {
            final int id = TestUtil.nextInt(random(), 0, i);
            deleted.set(id);
            w.deleteDocuments(new Term("id", Integer.toString(id)));
        }
    }
    // Check that segments are sorted
    DirectoryReader reader = w.getReader();
    for (LeafReaderContext ctx : reader.leaves()) {
        final SegmentReader leaf = (SegmentReader) ctx.reader();
        SegmentInfo info = leaf.getSegmentInfo().info;
        switch(info.getDiagnostics().get(IndexWriter.SOURCE)) {
            case IndexWriter.SOURCE_FLUSH:
            case IndexWriter.SOURCE_MERGE:
                assertEquals(indexSort, info.getIndexSort());
                final NumericDocValues values = leaf.getNumericDocValues("foo");
                long previous = Long.MIN_VALUE;
                for (int i = 0; i < leaf.maxDoc(); ++i) {
                    assertEquals(i, values.nextDoc());
                    final long value = values.longValue();
                    assertTrue(value >= previous);
                    previous = value;
                }
                break;
            default:
                fail();
        }
    }
    // Now check that the index is consistent
    IndexSearcher searcher = newSearcher(reader);
    for (int i = 0; i < numDocs; ++i) {
        TermQuery termQuery = new TermQuery(new Term("id", Integer.toString(i)));
        final TopDocs topDocs = searcher.search(termQuery, 1);
        if (deleted.get(i)) {
            assertEquals(0, topDocs.totalHits);
        } else {
            assertEquals(1, topDocs.totalHits);
            NumericDocValues values = MultiDocValues.getNumericValues(reader, "id");
            assertEquals(topDocs.scoreDocs[0].doc, values.advance(topDocs.scoreDocs[0].doc));
            assertEquals(i, values.longValue());
            Document document = reader.document(topDocs.scoreDocs[0].doc);
            assertEquals(Integer.toString(i), document.get("id"));
        }
    }
    reader.close();
    w.close();
    dir.close();
}

Example 68 with Sort

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

the class TestIndexSorting method testBadDVUpdate.

// docvalues fields involved in the index sort cannot be updated
public void testBadDVUpdate() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortField("foo", SortField.Type.LONG));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new StringField("id", new BytesRef("0"), Store.NO));
    doc.add(new NumericDocValuesField("foo", random().nextInt()));
    w.addDocument(doc);
    w.commit();
    IllegalArgumentException exc = expectThrows(IllegalArgumentException.class, () -> w.updateDocValues(new Term("id", "0"), new NumericDocValuesField("foo", -1)));
    assertEquals(exc.getMessage(), "cannot update docvalues field involved in the index sort, field=foo, sort=<long: \"foo\">");
    exc = expectThrows(IllegalArgumentException.class, () -> w.updateNumericDocValue(new Term("id", "0"), "foo", -1));
    assertEquals(exc.getMessage(), "cannot update docvalues field involved in the index sort, field=foo, sort=<long: \"foo\">");
    w.close();
    dir.close();
}

Example 69 with Sort

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

the class TestIndexSorting method testBasicString.

public void testBasicString() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortField("foo", SortField.Type.STRING));
    iwc.setIndexSort(indexSort);
    IndexWriter w = new IndexWriter(dir, iwc);
    Document doc = new Document();
    doc.add(new SortedDocValuesField("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 SortedDocValuesField("foo", new BytesRef("aaa")));
    w.addDocument(doc);
    w.commit();
    doc = new Document();
    doc.add(new SortedDocValuesField("foo", new BytesRef("mmm")));
    w.addDocument(doc);
    w.forceMerge(1);
    DirectoryReader r = DirectoryReader.open(w);
    LeafReader leaf = getOnlyLeafReader(r);
    assertEquals(3, leaf.maxDoc());
    SortedDocValues values = leaf.getSortedDocValues("foo");
    assertEquals(0, values.nextDoc());
    assertEquals("aaa", values.binaryValue().utf8ToString());
    assertEquals(1, values.nextDoc());
    assertEquals("mmm", values.binaryValue().utf8ToString());
    assertEquals(2, values.nextDoc());
    assertEquals("zzz", values.binaryValue().utf8ToString());
    r.close();
    w.close();
    dir.close();
}

Example 70 with Sort

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

the class TestIndexSorting method testBasicLong.

public void testBasicLong() throws Exception {
    Directory dir = newDirectory();
    IndexWriterConfig iwc = new IndexWriterConfig(new MockAnalyzer(random()));
    Sort indexSort = new Sort(new SortField("foo", SortField.Type.LONG));
    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();
}