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

Example 6 with Sort

use of org.apache.lucene.search.Sort in project elasticsearch by elastic.

the class NestedSortingTests method testNestedSorting.

public void testNestedSorting() throws Exception {
    List<Document> docs = new ArrayList<>();
    Document document = new Document();
    document.add(new StringField("field2", "a", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "b", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "c", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "a", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    writer.commit();
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "c", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "d", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "e", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "b", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "e", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "f", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "g", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "c", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "g", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "h", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "i", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "d", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    writer.commit();
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "i", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "j", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "k", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "f", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "k", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "l", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "m", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "g", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    // This doc will not be included, because it doesn't have nested docs
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "h", Field.Store.NO));
    writer.addDocument(document);
    docs.clear();
    document = new Document();
    document.add(new StringField("field2", "m", Field.Store.NO));
    document.add(new StringField("filter_1", "T", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "n", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("field2", "o", Field.Store.NO));
    document.add(new StringField("filter_1", "F", Field.Store.NO));
    docs.add(document);
    document = new Document();
    document.add(new StringField("__type", "parent", Field.Store.NO));
    document.add(new StringField("field1", "i", Field.Store.NO));
    docs.add(document);
    writer.addDocuments(docs);
    writer.commit();
    // Some garbage docs, just to check if the NestedFieldComparator can deal with this.
    document = new Document();
    document.add(new StringField("fieldXXX", "x", Field.Store.NO));
    writer.addDocument(document);
    document = new Document();
    document.add(new StringField("fieldXXX", "x", Field.Store.NO));
    writer.addDocument(document);
    document = new Document();
    document.add(new StringField("fieldXXX", "x", Field.Store.NO));
    writer.addDocument(document);
    MultiValueMode sortMode = MultiValueMode.MIN;
    DirectoryReader reader = DirectoryReader.open(writer);
    reader = ElasticsearchDirectoryReader.wrap(reader, new ShardId(indexService.index(), 0));
    IndexSearcher searcher = new IndexSearcher(reader);
    PagedBytesIndexFieldData indexFieldData = getForField("field2");
    Query parentFilter = new TermQuery(new Term("__type", "parent"));
    Query childFilter = Queries.not(parentFilter);
    BytesRefFieldComparatorSource nestedComparatorSource = new BytesRefFieldComparatorSource(indexFieldData, null, sortMode, createNested(searcher, parentFilter, childFilter));
    ToParentBlockJoinQuery query = new ToParentBlockJoinQuery(new ConstantScoreQuery(childFilter), new QueryBitSetProducer(parentFilter), ScoreMode.None);
    Sort sort = new Sort(new SortField("field2", nestedComparatorSource));
    TopFieldDocs topDocs = searcher.search(query, 5, sort);
    assertThat(topDocs.totalHits, equalTo(7));
    assertThat(topDocs.scoreDocs.length, equalTo(5));
    assertThat(topDocs.scoreDocs[0].doc, equalTo(3));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).utf8ToString(), equalTo("a"));
    assertThat(topDocs.scoreDocs[1].doc, equalTo(7));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).utf8ToString(), equalTo("c"));
    assertThat(topDocs.scoreDocs[2].doc, equalTo(11));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).utf8ToString(), equalTo("e"));
    assertThat(topDocs.scoreDocs[3].doc, equalTo(15));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).utf8ToString(), equalTo("g"));
    assertThat(topDocs.scoreDocs[4].doc, equalTo(19));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).utf8ToString(), equalTo("i"));
    sortMode = MultiValueMode.MAX;
    nestedComparatorSource = new BytesRefFieldComparatorSource(indexFieldData, null, sortMode, createNested(searcher, parentFilter, childFilter));
    sort = new Sort(new SortField("field2", nestedComparatorSource, true));
    topDocs = searcher.search(query, 5, sort);
    assertThat(topDocs.totalHits, equalTo(7));
    assertThat(topDocs.scoreDocs.length, equalTo(5));
    assertThat(topDocs.scoreDocs[0].doc, equalTo(28));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).utf8ToString(), equalTo("o"));
    assertThat(topDocs.scoreDocs[1].doc, equalTo(23));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).utf8ToString(), equalTo("m"));
    assertThat(topDocs.scoreDocs[2].doc, equalTo(19));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).utf8ToString(), equalTo("k"));
    assertThat(topDocs.scoreDocs[3].doc, equalTo(15));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).utf8ToString(), equalTo("i"));
    assertThat(topDocs.scoreDocs[4].doc, equalTo(11));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).utf8ToString(), equalTo("g"));
    BooleanQuery.Builder bq = new BooleanQuery.Builder();
    bq.add(parentFilter, Occur.MUST_NOT);
    bq.add(new TermQuery(new Term("filter_1", "T")), Occur.MUST);
    childFilter = bq.build();
    nestedComparatorSource = new BytesRefFieldComparatorSource(indexFieldData, null, sortMode, createNested(searcher, parentFilter, childFilter));
    query = new ToParentBlockJoinQuery(new ConstantScoreQuery(childFilter), new QueryBitSetProducer(parentFilter), ScoreMode.None);
    sort = new Sort(new SortField("field2", nestedComparatorSource, true));
    topDocs = searcher.search(query, 5, sort);
    assertThat(topDocs.totalHits, equalTo(6));
    assertThat(topDocs.scoreDocs.length, equalTo(5));
    assertThat(topDocs.scoreDocs[0].doc, equalTo(23));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).utf8ToString(), equalTo("m"));
    assertThat(topDocs.scoreDocs[1].doc, equalTo(28));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).utf8ToString(), equalTo("m"));
    assertThat(topDocs.scoreDocs[2].doc, equalTo(11));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).utf8ToString(), equalTo("g"));
    assertThat(topDocs.scoreDocs[3].doc, equalTo(15));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).utf8ToString(), equalTo("g"));
    assertThat(topDocs.scoreDocs[4].doc, equalTo(7));
    assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).utf8ToString(), equalTo("e"));
    searcher.getIndexReader().close();
}

Example 7 with Sort

use of org.apache.lucene.search.Sort in project elasticsearch by elastic.

the class AbstractNumberNestedSortingTestCase method assertAvgScoreMode.

protected void assertAvgScoreMode(Query parentFilter, IndexSearcher searcher) throws IOException {
    MultiValueMode sortMode = MultiValueMode.AVG;
    Query childFilter = Queries.not(parentFilter);
    XFieldComparatorSource nestedComparatorSource = createFieldComparator("field2", sortMode, -127, createNested(searcher, parentFilter, childFilter));
    Query query = new ToParentBlockJoinQuery(new ConstantScoreQuery(childFilter), new QueryBitSetProducer(parentFilter), ScoreMode.None);
    Sort sort = new Sort(new SortField("field2", nestedComparatorSource));
    TopDocs topDocs = searcher.search(query, 5, sort);
    assertThat(topDocs.totalHits, equalTo(7));
    assertThat(topDocs.scoreDocs.length, equalTo(5));
    assertThat(topDocs.scoreDocs[0].doc, equalTo(11));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).intValue(), equalTo(2));
    assertThat(topDocs.scoreDocs[1].doc, equalTo(3));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).intValue(), equalTo(3));
    assertThat(topDocs.scoreDocs[2].doc, equalTo(7));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).intValue(), equalTo(3));
    assertThat(topDocs.scoreDocs[3].doc, equalTo(15));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).intValue(), equalTo(3));
    assertThat(topDocs.scoreDocs[4].doc, equalTo(19));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).intValue(), equalTo(4));
}

Example 8 with Sort

use of org.apache.lucene.search.Sort in project elasticsearch by elastic.

the class FloatNestedSortingTests method assertAvgScoreMode.

protected void assertAvgScoreMode(Query parentFilter, IndexSearcher searcher, IndexFieldData.XFieldComparatorSource innerFieldComparator) throws IOException {
    MultiValueMode sortMode = MultiValueMode.AVG;
    Query childFilter = Queries.not(parentFilter);
    XFieldComparatorSource nestedComparatorSource = createFieldComparator("field2", sortMode, -127, createNested(searcher, parentFilter, childFilter));
    Query query = new ToParentBlockJoinQuery(new ConstantScoreQuery(childFilter), new QueryBitSetProducer(parentFilter), ScoreMode.None);
    Sort sort = new Sort(new SortField("field2", nestedComparatorSource));
    TopDocs topDocs = searcher.search(query, 5, sort);
    assertThat(topDocs.totalHits, equalTo(7));
    assertThat(topDocs.scoreDocs.length, equalTo(5));
    assertThat(topDocs.scoreDocs[0].doc, equalTo(11));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).intValue(), equalTo(2));
    assertThat(topDocs.scoreDocs[1].doc, equalTo(7));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).intValue(), equalTo(2));
    assertThat(topDocs.scoreDocs[2].doc, equalTo(3));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).intValue(), equalTo(3));
    assertThat(topDocs.scoreDocs[3].doc, equalTo(15));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).intValue(), equalTo(3));
    assertThat(topDocs.scoreDocs[4].doc, equalTo(19));
    assertThat(((Number) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).intValue(), equalTo(3));
}

Example 9 with Sort

use of org.apache.lucene.search.Sort in project elasticsearch by elastic.

the class SortBuilder method buildSort.

public static Optional<SortAndFormats> buildSort(List<SortBuilder<?>> sortBuilders, QueryShardContext context) throws IOException {
    List<SortField> sortFields = new ArrayList<>(sortBuilders.size());
    List<DocValueFormat> sortFormats = new ArrayList<>(sortBuilders.size());
    for (SortBuilder<?> builder : sortBuilders) {
        SortFieldAndFormat sf = builder.build(context);
        sortFields.add(sf.field);
        sortFormats.add(sf.format);
    }
    if (!sortFields.isEmpty()) {
        // optimize if we just sort on score non reversed, we don't really
        // need sorting
        boolean sort;
        if (sortFields.size() > 1) {
            sort = true;
        } else {
            SortField sortField = sortFields.get(0);
            if (sortField.getType() == SortField.Type.SCORE && !sortField.getReverse()) {
                sort = false;
            } else {
                sort = true;
            }
        }
        if (sort) {
            return Optional.of(new SortAndFormats(new Sort(sortFields.toArray(new SortField[sortFields.size()])), sortFormats.toArray(new DocValueFormat[sortFormats.size()])));
        }
    }
    return Optional.empty();
}

Example 10 with Sort

use of org.apache.lucene.search.Sort in project elasticsearch by elastic.

the class CollapsingTopDocsCollectorTests method assertSearchCollapse.

private <T extends Comparable> void assertSearchCollapse(CollapsingDocValuesProducer<T> dvProducers, boolean numeric, boolean multivalued) throws IOException {
    final int numDocs = randomIntBetween(1000, 2000);
    int maxGroup = randomIntBetween(2, 500);
    final Directory dir = newDirectory();
    final RandomIndexWriter w = new RandomIndexWriter(random(), dir);
    Set<T> values = new HashSet<>();
    int totalHits = 0;
    for (int i = 0; i < numDocs; i++) {
        final T value = dvProducers.randomGroup(maxGroup);
        values.add(value);
        Document doc = new Document();
        dvProducers.add(doc, value, multivalued);
        doc.add(new NumericDocValuesField("sort1", randomIntBetween(0, 10)));
        doc.add(new NumericDocValuesField("sort2", randomLong()));
        w.addDocument(doc);
        totalHits++;
    }
    List<T> valueList = new ArrayList<>(values);
    Collections.sort(valueList);
    final IndexReader reader = w.getReader();
    final IndexSearcher searcher = newSearcher(reader);
    final SortField collapseField = dvProducers.sortField(multivalued);
    final SortField sort1 = new SortField("sort1", SortField.Type.INT);
    final SortField sort2 = new SortField("sort2", SortField.Type.LONG);
    Sort sort = new Sort(sort1, sort2, collapseField);
    int expectedNumGroups = values.size();
    final CollapsingTopDocsCollector collapsingCollector;
    if (numeric) {
        collapsingCollector = CollapsingTopDocsCollector.createNumeric(collapseField.getField(), sort, expectedNumGroups, false);
    } else {
        collapsingCollector = CollapsingTopDocsCollector.createKeyword(collapseField.getField(), sort, expectedNumGroups, false);
    }
    TopFieldCollector topFieldCollector = TopFieldCollector.create(sort, totalHits, true, false, false);
    searcher.search(new MatchAllDocsQuery(), collapsingCollector);
    searcher.search(new MatchAllDocsQuery(), topFieldCollector);
    CollapseTopFieldDocs collapseTopFieldDocs = collapsingCollector.getTopDocs();
    TopFieldDocs topDocs = topFieldCollector.topDocs();
    assertEquals(collapseField.getField(), collapseTopFieldDocs.field);
    assertEquals(expectedNumGroups, collapseTopFieldDocs.scoreDocs.length);
    assertEquals(totalHits, collapseTopFieldDocs.totalHits);
    assertEquals(totalHits, topDocs.scoreDocs.length);
    assertEquals(totalHits, topDocs.totalHits);
    Set<Object> seen = new HashSet<>();
    // collapse field is the last sort
    int collapseIndex = sort.getSort().length - 1;
    int topDocsIndex = 0;
    for (int i = 0; i < expectedNumGroups; i++) {
        FieldDoc fieldDoc = null;
        for (; topDocsIndex < totalHits; topDocsIndex++) {
            fieldDoc = (FieldDoc) topDocs.scoreDocs[topDocsIndex];
            if (seen.contains(fieldDoc.fields[collapseIndex]) == false) {
                break;
            }
        }
        FieldDoc collapseFieldDoc = (FieldDoc) collapseTopFieldDocs.scoreDocs[i];
        assertNotNull(fieldDoc);
        assertEquals(collapseFieldDoc.doc, fieldDoc.doc);
        assertArrayEquals(collapseFieldDoc.fields, fieldDoc.fields);
        seen.add(fieldDoc.fields[fieldDoc.fields.length - 1]);
    }
    for (; topDocsIndex < totalHits; topDocsIndex++) {
        FieldDoc fieldDoc = (FieldDoc) topDocs.scoreDocs[topDocsIndex];
        assertTrue(seen.contains(fieldDoc.fields[collapseIndex]));
    }
    // check merge
    final IndexReaderContext ctx = searcher.getTopReaderContext();
    final SegmentSearcher[] subSearchers;
    final int[] docStarts;
    if (ctx instanceof LeafReaderContext) {
        subSearchers = new SegmentSearcher[1];
        docStarts = new int[1];
        subSearchers[0] = new SegmentSearcher((LeafReaderContext) ctx, ctx);
        docStarts[0] = 0;
    } else {
        final CompositeReaderContext compCTX = (CompositeReaderContext) ctx;
        final int size = compCTX.leaves().size();
        subSearchers = new SegmentSearcher[size];
        docStarts = new int[size];
        int docBase = 0;
        for (int searcherIDX = 0; searcherIDX < subSearchers.length; searcherIDX++) {
            final LeafReaderContext leave = compCTX.leaves().get(searcherIDX);
            subSearchers[searcherIDX] = new SegmentSearcher(leave, compCTX);
            docStarts[searcherIDX] = docBase;
            docBase += leave.reader().maxDoc();
        }
    }
    final CollapseTopFieldDocs[] shardHits = new CollapseTopFieldDocs[subSearchers.length];
    final Weight weight = searcher.createNormalizedWeight(new MatchAllDocsQuery(), false);
    for (int shardIDX = 0; shardIDX < subSearchers.length; shardIDX++) {
        final SegmentSearcher subSearcher = subSearchers[shardIDX];
        final CollapsingTopDocsCollector c;
        if (numeric) {
            c = CollapsingTopDocsCollector.createNumeric(collapseField.getField(), sort, expectedNumGroups, false);
        } else {
            c = CollapsingTopDocsCollector.createKeyword(collapseField.getField(), sort, expectedNumGroups, false);
        }
        subSearcher.search(weight, c);
        shardHits[shardIDX] = c.getTopDocs();
    }
    CollapseTopFieldDocs mergedFieldDocs = CollapseTopFieldDocs.merge(sort, 0, expectedNumGroups, shardHits);
    assertTopDocsEquals(mergedFieldDocs, collapseTopFieldDocs);
    w.close();
    reader.close();
    dir.close();
}