use of org.apache.lucene.search.TopFieldDocs in project elasticsearch by elastic.
the class AbstractStringFieldDataTestCase method testNestedSorting.
public void testNestedSorting(MultiValueMode sortMode) throws IOException {
final String[] values = new String[randomIntBetween(2, 20)];
for (int i = 0; i < values.length; ++i) {
values[i] = TestUtil.randomSimpleString(random());
}
final int numParents = scaledRandomIntBetween(10, 3072);
List<Document> docs = new ArrayList<>();
FixedBitSet parents = new FixedBitSet(64);
for (int i = 0; i < numParents; ++i) {
docs.clear();
final int numChildren = randomInt(4);
for (int j = 0; j < numChildren; ++j) {
final Document child = new Document();
final int numValues = randomInt(3);
for (int k = 0; k < numValues; ++k) {
final String value = RandomPicks.randomFrom(random(), values);
addField(child, "text", value);
}
docs.add(child);
}
final Document parent = new Document();
parent.add(new StringField("type", "parent", Store.YES));
final String value = RandomPicks.randomFrom(random(), values);
if (value != null) {
addField(parent, "text", value);
}
docs.add(parent);
int bit = parents.prevSetBit(parents.length() - 1) + docs.size();
parents = FixedBitSet.ensureCapacity(parents, bit);
parents.set(bit);
writer.addDocuments(docs);
if (randomInt(10) == 0) {
writer.commit();
}
}
DirectoryReader directoryReader = DirectoryReader.open(writer);
directoryReader = ElasticsearchDirectoryReader.wrap(directoryReader, new ShardId(indexService.index(), 0));
IndexSearcher searcher = new IndexSearcher(directoryReader);
IndexFieldData<?> fieldData = getForField("text");
final Object missingValue;
switch(randomInt(4)) {
case 0:
missingValue = "_first";
break;
case 1:
missingValue = "_last";
break;
case 2:
missingValue = new BytesRef(RandomPicks.randomFrom(random(), values));
break;
default:
missingValue = new BytesRef(TestUtil.randomSimpleString(random()));
break;
}
Query parentFilter = new TermQuery(new Term("type", "parent"));
Query childFilter = Queries.not(parentFilter);
Nested nested = createNested(searcher, parentFilter, childFilter);
BytesRefFieldComparatorSource nestedComparatorSource = new BytesRefFieldComparatorSource(fieldData, missingValue, sortMode, nested);
ToParentBlockJoinQuery query = new ToParentBlockJoinQuery(new ConstantScoreQuery(childFilter), new QueryBitSetProducer(parentFilter), ScoreMode.None);
Sort sort = new Sort(new SortField("text", nestedComparatorSource));
TopFieldDocs topDocs = searcher.search(query, randomIntBetween(1, numParents), sort);
assertTrue(topDocs.scoreDocs.length > 0);
BytesRef previous = null;
for (int i = 0; i < topDocs.scoreDocs.length; ++i) {
final int docID = topDocs.scoreDocs[i].doc;
assertTrue("expected " + docID + " to be a parent", parents.get(docID));
BytesRef cmpValue = null;
for (int child = parents.prevSetBit(docID - 1) + 1; child < docID; ++child) {
String[] sVals = searcher.doc(child).getValues("text");
final BytesRef[] vals;
if (sVals.length == 0) {
vals = new BytesRef[0];
} else {
vals = new BytesRef[sVals.length];
for (int j = 0; j < vals.length; ++j) {
vals[j] = new BytesRef(sVals[j]);
}
}
for (BytesRef value : vals) {
if (cmpValue == null) {
cmpValue = value;
} else if (sortMode == MultiValueMode.MIN && value.compareTo(cmpValue) < 0) {
cmpValue = value;
} else if (sortMode == MultiValueMode.MAX && value.compareTo(cmpValue) > 0) {
cmpValue = value;
}
}
}
if (cmpValue == null) {
if ("_first".equals(missingValue)) {
cmpValue = new BytesRef();
} else if ("_last".equals(missingValue) == false) {
cmpValue = (BytesRef) missingValue;
}
}
if (previous != null && cmpValue != null) {
assertTrue(previous.utf8ToString() + " / " + cmpValue.utf8ToString(), previous.compareTo(cmpValue) <= 0);
}
previous = cmpValue;
}
searcher.getIndexReader().close();
}
use of org.apache.lucene.search.TopFieldDocs in project elasticsearch by elastic.
the class AbstractStringFieldDataTestCase method testSortMissing.
public void testSortMissing(boolean first, boolean reverse) throws IOException {
final String[] values = new String[randomIntBetween(2, 10)];
for (int i = 1; i < values.length; ++i) {
values[i] = TestUtil.randomUnicodeString(random());
}
final int numDocs = scaledRandomIntBetween(10, 3072);
for (int i = 0; i < numDocs; ++i) {
final String value = RandomPicks.randomFrom(random(), values);
if (value == null) {
writer.addDocument(new Document());
} else {
Document d = new Document();
addField(d, "value", value);
writer.addDocument(d);
}
if (randomInt(10) == 0) {
writer.commit();
}
}
final IndexFieldData indexFieldData = getForField("value");
IndexSearcher searcher = new IndexSearcher(DirectoryReader.open(writer));
XFieldComparatorSource comparator = indexFieldData.comparatorSource(first ? "_first" : "_last", MultiValueMode.MIN, null);
TopFieldDocs topDocs = searcher.search(new MatchAllDocsQuery(), randomBoolean() ? numDocs : randomIntBetween(10, numDocs), new Sort(new SortField("value", comparator, reverse)));
assertEquals(numDocs, topDocs.totalHits);
BytesRef previousValue = first ? null : reverse ? UnicodeUtil.BIG_TERM : new BytesRef();
for (int i = 0; i < topDocs.scoreDocs.length; ++i) {
final String docValue = searcher.doc(topDocs.scoreDocs[i].doc).get("value");
if (first && docValue == null) {
assertNull(previousValue);
} else if (!first && docValue != null) {
assertNotNull(previousValue);
}
final BytesRef value = docValue == null ? null : new BytesRef(docValue);
if (previousValue != null && value != null) {
if (reverse) {
assertTrue(previousValue.compareTo(value) >= 0);
} else {
assertTrue(previousValue.compareTo(value) <= 0);
}
}
previousValue = value;
}
searcher.getIndexReader().close();
}
use of org.apache.lucene.search.TopFieldDocs in project elasticsearch by elastic.
the class AbstractFieldDataImplTestCase method testSortMultiValuesFields.
public void testSortMultiValuesFields() throws Exception {
fillExtendedMvSet();
IndexFieldData indexFieldData = getForField("value");
IndexSearcher searcher = new IndexSearcher(DirectoryReader.open(writer));
TopFieldDocs topDocs = searcher.search(new MatchAllDocsQuery(), 10, new Sort(new SortField("value", indexFieldData.comparatorSource(null, MultiValueMode.MIN, null))));
assertThat(topDocs.totalHits, equalTo(8));
assertThat(topDocs.scoreDocs.length, equalTo(8));
assertThat(topDocs.scoreDocs[0].doc, equalTo(7));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).utf8ToString(), equalTo("!08"));
assertThat(topDocs.scoreDocs[1].doc, equalTo(0));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).utf8ToString(), equalTo("02"));
assertThat(topDocs.scoreDocs[2].doc, equalTo(2));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).utf8ToString(), equalTo("03"));
assertThat(topDocs.scoreDocs[3].doc, equalTo(3));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).utf8ToString(), equalTo("04"));
assertThat(topDocs.scoreDocs[4].doc, equalTo(4));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).utf8ToString(), equalTo("06"));
assertThat(topDocs.scoreDocs[5].doc, equalTo(6));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[5]).fields[0]).utf8ToString(), equalTo("08"));
assertThat(topDocs.scoreDocs[6].doc, equalTo(1));
assertThat((BytesRef) ((FieldDoc) topDocs.scoreDocs[6]).fields[0], equalTo(null));
assertThat(topDocs.scoreDocs[7].doc, equalTo(5));
assertThat((BytesRef) ((FieldDoc) topDocs.scoreDocs[7]).fields[0], equalTo(null));
topDocs = searcher.search(new MatchAllDocsQuery(), 10, new Sort(new SortField("value", indexFieldData.comparatorSource(null, MultiValueMode.MAX, null), true)));
assertThat(topDocs.totalHits, equalTo(8));
assertThat(topDocs.scoreDocs.length, equalTo(8));
assertThat(topDocs.scoreDocs[0].doc, equalTo(6));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[0]).fields[0]).utf8ToString(), equalTo("10"));
assertThat(topDocs.scoreDocs[1].doc, equalTo(4));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[1]).fields[0]).utf8ToString(), equalTo("08"));
assertThat(topDocs.scoreDocs[2].doc, equalTo(3));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[2]).fields[0]).utf8ToString(), equalTo("06"));
assertThat(topDocs.scoreDocs[3].doc, equalTo(0));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[3]).fields[0]).utf8ToString(), equalTo("04"));
assertThat(topDocs.scoreDocs[4].doc, equalTo(2));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[4]).fields[0]).utf8ToString(), equalTo("03"));
assertThat(topDocs.scoreDocs[5].doc, equalTo(7));
assertThat(((BytesRef) ((FieldDoc) topDocs.scoreDocs[5]).fields[0]).utf8ToString(), equalTo("!10"));
assertThat(topDocs.scoreDocs[6].doc, equalTo(1));
assertThat(((FieldDoc) topDocs.scoreDocs[6]).fields[0], equalTo(null));
assertThat(topDocs.scoreDocs[7].doc, equalTo(5));
assertThat(((FieldDoc) topDocs.scoreDocs[7]).fields[0], equalTo(null));
}
use of org.apache.lucene.search.TopFieldDocs in project elasticsearch by elastic.
the class AbstractStringFieldDataTestCase method testActualMissingValue.
public void testActualMissingValue(boolean reverse) throws IOException {
// missing value is set to an actual value
final String[] values = new String[randomIntBetween(2, 30)];
for (int i = 1; i < values.length; ++i) {
values[i] = TestUtil.randomUnicodeString(random());
}
final int numDocs = scaledRandomIntBetween(10, 3072);
for (int i = 0; i < numDocs; ++i) {
final String value = RandomPicks.randomFrom(random(), values);
if (value == null) {
writer.addDocument(new Document());
} else {
Document d = new Document();
addField(d, "value", value);
writer.addDocument(d);
}
if (randomInt(10) == 0) {
writer.commit();
}
}
final IndexFieldData indexFieldData = getForField("value");
final String missingValue = values[1];
IndexSearcher searcher = new IndexSearcher(DirectoryReader.open(writer));
XFieldComparatorSource comparator = indexFieldData.comparatorSource(missingValue, MultiValueMode.MIN, null);
TopFieldDocs topDocs = searcher.search(new MatchAllDocsQuery(), randomBoolean() ? numDocs : randomIntBetween(10, numDocs), new Sort(new SortField("value", comparator, reverse)));
assertEquals(numDocs, topDocs.totalHits);
BytesRef previousValue = reverse ? UnicodeUtil.BIG_TERM : new BytesRef();
for (int i = 0; i < topDocs.scoreDocs.length; ++i) {
final String docValue = searcher.doc(topDocs.scoreDocs[i].doc).get("value");
final BytesRef value = new BytesRef(docValue == null ? missingValue : docValue);
if (reverse) {
assertTrue(previousValue.compareTo(value) >= 0);
} else {
assertTrue(previousValue.compareTo(value) <= 0);
}
previousValue = value;
}
searcher.getIndexReader().close();
}
use of org.apache.lucene.search.TopFieldDocs 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();
}
Aggregations