Andriy Redko created LUCENE-10466:
-------------------------------------
Summary: IndexSortSortedNumericDocValuesRangeQuery unconditionally
assumes the usage of the LONG-encoded SortField
Key: LUCENE-10466
URL: https://issues.apache.org/jira/browse/LUCENE-10466
Project: Lucene - Core
Issue Type: Bug
Components: core/query/scoring
Affects Versions: 9.0
Reporter: Andriy Redko
We have run into this issue while migrating to OpenSearch and making changes to
accommodate https://issues.apache.org/jira/browse/LUCENE-10040. It turned out
that *IndexSortSortedNumericDocValuesRangeQuery* unconditionally assumes the
usage of the LONG-encoded {*}SortField{*}, as could be seen inside *static
ValueComparator loadComparator* method
{noformat}
@SuppressWarnings("unchecked")
FieldComparator<Long> fieldComparator = (FieldComparator<Long>)
sortField.getComparator(1, 0);
fieldComparator.setTopValue(topValue);
{noformat}
Using the numeric range query (in case of sorted index) with anything but LONG
ends up with class cast exception:
{noformat}
> java.lang.ClassCastException: class java.lang.Long cannot be cast to
class java.lang.Integer (java.lang.Long and java.lang.Integer are in module
java.base of loader 'bootstrap')
> at
org.apache.lucene.search.comparators.IntComparator.setTopValue(IntComparator.java:29)
> at
org.apache.lucene.sandbox.search.IndexSortSortedNumericDocValuesRangeQuery.loadComparator(IndexSortSortedNumericDocValuesRangeQuery.java:251)
> at
org.apache.lucene.sandbox.search.IndexSortSortedNumericDocValuesRangeQuery.getDocIdSetIterator(IndexSortSortedNumericDocValuesRangeQuery.java:206)
> at
org.apache.lucene.sandbox.search.IndexSortSortedNumericDocValuesRangeQuery$1.scorer(IndexSortSortedNumericDocValuesRangeQuery.java:170)
{noformat}
Simple test case to reproduce (for
TestIndexSortSortedNumericDocValuesRangeQuery):
{noformat}
public void testIndexSortDocValuesWithIntRange() throws Exception {
Directory dir = newDirectory(); IndexWriterConfig iwc = new
IndexWriterConfig(new MockAnalyzer(random()));
Sort indexSort = new Sort(new SortedNumericSortField("field",
SortField.Type.INT, false));
iwc.setIndexSort(indexSort);
RandomIndexWriter writer = new RandomIndexWriter(random(), dir, iwc);
writer.addDocument(createDocument("field", -80)); DirectoryReader reader =
writer.getReader();
IndexSearcher searcher = newSearcher(reader); // Test ranges consisting
of one value.
assertEquals(1, searcher.count(createQuery("field", -80, -80)));
writer.close();
reader.close();
dir.close();
} {noformat}
The expectation is that *IndexSortSortedNumericDocValuesRangeQuery* should not
fail with class cast but correctly convert the numeric values.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
