[LUCENE-9017] GroupingSearch does not return all the groups when using SortedSetDocValuesField

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 8.2
• Component/s: core/search
• Lucene Fields: New
• Assignee: Unassigned
• Reporter: Juraj Kucera
• Created: 17/Oct/19 15:22
• Updated: 28/Aug/22 15:52

Description

I managed to retrieve the groups using the SortedSetDocValuesField in
GroupingSearch by initializing the groupsearch with SortedSetFieldSource

The problem is when a document has multiple values in the field
"SortedSetDocValuesField" than not the grouping query does not return all
the groups.

Let me demonstrate it in my example

// indexing, the first object has the category "one" and the second object
has category "two" and "three"

Document doc = new Document();
doc.add(new FacetField("Author", "Bob"));
doc.add(new SortedSetDocValuesField("category", new BytesRef("one")));
indexWriter.addDocument(config.build(taxoWriter, doc));

doc = new Document();
doc.add(new FacetField("Author", "Lisa"));
doc.add(new SortedSetDocValuesField("category", new BytesRef("two")));
doc.add(new SortedSetDocValuesField("category", new BytesRef("three")));
indexWriter.addDocument(config.build(taxoWriter, doc));

// initializing the grouping search
ValueSource vs = new SortedSetFieldSource(groupField);
groupingSearch = new GroupingSearch(vs, new HashMap<>());

// performing the group search
TopGroups groups = groupingSearch.search(searcher, new MatchAllDocsQuery(),
0, 100);

It returns 2 groups only and I would expect 3 groups ("one", "two" and
"three")

Is it a bug or am I using the API in a wrong way?