richardstartin commented on a change in pull request #8411:
URL: https://github.com/apache/pinot/pull/8411#discussion_r837782085
##########
File path:
pinot-core/src/main/java/org/apache/pinot/core/operator/filter/BitmapBasedFilterOperator.java
##########
@@ -100,6 +101,67 @@ protected FilterBlock getNextBlock() {
}
}
+ @Override
+ public boolean canOptimizeCount() {
+ return true;
+ }
+
+ @Override
+ public int getNumMatchingDocs() {
+ int count = 0;
+ if (_invertedIndexReader == null) {
+ count = _docIds.getCardinality();
+ } else {
+ int[] dictIds = _exclusive
+ ? _predicateEvaluator.getNonMatchingDictIds()
+ : _predicateEvaluator.getMatchingDictIds();
+ switch (dictIds.length) {
+ case 0:
+ break;
+ case 1: {
+ count = _invertedIndexReader.getDocIds(dictIds[0]).getCardinality();
+ break;
+ }
+ case 2: {
+ count =
ImmutableRoaringBitmap.orCardinality(_invertedIndexReader.getDocIds(dictIds[0]),
+ _invertedIndexReader.getDocIds(dictIds[1]));
+ break;
+ }
+ default: {
+ // this could be optimised if the bitmaps are known to be disjoint
(as in a single value bitmap index)
Review comment:
I didn't assume disjunction because it would break e.g. for inverted
indexes used to index tokenised text (values broken into several values). The
SV <=> disjoint invariant isn't enforced anywhere, so I don't want to rely on
it and end up with incorrect counts until there is enforcement.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
