mirror of
https://github.com/quickwit-oss/tantivy.git
synced 2026-10-06 03:42:36 +00:00
fix term aggregation u32::MAX overflow issue
(cherry picked from commit 1e859fd78d)
This commit is contained in:
@@ -264,7 +264,7 @@ impl SegmentCompositeCollector {
|
||||
dict.insert(
|
||||
key,
|
||||
IntermediateCompositeBucketEntry {
|
||||
doc_count: agg.count,
|
||||
doc_count: agg.count as u64,
|
||||
sub_aggregation: sub_aggregation_res,
|
||||
},
|
||||
);
|
||||
|
||||
@@ -977,12 +977,12 @@ where
|
||||
bucket.bucket_id,
|
||||
)?;
|
||||
Ok(IntermediateTermBucketEntry {
|
||||
doc_count: bucket.count,
|
||||
doc_count: bucket.count as u64,
|
||||
sub_aggregation: sub_aggregation_res,
|
||||
})
|
||||
} else {
|
||||
Ok(IntermediateTermBucketEntry {
|
||||
doc_count: bucket.count,
|
||||
doc_count: bucket.count as u64,
|
||||
sub_aggregation: Default::default(),
|
||||
})
|
||||
}
|
||||
|
||||
@@ -98,7 +98,7 @@ impl SegmentAggregationCollector for TermMissingAgg {
|
||||
|
||||
let missing_count = &self.missing_count_per_bucket[parent_bucket_id as usize];
|
||||
let mut missing_entry = IntermediateTermBucketEntry {
|
||||
doc_count: missing_count.missing_count,
|
||||
doc_count: missing_count.missing_count as u64,
|
||||
sub_aggregation: Default::default(),
|
||||
};
|
||||
if let Some(sub_agg) = &mut self.sub_agg {
|
||||
|
||||
@@ -915,7 +915,7 @@ impl IntermediateRangeBucketEntry {
|
||||
#[derive(Clone, Default, Debug, PartialEq, Serialize, Deserialize)]
|
||||
pub struct IntermediateTermBucketEntry {
|
||||
/// The number of documents in the bucket.
|
||||
pub doc_count: u32,
|
||||
pub doc_count: u64,
|
||||
/// The sub_aggregation in this bucket.
|
||||
pub sub_aggregation: IntermediateAggregationResults,
|
||||
}
|
||||
@@ -1229,6 +1229,24 @@ mod tests {
|
||||
assert_eq!(tree_left, tree_expected);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn test_term_bucket_doc_count_no_u32_overflow() {
|
||||
// Two segments each contributing (u32::MAX - 100) docs to the same term. Summing them
|
||||
// overflowed when doc_count was u32.
|
||||
let per_segment = u32::MAX as u64 - 100;
|
||||
let mut entry = IntermediateTermBucketEntry {
|
||||
doc_count: per_segment,
|
||||
sub_aggregation: Default::default(),
|
||||
};
|
||||
entry
|
||||
.merge_fruits(IntermediateTermBucketEntry {
|
||||
doc_count: per_segment,
|
||||
sub_aggregation: Default::default(),
|
||||
})
|
||||
.unwrap();
|
||||
assert_eq!(entry.doc_count, per_segment * 2);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn test_merge_fruits_tree_empty() {
|
||||
let mut tree_left = get_intermediate_tree_with_ranges(&[
|
||||
|
||||
Reference in New Issue
Block a user