mirrors/tantivy
2
0
Fork 1
mirror of https://github.com/quickwit-oss/tantivy.git synced 2026-06-19 09:16:45 +00:00
Code Issues Projects Releases Packages Wiki Activity

clarify counts/term_counts field docs

Browse source 
Spell out that `counts` is the flattened per-term × time-bucket grid (each
term's own contiguous slice) and that `term_counts` is only needed when the
per-term total can't be derived from that grid (i.e. with hard bounds).
This commit is contained in:
Pascal Seitz 2026-06-12 08:18:47 +02:00 • committed by PSeitz
commit 3cb400c300
1 changed files with 8 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

13
src/aggregation/bucket/term_agg/term_histogram.rs
View file

@ -41,13 +41,16 @@ const MAX_FUSED_GRID_BUCKETS: usize = 16384;
/// general path.
#[derive(Debug)]
pub(crate) struct SegmentTermHistogramCollector {
/// `[num_terms]` total doc count per term bucket (independent of the histogram bounds).
/// `u32` is enough: a per-segment count can't exceed the segment's doc count (`DocId` is
/// `u32`); the fused path is only taken when `num_docs < u32::MAX` (see
/// `maybe_build_collector`).
/// The counts, indexed by term id.
///
/// Only used when we can't derive the term counts from `counts`.
term_counts: Vec<u32>,
/// Flat row-major `[num_terms * num_time_buckets]` histogram counters (`u32`, see
/// Flattened `[num_terms * num_time_buckets]` histogram counters (`u32`, see
/// `term_counts`).
///
/// Each term id get its own contiguous slice of `num_time_buckets` histogram counter.
/// When we count all docs (#nofilter), we can derive the per-term total as the sum over that
/// term's slice.
counts: Vec<u32>,
/// Histogram buckets per term (the dense time-range length).
num_time_buckets: usize,