mirror of
https://github.com/meilisearch/meilisearch.git
synced 2024-11-26 03:55:07 +08:00
Facet distribution: fix issue where truncated facet distribution would have a wrong order
This commit is contained in:
parent
02c2b660f8
commit
e44325683a
@ -100,7 +100,6 @@ impl<'a> FacetDistribution<'a> {
|
|||||||
let mut lexicographic_distribution = BTreeMap::new();
|
let mut lexicographic_distribution = BTreeMap::new();
|
||||||
let mut key_buffer: Vec<_> = field_id.to_be_bytes().to_vec();
|
let mut key_buffer: Vec<_> = field_id.to_be_bytes().to_vec();
|
||||||
|
|
||||||
let distribution_prelength = distribution.len();
|
|
||||||
let db = self.index.field_id_docid_facet_f64s;
|
let db = self.index.field_id_docid_facet_f64s;
|
||||||
for docid in candidates {
|
for docid in candidates {
|
||||||
key_buffer.truncate(mem::size_of::<FieldId>());
|
key_buffer.truncate(mem::size_of::<FieldId>());
|
||||||
@ -113,23 +112,21 @@ impl<'a> FacetDistribution<'a> {
|
|||||||
for result in iter {
|
for result in iter {
|
||||||
let ((_, _, value), ()) = result?;
|
let ((_, _, value), ()) = result?;
|
||||||
*lexicographic_distribution.entry(value.to_string()).or_insert(0) += 1;
|
*lexicographic_distribution.entry(value.to_string()).or_insert(0) += 1;
|
||||||
|
|
||||||
if lexicographic_distribution.len() - distribution_prelength
|
|
||||||
== self.max_values_per_facet
|
|
||||||
{
|
|
||||||
break;
|
|
||||||
}
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
distribution.extend(lexicographic_distribution);
|
distribution.extend(
|
||||||
|
lexicographic_distribution
|
||||||
|
.into_iter()
|
||||||
|
.take(self.max_values_per_facet.saturating_sub(distribution.len())),
|
||||||
|
);
|
||||||
}
|
}
|
||||||
FacetType::String => {
|
FacetType::String => {
|
||||||
let mut normalized_distribution = BTreeMap::new();
|
let mut normalized_distribution = BTreeMap::new();
|
||||||
let mut key_buffer: Vec<_> = field_id.to_be_bytes().to_vec();
|
let mut key_buffer: Vec<_> = field_id.to_be_bytes().to_vec();
|
||||||
|
|
||||||
let db = self.index.field_id_docid_facet_strings;
|
let db = self.index.field_id_docid_facet_strings;
|
||||||
'outer: for docid in candidates {
|
for docid in candidates {
|
||||||
key_buffer.truncate(mem::size_of::<FieldId>());
|
key_buffer.truncate(mem::size_of::<FieldId>());
|
||||||
key_buffer.extend_from_slice(&docid.to_be_bytes());
|
key_buffer.extend_from_slice(&docid.to_be_bytes());
|
||||||
let iter = db
|
let iter = db
|
||||||
@ -144,14 +141,14 @@ impl<'a> FacetDistribution<'a> {
|
|||||||
.or_insert_with(|| (original_value, 0));
|
.or_insert_with(|| (original_value, 0));
|
||||||
*count += 1;
|
*count += 1;
|
||||||
|
|
||||||
if normalized_distribution.len() == self.max_values_per_facet {
|
// we'd like to break here if we have enough facet values, but we are collecting them by increasing docid,
|
||||||
break 'outer;
|
// so higher ranked facets could be in later docids
|
||||||
}
|
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
let iter = normalized_distribution
|
let iter = normalized_distribution
|
||||||
.into_iter()
|
.into_iter()
|
||||||
|
.take(self.max_values_per_facet.saturating_sub(distribution.len()))
|
||||||
.map(|(_normalized, (original, count))| (original.to_string(), count));
|
.map(|(_normalized, (original, count))| (original.to_string(), count));
|
||||||
distribution.extend(iter);
|
distribution.extend(iter);
|
||||||
}
|
}
|
||||||
@ -467,7 +464,7 @@ mod tests {
|
|||||||
.execute()
|
.execute()
|
||||||
.unwrap();
|
.unwrap();
|
||||||
|
|
||||||
milli_snap!(format!("{map:?}"), @r###"{"colour": {"Blue": 1}}"###);
|
milli_snap!(format!("{map:?}"), @r###"{"colour": {"Blue": 2}}"###);
|
||||||
|
|
||||||
let map = FacetDistribution::new(&txn, &index)
|
let map = FacetDistribution::new(&txn, &index)
|
||||||
.facets(iter::once(("colour", OrderBy::Count)))
|
.facets(iter::once(("colour", OrderBy::Count)))
|
||||||
|
Loading…
Reference in New Issue
Block a user