rust/lancedb
1
0
Fork 0
mirror of https://github.com/lancedb/lancedb.git synced 2026-11-13 14:16:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(python): add CombinedFieldsQuery for BM25F full text search

Browse Source 
Add a typed Python `CombinedFieldsQuery` for BM25F search across multiple columns.

The query mirrors Lance's combined-fields semantics with ordered per-column boosts and AND/OR term matching. Convert it in both directions at the PyO3 boundary, including nested Boolean queries, and document the Python typed convenience entry.

Closes #4652.
This commit is contained in:
BHAWESHBHASKAR
2026-10-29 17:29:22 +08:00
committed by GitHub
parent 06b6625a94
commit 891774343e
5 changed files with 227 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/src/query.rs
+34 -5
View File
@@ -17,8 +17,8 @@ use arrow::pyarrow::FromPyArrow;
use arrow::pyarrow::IntoPyArrow;
use arrow::pyarrow::ToPyArrow;
use lancedb::index::scalar::{
BooleanQuery, BoostQuery, DocumentGranularity, FtsQuery, FullTextSearchQuery, MatchQuery,
MultiMatchQuery, Occur, Operator, PhraseQuery,
BooleanQuery, BoostQuery, CombinedFieldsQuery, DocumentGranularity, FtsQuery,
FullTextSearchQuery, MatchQuery, MultiMatchQuery, Occur, Operator, PhraseQuery,
};
use lancedb::query::AnalyzePlanDistributedMetrics;
use lancedb::query::QueryBase;
@@ -148,6 +148,26 @@ impl<'a, 'py> FromPyObject<'a, 'py> for PyLanceDB<FtsQuery> {
Ok(Self(q.with_operator(op).into()))
}
"CombinedFieldsQuery" => {
let query = ob.getattr("query")?.extract()?;
let columns = ob.getattr("columns")?.extract()?;
let boosts: Option<Vec<f32>> = ob.getattr("boosts")?.extract()?;
let operator: String = ob.getattr("operator")?.extract()?;
let q = CombinedFieldsQuery::try_new(query, columns)
.map_err(|e| PyValueError::new_err(format!("Invalid query: {}", e)))?;
let q = if let Some(boosts) = boosts {
q.try_with_boosts(boosts)
.map_err(|e| PyValueError::new_err(format!("Invalid boosts: {}", e)))?
} else {
q
};
let op = Operator::try_from(operator.as_str())
.map_err(|e| PyValueError::new_err(format!("Invalid operator: {}", e)))?;
Ok(Self(q.with_operator(op).into()))
}
"BooleanQuery" => {
let queries: Vec<(String, Self)> = ob.getattr("queries")?.extract()?;
let mut sub_queries = Vec::with_capacity(queries.len());
@@ -178,9 +198,6 @@ impl<'py> IntoPyObject<'py> for PyLanceDB<FtsQuery> {
.expect("Failed to import namespace");
match self.0 {
FtsQuery::CombinedFields(_) => Err(PyValueError::new_err(
"Combined-fields FTS queries are not yet supported in Python",
)),
FtsQuery::Match(query) => {
let kwargs = PyDict::new(py);
kwargs.set_item("boost", query.boost)?;
@@ -236,6 +253,18 @@ impl<'py> IntoPyObject<'py> for PyLanceDB<FtsQuery> {
.getattr(intern!(py, "MultiMatchQuery"))?
.call((first.terms.clone(), columns), Some(&kwargs))
}
FtsQuery::CombinedFields(query) => {
let (columns, boosts): (Vec<_>, Vec<_>) = query
.weighted_columns()
.map(|(column, boost)| (column.to_string(), boost))
.unzip();
let kwargs = PyDict::new(py);
kwargs.set_item("boosts", boosts)?;
kwargs.set_item::<_, &str>("operator", query.operator().into())?;
namespace
.getattr(intern!(py, "CombinedFieldsQuery"))?
.call((query.terms(), columns), Some(&kwargs))
}
FtsQuery::Boolean(query) => {
let mut queries: Vec<(&str, Bound<'py, PyAny>)> = Vec::with_capacity(
query.should.len() + query.must.len() + query.must_not.len(),