fix: improve unpacking of keyword search results (#46)

Author: lsorber

src/raglite/_search.py

@@ -129,9 +129,10 @@ def keyword_search(
             """)
             results = session.execute(statement, params={"match": fts5_query, "limit": num_results})
         # Unpack the results.
-        chunk_ids, keyword_score = zip(*results, strict=True)
-        chunk_ids, keyword_score = list(chunk_ids), list(keyword_score)  # type: ignore[assignment]
-    return chunk_ids, keyword_score  # type: ignore[return-value]
+        results = list(results)  # type: ignore[assignment]
+        chunk_ids = [result.chunk_id for result in results]
+        keyword_score = [result.score for result in results]
+    return chunk_ids, keyword_score
 
 
 def reciprocal_rank_fusion(

tests/test_search.py

@@ -46,3 +46,14 @@ def test_search(raglite_test_config: RAGLiteConfig, search_method: SearchMethod)
     # Extend the chunks with their neighbours and group them into contiguous segments.
     segments = retrieve_segments(chunk_ids, neighbors=(-1, 1), config=raglite_test_config)
     assert all(isinstance(segment, str) for segment in segments)
+
+
+def test_search_no_results(raglite_test_config: RAGLiteConfig, search_method: SearchMethod) -> None:
+    """Test searching for a query with no keyword search results."""
+    query = "supercalifragilisticexpialidocious"
+    num_results = 5
+    chunk_ids, scores = search_method(query, num_results=num_results, config=raglite_test_config)
+    num_results_expected = 0 if search_method == keyword_search else num_results
+    assert len(chunk_ids) == len(scores) == num_results_expected
+    assert all(isinstance(chunk_id, str) for chunk_id in chunk_ids)
+    assert all(isinstance(score, float) for score in scores)