Fix #480: Multi Vector > 10000 documents throws scoring error

redisvl/query/aggregate.py

@@ -344,8 +344,13 @@ def __init__(
         super().__init__(query_string)
 
         # calculate the respective vector similarities
+        # use exists() guard so documents where Redis did not yield a distance
+        # (e.g. when the index contains >10 000 entries) get score 0 instead
+        # of raising "Could not find the value for a parameter name".
         for i in range(len(self._vectors)):
-            self.apply(**{f"score_{i}": f"(2 - @distance_{i})/2"})
+            self.apply(
+                **{f"score_{i}": f"if(exists(@distance_{i}), (2 - @distance_{i})/2, 0)"}
+            )
 
         # construct the scoring string based on the vector similarity scores and weights
         combined_scores = []

tests/unit/test_aggregation_types.py

@@ -391,7 +391,7 @@ def test_multi_vector_query_string():
 
     assert (
         str(multi_vector_query)
-        == f"@{field_1}:[VECTOR_RANGE {max_distance_1} $vector_0]=>{{$YIELD_DISTANCE_AS: distance_0}} AND @{field_2}:[VECTOR_RANGE {max_distance_2} $vector_1]=>{{$YIELD_DISTANCE_AS: distance_1}} SCORER TFIDF DIALECT 2 APPLY (2 - @distance_0)/2 AS score_0 APPLY (2 - @distance_1)/2 AS score_1 APPLY @score_0 * {weight_1} + @score_1 * {weight_2} AS combined_score SORTBY 2 @combined_score DESC MAX 10"
+        == f"@{field_1}:[VECTOR_RANGE {max_distance_1} $vector_0]=>{{$YIELD_DISTANCE_AS: distance_0}} AND @{field_2}:[VECTOR_RANGE {max_distance_2} $vector_1]=>{{$YIELD_DISTANCE_AS: distance_1}} SCORER TFIDF DIALECT 2 APPLY if(exists(@distance_0), (2 - @distance_0)/2, 0) AS score_0 APPLY if(exists(@distance_1), (2 - @distance_1)/2, 0) AS score_1 APPLY @score_0 * {weight_1} + @score_1 * {weight_2} AS combined_score SORTBY 2 @combined_score DESC MAX 10"
     )