cleans text and hybrid tests

Author: justin-cechmanek

redisvl/index/index.py

@@ -42,6 +42,7 @@
 from redisvl.index.storage import BaseStorage, HashStorage, JsonStorage
 from redisvl.query import (
     AggregationQuery,
+    BaseVectorQuery,
     BaseQuery,
     CountQuery,
     FilterQuery,

redisvl/query/query.py

@@ -4,8 +4,9 @@
 from redis.commands.search.query import Query as RedisQuery
 
 from redisvl.query.filter import FilterExpression
-from redisvl.redis.utils import array_to_buffer, denorm_cosine_distance
+from redisvl.redis.utils import array_to_buffer
 from redisvl.utils.token_escaper import TokenEscaper
+from redisvl.utils.utils import denorm_cosine_distance
 
 
 class BaseQuery(RedisQuery):
@@ -811,7 +812,7 @@ def _build_query_string(self) -> str:
         else:
             filter_expression = ""
 
-        text = f"~(@{self._text_field}:({self.tokenize_and_escape_query(self._text)}))"
+        text = f"(~@{self._text_field}:({self.tokenize_and_escape_query(self._text)}))"
         if filter_expression and filter_expression != "*":
             text += f"({filter_expression})"
         return text

tests/unit/test_aggregation_types.py

@@ -13,8 +13,6 @@
 
 
 # Test Cases
-
-
 def test_aggregate_hybrid_query():
     text_field_name = "description"
     vector_field_name = "embedding"
@@ -34,10 +32,6 @@ def test_aggregate_hybrid_query():
     assert hybrid_query._vector == sample_vector
     assert hybrid_query._vector_field == vector_field_name
     assert hybrid_query._scorer == "BM25STD"
-    # assert (
-    #    hybrid_query.filter
-    #    == f"(@{text_field_name}:({hybrid_query.tokenize_and_escape_query(text_string)}))"
-    # )
     assert hybrid_query._filter_expression == None
     assert hybrid_query._alpha == 0.7
     assert hybrid_query._num_results == 10
@@ -72,47 +66,50 @@ def test_aggregate_hybrid_query():
     assert hybrid_query._vector == sample_vector
     assert hybrid_query._vector_field == vector_field_name
     assert hybrid_query._scorer == scorer
-    # assert (
-    #    hybrid_query.filter
-    #    == f"(@{text_field_name}:({hybrid_query.tokenize_and_escape_query(text_string)}))"
-    # )
     assert hybrid_query._filter_expression == filter_expression
     assert hybrid_query._alpha == 0.5
     assert hybrid_query._num_results == 8
     assert hybrid_query._loadfields == return_fields
     assert hybrid_query._dialect == 2
     assert hybrid_query.stopwords == set()
 
-    return
-
     # Test stopwords are configurable
-    hybrid_query = HybridAggregationQuery(text_string, text_field_name, stopwords=None)
+    hybrid_query = HybridAggregationQuery(
+        sample_text, text_field_name, sample_vector, vector_field_name, stopwords=None
+    )
     assert hybrid_query.stopwords == set([])
-    # assert (
-    #    hyrid_query.filter
-    #    == f"(@{text_field_name}:({hyrid_query.tokenize_and_escape_query(text_string)}))"
-    # )
 
-    hyrid_query = HybridAggregationQuery(
-        text_string, text_field_name, stopwords=["the", "a", "of"]
+    hybrid_query = HybridAggregationQuery(
+        sample_text,
+        text_field_name,
+        sample_vector,
+        vector_field_name,
+        stopwords=["the", "a", "of"],
     )
     assert hybrid_query.stopwords == set(["the", "a", "of"])
-    assert (
-        hybrid_query.filter
-        == f"(@{text_field_name}:({hybrid_query.tokenize_and_escape_query(text_string)}))"
-    )
-
     hybrid_query = HybridAggregationQuery(
-        text_string, text_field_name, stopwords="german"
+        sample_text,
+        text_field_name,
+        sample_vector,
+        vector_field_name,
+        stopwords="german",
     )
     assert hybrid_query.stopwords != set([])
 
     with pytest.raises(ValueError):
         hybrid_query = HybridAggregationQuery(
-            text_string, text_field_name, stopwords="gibberish"
+            sample_text,
+            text_field_name,
+            sample_vector,
+            vector_field_name,
+            stopwords="gibberish",
         )
 
     with pytest.raises(TypeError):
         hybrid_query = HybridAggregationQuery(
-            text_string, text_field_name, stopwords=[1, 2, 3]
+            sample_text,
+            text_field_name,
+            sample_vector,
+            vector_field_name,
+            stopwords=[1, 2, 3],
         )

tests/unit/test_query_types.py

@@ -270,7 +270,6 @@ def test_text_query():
     with pytest.raises(TypeError):
         text_query = TextQuery(text_string, text_field_name, stopwords=[1, 2, 3])
 
-
     text_query = TextQuery(text_string, text_field_name, stopwords=["the", "a", "of"])
     assert text_query.stopwords == set(["the", "a", "of"])
     assert (