fix: format and tests

Author: ahgraber

ragas/src/ragas/testset/transforms/relationship_builders/cosine.py

@@ -14,17 +14,6 @@ class CosineSimilarityBuilder(RelationshipBuilder):
     threshold: float = 0.9
     block_size: int = 1024
 
-    def _validate_embedding_shapes(self, embeddings: t.List[t.Any]):
-        if not embeddings:
-            raise ValueError(f"No nodes have a valid {self.property_name}")
-        first_len = len(embeddings[0])
-        for idx, emb in enumerate(embeddings):
-            if len(emb) != first_len:
-                raise ValueError(
-                    f"Embedding at index {idx} has length {len(emb)}, expected {first_len}. "
-                    "All embeddings must have the same length."
-                )
-
     def _block_cosine_similarity(self, i: np.ndarray, j: np.ndarray):
         """Calculate cosine similarity matrix between two sets of embeddings."""
         i_norm = i / np.linalg.norm(i, axis=1, keepdims=True)

ragas/tests/unit/test_cosine_relationship_builders.py

@@ -1,6 +1,7 @@
 import asyncio
 import copy
 import random
+from typing import Optional
 from uuid import UUID
 
 import numpy as np
@@ -18,7 +19,7 @@ def generate_test_vectors(
     d: int = 32,
     min_similarity: float = 0.5,
     similar_fraction: float = 0.3,
-    seed: int | None = None,
+    seed: Optional[int] = None,
 ) -> np.ndarray:
     """
     Generate `n` unit vectors of dimension `d`, where at least `similar_fraction` of them
@@ -287,7 +288,7 @@ async def test_malformed_embedding_raises(self):
     async def test_cosine_similarity_builder_empty_graph(self):
         kg = KnowledgeGraph(nodes=[])
         builder = CosineSimilarityBuilder(property_name="embedding")
-        with pytest.raises(ValueError, match="No nodes have a valid embedding"):
+        with pytest.raises(ValueError):
             await builder.transform(kg)
 
     @pytest.mark.asyncio

ragas/tests/unit/test_traditional_relationship_builders.py

@@ -262,9 +262,9 @@ def test__find_similar_embedding_pairs_jaccard(n_test_sets, max_len, threshold):
     assert len(result) == len(expected)
     for i, j, similarity_float in result:
         assert i < j, "Pairs should be ordered (i < j)"
-        assert similarity_float >= threshold, (
-            f"Similarity {similarity_float} should be >= {threshold}"
-        )
+        assert (
+            similarity_float >= threshold
+        ), f"Similarity {similarity_float} should be >= {threshold}"
         for x, y, expected_similarity in expected:
             if i == x and j == y:
                 assert similarity_float == pytest.approx(expected_similarity)
@@ -277,9 +277,9 @@ async def test_no_self_similarity_relationships(self, simple_kg):
         builder = JaccardSimilarityBuilder(property_name="entities", threshold=0.1)
         relationships = await builder.transform(copy.deepcopy(simple_kg))
         for r in relationships:
-            assert r.source.id != r.target.id, (
-                "Self-relationships should not be created"
-            )
+            assert (
+                r.source.id != r.target.id
+            ), "Self-relationships should not be created"
 
     @pytest.mark.asyncio
     async def test_no_duplicate_relationships(self, simple_kg):
@@ -307,9 +307,9 @@ async def test_all_below_threshold(self):
         kg = KnowledgeGraph(nodes=[node1, node2])
         builder = JaccardSimilarityBuilder(property_name="entities", threshold=0.1)
         relationships = await builder.transform(kg)
-        assert len(relationships) == 0, (
-            "No relationships should be created below threshold"
-        )
+        assert (
+            len(relationships) == 0
+        ), "No relationships should be created below threshold"
 
     @pytest.mark.asyncio
     async def test_all_above_threshold(self):
@@ -379,9 +379,9 @@ async def test_apply_transforms_cosine_similarity_builder(self, simple_kg):
         # Should mutate kg in-place
         apply_transforms(kg, builder, run_config=RunConfig(max_workers=2))
         # Check that relationships were added
-        assert any(r.type == "jaccard_similarity" for r in kg.relationships), (
-            "No jaccard_similarity relationships found after apply_transforms"
-        )
+        assert any(
+            r.type == "jaccard_similarity" for r in kg.relationships
+        ), "No jaccard_similarity relationships found after apply_transforms"
         # Check that expected relationship exists
         assert any(
             str(r.source.id) == "f353e5c2-e432-4d1e-84a8-d750c93d4edf"