refactor: trim tests to high-signal only

Remove tests that just verify mock plumbing (count, write, delete calling
the mock client). Keep tests that verify our actual logic:
- Serialization roundtrip (full dict structure)
- Score conversion (cosine + euclidean)
- Filter conversion (pure function with real logic)
- Duplicate policy batch checks (SKIP/NONE)
- Document <-> S3 vector conversion
- Input validation

Before: 49 unit tests (many testing mock behavior)
After: 26 unit tests (all testing our code) + 12 integration tests

Author: dotKokott

integrations/amazon_s3_vectors/tests/test_document_store.py

@@ -89,64 +89,9 @@ def test_from_dict(_mock_boto3):
     assert store.create_bucket_and_index is False
 
 
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_count_documents_empty(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    client.list_vectors.return_value = {"vectors": []}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    assert store.count_documents() == 0
-
-
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_count_documents_pagination(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    client.list_vectors.side_effect = [
-        {"vectors": [{"key": "1"}, {"key": "2"}], "nextToken": "tok"},
-        {"vectors": [{"key": "3"}]},
-    ]
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    assert store.count_documents() == 3
-
-
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_write_documents(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    client.put_vectors.return_value = {}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    docs = [
-        Document(id="1", content="Hello", embedding=[0.1] * 4),
-        Document(id="2", content="World", embedding=[0.2] * 4),
-    ]
-    assert store.write_documents(docs) == 2
-    client.put_vectors.assert_called_once()
-
-
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_write_documents_empty(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    assert store.write_documents([]) == 0
-    client.put_vectors.assert_not_called()
-
-
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
 def test_write_documents_no_embedding_raises(mock_boto3):
+    """S3 Vectors requires embeddings — this tests our validation, not the store."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
@@ -159,98 +104,82 @@ def test_write_documents_no_embedding_raises(mock_boto3):
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
 def test_write_documents_skip_existing(mock_boto3):
+    """Tests our batch existence check logic for SKIP policy."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
     client.get_vectors.return_value = {"vectors": [{"key": "1"}]}
     mock_boto3.client.return_value = client
 
     store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    docs = [Document(id="1", content="Hello", embedding=[0.1] * 4)]
-    assert store.write_documents(docs, policy=DuplicatePolicy.SKIP) == 0
+    result = store.write_documents(
+        [Document(id="1", content="Hello", embedding=[0.1] * 4)],
+        policy=DuplicatePolicy.SKIP,
+    )
+    assert result == 0
+    client.put_vectors.assert_not_called()
 
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
 def test_write_documents_none_policy_raises(mock_boto3):
+    """Tests our batch existence check logic for NONE policy."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
     client.get_vectors.return_value = {"vectors": [{"key": "1"}]}
     mock_boto3.client.return_value = client
 
     store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    docs = [Document(id="1", content="Hello", embedding=[0.1] * 4)]
     with pytest.raises(DocumentStoreError, match="already exist"):
-        store.write_documents(docs, policy=DuplicatePolicy.NONE)
+        store.write_documents(
+            [Document(id="1", content="Hello", embedding=[0.1] * 4)],
+            policy=DuplicatePolicy.NONE,
+        )
 
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_write_documents_metadata(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    client.put_vectors.return_value = {}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    docs = [Document(id="1", content="Hello", embedding=[0.1] * 4, meta={"category": "test", "year": 2024})]
-    store.write_documents(docs)
-
-    vectors = client.put_vectors.call_args[1]["vectors"]
-    assert len(vectors) == 1
-    assert vectors[0]["key"] == "1"
-    assert vectors[0]["metadata"]["_content"] == "Hello"
-    assert vectors[0]["metadata"]["category"] == "test"
-    assert vectors[0]["metadata"]["year"] == 2024
-
-
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_delete_documents(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    client.delete_vectors.return_value = {}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    store.delete_documents(["1", "2"])
-    client.delete_vectors.assert_called_once_with(vectorBucketName="b", indexName="i", keys=["1", "2"])
-
-
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_delete_documents_empty(mock_boto3):
+def test_embedding_retrieval_score_conversion(mock_boto3):
+    """Tests our distance-to-score conversion logic — the only non-trivial transform in retrieval."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
+    client.query_vectors.return_value = {
+        "vectors": [{"key": "1", "distance": 0.05, "metadata": {"_content": "Hello", "category": "news"}}],
+        "distanceMetric": "cosine",
+    }
     mock_boto3.client.return_value = client
 
     store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
-    store.delete_documents([])
-    client.delete_vectors.assert_not_called()
+    docs = store._embedding_retrieval(query_embedding=[0.1] * 4, top_k=5)
+    assert len(docs) == 1
+    assert docs[0].id == "1"
+    assert docs[0].content == "Hello"
+    assert docs[0].score == pytest.approx(0.95)  # cosine: 1.0 - 0.05
+    assert docs[0].meta == {"category": "news"}
 
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_embedding_retrieval(mock_boto3):
+def test_embedding_retrieval_euclidean_score(mock_boto3):
+    """Tests euclidean distance-to-score conversion (negated)."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
     client.query_vectors.return_value = {
-        "vectors": [{"key": "1", "distance": 0.05, "metadata": {"_content": "Hello", "category": "news"}}],
-        "distanceMetric": "cosine",
+        "vectors": [{"key": "1", "distance": 1.5, "metadata": {}}],
+        "distanceMetric": "euclidean",
     }
     mock_boto3.client.return_value = client
 
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
+    store = S3VectorsDocumentStore(
+        vector_bucket_name="b", index_name="i", dimension=4, distance_metric="euclidean", region_name="us-east-1"
+    )
     docs = store._embedding_retrieval(query_embedding=[0.1] * 4, top_k=5)
-    assert len(docs) == 1
-    assert docs[0].id == "1"
-    assert docs[0].content == "Hello"
-    assert docs[0].score == pytest.approx(0.95)  # 1 - 0.05
-    assert docs[0].meta == {"category": "news"}
+    assert docs[0].score == pytest.approx(-1.5)  # euclidean: negated
 
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_embedding_retrieval_with_filters(mock_boto3):
+def test_embedding_retrieval_passes_filters(mock_boto3):
+    """Tests that Haystack filters are converted and passed to query_vectors."""
     client = MagicMock()
     client.get_vector_bucket.return_value = {}
     client.get_index.return_value = {}
@@ -265,19 +194,15 @@ def test_embedding_retrieval_with_filters(mock_boto3):
     assert call_args["filter"] == {"$and": [{"category": {"$eq": "news"}}]}
 
 
-@patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
-def test_embedding_retrieval_empty_embedding_raises(mock_boto3):
-    client = MagicMock()
-    client.get_vector_bucket.return_value = {}
-    client.get_index.return_value = {}
-    mock_boto3.client.return_value = client
-
-    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, region_name="us-east-1")
+def test_embedding_retrieval_empty_embedding_raises():
+    """Tests our input validation — no mocking needed."""
+    store = S3VectorsDocumentStore(vector_bucket_name="b", index_name="i", dimension=4, create_bucket_and_index=False)
     with pytest.raises(ValueError, match="non-empty"):
         store._embedding_retrieval(query_embedding=[])
 
 
 def test_document_to_s3_vector():
+    """Tests our Document → S3 vector conversion (pure function)."""
     doc = Document(
         id="test-1", content="Hello world", embedding=[0.1, 0.2, 0.3], meta={"category": "test", "year": 2024}
     )
@@ -290,6 +215,7 @@ def test_document_to_s3_vector():
 
 
 def test_s3_vector_to_document():
+    """Tests our S3 vector → Document conversion (pure function)."""
     vector = {
         "key": "test-1",
         "data": {"float32": [0.1, 0.2, 0.3]},
@@ -303,6 +229,7 @@ def test_s3_vector_to_document():
 
 
 def test_document_roundtrip():
+    """Tests Document → S3 vector → Document is lossless."""
     doc = Document(
         id="test-1", content="Hello world", embedding=[0.1, 0.2, 0.3], meta={"category": "test", "year": 2024}
     )

integrations/amazon_s3_vectors/tests/test_embedding_retriever.py

@@ -27,11 +27,6 @@ def test_init_default():
         S3VectorsEmbeddingRetriever(document_store=mock_store, filter_policy="invalid")
 
 
-def test_init_invalid_store():
-    with pytest.raises(ValueError, match="must be an instance"):
-        S3VectorsEmbeddingRetriever(document_store="not a store")
-
-
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
 def test_to_dict(_mock_boto3):
     store = S3VectorsDocumentStore(
@@ -102,6 +97,7 @@ def test_from_dict(_mock_boto3):
 
 @patch("haystack_integrations.document_stores.amazon_s3_vectors.document_store.boto3")
 def test_from_dict_no_filter_policy(_mock_boto3):
+    """Pipelines serialized with older versions may not have filter_policy."""
     data = {
         "type": "haystack_integrations.components.retrievers.amazon_s3_vectors.embedding_retriever.S3VectorsEmbeddingRetriever",
         "init_parameters": {
@@ -134,28 +130,3 @@ def test_run():
     )
     assert len(res["documents"]) == 1
     assert res["documents"][0].content == "Test doc"
-
-
-def test_run_with_top_k_override():
-    mock_store = Mock(spec=S3VectorsDocumentStore)
-    mock_store._embedding_retrieval.return_value = []
-    retriever = S3VectorsEmbeddingRetriever(document_store=mock_store, top_k=10)
-    retriever.run(query_embedding=[0.1, 0.2], top_k=3)
-    mock_store._embedding_retrieval.assert_called_once_with(
-        query_embedding=[0.1, 0.2],
-        filters={},
-        top_k=3,
-    )
-
-
-def test_run_with_filters():
-    mock_store = Mock(spec=S3VectorsDocumentStore)
-    mock_store._embedding_retrieval.return_value = []
-    retriever = S3VectorsEmbeddingRetriever(document_store=mock_store)
-    filters = {"operator": "AND", "conditions": [{"field": "meta.category", "operator": "==", "value": "news"}]}
-    retriever.run(query_embedding=[0.1, 0.2], filters=filters)
-    mock_store._embedding_retrieval.assert_called_once_with(
-        query_embedding=[0.1, 0.2],
-        filters=filters,
-        top_k=10,
-    )