test: replacing each DocumentStore specific tests and used the generalised ones from haystack.testing.document_store (#2812)

* removing test_delete_all_documents from PgVector - already in haystack.testing

* removing duplicated tests from OpenSearchDocumentStore

* removing duplicated tests from ElasticStoreDocumentTest

* removing test_delete_all_documents from weaviate

* removing test_delete_all_documents from qdrant

* removing delete_all_documents tests from PineCone

* removing delete_all_documents tests from MondoDBDocumentStore

* adding check for empty list for delete_all_documents on Chroma

* removing delete_all_documents tests from AzureAISearchDocumentStore

* fixing Pinecone tests

* cleaning up tests in qdrant and weaviate

* removing more weavite delete_by_filter tests

* cleaning up tests in pgvector

* updating OpenSearch: delete_by_filter_async refresh=True as default

* updating OpenSearch: update_by_filter refresh=True as default

* removing standard test from MongoDBDocumentStore

* reverting default refresh param in OpenSearch

* removing standard test from Chroma

* updating AzuresAISearchDocumentStore to make use of base tests

* removing duplicated tests from Astra

* adding UpdateByFilterTest to Weavite

* fixing Azure linting issues

* formatting

* formatting

* temporarly using haystack branch with new standard tests

* temporarly using haystack branch with new standard tests

* formatting

* removing duplicated tests from Pinecone

* updating all tests to use new test class from haystack core

* fixing imports for 3.10

* fixing linting issues

* fixing azure_ai_search overrides due to class changes in haystack core

* fixing linting

* chaning imports due to MRO errors + changing document_stores fixtures when needed due to new method from ExtendedTests

* fixing weaviate tests and imports

* fixing Qdrandt tests and imports

* fixing PGVector tests

* fixing Chroma tests + imports

* pointing all pyproject.toml to main

* fixing elasticsearch 3.10 arg not used

* fixing mongodb imports

* fixing chroma formatting

* fixing weaviate tests and imports

* fixing pinecone imports

* attending PR comments

* removing white space

* adding filterable docs fixture to azure_ai_search

* fixing linter

* updating pyproject.toml to reference pypi instead of main branch

* updating pyproject.toml to reference latest pypi instead of main branch

* Update integrations/qdrant/pyproject.toml

Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

* Update integrations/pgvector/pyproject.toml

Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

* Update integrations/mongodb_atlas/pyproject.toml

Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

* Update integrations/weaviate/pyproject.toml

Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

---------

Co-authored-by: Stefano Fiorucci <stefanofiorucci@gmail.com>

Author: davidsbatista

integrations/astra/pyproject.toml

@@ -22,7 +22,12 @@ classifiers = [
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",
 ]
-dependencies = ["haystack-ai>=2.22.0", "pydantic", "typing_extensions", "astrapy>=1.5.0,<2.0"]
+dependencies = [
+  "astrapy>=1.5.0,<2.0",
+  "haystack-ai>=2.24.0",
+  "pydantic",
+  "typing_extensions",
+]
 
 [project.urls]
 Documentation = "https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/astra#readme"

integrations/astra/tests/test_document_store.py

@@ -10,7 +10,7 @@
 from haystack import Document
 from haystack.document_stores.errors import MissingDocumentError
 from haystack.document_stores.types import DuplicatePolicy
-from haystack.testing.document_store import DocumentStoreBaseTests
+from haystack.testing.document_store import DocumentStoreBaseExtendedTests
 
 from haystack_integrations.document_stores.astra import AstraDocumentStore
 
@@ -48,9 +48,9 @@ def test_to_dict(mock_auth):  # noqa
     os.environ.get("ASTRA_DB_APPLICATION_TOKEN", "") == "", reason="ASTRA_DB_APPLICATION_TOKEN env var not set"
 )
 @pytest.mark.skipif(os.environ.get("ASTRA_DB_API_ENDPOINT", "") == "", reason="ASTRA_DB_API_ENDPOINT env var not set")
-class TestDocumentStore(DocumentStoreBaseTests):
+class TestDocumentStore(DocumentStoreBaseExtendedTests):
     """
-    Common test cases will be provided by `DocumentStoreBaseTests` but
+    Common test cases will be provided by `DocumentStoreBaseExtendedTests` but
     you can add more to this class.
     """
 
@@ -204,74 +204,6 @@ def test_filter_documents_by_in_operator(self, document_store):
         self.assert_documents_are_equal([result[0]], [docs[0]])
         self.assert_documents_are_equal([result[1]], [docs[1]])
 
-    def test_delete_all_documents(self, document_store: AstraDocumentStore):
-        """
-        Test delete_all_documents() on an Astra.
-        """
-        document_store.delete_all_documents()
-        assert document_store.count_documents() == 0
-
-    def test_delete_by_filter(self, document_store: AstraDocumentStore, filterable_docs):
-        document_store.write_documents(filterable_docs)
-        initial_count = document_store.count_documents()
-        assert initial_count > 0
-
-        # count documents that match the filter before deletion
-        matching_docs = [d for d in filterable_docs if d.meta.get("chapter") == "intro"]
-        expected_deleted_count = len(matching_docs)
-
-        # delete all documents with chapter="intro"
-        deleted_count = document_store.delete_by_filter(
-            filters={"field": "meta.chapter", "operator": "==", "value": "intro"}
-        )
-
-        assert deleted_count == expected_deleted_count
-        assert document_store.count_documents() == initial_count - deleted_count
-
-        # remaining documents don't have chapter="intro"
-        remaining_docs = document_store.filter_documents()
-        for doc in remaining_docs:
-            assert doc.meta.get("chapter") != "intro"
-
-        # all documents with chapter="intro" were deleted
-        intro_docs = document_store.filter_documents(
-            filters={"field": "meta.chapter", "operator": "==", "value": "intro"}
-        )
-        assert len(intro_docs) == 0
-
-    def test_update_by_filter(self, document_store: AstraDocumentStore, filterable_docs):
-        document_store.write_documents(filterable_docs)
-        initial_count = document_store.count_documents()
-        assert initial_count > 0
-
-        # count documents that match the filter before update
-        matching_docs = [d for d in filterable_docs if d.meta.get("chapter") == "intro"]
-        expected_updated_count = len(matching_docs)
-
-        # update all documents with chapter="intro" to have status="updated"
-        updated_count = document_store.update_by_filter(
-            filters={"field": "meta.chapter", "operator": "==", "value": "intro"},
-            meta={"status": "updated"},
-        )
-
-        assert updated_count == expected_updated_count
-        assert document_store.count_documents() == initial_count
-
-        # verify the updated documents have the new metadata
-        updated_docs = document_store.filter_documents(
-            filters={"field": "meta.status", "operator": "==", "value": "updated"}
-        )
-        assert len(updated_docs) == expected_updated_count
-        for doc in updated_docs:
-            assert doc.meta.get("chapter") == "intro"
-            assert doc.meta.get("status") == "updated"
-
-        # verify other documents weren't affected
-        all_docs = document_store.filter_documents()
-        for doc in all_docs:
-            if doc.meta.get("chapter") != "intro":
-                assert doc.meta.get("status") != "updated"
-
     @pytest.mark.skip(reason="Unsupported filter operator not.")
     def test_not_operator(self, document_store, filterable_docs):
         pass

integrations/azure_ai_search/pyproject.toml

@@ -22,7 +22,11 @@ classifiers = [
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",
 ]
-dependencies = ["haystack-ai>=2.22.0", "azure-search-documents>=11.5", "azure-identity"]
+dependencies = [
+  "haystack-ai>=2.24.0",
+  "azure-search-documents>=11.5",
+  "azure-identity"
+]
 
 [project.urls]
 Documentation = "https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/azure_ai_search#readme"

integrations/azure_ai_search/src/haystack_integrations/document_stores/azure_ai_search/filters.py

@@ -68,11 +68,19 @@ def _parse_comparison_condition(condition: dict[str, Any]) -> str:
 
 
 def _eq(field: str, value: Any) -> str:
-    return f"{field} eq '{value}'" if isinstance(value, str) and value != "null" else f"{field} eq {value}"
+    if isinstance(value, str) and value != "null":
+        return f"{field} eq '{value}'"
+    if isinstance(value, bool):
+        return f"{field} eq {str(value).lower()}"
+    return f"{field} eq {value}"
 
 
 def _ne(field: str, value: Any) -> str:
-    return f"not ({field} eq '{value}')" if isinstance(value, str) and value != "null" else f"not ({field} eq {value})"
+    if isinstance(value, str) and value != "null":
+        return f"not ({field} eq '{value}')"
+    if isinstance(value, bool):
+        return f"not ({field} eq {str(value).lower()})"
+    return f"not ({field} eq {value})"
 
 
 def _in(field: str, value: Any) -> str:

integrations/azure_ai_search/tests/conftest.py

@@ -70,8 +70,9 @@ def delete_by_filter_and_wait(filters):
         time.sleep(SLEEP_TIME_IN_SECONDS)
         return deleted_count
 
-    def update_by_filter_and_wait(filters, fields):
-        updated_count = original_update_by_filter(filters, fields)
+    def update_by_filter_and_wait(filters, meta=None, fields=None):
+        updates = meta if meta is not None else fields or {}
+        updated_count = original_update_by_filter(filters, updates)
         time.sleep(SLEEP_TIME_IN_SECONDS)
         return updated_count
 

integrations/azure_ai_search/tests/test_document_store.py

@@ -13,8 +13,12 @@
 from haystack.errors import FilterError
 from haystack.testing.document_store import (
     CountDocumentsTest,
+    DeleteAllTest,
+    DeleteByFilterTest,
     DeleteDocumentsTest,
+    FilterableDocsFixtureMixin,
     FilterDocumentsTest,
+    UpdateByFilterTest,
     WriteDocumentsTest,
 )
 from haystack.utils.auth import EnvVarSecret, Secret
@@ -256,7 +260,15 @@ def _assert_documents_are_equal(received: list[Document], expected: list[Documen
     not os.environ.get("AZURE_AI_SEARCH_ENDPOINT", None) and not os.environ.get("AZURE_AI_SEARCH_API_KEY", None),
     reason="Missing AZURE_AI_SEARCH_ENDPOINT or AZURE_AI_SEARCH_API_KEY.",
 )
-class TestDocumentStore(CountDocumentsTest, WriteDocumentsTest, DeleteDocumentsTest):
+class TestDocumentStore(
+    CountDocumentsTest,
+    DeleteDocumentsTest,
+    DeleteAllTest,
+    DeleteByFilterTest,
+    FilterableDocsFixtureMixin,
+    WriteDocumentsTest,
+    UpdateByFilterTest,
+):
     def assert_documents_are_equal(self, received: list[Document], expected: list[Document]):
         _assert_documents_are_equal(received, expected)
 
@@ -290,120 +302,80 @@ def test_write_documents_duplicate_fail(self, document_store: AzureAISearchDocum
     @pytest.mark.skip(reason="Azure AI search index overwrites duplicate documents by default")
     def test_write_documents_duplicate_skip(self, document_store: AzureAISearchDocumentStore): ...
 
-    def test_delete_all_documents(self, document_store: AzureAISearchDocumentStore):
-        docs = [Document(content="first doc"), Document(content="second doc")]
-        document_store.write_documents(docs)
-        assert document_store.count_documents() == 2
-        document_store.delete_all_documents()
-        assert document_store.count_documents() == 0
-
-    def test_delete_all_documents_empty_index(self, document_store: AzureAISearchDocumentStore):
-        assert document_store.count_documents() == 0
-        document_store.delete_all_documents()
-        assert document_store.count_documents() == 0
-
     @pytest.mark.parametrize(
         "document_store",
         [{"metadata_fields": {"category": str}}],
         indirect=True,
     )
     def test_delete_by_filter(self, document_store: AzureAISearchDocumentStore):
-        docs = [
-            Document(content="Doc 1", meta={"category": "A"}),
-            Document(content="Doc 2", meta={"category": "B"}),
-            Document(content="Doc 3", meta={"category": "A"}),
-        ]
-        document_store.write_documents(docs)
-        assert document_store.count_documents() == 3
-
-        # Delete documents with category="A"
-        deleted_count = document_store.delete_by_filter(
-            filters={"field": "meta.category", "operator": "==", "value": "A"}
-        )
-        assert deleted_count == 2
-        assert document_store.count_documents() == 1
-
-        # Verify only category B remains
-        remaining_docs = document_store.filter_documents()
-        assert len(remaining_docs) == 1
-        assert remaining_docs[0].meta["category"] == "B"
+        """Override to use a document_store with category metadata field."""
+        DeleteByFilterTest.test_delete_by_filter(document_store)
 
     @pytest.mark.parametrize(
         "document_store",
         [{"metadata_fields": {"category": str}}],
         indirect=True,
     )
     def test_delete_by_filter_no_matches(self, document_store: AzureAISearchDocumentStore):
-        docs = [
-            Document(content="Doc 1", meta={"category": "A"}),
-            Document(content="Doc 2", meta={"category": "B"}),
-        ]
-        document_store.write_documents(docs)
-        assert document_store.count_documents() == 2
-
-        # Try to delete documents with category="C" (no matches)
-        deleted_count = document_store.delete_by_filter(
-            filters={"field": "meta.category", "operator": "==", "value": "C"}
-        )
-        assert deleted_count == 0
-        assert document_store.count_documents() == 2
+        """Override to use a document_store with category metadata field."""
+        DeleteByFilterTest.test_delete_by_filter_no_matches(document_store)
 
     @pytest.mark.parametrize(
         "document_store",
-        [{"metadata_fields": {"category": str, "status": str}}],
+        [{"metadata_fields": {"category": str, "year": int, "status": str}}],
         indirect=True,
     )
-    def test_update_by_filter(self, document_store: AzureAISearchDocumentStore):
-        docs = [
-            Document(content="Doc 1", meta={"category": "A", "status": "draft"}),
-            Document(content="Doc 2", meta={"category": "B", "status": "draft"}),
-            Document(content="Doc 3", meta={"category": "A", "status": "draft"}),
-        ]
-        document_store.write_documents(docs)
-        assert document_store.count_documents() == 3
+    def test_delete_by_filter_advanced_filters(self, document_store: AzureAISearchDocumentStore):
+        """Override to use a document_store with category, year, status metadata fields."""
+        DeleteByFilterTest.test_delete_by_filter_advanced_filters(document_store)
+
+    # Metadata fields required by haystack UpdateByFilterTest filterable_docs (chapter, name, page, number, date, etc.)
+    _FILTERABLE_DOCS_METADATA = {  # noqa: RUF012
+        "name": str,
+        "page": str,
+        "chapter": str,
+        "number": int,
+        "date": str,
+        "no_embedding": bool,
+        "updated": bool,
+        "extra_field": str,
+    }
 
-        # Update status for category="A" documents
-        updated_count = document_store.update_by_filter(
-            filters={"field": "meta.category", "operator": "==", "value": "A"},
-            fields={"status": "published"},
-        )
-        assert updated_count == 2
+    @pytest.mark.parametrize(
+        "document_store",
+        [{"metadata_fields": _FILTERABLE_DOCS_METADATA}],
+        indirect=True,
+    )
+    def test_update_by_filter(self, document_store: AzureAISearchDocumentStore, filterable_docs):
+        """Override to use a document_store with metadata fields for filterable_docs."""
+        UpdateByFilterTest.test_update_by_filter(document_store, filterable_docs)
 
-        # Verify the updates
-        published_docs = document_store.filter_documents(
-            filters={"field": "meta.status", "operator": "==", "value": "published"}
-        )
-        assert len(published_docs) == 2
-        for doc in published_docs:
-            assert doc.meta["category"] == "A"
-            assert doc.meta["status"] == "published"
-
-        # Verify category B still has draft status
-        draft_docs = document_store.filter_documents(
-            filters={"field": "meta.status", "operator": "==", "value": "draft"}
-        )
-        assert len(draft_docs) == 1
-        assert draft_docs[0].meta["category"] == "B"
+    @pytest.mark.parametrize(
+        "document_store",
+        [{"metadata_fields": _FILTERABLE_DOCS_METADATA}],
+        indirect=True,
+    )
+    def test_update_by_filter_no_matches(self, document_store: AzureAISearchDocumentStore, filterable_docs):
+        """Override to use a document_store with metadata fields for filterable_docs."""
+        UpdateByFilterTest.test_update_by_filter_no_matches(document_store, filterable_docs)
 
     @pytest.mark.parametrize(
         "document_store",
-        [{"metadata_fields": {"category": str, "status": str}}],
+        [{"metadata_fields": _FILTERABLE_DOCS_METADATA}],
         indirect=True,
     )
-    def test_update_by_filter_no_matches(self, document_store: AzureAISearchDocumentStore):
-        docs = [
-            Document(content="Doc 1", meta={"category": "A", "status": "draft"}),
-            Document(content="Doc 2", meta={"category": "B", "status": "draft"}),
-        ]
-        document_store.write_documents(docs)
-        assert document_store.count_documents() == 2
+    def test_update_by_filter_multiple_fields(self, document_store: AzureAISearchDocumentStore, filterable_docs):
+        """Override to use a document_store with metadata fields for filterable_docs."""
+        UpdateByFilterTest.test_update_by_filter_multiple_fields(document_store, filterable_docs)
 
-        # Try to update documents with category="C" (no matches)
-        updated_count = document_store.update_by_filter(
-            filters={"field": "meta.category", "operator": "==", "value": "C"},
-            fields={"status": "published"},
-        )
-        assert updated_count == 0
+    @pytest.mark.parametrize(
+        "document_store",
+        [{"metadata_fields": {"category": str, "year": int, "status": str, "featured": bool}}],
+        indirect=True,
+    )
+    def test_update_by_filter_advanced_filters(self, document_store: AzureAISearchDocumentStore):
+        """Override to use a document_store with category, year, status, featured metadata fields."""
+        UpdateByFilterTest.test_update_by_filter_advanced_filters(document_store)
 
     @pytest.mark.parametrize(
         "document_store",

integrations/chroma/pyproject.toml

@@ -22,7 +22,10 @@ classifiers = [
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",
 ]
-dependencies = ["haystack-ai>=2.22.0", "chromadb>=1.0.2"]
+dependencies = [
+  "haystack-ai>=2.24.0",
+  "chromadb>=1.0.2"
+]
 
 [project.urls]
 Documentation = "https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/chroma#readme"