GraphRAG CI Fixes

Author: caseyclements

libs/langchain-mongodb/langchain_mongodb/graphrag/graph.py

@@ -14,6 +14,7 @@
 from pymongo import MongoClient, UpdateOne
 from pymongo.collection import Collection
 from pymongo.driver_info import DriverInfo
+from pymongo.errors import OperationFailure
 from pymongo.results import BulkWriteResult
 
 from langchain_mongodb.graphrag import example_templates, prompts
@@ -145,12 +146,18 @@ def __init__(
         else:
             self.allowed_relationship_types = []
         if validate:
-            collection.database.command(
-                "collMod",
-                collection.name,
-                validator={"$jsonSchema": self._schema},
-                validationAction=validation_action,
-            )
+            try:
+                collection.database.command(
+                    "collMod",
+                    collection.name,
+                    validator={"$jsonSchema": self._schema},
+                    validationAction=validation_action,
+                )
+            except OperationFailure:
+                logger.warning(
+                    "Validation will NOT be performed. User must be DB Admin to add validation **after** a Collection is created. \n"
+                    "Please add validator when you create collection: db.create_collection.(coll_name, validator={'$jsonSchema': self._schema})"
+                )
         self.collection = collection
 
         # Include examples

libs/langchain-mongodb/langchain_mongodb/graphrag/prompts.py

@@ -21,6 +21,7 @@
 INPUT: You will be provided a text document.
 OUTPUT:
 - You will produce valid json according the "Output Schema" section below.
+- Your response **must be** a **valid JSON document** with NO extra text, explanations, or markdown formatting.
 - The extracted entities and relationships **MUST STRICTLY CONFORM** to the constraints outlined below.
 - Any entities or relationships not matching the allowed types must be **EXCLUDED**.
 

libs/langchain-mongodb/tests/integration_tests/test_graphrag.py

@@ -17,7 +17,7 @@
 COLLECTION_NAME = "langchain_test_graphrag"
 
 
-@pytest.fixture(scope="function")
+@pytest.fixture(scope="module")
 def collection() -> Collection:
     client = MongoClient(MONGODB_URI)
     db = client[DB_NAME]
@@ -36,7 +36,7 @@ def collection() -> Collection:
 def entity_extraction_model() -> BaseChatModel:
     """LLM for converting documents into Graph of Entities and Relationships"""
     try:
-        return ChatOpenAI(model="gpt-4o", temperature=0.0)
+        return ChatOpenAI(model="gpt-4o", temperature=0.0, cache=False)
     except Exception:
         pass
 
@@ -114,7 +114,7 @@ def entity_example():
 """
 
 
-@pytest.fixture(scope="function")
+@pytest.fixture(scope="module")
 def graph_store(collection, entity_extraction_model, documents) -> MongoDBGraphStore:
     store = MongoDBGraphStore(
         collection, entity_extraction_model, entity_prompt, query_prompt