update langchain rag app final

langchain-rag-app/source_code_final/chatbot_api/pyproject.toml

@@ -4,12 +4,12 @@ version = "0.1"
 dependencies = [
     "asyncio==3.4.3",
     "fastapi==0.109.0",
-    "langchain==0.1.0",
-    "langchain-openai==0.0.2",
-    "langchainhub==0.1.14",
+    "langchain==0.2.17",
+    "langchain-community==0.2.19",
+    "langchain-openai==0.1.25",
+    "langchainhub==0.1.21",
     "neo4j==5.14.1",
     "numpy==1.26.2",
-    "openai==1.7.2",
     "opentelemetry-api==1.22.0",
     "pydantic==2.5.1",
     "uvicorn==0.25.0"

langchain-rag-app/source_code_final/chatbot_api/src/chains/hospital_cypher_chain.py

@@ -12,10 +12,9 @@
     url=os.getenv("NEO4J_URI"),
     username=os.getenv("NEO4J_USERNAME"),
     password=os.getenv("NEO4J_PASSWORD"),
+    database=os.getenv("NEO4J_DATABASE", "neo4j"),
 )
 
-graph.refresh_schema()
-
 cypher_generation_template = """
 Task:
 Generate Cypher query for a Neo4j graph database.
@@ -154,4 +153,6 @@
     cypher_prompt=cypher_generation_prompt,
     validate_cypher=True,
     top_k=100,
+    # acknowledges the risk of letting an LLM generate and execute database queries
+    allow_dangerous_requests=True,
 )

langchain-rag-app/source_code_final/chatbot_api/src/chains/hospital_review_chain.py

@@ -7,16 +7,38 @@
     PromptTemplate,
     SystemMessagePromptTemplate,
 )
-from langchain.vectorstores.neo4j_vector import Neo4jVector
+from langchain_community.vectorstores import Neo4jVector
 from langchain_openai import ChatOpenAI, OpenAIEmbeddings
+from neo4j import GraphDatabase
 
 HOSPITAL_QA_MODEL = os.getenv("HOSPITAL_QA_MODEL")
 
+# langchain-community's create_new_index uses the deprecated
+# db.index.vector.createNodeIndex procedure removed in Neo4j 5.27.
+# Pre-create the index with the current syntax so from_existing_graph
+# detects it and skips create_new_index.
+_driver = GraphDatabase.driver(
+    os.getenv("NEO4J_URI"),
+    auth=(os.getenv("NEO4J_USERNAME"), os.getenv("NEO4J_PASSWORD")),
+)
+with _driver.session(
+    database=os.getenv("NEO4J_DATABASE", "neo4j")
+) as _session:
+    _session.run(
+        "CREATE VECTOR INDEX reviews IF NOT EXISTS "
+        "FOR (n:Review) ON (n.embedding) "
+        "OPTIONS {indexConfig: {`vector.dimensions`: 1536, "
+        "`vector.similarity_function`: 'cosine'}}"
+    )
+_driver.close()
+del _driver
+
 neo4j_vector_index = Neo4jVector.from_existing_graph(
     embedding=OpenAIEmbeddings(),
     url=os.getenv("NEO4J_URI"),
     username=os.getenv("NEO4J_USERNAME"),
     password=os.getenv("NEO4J_PASSWORD"),
+    database=os.getenv("NEO4J_DATABASE", "neo4j"),
     index_name="reviews",
     node_label="Review",
     text_node_properties=[

langchain-rag-app/source_code_final/docker-compose.yml

@@ -13,7 +13,8 @@ services:
     env_file:
       - .env
     depends_on:
-      - hospital_neo4j_etl
+      hospital_neo4j_etl:
+        condition: service_completed_successfully
     ports:
       - "8000:8000"
 

langchain-rag-app/source_code_final/hospital_neo4j_etl/src/hospital_bulk_csv_write.py

@@ -42,12 +42,13 @@ def load_hospital_graph_from_csv() -> None:
     )
 
     LOGGER.info("Setting uniqueness constraints on nodes")
-    with driver.session(database="neo4j") as session:
+
+    with driver.session() as session:
         for node in NODES:
             session.execute_write(_set_uniqueness_constraints, node)
 
     LOGGER.info("Loading hospital nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS
         FROM '{HOSPITALS_CSV_PATH}' AS hospitals
@@ -58,7 +59,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading payer nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS
         FROM '{PAYERS_CSV_PATH}' AS payers
@@ -68,7 +69,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading physician nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS
         FROM '{PHYSICIANS_CSV_PATH}' AS physicians
@@ -83,7 +84,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading visit nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS visits
         MERGE (v:Visit {{id: toInteger(visits.visit_id),
@@ -106,7 +107,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading patient nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS
         FROM '{PATIENTS_CSV_PATH}' AS patients
@@ -120,7 +121,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading review nodes")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS
         FROM '{REVIEWS_CSV_PATH}' AS reviews
@@ -134,7 +135,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'AT' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS row
         MATCH (source: `Visit` {{ `id`: toInteger(trim(row.`visit_id`)) }})
@@ -145,7 +146,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'WRITES' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{REVIEWS_CSV_PATH}' AS reviews
             MATCH (v:Visit {{id: toInteger(reviews.visit_id)}})
@@ -155,7 +156,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'TREATS' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS visits
             MATCH (p:Physician {{id: toInteger(visits.physician_id)}})
@@ -165,7 +166,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'COVERED_BY' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS visits
             MATCH (v:Visit {{id: toInteger(visits.visit_id)}})
@@ -178,7 +179,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'HAS' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS visits
             MATCH (p:Patient {{id: toInteger(visits.patient_id)}})
@@ -188,7 +189,7 @@ def load_hospital_graph_from_csv() -> None:
         _ = session.run(query, {})
 
     LOGGER.info("Loading 'EMPLOYS' relationships")
-    with driver.session(database="neo4j") as session:
+    with driver.session() as session:
         query = f"""
         LOAD CSV WITH HEADERS FROM '{VISITS_CSV_PATH}' AS visits
             MATCH (h:Hospital {{id: toInteger(visits.hospital_id)}})