langchain-postgres

import os from langchain_core.documents import Document from langchain_core.embeddings import DeterministicFakeEmbedding from langchain_postgres import PGEngine, PGVectorStore # Replace with your PostgreSQL connection string CONNECTION_STRING = os.environ.get('POSTGRES_CONNECTION_STRING', 'postgresql+psycopg://langchain:langchain@localhost:6024/langchain') # Initialize PGEngine engine = PGEngine.from_connection_string(url=CONNECTION_STRING) # Define vector size and embedding service VECTOR_SIZE = 768 # Adjust based on your embedding model embedding = DeterministicFakeEmbedding(size=VECTOR_SIZE) TABLE_NAME = "my_doc_collection" # Initialize the vector store table (if it doesn't exist) engine.init_vectorstore_table( table_name=TABLE_NAME, vector_size=VECTOR_SIZE, ) # Create a synchronous PGVectorStore instance store = PGVectorStore.create_sync( engine=engine, table_name=TABLE_NAME, embedding_service=embedding, ) # Add documents docs = [ Document(page_content="Apples and oranges"), Document(page_content="Cars and airplanes"), Document(page_content="Dogs and cats"), ] store.add_documents(docs) # Perform a similarity search query = "fruits" results = store.similarity_search(query, k=1) print(f"Similarity search for '{query}': {results[0].page_content}") # Example for Chat Message History from langchain_postgres.chat_message_histories import PostgresChatMessageHistory import uuid session_id = str(uuid.uuid4()) chat_history = PostgresChatMessageHistory(session_id=session_id, table_name="chat_messages", connection=engine.get_connection()) chat_history.add_user_message("Hello LangChain Postgres!") chat_history.add_ai_message("Hi there!") print(f"Chat history for session {session_id}: {chat_history.messages}")