langchain/templates/rag-pinecone-multi-query/rag_pinecone_multi_query/chain.py

import os

from langchain.retrievers.multi_query import MultiQueryRetriever
from langchain_community.chat_models import ChatOpenAI
from langchain_community.embeddings import OpenAIEmbeddings
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.pydantic_v1 import BaseModel
from langchain_core.runnables import RunnableParallel, RunnablePassthrough
from langchain_pinecone import PineconeVectorStore

if os.environ.get("PINECONE_API_KEY", None) is None:
    raise Exception("Missing `PINECONE_API_KEY` environment variable.")

if os.environ.get("PINECONE_ENVIRONMENT", None) is None:
    raise Exception("Missing `PINECONE_ENVIRONMENT` environment variable.")

PINECONE_INDEX_NAME = os.environ.get("PINECONE_INDEX", "langchain-test")

### Ingest code - you may need to run this the first time
# Load
# from langchain_community.document_loaders import WebBaseLoader
# loader = WebBaseLoader("https://lilianweng.github.io/posts/2023-06-23-agent/")
# data = loader.load()

# # Split
# from langchain.text_splitter import RecursiveCharacterTextSplitter
# text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=0)
# all_splits = text_splitter.split_documents(data)

# # Add to vectorDB
# vectorstore = PineconeVectorStore.from_documents(
#     documents=all_splits, embedding=OpenAIEmbeddings(), index_name=PINECONE_INDEX_NAME
# )
# retriever = vectorstore.as_retriever()

# Set up index with multi query retriever
vectorstore = PineconeVectorStore.from_existing_index(
    PINECONE_INDEX_NAME, OpenAIEmbeddings()
)
model = ChatOpenAI(temperature=0)
retriever = MultiQueryRetriever.from_llm(
    retriever=vectorstore.as_retriever(), llm=model
)

# RAG prompt
template = """Answer the question based only on the following context:
{context}
Question: {question}
"""
prompt = ChatPromptTemplate.from_template(template)

# RAG
model = ChatOpenAI()
chain = (
    RunnableParallel({"context": retriever, "question": RunnablePassthrough()})
    | prompt
    | model
    | StrOutputParser()
)


# Add typing for input
class Question(BaseModel):
    __root__: str


chain = chain.with_types(input_type=Question)