app.py

import chainlit as cl
from langchain.retrievers import BM25Retriever, EnsembleRetriever
from langchain.vectorstores import FAISS
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.embeddings import CacheBackedEmbeddings
from langchain.storage import LocalFileStore
from langchain.agents.agent_toolkits import create_retriever_tool
from langchain.agents.agent_toolkits import create_conversational_retrieval_agent
from langchain.document_loaders import WikipediaLoader, CSVLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.prompts import ChatPromptTemplate
from langchain.agents import Tool
from langchain.agents import ZeroShotAgent, AgentExecutor
from langchain.chat_models import ChatOpenAI
from langchain import LLMChain

@cl.author_rename
def rename(orig_author: str):
    rename_dict = {"RetrievalQA": "Consulting The Barbenheimer"}
    return rename_dict.get(orig_author, orig_author)

@cl.on_chat_start
async def init():
    
    msg = cl.Message(content=f"Building Index...")
    await msg.send()
    
    llm = ChatOpenAI(model="gpt-3.5-turbo", temperature = 0)
    
    # set up text splitters
    wikipedia_text_splitter = RecursiveCharacterTextSplitter(
        chunk_size = 1024,
        chunk_overlap = 512,
        length_function = len,
        is_separator_regex= False,
        separators = ["\n==", "\n", " "] # keep headings, then paragraphs, then sentences
    )
    csv_text_splitter = RecursiveCharacterTextSplitter(
        chunk_size = 1024,
        chunk_overlap = 512,
        length_function = len,
        is_separator_regex= False,
        separators = ["\n", " "] # keep paragraphs, then sentences
    )
    
    
    # set up cached embeddings store
    store = LocalFileStore("./.cache/")
    core_embeddings_model = OpenAIEmbeddings()
    embedder = CacheBackedEmbeddings.from_bytes_store(core_embeddings_model, 
                                                      store, 
                                                      namespace=core_embeddings_model.model)


    # Barbie retrieval system (Wikipedia, CSV)
    # load the multiple source documents for Barbie and build FAISS index
    barbie_wikipedia_docs = WikipediaLoader(
        query="Barbie (film)", 
        load_max_docs= 1, # YOUR CODE HERE, 
        doc_content_chars_max=10000000 
        ).load()
    barbie_csv_docs = CSVLoader(
        file_path= "./barbie_data/barbie.csv", 
        source_column="Review"   
        ).load()
    # chunk the loaded documents using the text splitters
    chunked_barbie_wikipedia_docs = wikipedia_text_splitter.transform_documents(barbie_wikipedia_docs)
    chunked_barbie_csv_docs = csv_text_splitter.transform_documents(barbie_csv_docs)
    # set up FAISS vector store and create retriever for CSV docs
    barbie_csv_faiss_retriever = FAISS.from_documents(chunked_barbie_csv_docs, embedder)
    # set up BM25 retriever
    barbie_wikipedia_bm25_retriever = BM25Retriever.from_documents(
        chunked_barbie_wikipedia_docs 
    )
    barbie_wikipedia_bm25_retriever.k = 1
    # set up FAISS vector store and create retriever
    barbie_wikipedia_faiss_store = FAISS.from_documents(
        chunked_barbie_wikipedia_docs,
        embedder 
    )
    barbie_wikipedia_faiss_retriever = barbie_wikipedia_faiss_store.as_retriever(search_kwargs={"k": 1})
    # set up ensemble retriever
    barbie_ensemble_retriever = EnsembleRetriever(
        retrievers=[barbie_wikipedia_bm25_retriever, barbie_wikipedia_faiss_retriever],
        weights= [0.25, 0.75]  # should sum to 1
    )
    # create retriever tools
    barbie_wikipedia_retrieval_tool = create_retriever_tool(
        retriever=barbie_ensemble_retriever,  
        name='Search_Wikipedia', 
        description='Useful for when you need to answer questions about plot, cast, production, release, music, marketing, reception, themes and analysis of the Barbie movie.' 
    )
    barbie_csv_retrieval_tool = create_retriever_tool(
        retriever=barbie_csv_faiss_retriever.as_retriever(),  
        name='Search_Reviews', 
        description='Useful for when you need to answer questions about public reviews of the Barbie movie.'
    )
    barbie_retriever_tools = [barbie_wikipedia_retrieval_tool, barbie_csv_retrieval_tool] 
    # retrieval agent
    barbie_retriever_agent_executor = create_conversational_retrieval_agent(llm=llm, tools=barbie_retriever_tools, verbose=True)


    # Oppenheimer retrieval system (Wikipedia, CSV)
    # load the multiple source documents for Oppenheimer and build FAISS index
    oppenheimer_wikipedia_docs = WikipediaLoader(
        query="Oppenheimer",
        load_max_docs=1,
        doc_content_chars_max=10000000
    ).load()
    oppenheimer_csv_docs = CSVLoader(
        file_path="./oppenheimer_data/oppenheimer.csv",
        source_column="Review"
    ).load()
    # chunk the loaded documents using the text splitters
    chunked_opp_wikipedia_docs = wikipedia_text_splitter.transform_documents(oppenheimer_wikipedia_docs)
    chunked_opp_csv_docs = csv_text_splitter.transform_documents(oppenheimer_csv_docs)
    # set up FAISS vector store and create retriever for CSV docs
    opp_csv_faiss_retriever = FAISS.from_documents(chunked_opp_csv_docs, embedder).as_retriever()
    # set up BM25 retriever
    opp_wikipedia_bm25_retriever = BM25Retriever.from_documents(chunked_opp_wikipedia_docs)
    opp_wikipedia_bm25_retriever.k = 1
    # set up FAISS vector store and create retriever
    opp_wikipedia_faiss_store = FAISS.from_documents(
        chunked_opp_wikipedia_docs,
        embedder
    )
    opp_wikipedia_faiss_retriever = opp_wikipedia_faiss_store.as_retriever(search_kwargs={"k": 1})
    # set up ensemble retriever
    opp_ensemble_retriever = EnsembleRetriever(
        retrievers=[opp_wikipedia_bm25_retriever, opp_wikipedia_faiss_retriever],
        weights= [0.25, 0.75]  # should sum to 1
    )
    # setup prompt
    system_message = """Use the information from the below two sources to answer any questions.
    
    Source 1: public user reviews about the Oppenheimer movie
    <source1>
    {source1}
    </source1>

    Source 2: the wikipedia page for the Oppenheimer movie including the plot summary, cast, and production information
    <source2>
    {source2}
    </source2>
    """
    prompt = ChatPromptTemplate.from_messages([("system", system_message), ("human", "{question}")])
    # build multi-source chain
    oppenheimer_multisource_chain = {
        "source1": (lambda x: x["question"]) | opp_ensemble_retriever,
        "source2": (lambda x: x["question"]) | opp_csv_faiss_retriever,
        "question": lambda x: x["question"],
    } | prompt | llm

    
    # Agent creation
    # set up tools
    def query_barbie(input):
        return barbie_retriever_agent_executor({"input" : input})
    def query_oppenheimer(input):
        return oppenheimer_multisource_chain.invoke({"question" : input})
    tools = [
        Tool(
            name="BarbieInfo",
            func=query_barbie,
            description='Useful when you need to answer questions about the Barbie movie'
        ),
        Tool(
            name="OppenheimerInfo",
            func=query_oppenheimer,
            description='Useful when you need to answer questions about the Oppenheimer movie' 
        ),
    ]
    # create prompt
    prefix = """Have a conversation with a human, answering the following questions as best you can. You have access to the following tools:"""
    suffix = """Begin!"

    Question: {input}
    {agent_scratchpad}"""
    prompt = ZeroShotAgent.create_prompt(
        tools=tools, 
        prefix=prefix, 
        suffix=suffix, 
        input_variables=['input', 'agent_scratchpad'] 
    )
    # chain llm with prompt
    llm_chain = LLMChain(llm=llm, prompt=prompt, verbose=True)
    # create reasoning agent
    barbenheimer_agent = ZeroShotAgent(
        llm_chain=llm_chain,
        tools=tools,
        verbose=True )
    # create execution agent 
    barbenheimer_agent_chain = AgentExecutor.from_agent_and_tools(
        agent=barbenheimer_agent,
        tools=tools, 
        verbose=True )
    
    cl.user_session.set("chain", barbenheimer_agent_chain)

    msg.content = f"Agent ready!"
    await msg.send()

@cl.on_message
async def main(message):
       
    # msg = cl.Message(content=f"Thinking...")
    # await msg.send()
    
    chain = cl.user_session.get("chain")
    cb = cl.LangchainCallbackHandler(
        stream_final_answer=False, answer_prefix_tokens=["FINAL", "ANSWER"]
    )
    cb.answer_reached = True
    res = chain.__call__(message, callbacks=[cb], )
    
    # print(res.keys()) # keys are "input" and "output"
    
    answer = res["output"]
    source_elements = []
    # visited_sources = set()

    # # Get the documents from the user session
    # docs = res["source_documents"]
    # metadatas = [doc.metadata for doc in docs]
    # all_sources = [m["source"] for m in metadatas]

    # for source in all_sources:
    #     if source in visited_sources:
    #         continue
    #     visited_sources.add(source)
    #     # Create the text element referenced in the message
    #     source_elements.append(
    #         cl.Text(content="https://www.imdb.com" + source, name="Review URL")
    #     )

    # if source_elements:
    #     answer += f"\nSources: {', '.join([e.content.decode('utf-8') for e in source_elements])}"
    # else:
    #     answer += "\nNo sources found"

    await cl.Message(content=answer, elements=source_elements).send()