app.py

import streamlit as st
import ollama
import os
import logging

from langchain_ollama import ChatOllama
from langchain_community.llms import Ollama

from langchain_community.document_loaders import PyPDFLoader

from langchain_text_splitters import RecursiveCharacterTextSplitter

from langchain.embeddings import HuggingFaceEmbeddings

import faiss
from langchain_community.vectorstores import FAISS
from langchain_community.docstore.in_memory import InMemoryDocstore

from langchain import hub
from langchain_core.output_parsers import StrOutputParser
from langchain_core.runnables import RunnablePassthrough
from langchain_core.prompts import ChatPromptTemplate

from typing import List, Tuple, Dict, Any, Optional


# pip install -qU langchain-ollama
# pip install langchain

        ##### Logging



def format_docs(docs):
    return "\n\n".join([doc.page_content for doc in docs])

@st.cache_resource(show_spinner=True)
def extract_model_names(
    models_info: Dict[str, List[Dict[str, Any]]],
) -> Tuple[str, ...]:
    """
    Extract model names from the provided models information.

    Args:
        models_info (Dict[str, List[Dict[str, Any]]]): Dictionary containing information about available models.

    Returns:
        Tuple[str, ...]: A tuple of model names.
    """

    # Logging configuration
    logging.basicConfig(
        level=logging.INFO,
        format="%(asctime)s - %(levelname)s - %(message)s",
        datefmt="%Y-%m-%d %H:%M:%S",
    )

    logger = logging.getLogger(__name__)

    logger.info("Extracting model names from models_info")
    model_names = tuple(model["name"] for model in models_info["models"])
    logger.info(f"Extracted model names: {model_names}")
    return model_names


def generate_response(rag_chain, input_text):

    response = rag_chain.invoke(input_text)

    return response

def get_pdf(uploaded_file):
    temp_file = "./temp.pdf"
    if uploaded_file :
        #temp_file = "./temp.pdf"
        # Delete the existing temp.pdf file if it exists
        if os.path.exists(temp_file):
            os.remove(temp_file)
        with open(temp_file, "wb") as file:
            file.write(uploaded_file.getvalue())
            file_name = uploaded_file.name
    loader = PyPDFLoader(temp_file)
    docs = loader.load()
    return docs

def inference(chain, input_query):
    """Invoke the processing chain with the input query."""
    result = chain.invoke(input_query)
    return result


def main() -> None:

    st.title("🧠 This is a RAG Chatbot with Ollama and Langchain !!!")

    st.write("The LLM model Llama-3.2 is used")
    st.write("You can upload a PDF to chat with !!!")

    with st.sidebar:
        st.title("PDF FILE UPLOAD:")
        docs = st.file_uploader("Upload your PDF File and Click on the Submit & Process Button", accept_multiple_files=False, key="pdf_uploader")

    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)

    raw_text = get_pdf(docs)

    chunks = text_splitter.split_documents(raw_text)

    #embeddings = OllamaEmbeddings(model='nomic-embed-text', base_url="http://localhost:11434")
    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")

    single_vector = embeddings.embed_query("this is some text data")

    index = faiss.IndexFlatL2(len(single_vector))

    vector_store = FAISS(
            embedding_function=embeddings,
            index=index,
            docstore=InMemoryDocstore(),
            index_to_docstore_id={}
    )

    ids = vector_store.add_documents(documents=chunks)

    ## Retreival

    retriever = vector_store.as_retriever(search_type="mmr", search_kwargs = {'k': 3, 
                                                                            'fetch_k': 100,
                                                                            'lambda_mult': 1})


    prompt = """
        You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question.
        If you don't know the answer, just say that you don't know.
        Answer in bullet points. Make sure your answer is relevant to the question and it is answered from the context only.
        Question: {question} 
        Context: {context} 
        Answer:
    """

    prompt = ChatPromptTemplate.from_template(prompt)

    ## from llama3.2:latest to unsloth/Llama-3.2-3B mistralai/Mistral-Nemo-Instruct-2407
    model = ChatOllama(model="mistralai/Mistral-Nemo-Instruct-2407")


    
    rag_chain = (
        {"context": retriever|format_docs, "question": RunnablePassthrough()}
        | prompt
        | model
        | StrOutputParser()
    )


    with st.form("llm-form"):
        text = st.text_area("Enter your question or statement:")
        submit = st.form_submit_button("Submit")

    if "chat_history" not in st.session_state:
        st.session_state['chat_history'] = []

    if submit and text:
        with st.spinner("Generating response..."):
            # Ken (12/11/2024): modify start
            response = generate_response(rag_chain, text)
            # Ken (12/11/2024): modify end
            st.session_state['chat_history'].append({"user": text, "ollama": response})
            st.write(response)

    st.write("## Chat History")
    for chat in reversed(st.session_state['chat_history']):
        st.write(f"**🧑 User**: {chat['user']}")
        st.write(f"**🧠 Assistant**: {chat['ollama']}")
        st.write("---")

if __name__ == "__main__":
    main()