app.py

import gradio as gr
import ollama
import bs4
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.document_loaders import WebBaseLoader
from langchain_community.document_loaders import PyPDFLoader
from langchain_community.vectorstores import Chroma
from langchain_community.embeddings import OllamaEmbeddings

# Check if user has inputted a URL or uploaded a document and load, split, and retrieve documents
def load_and_retrieve(url, document):
    
    # If user has inputted a URL
    if url:
        loader = WebBaseLoader(
            web_paths=(url,),
            bs_kwargs=dict() 
        )
        docs = loader.load()
        text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=200)
        splits = text_splitter.split_documents(docs)
        embeddings = OllamaEmbeddings(model="nomic-embed-text")
        vectorstore = Chroma.from_documents(documents=splits, embedding=embeddings)
        return vectorstore.as_retriever()
    
    # If user has uploaded a document
    if document:
        loader = PyPDFLoader(document)
        docs = loader.load_and_split()
        text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=200)
        splits = text_splitter.split_documents(docs)
        embeddings = OllamaEmbeddings(model="nomic-embed-text")
        vectorstore = Chroma.from_documents(documents=splits, embedding=embeddings)
        return vectorstore.as_retriever()

# Function to format documents
def format_docs(docs):
    # Return the page content of each document
    return "\n\n".join(doc.page_content for doc in docs)

# Function that defines the RAG chain
def rag_chain(url = False, document = False, question = ''):
    retriever = load_and_retrieve(url, document)
    retrieved_docs = retriever.invoke(question)
    formatted_context = format_docs(retrieved_docs)
    formatted_prompt = f"Question: {question}\n\nContext: {formatted_context}"
    print("==============")
    print(formatted_prompt)
    print("==============")
    response = ollama.chat(model='llama3', messages=[{'role': 'user', 'content': formatted_prompt}])
    return response['message']['content']

# Gradio interface
iface = gr.Interface(
    fn=rag_chain,
    inputs=["text", "file", "text"],
    outputs="text",
    title="RAG Chain Question Answering",
    description="Enter a URL or upload a document and a query to get answers from the RAG chain."
)

# Launch the app
iface.launch(share=True)