app.py

import os 
import gradio as gr

from langchain.document_loaders import PDFMinerLoader,CSVLoader ,UnstructuredWordDocumentLoader,TextLoader,OnlinePDFLoader
from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings import SentenceTransformerEmbeddings
from langchain.vectorstores import FAISS
from langchain import HuggingFaceHub
from langchain.chains import RetrievalQA
from langchain.prompts import PromptTemplate

DEVICE = 'cpu'
FILE_EXT = ['pdf','text','csv','word','wav']


def loading_file():
    return "Loading..."


def get_openai_chat_model(API_key):
    try:
        from langchain.llms import OpenAI
    except ImportError as err:
        raise "{}, unable to load openAI. Please install openai and add OPENAIAPI_KEY"
    os.environ["OPENAI_API_KEY"] = API_key
    llm = OpenAI()
    return llm

def process_documents(documents,data_chunk=1000,chunk_overlap=50):
    text_splitter = CharacterTextSplitter(chunk_size=data_chunk, chunk_overlap=chunk_overlap,separator='\n')
    texts = text_splitter.split_documents(documents)
    return texts

def get_hugging_face_model(model_id,API_key,temperature=0.1):
    chat_llm = HuggingFaceHub(huggingfacehub_api_token=API_key,
                                      repo_id=model_id,
                                      model_kwargs={"temperature": temperature, "max_new_tokens": 2048})
    return chat_llm

def chat_application(llm_service,key):
    if llm_service == 'HuggingFace':
        llm = get_hugging_face_model(model_id='tiiuae/falcon-7b-instruct',API_key=key)
    else:
        llm = get_openai_chat_model(API_key=key)
    return llm 

def summarize_contents():
    question = "Generate a summary of the contents. Do not return the response in json format"
    return qa.run(question)

def document_loader(file_path,api_key,doc_type='pdf',llm='Huggingface'):
    document = None
    if doc_type == 'pdf':
        document = process_pdf_document(document_file=file_path)
    elif doc_type == 'text':
        document = process_text_document(document_file=file_path)
    elif doc_type == 'csv':
        document = process_csv_document(document_file=file_path)
    elif doc_type == 'word':
        document = process_word_document(document_file=file_path)
    if document:
        embedding_model = SentenceTransformerEmbeddings(model_name='thenlper/gte-base',model_kwargs={"device": DEVICE})
        texts = process_documents(documents=document)
        vector_db = FAISS.from_documents(documents=texts, embedding= embedding_model)
        global qa 
        qa = RetrievalQA.from_chain_type(llm=chat_application(llm_service=llm,key=api_key),
                                     chain_type='stuff',
                                     retriever=vector_db.as_retriever(),
                                    #  chain_type_kwargs=chain_type_kwargs,
                                    #  return_source_documents=True
                                    )
    else:
        return "Error in loading Documents "
    
    return "Document processing complete-Embeddings Created "

        
def process_text_document(document_file):
    loader = TextLoader(document_file.name)
    document = loader.load()
    return document

def process_csv_document(document_file):
    loader = CSVLoader(file_path=document_file.name)
    document = loader.load()
    return document


def process_word_document(document_file):
    loader = UnstructuredWordDocumentLoader(file_path=document_file.name)
    document = loader.load()
    return document


def process_pdf_document(document_file):
    print("Document File Name :",document_file.name)
    loader = PDFMinerLoader(document_file.name)
    document = loader.load()
    return document



css="""
#col-container {max-width: 700px; margin-left: auto; margin-right: auto;}
"""

title = """
<div style="text-align: center;max-width: 700px;">
    <h1>Chat with Data • OpenAI/HuggingFace</h1>
    <p style="text-align: center;">Upload a file from your computer, click the "Load data to LangChain" button, <br />
    when everything is ready, you can start asking questions about the data you uploaded ;) <br />
    This version is just for QA retrival so it will not use chat history, and uses Hugging face as LLM, 
    so you don't need any key</p>
</div>
"""

with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.HTML(title)
        
        with gr.Column():
            with gr.Box():
                gr.Row()
                LLM_option = gr.Dropdown(['HuggingFace','OpenAI'],label='Large Language Model Selection',info='LLM Service')
                file_extension = gr.Dropdown(FILE_EXT, label="File Extensions", info="Select your files extensions!")
            API_key = gr.Textbox(label="Add API key", type="password")
            with gr.Column():
                with gr.Box():
                    pdf_doc = gr.File(label="Upload File to start QA", file_types=FILE_EXT, type="file")
                    with gr.Row():
                        langchain_status = gr.Textbox(label="Status", placeholder="", interactive=True)
                        load_pdf = gr.Button("Upload File & Generate Embeddings",).style(full_width=False)
            with gr.Row():
                summary = gr.Textbox(label="Summary")
                summarize_pdf = gr.Button("Summarize the Contents").style(full_width=False)

        chatbot = gr.Chatbot()
        question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter")
        submit_button = gr.Button("Send Message")

    load_pdf.click(loading_file, None, langchain_status, queue=False)    
    load_pdf.click(document_loader, inputs=[pdf_doc,API_key,file_extension,LLM_option], outputs=[langchain_status], queue=False)
    
    summarize_pdf.click(summarize_contents,outputs=summary)
    
    with gr.Column():
        with gr.Row():
            chatbot = gr.Chatbot(height=300)
            sources = gr.HTML(value = "Source paragraphs where I looked for answers will appear here", height=300)

        with gr.Row():
            message = gr.Textbox(label="Type your question?",lines=1).style(full_width=False)
        submit_query = gr.Button(value="Send message", variant="secondary", scale = 1)


demo.launch()