app.py

import gradio as gr
import os
from langchain.vectorstores import Chroma  # Chroma als Vektordatenbank
from langchain.document_loaders import PyPDFLoader
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.chains import ConversationalRetrievalChain
from langchain.memory import ConversationBufferMemory
from langchain.llms import HuggingFaceHub
from langchain.text_splitter import RecursiveCharacterTextSplitter

list_llm = ["google/flan-t5-small", "sentence-transformers/all-MiniLM-L6-v2"]

def load_doc(list_file_path):
    loaders = [PyPDFLoader(x) for x in list_file_path]
    pages = []
    for loader in loaders:
        pages.extend(loader.load())
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=32)
    doc_splits = text_splitter.split_documents(pages)
    return doc_splits

def create_db(splits):
    embeddings = HuggingFaceEmbeddings()
    vectordb = Chroma.from_documents(splits, embeddings)
    return vectordb

def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db):
    llm = HuggingFaceHub(
        repo_id=llm_model,
        model_kwargs={
            "temperature": temperature,
            "max_length": max_tokens,
            "top_k": top_k,
        }
    )
    memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
    retriever = vector_db.as_retriever()
    qa_chain = ConversationalRetrievalChain.from_llm(
        llm,
        retriever=retriever,
        chain_type="stuff",
        memory=memory,
        return_source_documents=True,
        verbose=False
    )
    return qa_chain

def initialize_database(list_file_obj):
    list_file_path = [x.name for x in list_file_obj if x is not None]
    doc_splits = load_doc(list_file_path)
    vector_db = create_db(doc_splits)
    return vector_db, "Datenbank erfolgreich erstellt!"

def initialize_LLM(llm_option, llm_temperature, max_tokens, top_k, vector_db):
    llm_name = list_llm[llm_option]
    qa_chain = initialize_llmchain(llm_name, llm_temperature, max_tokens, top_k, vector_db)
    return qa_chain, "LLM erfolgreich initialisiert! Chatbot ist bereit."

def conversation(qa_chain, message, history):
    formatted_chat_history = [(f"User: {m}", f"Assistant: {r}") for m, r in history]
    response = qa_chain({"question": message, "chat_history": formatted_chat_history})
    response_answer = response["answer"]
    new_history = history + [(message, response_answer)]
    return qa_chain, gr.update(value=""), new_history

def demo():
    with gr.Blocks() as demo:
        vector_db = gr.State()
        qa_chain = gr.State()
        gr.HTML("<center><h1>RAG PDF Chatbot (Kostenlose Version)</h1></center>")
        with gr.Row():
            with gr.Column():
                document = gr.Files(height=300, file_count="multiple", file_types=[".pdf"], interactive=True)
                db_btn = gr.Button("Erstelle Vektordatenbank")
                db_progress = gr.Textbox(value="Nicht initialisiert", show_label=False)
                llm_btn = gr.Radio(["Flan-T5-Small", "MiniLM"], label="Verfügbare Modelle")
                slider_temperature = gr.Slider(0.01, 1.0, value=0.5, label="Temperature")
                slider_maxtokens = gr.Slider(64, 512, value=256, label="Max Tokens")
                qachain_btn = gr.Button("Initialisiere QA-Chatbot")

            with gr.Column():
                chatbot = gr.Chatbot(height=400, type="messages")
                msg = gr.Textbox(placeholder="Frage stellen...")
                submit_btn = gr.Button("Absenden")

        db_btn.click(initialize_database, [document], [vector_db, db_progress])
        qachain_btn.click(initialize_LLM, [llm_btn, slider_temperature, slider_maxtokens, vector_db], [qa_chain])
        submit_btn.click(conversation, [qa_chain, msg, chatbot], [qa_chain, msg, chatbot])
    demo.launch(debug=True)

if __name__ == "__main__":
    demo()