import os
import gradio as gr
from llama_index.readers.file import PDFReader
from llama_index.core import VectorStoreIndex
from llama_index.core.chat_engine.types import BaseChatEngine

# Set your OpenAI API key from environment (set this in Hugging Face Secrets)
os.environ['OPENAI_API_KEY'] = os.getenv("OPENAI_API_KEY")

# Globals
chat_engine: BaseChatEngine = None

# Function to process uploaded resume
def process_resume(file):
    global chat_engine
    if file is None:
        return "⚠️ Please upload a PDF file."
    reader = PDFReader()
    documents = reader.load_data(file=file)
    index = VectorStoreIndex.from_documents(documents)
    chat_engine = index.as_chat_engine(chat_mode="condense_question", verbose=False)
    return "✅ Resume uploaded and indexed! You can now ask questions."

# Chat function
def chat_with_resume(message, chat_history):
    global chat_engine
    if not chat_engine:
        return "⚠️ Please upload a resume first.", chat_history
    response = chat_engine.chat(message)
    chat_history.append({"role": "user", "content": message})
    chat_history.append({"role": "assistant", "content": response.response})
    return "", chat_history

# Gradio UI
with gr.Blocks() as demo:
    gr.Markdown("# 📄 Resume Chatbot\nUpload your resume and ask questions about your experience, skills, and more.")
    
    with gr.Row():
        file_input = gr.File(label="Upload Resume (PDF)", file_types=[".pdf"])
        upload_button = gr.Button("Process Resume")

    upload_output = gr.Textbox(label="Status")

    upload_button.click(fn=process_resume, inputs=file_input, outputs=upload_output)

    chatbot = gr.Chatbot(label="Chat with Resume", type="messages")
    message = gr.Textbox(placeholder="Ask something like: What are my key skills?", label="Your Question")
    send = gr.Button("Send")

    send.click(chat_with_resume, inputs=[message, chatbot], outputs=[message, chatbot])

demo.launch()