app.py

import os
import json
import requests
import gradio as gr
from bs4 import BeautifulSoup
from groq import Groq
from youtube_transcript_api import YouTubeTranscriptApi

BRIGHTDATA_API_KEY = os.getenv("BRIGHTDATA_API_KEY")
GROQ_API_KEY = os.getenv("GROQ_API_KEY")

# --- Persistent storage ---
HISTORY_FILE = "chat_history.json"
if os.path.exists(HISTORY_FILE):
    with open(HISTORY_FILE, "r") as f:
        conversation_history = json.load(f)
else:
    conversation_history = []

user_preferences = {
    "response_style": "formal",
    "cite_sources": True
}

client = Groq(api_key=GROQ_API_KEY)

def ask_llm(context, history):
    system_prompt = f"""
You are an AI assistant.
User Preferences:
{json.dumps(user_preferences)}

Context:
{context}
"""
    messages = [{"role": "system", "content": system_prompt}] + history
    response = client.chat.completions.create(
        model="llama-3.1-8b-instant",
        messages=messages,
        temperature=0.3
    )
    return response.choices[0].message.content

# --- Website Scraper ---
def scrape_website(url, question):
    try:
        headers = {"Authorization": f"Bearer {BRIGHTDATA_API_KEY}"}
        payload = {"zone": "web_unlocker1", "url": url, "format": "raw"}
        r = requests.post("https://api.brightdata.com/request", headers=headers, json=payload, timeout=60)
        if r.status_code != 200:
            return f"Bright Data Error: {r.status_code}"
        soup = BeautifulSoup(r.text, "html.parser")
        text = soup.get_text(separator=" ", strip=True)
        if not text:
            return "⚠️ Could not extract content from the website."
        # Build temporary history for this turn
        temp_history = conversation_history + [{"role":"user","content":question}]
        answer = ask_llm(text[:12000], temp_history)
        # Update session & persistent history
        conversation_history.append({"role":"user","content":question})
        conversation_history.append({"role":"assistant","content":answer})
        with open(HISTORY_FILE, "w") as f:
            json.dump(conversation_history, f, indent=2)
        return answer
    except Exception as e:
        return f"Error scraping website: {str(e)}"

# --- YouTube Q&A ---
from youtube_transcript_api import YouTubeTranscriptApi

def youtube_qa(video_id, question):
    try:
        video_id = video_id.strip()

        api = YouTubeTranscriptApi()  # create instance
        transcript_list = api.list(video_id)

        # Try English transcript
        transcript = None
        for t in transcript_list:
            if t.language_code == "en":
                transcript = t
                break

        if transcript is None:
            return "❌ No English transcript available for this video."

        transcript_data = transcript.fetch()
        full_text = " ".join([item.text for item in transcript_data])

        if not full_text:
            return "⚠️ Transcript found but empty."

        return ask_llm(full_text[:12000], question)

    except Exception as e:
        return f"❌ Could not retrieve transcript.\nDetails: {str(e)}"
    
# --- User Preferences Update ---
def update_preferences(style, cite_sources):
    user_preferences["response_style"] = style
    user_preferences["cite_sources"] = cite_sources
    return f"Preferences updated: {json.dumps(user_preferences)}"

# --- Gradio Interface ---
with gr.Blocks() as demo:
    gr.Markdown("# 🤖 Version 3: Persistent Multi-Turn AI Chatbot")

    with gr.Tabs():
        # TAB 1: Website
        with gr.Tab("🌐 Website Q&A"):
            url_input = gr.Textbox(label="Enter Website URL")
            website_question = gr.Textbox(label="Ask a Question")
            website_output = gr.Textbox(label="Answer")
            website_btn = gr.Button("Ask")
            website_btn.click(scrape_website, inputs=[url_input, website_question], outputs=website_output)

        # TAB 2: YouTube
        with gr.Tab("🎥 YouTube Transcript Q&A"):
            video_id_input = gr.Textbox(label="Enter YouTube Video ID")
            youtube_question = gr.Textbox(label="Ask a Question")
            youtube_output = gr.Textbox(label="Answer")
            youtube_btn = gr.Button("Ask")
            youtube_btn.click(youtube_qa, inputs=[video_id_input, youtube_question], outputs=youtube_output)

        # TAB 3: Preferences
        with gr.Tab("⚙️ Preferences"):
            style_input = gr.Dropdown(choices=["formal", "informal"], label="Response Style")
            cite_input = gr.Checkbox(label="Cite Sources?")
            pref_btn = gr.Button("Update Preferences")
            pref_output = gr.Textbox(label="Status")
            pref_btn.click(update_preferences, inputs=[style_input, cite_input], outputs=pref_output)

demo.launch()