import streamlit as st
import requests
import torch
from transformers import pipeline
from deepface import DeepFace
from PIL import Image
import cv2
import numpy as np
from bs4 import BeautifulSoup

# Load Fake News Detection Models
fake_news_pipeline = pipeline("text-classification", model="mrm8488/bert-tiny-finetuned-fake-news-detection")
sentiment_pipeline = pipeline("sentiment-analysis", model="nlptown/bert-base-multilingual-uncased-sentiment")
ai_detector_pipeline = pipeline("text-classification", model="roberta-base-openai-detector")

def classify_text(news_text):
    fake_news_result = fake_news_pipeline(news_text)[0]
    sentiment_result = sentiment_pipeline(news_text)[0]
    ai_result = ai_detector_pipeline(news_text)[0]
    
    fake_label = fake_news_result['label'].lower()
    sentiment_label = sentiment_result['label'].lower()
    ai_label = ai_result['label'].lower()
    
    fake_score = fake_news_result['score'] * 100
    sentiment_score = sentiment_result['score'] * 100
    ai_score = ai_result['score'] * 100
    
    final_label = "Fake" if fake_label == "fake" or ai_label == "ai-generated" else "Real"
    return final_label, round(fake_score, 2), round(sentiment_score, 2), round(ai_score, 2)

def analyze_image(image):
    try:
        analysis = DeepFace.analyze(image, actions=["emotion"])
        dominant_emotion = analysis[0]["dominant_emotion"]
        return "Fake" if dominant_emotion in ["fear", "surprise"] else "Real"
    except Exception as e:
        return "Error: " + str(e)

def analyze_video(video_path):
    try:
        cap = cv2.VideoCapture(video_path)
        frame_count = 0
        results = []
        while cap.isOpened():
            ret, frame = cap.read()
            if not ret or frame_count >= 10:
                break
            image = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
            result = analyze_image(image)
            results.append(result)
            frame_count += 1
        cap.release()
        return "Fake" if results.count("Fake") > results.count("Real") else "Real"
    except Exception as e:
        return "Error: " + str(e)

def verify_news(news_text):
    search_url = f"https://www.google.com/search?q={'%20'.join(news_text.split())}"
    sources = [
        f"https://www.bbc.co.uk/search?q={'%20'.join(news_text.split())}",
        f"https://www.cnn.com/search?q={'%20'.join(news_text.split())}",
        f"https://www.reuters.com/search/news?blob={'%20'.join(news_text.split())}",
        f"https://www.factcheck.org/?s={'%20'.join(news_text.split())}",
        f"https://www.snopes.com/?s={'%20'.join(news_text.split())}",
        f"https://www.politifact.com/search/?q={'%20'.join(news_text.split())}"
    ]
    return search_url, sources

st.set_page_config(page_title="Fake News Detector", layout="wide")
st.title("📰 Fake News Detector")

col1, col2, col3 = st.columns(3)

with col1:
    st.header("Text News Analysis")
    news_text = st.text_area("Enter the news content to check:", height=200)
    if st.button("Analyze News", key="text_analyze"):
        if news_text.strip():
            result, fake_acc, sentiment_acc, ai_acc = classify_text(news_text)
            verification_link, sources = verify_news(news_text)
            st.write(f"**Result:** {result}")
            st.write(f"**Fake News Score:** {fake_acc}%")
            st.write(f"**Sentiment Score:** {sentiment_acc}%")
            st.write(f"**AI Detection Score:** {ai_acc}%")
            st.markdown(f"[Verify on Google]({verification_link})")
            for source in sources:
                st.markdown(f"[Check Source]({source})")
        else:
            st.warning("Please enter some text.")

with col2:
    st.header("Image News Analysis")
    uploaded_image = st.file_uploader("Upload a news image", type=["jpg", "png", "jpeg"])
    if uploaded_image and st.button("Analyze Image", key="image_analyze"):
        image = Image.open(uploaded_image)
        result = analyze_image(image)
        st.image(image, caption="Uploaded Image", use_container_width=True)
        st.write(f"**Result:** {result}")
        verification_link, sources = verify_news("Fake news image verification")
        st.markdown(f"[Verify on Google]({verification_link})")
        for source in sources:
            st.markdown(f"[Check Source]({source})")

with col3:
    st.header("Video News Analysis")
    video_url = st.text_input("Enter the video link:")
    if st.button("Analyze Video", key="video_analyze"):
        if video_url.strip():
            result = analyze_video(video_url)
            st.video(video_url)
            st.write(f"**Result:** {result}")
            verification_link, sources = verify_news("Fake news video verification")
            st.markdown(f"[Verify on Google]({verification_link})")
            for source in sources:
                st.markdown(f"[Check Source]({source})")
        else:
            st.warning("Please enter a valid video link.")