app.py

import requests
from collections import Counter
from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification
import re
import gradio as gr

# Load model and tokenizer
model_name = "hamzab/roberta-fake-news-classification"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)
classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

# API Keys (consider moving these to environment variables in production)
GOOGLE_API_KEY = "AIzaSyCKnDlhWih34GdCuheNusnrEw_YE_q6GWQ"
NEWSAPI_KEY = "fcb304918fce4fb29b17b6c95dbc7518"

# Keyword extractor
def extract_keywords(text, max_words=5):
    words = re.findall(r'\b\w+\b', text.lower())
    stopwords = {
        "the", "and", "is", "in", "to", "of", "a", "on", "for", "with", 
        "as", "by", "at", "an", "be", "are", "from", "this", "it", "that"
    }
    filtered = [w for w in words if w not in stopwords and len(w) > 2]
    return " ".join([w for w, _ in Counter(filtered).most_common(max_words)])

# Google Fact Check
def check_google_fact_check(query):
    try:
        response = requests.get(
            "https://factchecktools.googleapis.com/v1alpha1/claims:search",
            params={"query": query, "key": GOOGLE_API_KEY}
        ).json()
        if "claims" in response and response["claims"]:
            result = ""
            for claim in response["claims"][:3]:
                text = claim.get("text", "")
                review = claim.get("claimReview", [{}])[0]
                rating = review.get("textualRating", "Unrated")
                source = review.get("publisher", {}).get("name", "Unknown")
                url = review.get("url", "")
                result += f"✅ Claim: {text}\n📊 Verdict: {rating}\n📰 Source: {source}\n🔗 {url}\n\n"
            return result.strip()
        return None
    except Exception as e:
        return f"❌ Google API error: {e}"

# NewsAPI fallback
def search_newsapi(query):
    try:
        response = requests.get("https://newsapi.org/v2/everything", params={
            "q": query, "apiKey": NEWSAPI_KEY, "language": "en", "sortBy": "relevancy", "pageSize": 3
        }).json()
        if response.get("status") != "ok":
            return f"❌ NewsAPI error: {response.get('message')}"
        articles = response.get("articles", [])
        if not articles:
            return "ℹ️ No similar real news articles found."
        output = "📰 No official fact-check found.\n\nBut here are similar real news articles:\n\n"
        for article in articles:
            title = article.get("title", "No title")
            source = article.get("source", {}).get("name", "Unknown")
            url = article.get("url", "#")
            output += f"• 📰 Source: {source}\n  📌 Title: {title}\n  🔗 {url}\n\n"
        return output.strip()
    except Exception as e:
        return f"❌ NewsAPI error: {e}"

# Chatbot function
def fake_news_chatbot(message, history):
    try:
        prediction = classifier(message)[0]
        label = prediction['label']
        confidence = round(prediction['score'], 2)

        if confidence < 0.7:
            verdict = "🟡 UNSURE"
        elif label == "FAKE":
            verdict = "🟥 FAKE"
        else:
            verdict = "🟩 REAL"

        ml_result = f"📊 ML Prediction: {verdict}\n🔢 Confidence: {confidence}"

        keywords = extract_keywords(message)
        fact_result = check_google_fact_check(keywords)
        if fact_result:
            return f"{ml_result}\n\n🔍 Keywords Used: {keywords}\n\n{fact_result}"

        newsapi_result = search_newsapi(keywords)
        if "📰 Source:" in newsapi_result and verdict == "🟥 FAKE":
            ml_result += " ⚠️ ML says FAKE but similar real news found"
        elif "No similar" in newsapi_result and verdict == "🟩 REAL":
            ml_result += " ⚠️ ML says REAL but no matching news found"

        return f"{ml_result}\n\n🔍 Keywords Used: {keywords}\n\n{newsapi_result}"

    except Exception as e:
        return f"❌ Error: {str(e)}"

# Gradio interface
gr.ChatInterface(
    fn=fake_news_chatbot,
    title="🧠 Fake News Detection Chatbot",
    description="Ask me if a news item is real or fake. I’ll use an ML model + Google Fact Check + NewsAPI!",
    examples=[
        "The Prime Minister announced a new moon mission",
        "Aliens landed in New York yesterday",
        "COVID vaccine turns you into a lizard"
    ],
).launch()