app.py

import streamlit as st
import joblib
import re
from related_news import fetch_related_articles

# Load model and vectorizer
model = joblib.load("model.pkl")
vectorizer = joblib.load("vectorizer.pkl")

# 🧼 Text cleaning function
def clean_text(text):
    text = re.sub(r"http\S+", "", text)
    text = re.sub(r"[^a-zA-Z\s]", "", text)
    text = text.lower()
    return text

# 💅 Page setup
st.set_page_config(
    page_title="TruthRadar 🧠",
    page_icon="🛰️",
    layout="centered",
    initial_sidebar_state="auto"
)

# 🎨 Custom styles for light theme
st.markdown("""
    <style>
        body, .reportview-container, .main {
            background-color: white !important;
            color: black !important;
        }
        .stTextArea textarea {
            background-color: #ffffff !important;
            color: #111827 !important;
            font-size: 16px !important;
            border: 1px solid #d1d5db !important;
            border-radius: 10px !important;
            padding: 12px !important;
        }
        .stButton > button {
            background-color: #2563eb;
            color: white;
            border-radius: 0.5rem;
            padding: 0.6rem 1.2rem;
        }
        .stButton > button:hover {
            background-color: #1e40af;
        }
    </style>
""", unsafe_allow_html=True)

# 🛰️ Header
st.markdown("<h1 style='text-align: center; color: #1f2937;'>🛰️ TruthRadar</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center; color: #4b5563;'>Detect fake news in a flash — headlines or full articles 🔍</p>", unsafe_allow_html=True)
st.markdown("---")
st.markdown("<p style='color: #4b5563;'>Currently optimised for US-based news articles.</p>", unsafe_allow_html=True)

#input
st.markdown("<label style='font-weight:600; font-size:16px; color:#1f2937;'>📝 Paste any news headline or article below:</label>", unsafe_allow_html=True)
user_input = st.text_area("", height=200)

# 🚀 Analyze button
if st.button("🚀 Analyze"):
    if not user_input.strip():
        st.warning("Bruhhh paste *something* to analyze 😅")
    else:
        with st.spinner("🧠 Scanning for truth..."):
            try:
                cleaned = clean_text(user_input)
                transformed = vectorizer.transform([cleaned])
                prediction = model.predict(transformed)[0]
                proba = model.predict_proba(transformed)[0]
                confidence = max(proba) * 100

                label = prediction.upper()
                articles = fetch_related_articles(user_input)

                # ✨ Show prediction result using markdown for HTML formatting
                if label == "REAL":
                    st.markdown("<div style='background-color:#dcfce7; padding:10px; border-radius:8px;'><strong>✅ Prediction:</strong> <span style='color:green;'>REAL</span></div>", unsafe_allow_html=True)
                else:
                    st.markdown("<div style='background-color:#fee2e2; padding:10px; border-radius:8px;'><strong>❌ Prediction:</strong> <span style='color:#b91c1c;'>FAKE</span></div>", unsafe_allow_html=True)

                st.markdown(f"<div style='background-color:#e0f2fe; padding:10px; border-radius:8px;'><strong>🔎 Confidence:</strong> <span style='color:#2563eb;'>{confidence:.2f}%</span></div>", unsafe_allow_html=True)

                # 🧐 Improved keyword matching for warning
                if label == "FAKE":
                    input_keywords = set(re.findall(r'\b\w{4,}\b', user_input.lower()))
                    similar_found = any(
                        any(word in (article.get("title", "") + article.get("description", "")).lower()
                            for word in input_keywords)
                        for article in articles
                    )
                    if similar_found:
                        st.markdown(
                            "<div style='background-color:#fef9c3; padding:10px; border-radius:8px;'><strong>🧐 Warning:</strong> <span style='color:#92400e;'> Similar stories were found from trusted sources. Cross-check below.</span></div>",
                            unsafe_allow_html=True
                        )

                # 📰 Related News Display
                st.markdown("---")
                st.markdown("<h3 style='color:#1f2937;'>📰 Related News from Trusted Sources</h3>", unsafe_allow_html=True)

                if not articles:
                    st.write("No related articles found.")
                else:
                    for article in articles:
                        title = article.get("title", "No title")
                        url = article.get("url", "#")
                        desc = article.get("description", "No description available.")

                        st.markdown(f"**[{title}]({url})**")
                        st.caption(desc)

            except Exception as e:
                st.error(f"🚨 Error: {e}")

# 📎 Footer
st.markdown("---")
st.markdown("<p style='text-align: center; font-size: 0.9em; color: gray;'>Made with 💻 by Sai Srikar • TruthRadar AI</p>", unsafe_allow_html=True)