Update app.py

app.py

@@ -3,17 +3,17 @@ import torch
 from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
 from deep_translator import GoogleTranslator
 import requests
-import os
+import plotly.graph_objects as go
 
-# Load tokenizer and fake news model
+# Load models
 tokenizer = AutoTokenizer.from_pretrained("hamzab/roberta-fake-news-classification")
 model = AutoModelForSequenceClassification.from_pretrained("hamzab/roberta-fake-news-classification")
 
-# Load summarizer pipeline
-summarizer = pipeline("summarization", model="sshleifer/distilbart-cnn-12-6")
+summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
 
-# Google Fact Check API key (replace with your actual key or use st.secrets)
-GOOGLE_API_KEY = "AIzaSyAf5v5380xkpo0Rk3kBiSxpxYVBQwcDi2A"
+st.set_page_config(page_title="Fake News Detector", page_icon="📰")
+
+# --- Functions ---
 
 def translate_to_english(text):
     try:
@@ -22,11 +22,10 @@ def translate_to_english(text):
         return f"Error in translation: {e}"
 
 def summarize_text(text):
-    try:
-        summary = summarizer(text, max_length=60, min_length=15, do_sample=False)
+    if len(text) > 100:
+        summary = summarizer(text[:1024], max_length=80, min_length=30, do_sample=False)
         return summary[0]['summary_text']
-    except Exception as e:
-        return f"Error in summarization: {e}"
+    return text
 
 def predict_fake(title, text):
     input_str = "<title>" + title + "<content>" + text + "<end>"
@@ -35,59 +34,109 @@ def predict_fake(title, text):
     model.to(device)
     with torch.no_grad():
         output = model(input_ids["input_ids"].to(device), attention_mask=input_ids["attention_mask"].to(device))
-    return dict(zip(["Fake", "Real"], [x.item() for x in list(torch.nn.Softmax()(output.logits)[0])]))
+    probs = torch.nn.Softmax(dim=1)(output.logits)[0]
+    return {"Fake": probs[0].item(), "Real": probs[1].item()}
+
+def render_confidence_chart(confidence):
+    fig = go.Figure(data=[go.Pie(
+        values=[confidence, 100 - confidence],
+        labels=['Confidence', 'Uncertainty'],
+        hole=0.6,
+        marker_colors=[
+            f'hsl({confidence * 1.2}, 70%, 50%)',
+            'rgb(240, 240, 240)'
+        ],
+        textinfo='label+percent',
+        hoverinfo='label+value'
+    )])
+    fig.update_layout(
+        showlegend=False,
+        margin=dict(t=10, b=10, l=10, r=10),
+        annotations=[dict(text=f'{confidence}%', x=0.5, y=0.5, font_size=20, showarrow=False)],
+        height=300
+    )
+    st.plotly_chart(fig, use_container_width=True)
+
+def simulate_detected_patterns(text):
+    patterns = []
+    if "breaking" in text.lower():
+        patterns.append({"phrase": "breaking", "category": "clickbait", "impact": -5})
+    if "confirmed" in text.lower():
+        patterns.append({"word": "confirmed", "category": "assertive", "impact": 5})
+    if "shocking" in text.lower():
+        patterns.append({"word": "shocking", "category": "exaggeration", "impact": -10})
+    return patterns
 
 def fact_check_with_google(api_key, query):
     url = f"https://factchecktools.googleapis.com/v1alpha1/claims:search"
     params = {"query": query, "key": api_key}
     response = requests.get(url, params=params)
-    if response.status_code == 200:
-        return response.json()
+    return response.json() if response.status_code == 200 else {"error": f"Error: {response.status_code}"}
+
+# --- App UI ---
+
+st.title("📰 Fake News Detection App")
+st.markdown("Enter a news article to predict its credibility and view confidence metrics.")
+
+with st.form("news_form"):
+    title = st.text_input("📝 Title")
+    text = st.text_area("📜 Content")
+    language = st.selectbox("🌐 Language of Input", ["English", "Other"])
+    summarize_option = st.checkbox("🧠 Summarize before fact check")
+    check_fact = st.checkbox("🔍 Check with Google Fact Check")
+    GOOGLE_API_KEY = st.text_input("🔑 Google API Key (optional)", type="password")
+    submit_button = st.form_submit_button("🚀 Predict")
+
+if submit_button:
+    if language != "English":
+        title = translate_to_english(title)
+        text = translate_to_english(text)
+    
+    if summarize_option:
+        summary = summarize_text(title + " " + text)
+        st.markdown("### ✂️ Summary Used for Fact Check")
+        st.info(summary)
     else:
-        return {"error": f"Unable to fetch fact-checks. HTTP {response.status_code}: {response.text}"}
-
-def main():
-    st.set_page_config(page_title="News Credibility Checker", page_icon="🧠")
-    st.title("🧠 Fake News Detection & Fact Check")
-
-    with st.form("news_form"):
-        title = st.text_input("News Title")
-        text = st.text_area("News Content")
-        language = st.selectbox("Select Language", ["English", "Other"])
-        check_fact = st.checkbox("Check with Google Fact Check API")
-        submit_button = st.form_submit_button("Analyze")
-
-    if submit_button:
-        if language == "Other":
-            title = translate_to_english(title)
-            text = translate_to_english(text)
-        
-        prediction = predict_fake(title, text)
-
-        st.subheader("Prediction Result:")
-        st.write("Prediction Score:", prediction)
-        verdict = "real" if prediction.get("Real") > 0.5 else "fake"
-        st.success(f"This news is predicted to be **{verdict}**.")
-
-        if check_fact and GOOGLE_API_KEY:
-            # Generate summary for fact-checking
-            summary_text = summarize_text(title + ". " + text)
-            st.markdown("**Fact-check Query (Summary):** " + summary_text)
-
-            # Call Google Fact Check API
-            fact_check_data = fact_check_with_google(GOOGLE_API_KEY, summary_text)
-            st.subheader("Google Fact Check Results")
-
-            if "claims" in fact_check_data:
-                for claim in fact_check_data["claims"]:
-                    st.markdown(f"**Claim:** {claim.get('text', 'N/A')}")
-                    for review in claim.get("claimReview", []):
-                        st.write(f"- **Publisher**: {review.get('publisher', {}).get('name', 'N/A')}")
-                        st.write(f"- **Rating**: {review.get('textualRating', 'N/A')}")
-                        st.write(f"- **URL**: {review.get('url', 'N/A')}")
-                        st.write("---")
-            else:
-                st.info("No fact-check results found.")
-
-if __name__ == "__main__":
-    main()
+        summary = title + " " + text
+
+    prediction = predict_fake(title, text)
+    confidence = round(prediction["Real"] * 100)
+    verdict = "Real" if confidence > 60 else "Fake" if confidence < 40 else "Uncertain"
+    color = "green" if verdict == "Real" else "red" if verdict == "Fake" else "orange"
+
+    # Output
+    st.subheader("✅ Prediction Result")
+    st.markdown(f"""
+    <div style='background-color:#f7f9fc;padding:1rem;border-radius:10px'>
+        <b>Verdict:</b> <span style='color:{color}; font-weight:600'>{verdict}</span><br>
+        <b>Confidence:</b> {confidence}%
+    </div>
+    """, unsafe_allow_html=True)
+
+    render_confidence_chart(confidence)
+
+    # Detected Patterns
+    patterns = simulate_detected_patterns(title + " " + text)
+    if patterns:
+        st.subheader("🔎 Detected Language Patterns")
+        for p in patterns:
+            word = p.get("word") or p.get("phrase")
+            st.markdown(f"- **{word}** ({p['category']}) — "
+                        f"<span style='color:{'green' if p['impact']>0 else 'red'};'>{p['impact']:+}</span>",
+                        unsafe_allow_html=True)
+
+    GOOGLE_API_KEY = "AIzaSyAf5v5380xkpo0Rk3kBiSxpxYVBQwcDi2A"  
+    # Google Fact Check
+    if check_fact and GOOGLE_API_KEY:
+        st.subheader("🌐 Google Fact Check Results")
+        facts = fact_check_with_google(GOOGLE_API_KEY, summary)
+        if "claims" in facts:
+            for claim in facts["claims"]:
+                st.markdown(f"**Claim:** {claim.get('text', 'N/A')}")
+                for review in claim.get("claimReview", []):
+                    st.write(f"- **Publisher**: {review.get('publisher', {}).get('name', 'N/A')}")
+                    st.write(f"- **Rating**: {review.get('textualRating', 'N/A')}")
+                    st.write(f"- **URL**: {review.get('url', 'N/A')}")
+                    st.write("---")
+        else:
+            st.warning("No fact-check results found.")