Update app.py

app.py

@@ -3,14 +3,12 @@ import pandas as pd
 from detoxify import Detoxify
 from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import torch
-import io
 
 # Load models
-tox_model = Detoxify('multilingual')  # 🌍 Multilingual toxicity classifier
+tox_model = Detoxify('multilingual')
 ai_tokenizer = AutoTokenizer.from_pretrained("openai-community/roberta-base-openai-detector")
 ai_model = AutoModelForSequenceClassification.from_pretrained("openai-community/roberta-base-openai-detector")
 
-# Thresholds
 TOXICITY_THRESHOLD = 0.7
 AI_THRESHOLD = 0.5
 
@@ -24,60 +22,49 @@ def detect_ai(text):
 def classify_comments(comment_list):
     results = tox_model.predict(comment_list)
     df = pd.DataFrame(results, index=comment_list).round(4)
-
-    # Capitalize columns
     df.columns = [col.replace("_", " ").title().replace(" ", "_") for col in df.columns]
     df.columns = [col.replace("_", " ") for col in df.columns]
-
-    # Add warning & AI detection
     df["⚠️ Warning"] = df.apply(lambda row: "⚠️ High Risk" if any(score > TOXICITY_THRESHOLD for score in row) else "✅ Safe", axis=1)
     df["🧪 AI Probability"] = [detect_ai(c) for c in df.index]
     df["🧪 AI Detection"] = df["🧪 AI Probability"].apply(lambda x: "🤖 Likely AI" if x > AI_THRESHOLD else "🧍 Human")
-
     return df
 
-def classify_from_textbox(text_input):
-    comment_list = [c.strip() for c in text_input.strip().split('\n') if c.strip()]
+def run_classification(text_input, csv_file):
+    comment_list = []
+
+    # From text input
+    if text_input.strip():
+        comment_list += [c.strip() for c in text_input.strip().split('\n') if c.strip()]
+
+    # From CSV
+    if csv_file:
+        df = pd.read_csv(csv_file.name)
+        if 'comment' not in df.columns:
+            return "CSV must contain a 'comment' column.", None
+        comment_list += df['comment'].astype(str).tolist()
+
     if not comment_list:
-        return "Please enter at least one comment.", None
-    df = classify_comments(comment_list)
-    csv_data = df.copy()
-    csv_data.insert(0, "Comment", df.index)
-    return df, ("toxicity_predictions.csv", csv_data.to_csv(index=False).encode())
+        return "Please provide comments via text or CSV.", None
 
-def classify_from_csv(file_obj):
-    df = pd.read_csv(file_obj.name)
-    if 'comment' not in df.columns:
-        return "CSV must contain a 'comment' column.", None
-    comment_list = df['comment'].astype(str).tolist()
     df = classify_comments(comment_list)
     csv_data = df.copy()
     csv_data.insert(0, "Comment", df.index)
     return df, ("toxicity_predictions.csv", csv_data.to_csv(index=False).encode())
 
-# Gradio Interface
-text_input = gr.Textbox(lines=8, label="💬 Paste Comments (one per line)")
-csv_input = gr.File(label="📥 Or Upload .CSV with 'comment' column")
-output_table = gr.Dataframe(label="📊 Predictions")
-download_button = gr.File(label="📤 Download CSV")
+# UI layout
+with gr.Blocks(title="🌍 Toxic Comment & AI Detector") as app:
+    gr.Markdown("## 🌍 Toxic Comment & AI Detector")
+    gr.Markdown("Detects multilingual toxicity and whether the comment is AI-generated. Paste text or upload CSV.")
 
-with gr.Blocks(title="Toxicity & AI Comment Detector") as app:
-    gr.Markdown("## 🌍 Toxic Comment & AI Detector\nDetects multilingual toxicity and whether the text is AI-generated.")
-    
-    with gr.Tab("📝 Paste Text"):
-        text = text_input
-        btn1 = gr.Button("Analyze Text Comments")
-        output1 = output_table
-        download1 = download_button
+    with gr.Row():
+        text_input = gr.Textbox(lines=8, label="💬 Enter Comments (one per line)")
+        file_input = gr.File(label="📥 Upload CSV (with 'comment' column)")
 
-    with gr.Tab("📁 Upload CSV"):
-        csv = csv_input
-        btn2 = gr.Button("Analyze CSV File")
-        output2 = output_table
-        download2 = download_button
+    submit_btn = gr.Button("🔍 Analyze Comments")
+    output_table = gr.Dataframe(label="📊 Results")
+    download_btn = gr.File(label="📤 Download CSV")
 
-    btn1.click(fn=classify_from_textbox, inputs=text, outputs=[output1, download1])
-    btn2.click(fn=classify_from_csv, inputs=csv, outputs=[output2, download2])
+    submit_btn.click(fn=run_classification, inputs=[text_input, file_input], outputs=[output_table, download_btn])
 
 if __name__ == "__main__":
     app.launch()