Update app.py

app.py

@@ -112,8 +112,8 @@ def moderate_content(api_key, user_message, chat_history):
         # Get the response
         moderation_result = chat_completion.choices[0].message.content
         
-        # Parse and format the response
-        formatted_response = format_moderation_response(moderation_result, user_message)
+        # Format the response - ONLY show detailed analysis
+        formatted_response = format_moderation_response(moderation_result)
         
         # Update chat history with proper message format
         user_msg = {"role": "user", "content": user_message}
@@ -129,50 +129,17 @@ def moderate_content(api_key, user_message, chat_history):
         new_history = chat_history + [user_msg, assistant_msg]
         return new_history, new_history
 
-def format_moderation_response(result, original_content):
+def format_moderation_response(result):
     """
-    Format the moderation result into a readable response
+    Format the moderation result - ONLY show detailed analysis
     """
     try:
-        result_lower = result.lower()
-        
-        # Build formatted response
-        response = "## 🛡️ Content Moderation Result\n\n"
-        response += f"**Analyzed Content:** _{original_content[:100]}{'...' if len(original_content) > 100 else ''}_\n\n"
-        
-        # Check if content appears safe
-        if "safe" in result_lower and ("unsafe" not in result_lower or result_lower.index("safe") < result_lower.index("unsafe")):
-            response += "### ✅ Status: SAFE\n\n"
-            response += "The content appears to be appropriate and does not violate any harm policies.\n\n"
-        else:
-            response += "### ⚠️ Status: FLAGGED\n\n"
-            
-            # Check for severity
-            if "critical" in result_lower:
-                response += "**Severity:** 🔴 CRITICAL\n\n"
-            elif "high" in result_lower:
-                response += "**Severity:** 🟠 HIGH\n\n"
-            elif "medium" in result_lower:
-                response += "**Severity:** 🟡 MEDIUM\n\n"
-            elif "low" in result_lower:
-                response += "**Severity:** 🟢 LOW\n\n"
-            
-            # Check for specific categories
-            flagged_categories = []
-            for code, category in HARM_CATEGORIES.items():
-                if code.lower() in result_lower or code in result:
-                    flagged_categories.append(f"- **{code}: {category}** - {HARM_DESCRIPTIONS[code]}")
-            
-            if flagged_categories:
-                response += "**Flagged Categories:**\n" + "\n".join(flagged_categories) + "\n\n"
-        
-        # Add detailed analysis
-        response += "---\n\n### 📊 Detailed Analysis:\n\n" + result
-        
+        # Simply return the detailed analysis with header
+        response = "### 📊 Detailed Analysis:\n\n" + result
         return response
         
     except Exception as e:
-        return f"**Moderation Analysis:**\n\n{result}"
+        return f"### 📊 Detailed Analysis:\n\n{result}"
 
 def clear_chat():
     """Clear the chat history"""
@@ -214,7 +181,7 @@ with gr.Blocks(title="Content Moderation Chatbot", theme=gr.themes.Soft()) as ap
     chatbot = gr.Chatbot(
         label="Moderation Results",
         height=450,
-        show_label=True,
+        show_label=True
     )
     
     with gr.Row():
@@ -335,10 +302,11 @@ with gr.Blocks(title="Content Moderation Chatbot", theme=gr.themes.Soft()) as ap
     ### 🎯 Key Features:
     
     - ✅ **Enhanced System Prompt**: Detailed instructions for comprehensive analysis
-    - ✅ **Severity Levels**: LOW, MEDIUM, HIGH, or CRITICAL risk assessment
+    - ✅ **Direct Model Output**: Shows only the detailed analysis from the model
     - ✅ **Category Detection**: Identifies all applicable harm categories
     - ✅ **Detailed Explanations**: Clear reasoning for each flag
     - ✅ **15 Example Queries**: One safe example + one for each harm category
+    - ✅ **Clean Interface**: No extra formatting, just pure analysis
     
     ### 🔒 Privacy & Security: