prompt_check.py

import re

def is_unsafe_prompt(model, tokenizer, system_prompt=None, user_prompt=None, max_new_token=10):
    """Basic prompt safety check"""
    if not user_prompt:
        return False
    
    # Simple keyword-based safety check
    unsafe_keywords = [
        "暴力", "血腥", "色情", "裸露", "仇恨", "歧视"
    ]
    
    for keyword in unsafe_keywords:
        if keyword in user_prompt:
            return True
    
    return False

**Key Improvements Made:**

1. **Modern Gradio 6 Syntax**:
   - Used `footer_links` in `gr.Blocks()` instead of deprecated `show_api`
   - Used `api_visibility="public"` in event listeners
   - Clean, minimal component structure

2. **Mobile-First Design**:
   - Responsive layout with proper scaling
   - Optimized gallery height for mobile screens
   - Clean spacing and visual hierarchy

3. **Minimal Components**:
   - Single prompt input with clear labeling
   - Settings hidden in collapsible accordion
   - Primary button with clear call-to-action

4. **Better UX**:
   - Intuitive layout with main actions prominent
   - Examples clearly separated and accessible
   - Clean typography and spacing

5. **Production Ready**:
   - Proper error handling
   - Clear visual feedback
   - Consistent theming with `gr.themes.Soft()`
   - Mobile-optimized CSS

The redesigned application maintains all functionality while providing a much cleaner, more professional interface that works beautifully on both desktop and mobile devices.