Update app.py

app.py

@@ -1,6 +1,7 @@
 import gradio as gr
 import torch
 from transformers import pipeline
+import re
 
 # Global variable to store the model
 pipe = None
@@ -19,17 +20,157 @@ def load_model():
         print("✅ Model loaded successfully!")
     return pipe
 
+def detect_arabizi(text):
+    """
+    Detect if input text is written in Arabizi (Latin script with numbers)
+    Returns True if Arabizi is detected
+    """
+    if not text or len(text.strip()) < 2:
+        return False
+    
+    # Remove spaces and convert to lowercase for analysis
+    clean_text = text.lower().replace(" ", "")
+    
+    # Check for Arabic script - if present, it's NOT Arabizi
+    arabic_pattern = r'[\u0600-\u06FF\u0750-\u077F\u08A0-\u08FF\uFB50-\uFDFF\uFE70-\uFEFF]'
+    if re.search(arabic_pattern, text):
+        return False
+    
+    # Arabizi indicators
+    arabizi_numbers = ['2', '3', '7', '9']  # Common Arabic letter substitutions
+    arabizi_patterns = [
+        'wach', 'wash', 'ach', 'achno', 'chno', 'shno',  # What
+        'kif', 'kifash', 'ki', 'kayf',  # How
+        'feen', 'fin', 'fen',  # Where  
+        'imta', 'meta', 'waqt',  # When
+        '3la', '3ala', 'ala',  # On/about
+        'hna', '7na', 'ahna',  # We/us
+        'nta', 'nti', 'ntuma',  # You
+        'howa', 'hiya', 'huma',  # He/she/they
+        'ma3', 'maa3', 'maak', 'maaki',  # With
+        'had', 'hadchi', 'hada', 'hadi',  # This
+        'bghit', 'bghiti', 'bgha',  # Want
+        'galt', 'galti', 'gal',  # Said
+        'rah', 'raha', 'rahi',  # Going
+        'kan', 'kanu', 'kana',  # Was/were
+        'ghadi', 'ghad', 'gha',  # Will/going to
+        'daba', 'dak', 'dakchi',  # Now/that
+        'bzf', 'bzzaf', 'bezzaf',  # A lot
+        'chway', 'chwiya', 'shwiya',  # A little
+        'khoya', 'khuya', 'akhi',  # Brother
+        'khti', 'khtiya', 'ukhti',  # Sister
+        'allah', 'llah', 'rabi',  # God
+        'inchallah', 'insha allah',  # God willing
+        'hamdulillah', 'alhamdulillah',  # Praise God
+        'salam', 'salamu aleikum',  # Peace
+        'baraka', 'barakallahu',  # Blessing
+        'yallah', 'yalla', 'hya'  # Come on/let's go
+    ]
+    
+    # Count Latin letters
+    latin_letters = sum(1 for c in clean_text if c.isalpha() and ord(c) < 128)
+    
+    # Count Arabizi number substitutions
+    arabizi_number_count = sum(1 for num in arabizi_numbers if num in clean_text)
+    
+    # Count Arabizi word patterns
+    arabizi_word_count = sum(1 for pattern in arabizi_patterns if pattern in clean_text)
+    
+    # Decision logic
+    total_chars = len(clean_text)
+    
+    # Strong indicators
+    if arabizi_number_count >= 2:  # Multiple number substitutions
+        return True
+    
+    if arabizi_word_count >= 2:  # Multiple Arabizi words
+        return True
+    
+    # Medium indicators 
+    if arabizi_number_count >= 1 and latin_letters > total_chars * 0.7:
+        return True
+    
+    if arabizi_word_count >= 1 and latin_letters > total_chars * 0.8:
+        return True
+    
+    # Weak but possible indicators
+    if latin_letters > total_chars * 0.9 and total_chars > 10:
+        # Mostly Latin letters in longer text - could be Arabizi
+        if arabizi_number_count >= 1 or arabizi_word_count >= 1:
+            return True
+    
+    return False
+
+def determine_response_language(user_input):
+    """
+    Determine what language/script the response should be in
+    Returns: 'arabizi', 'arabic', or 'english'
+    """
+    if detect_arabizi(user_input):
+        return 'arabizi'
+    
+    # Check for Arabic script
+    arabic_pattern = r'[\u0600-\u06FF\u0750-\u077F\u08A0-\u08FF\uFB50-\uFDFF\uFE70-\uFEFF]'
+    if re.search(arabic_pattern, user_input):
+        return 'arabic'
+    
+    # Default to English for Latin-only text without Arabizi indicators
+    return 'english'
+
+def create_system_prompt(response_lang):
+    """Create appropriate system prompt based on desired response language"""
+    
+    if response_lang == 'arabizi':
+        return """You are Atlas-Chat, an AI assistant specialized in Moroccan Arabic (Darija). 
+
+CRITICAL INSTRUCTION: The user has written in Arabizi (Latin script), so you MUST respond ONLY in Arabizi using Latin letters and numbers.
+
+ARABIZI RULES YOU MUST FOLLOW:
+- Use ONLY Latin letters (a-z) and numbers for Arabic sounds
+- Use these number substitutions: 3=ع, 7=ح, 9=ق, 2=ء, 5=خ, 6=ط, 8=غ
+- Write naturally in Moroccan Darija but with Latin script
+- Examples: "ana" (أنا), "hna" (حنا), "3la" (على), "7na" (حنا), "wach" (واش)
+- Do NOT use any Arabic script characters
+- Do NOT switch to English unless the user specifically asks for translation
+
+Respond naturally in Arabizi about Moroccan culture, language, and general topics."""
+
+    elif response_lang == 'arabic':
+        return """You are Atlas-Chat, an AI assistant specialized in Moroccan Arabic (Darija). Respond in Arabic script (Darija) as this is what the user is using. Be helpful and culturally aware about Morocco and its traditions."""
+    
+    else:  # English
+        return """You are Atlas-Chat, an AI assistant specialized in Moroccan Arabic (Darija) but also fluent in English. The user has written in English, so respond in English while being knowledgeable about Moroccan culture and language."""
+
 def chat_with_atlas(message, history):
-    """Generate response from Atlas-Chat model"""
+    """Generate response from Atlas-Chat model with language detection"""
     if not message.strip():
-        return "مرحبا! أهلا وسهلا. Please enter a message!"
+        return "مرحبا! أهلا وسهلا. Please enter a message! / Ahlan wa sahlan!"
     
     try:
         # Load model if not already loaded
         model = load_model()
         
-        # Prepare the message
-        messages = [{"role": "user", "content": message}]
+        # Determine response language
+        response_lang = determine_response_language(message)
+        
+        # Create appropriate system prompt
+        system_prompt = create_system_prompt(response_lang)
+        
+        # Prepare messages with system context
+        if response_lang == 'arabizi':
+            # Extra emphasis for Arabizi responses
+            enhanced_message = f"""System: {system_prompt}
+
+User message (in Arabizi): {message}
+
+Remember: Respond ONLY in Arabizi (Latin letters + numbers). Do not use Arabic script."""
+            
+            messages = [{"role": "user", "content": enhanced_message}]
+        else:
+            messages = [
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": message}
+            ]
         
         # Generate response
         outputs = model(
@@ -42,10 +183,19 @@ def chat_with_atlas(message, history):
         
         # Extract the response
         response = outputs[0]["generated_text"][-1]["content"].strip()
+        
+        # Post-process for Arabizi if needed
+        if response_lang == 'arabizi':
+            # Remove any Arabic script that might have leaked through
+            arabic_pattern = r'[\u0600-\u06FF\u0750-\u077F\u08A0-\u08FF\uFB50-\uFDFF\uFE70-\uFEFF]'
+            if re.search(arabic_pattern, response):
+                # If Arabic script is detected, provide a fallback Arabizi response
+                response = "ana Atlas-Chat, kay3jebni n7der m3ak! chno bghiti t3ref 3la lmaghrib? (I'm Atlas-Chat, I'd love to chat with you! What do you want to know about Morocco?)"
+        
         return response
         
     except Exception as e:
-        return f"عذراً، واجهت خطأ: {str(e)}. جرب مرة أخرى!"
+        return f"عذراً، واجهت خطأ: {str(e)}. جرب مرة أخرى! / Sorry, error occurred. Try again!"
 
 # Create the Gradio interface
 demo = gr.ChatInterface(
@@ -54,18 +204,25 @@ demo = gr.ChatInterface(
     description="""
     **مرحبا بك في أطلس شات!** Welcome to Atlas-Chat! 🇲🇦
     
-    I'm an AI assistant specialized in **Moroccan Arabic (Darija)** and English.
-    Ask me questions about Morocco, culture, or just have a chat!
+    I'm an AI assistant specialized in **Moroccan Arabic (Darija)** with smart language detection:
+    
+    - **Arabic Script (العربية)** → I respond in Arabic
+    - **Arabizi (3arabi bi 7oruf latin)** → I respond in Arabizi  
+    - **English** → I respond in English
     
     **جرب هذه الأسئلة / Try these questions:**
     """,
     examples=[
         "شكون لي صنعك؟",
-        "اشنو هو الطاجين؟", 
+        "shkoun li sna3ek?",
+        "اشنو هو الطاجين؟",
+        "achno howa tajine?", 
         "شنو كيتسمى المنتخب المغربي؟",
+        "chno kaytsma lmontakhab lmaghribi?",
         "What is Morocco famous for?",
         "Tell me about Casablanca",
-        "كيفاش نقدر نتعلم الدارجة؟"
+        "كيفاش نقدر نتعلم الدارجة؟",
+        "kifash n9der nt3elem darija?"
     ],
     cache_examples=False
 )