imagen generation added

requirements.txt

@@ -1,2 +1,3 @@
 openai==2.8.0
-gradio[mcp]>=5.49.0
+gradio[mcp]>=5.49.0
+huggingface-hub>=0.20.0

src/learnbee/constants.py

@@ -39,15 +39,8 @@ LANGUAGES = [
     "French",
     "German",
     "Italian",
-    "Portuguese",
     "Chinese",
     "Japanese",
-    "Korean",
-    "Arabic",
-    "Russian",
-    "Dutch",
-    "Polish",
-    "Turkish",
     "Hindi"
 ]
 

src/learnbee/image_generator.py

@@ -0,0 +1,121 @@
+"""Image generation module using Hugging Face Inference API."""
+
+import os
+from typing import Optional
+from dotenv import load_dotenv
+from huggingface_hub import InferenceClient
+
+load_dotenv()
+
+
+class ImageGenerator:
+    """Generate images using Hugging Face Inference API."""
+    
+    def __init__(self, model: str = "black-forest-labs/FLUX.1-schnell"):
+        """
+        Initialize the image generator.
+        
+        Args:
+            model: Hugging Face model ID for image generation
+        """
+        self.api_token = os.getenv("HF_TOKEN")
+        self.model = model
+        
+        if not self.api_token:
+            raise ValueError("HF_TOKEN not found in environment variables")
+        
+        # Initialize Hugging Face Inference Client
+        self.client = InferenceClient(token=self.api_token)
+    
+    def generate_image(self, prompt: str) -> Optional[bytes]:
+        """
+        Generate an image from a text prompt.
+        
+        Args:
+            prompt: Text description of the image to generate
+            
+        Returns:
+            Image bytes if successful, None otherwise
+        """
+        # Enhance prompt for child-friendly, educational content
+        enhanced_prompt = self._enhance_prompt_for_children(prompt)
+        
+        try:
+            # Use the text_to_image method from InferenceClient
+            image = self.client.text_to_image(
+                enhanced_prompt,
+                model=self.model
+            )
+            
+            # Convert PIL Image to bytes
+            from io import BytesIO
+            img_byte_arr = BytesIO()
+            image.save(img_byte_arr, format='PNG')
+            img_byte_arr.seek(0)
+            
+            return img_byte_arr.read()
+                
+        except Exception as e:
+            print(f"Error generating image: {str(e)}")
+            return None
+    
+    def _enhance_prompt_for_children(self, prompt: str) -> str:
+        """
+        Enhance the prompt to ensure child-friendly, educational images.
+        
+        Args:
+            prompt: Original prompt
+            
+        Returns:
+            Enhanced prompt
+        """
+        # Add style modifiers for child-friendly content
+        enhancements = [
+            "child-friendly",
+            "colorful",
+            "educational illustration",
+            "cartoon style",
+            "bright and cheerful"
+        ]
+        
+        # Combine original prompt with enhancements
+        enhanced = f"{prompt}, {', '.join(enhancements)}"
+        
+        return enhanced
+    
+    def detect_image_request(self, message: str) -> Optional[str]:
+        """
+        Detect if a message contains an image request and extract the subject.
+        
+        Args:
+            message: User's message
+            
+        Returns:
+            Subject to generate image for, or None if no request detected
+        """
+        message_lower = message.lower()
+        
+        # Keywords that indicate image request
+        image_keywords = [
+            "show me", "muéstrame", "muestra",
+            "draw", "dibuja", "dibujar",
+            "picture of", "imagen de", "foto de",
+            "what does", "cómo es", "como es",
+            "i want to see", "quiero ver",
+            "can you show", "puedes mostrar"
+        ]
+        
+        # Check if message contains image request keywords
+        for keyword in image_keywords:
+            if keyword in message_lower:
+                # Extract subject (simplified - could be improved with NLP)
+                # Remove the keyword and get the remaining text
+                subject = message_lower.replace(keyword, "").strip()
+                # Remove common words
+                subject = subject.replace("a ", "").replace("an ", "").replace("the ", "")
+                subject = subject.replace("un ", "").replace("una ", "").replace("el ", "").replace("la ", "")
+                
+                if subject:
+                    return subject
+        
+        return None

src/learnbee/prompts.py

@@ -76,6 +76,13 @@ def generate_tutor_system_prompt(
         "- Be warm, enthusiastic, and patient. Show excitement about problem-solving!\n"
         "- Use the child's name when possible (refer to them as 'you' or 'little learner').\n\n"
         
+        "IMAGE GENERATION CAPABILITY:\n"
+        "- You can suggest the child ask to see images of things they're learning about!\n"
+        "- When discussing visual concepts (animals, objects, places), encourage them: 'Would you like to see what a [subject] looks like?'\n"
+        "- The child can request images by saying things like 'show me a dinosaur' or 'I want to see a rocket'\n"
+        "- Images help visual learners understand concepts better - use this feature to enhance learning!\n"
+        "- After an image is shown, ask questions about what they see in the image.\n\n"
+        
         "TEACHING STRATEGIES BY DIFFICULTY LEVEL:\n"
         f"- {difficulty_level.upper()} level:\n"
         f"{difficulty_instruction}\n"

src/learnbee/tutor_handlers.py

@@ -11,6 +11,9 @@ from learnbee.prompts import generate_tutor_system_prompt
 
 from learnbee.session_state import SessionState
 from learnbee.gamification import GamificationTracker
+from learnbee.image_generator import ImageGenerator
+import os
+import time
 
 
 def load_lesson_content(lesson_name, selected_tutor, selected_language, progress=gr.Progress()):
@@ -239,7 +242,7 @@ def custom_respond(
     session_state, gamification_tracker
 ):
     """
-    Custom respond function with educational system prompt, adaptive personalization, and gamification.
+    Custom respond function with educational system prompt, adaptive personalization, gamification, and image generation.
     
     Args:
         message: User's message
@@ -260,6 +263,43 @@ def custom_respond(
 
     if not lesson_content:
         lesson_content = get_lesson_content(lesson_name, LESSON_CONTENT_MAX_LENGTH)
+    
+    # Check if message contains an image request
+    image_gen = ImageGenerator()
+    image_subject = image_gen.detect_image_request(message)
+    
+    if image_subject:
+        # Generate image
+        yield "🎨 Generating image of {}... Please wait!".format(image_subject), gamification_tracker.get_progress_html()
+        
+        image_bytes = image_gen.generate_image(image_subject)
+        
+        if image_bytes:
+            # Save image to generated_images directory with absolute path
+            images_dir = os.path.abspath("./generated_images")
+            os.makedirs(images_dir, exist_ok=True)
+            timestamp = int(time.time())
+            image_filename = f"image_{timestamp}.png"
+            image_path = os.path.join(images_dir, image_filename)
+            
+            with open(image_path, "wb") as f:
+                f.write(image_bytes)
+            
+            # Award star for creative request
+            gamification_tracker.award_star("Creative image request!")
+            session_state.total_messages += 1
+            
+            # Return message with gr.Image component for proper display
+            # First yield the text message
+            yield f"Here's your image of {image_subject}! 🎨", gamification_tracker.get_progress_html()
+            
+            # Then yield the image as a separate message using gr.Image
+            import gradio as gr
+            yield gr.Image(value=image_path, label=image_subject, show_label=False, height=400), gamification_tracker.get_progress_html()
+            return
+        else:
+            yield f"Sorry, I couldn't generate the image right now. Let's continue learning! 😊", gamification_tracker.get_progress_html()
+            return
 
     # Analyze the child's message for adaptive personalization
     message_analysis = session_state.analyze_message(message)
@@ -291,12 +331,25 @@ def custom_respond(
         adaptive_context=adaptive_context
     )
 
+    # Filter history to remove gr.Image components (they can't be sent to LLM)
+    # Keep only text messages for LLM context
+    filtered_history = []
+    for msg in history:
+        if isinstance(msg, dict):
+            # Check if content is a string (text message)
+            if isinstance(msg.get("content"), str):
+                filtered_history.append(msg)
+            # Skip messages with gr.Image or other components
+        elif isinstance(msg, (list, tuple)) and len(msg) == 2:
+            # Old tuple format - keep it
+            filtered_history.append(msg)
+
     # Call the respond method with educational system prompt
     call_llm = LLMCall()
     response_text = ""
     for response in call_llm.respond(
         message, 
-        history, 
+        filtered_history,  # Use filtered history without images
         system_prompt=system_prompt,
         tutor_name=selected_tutor,
         difficulty_level=difficulty_level