quick_start.py

"""

Quick Start Script for Hugging Face Integration

Helps you upload/download models easily

"""

import sys
from pathlib import Path


def print_banner(text):
    """Print formatted banner"""
    print("\n" + "="*60)
    print(text.center(60))
    print("="*60 + "\n")


def check_model_exists():
    """Check if trained model exists"""
    model_path = Path("./models/bytedream")
    
    if not model_path.exists():
        print("❌ Model directory not found!")
        print("\nPlease train the model first:")
        print("  python train.py")
        print("\nOr download from Hugging Face:")
        print("  python infer.py --hf_repo username/repo --prompt 'test'")
        return False
    
    # Check for weights
    unet_weights = model_path / "unet" / "pytorch_model.bin"
    vae_weights = model_path / "vae" / "pytorch_model.bin"
    
    if not (unet_weights.exists() or (model_path / "pytorch_model.bin").exists()):
        print("⚠ Model directory exists but no weights found!")
        print("Please train the model first.")
        return False
    
    return True


def upload_to_hf():
    """Upload model to Hugging Face"""
    print_banner("UPLOAD TO HUGGING FACE HUB")
    
    # Check model exists
    if not check_model_exists():
        return
    
    # Get token
    token = input("Enter your Hugging Face token (hf_...): ").strip()
    if not token:
        print("❌ Token is required!")
        return
    
    # Get repo ID
    repo_id = input("Enter repository ID (e.g., username/ByteDream): ").strip()
    if not repo_id:
        print("❌ Repository ID is required!")
        return
    
    print(f"\n📤 Uploading to {repo_id}...")
    
    try:
        from bytedream.generator import ByteDreamGenerator
        
        # Load model
        print("\nLoading model...")
        generator = ByteDreamGenerator(
            model_path="./models/bytedream",
            config_path="config.yaml",
            device="cpu",
        )
        
        # Upload
        generator.push_to_hub(
            repo_id=repo_id,
            token=token,
            private=False,
            commit_message="Upload Byte Dream model",
        )
        
        print("\n✅ SUCCESS!")
        print(f"\n📦 Your model is available at:")
        print(f"https://huggingface.co/{repo_id}")
        print(f"\nTo use this model:")
        print(f"  python infer.py --prompt 'your prompt' --hf_repo '{repo_id}'")
        print("="*60)
        
    except Exception as e:
        print(f"\n❌ Error: {e}")
        import traceback
        traceback.print_exc()


def download_from_hf():
    """Download model from Hugging Face"""
    print_banner("DOWNLOAD FROM HUGGING FACE HUB")
    
    # Get repo ID
    repo_id = input("Enter repository ID (e.g., username/ByteDream): ").strip()
    if not repo_id:
        print("❌ Repository ID is required!")
        return
    
    print(f"\n📥 Downloading from {repo_id}...")
    
    try:
        from bytedream.generator import ByteDreamGenerator
        
        # Load from HF
        generator = ByteDreamGenerator(
            hf_repo_id=repo_id,
            config_path="config.yaml",
            device="cpu",
        )
        
        print("\n✅ Model loaded successfully!")
        
        # Test generation
        test = input("\nGenerate test image? (y/n): ").strip().lower()
        if test == 'y':
            print("\nGenerating test image...")
            image = generator.generate(
                prompt="test pattern, simple colors",
                width=256,
                height=256,
                num_inference_steps=10,
            )
            
            output = "test_output.png"
            image.save(output)
            print(f"✓ Test image saved to: {output}")
        
        print("\nTo generate images:")
        print(f"  python infer.py --prompt 'your prompt' --hf_repo '{repo_id}'")
        print(f"  HF_REPO_ID={repo_id} python app.py")
        print("="*60)
        
    except Exception as e:
        print(f"\n❌ Error: {e}")
        import traceback
        traceback.print_exc()


def test_local_model():
    """Test local model"""
    print_banner("TEST LOCAL MODEL")
    
    if not check_model_exists():
        return
    
    print("Loading local model...")
    
    try:
        from bytedream.generator import ByteDreamGenerator
        
        generator = ByteDreamGenerator(
            model_path="./models/bytedream",
            config_path="config.yaml",
            device="cpu",
        )
        
        print("\n✅ Model loaded successfully!")
        
        # Generate test image
        print("\nGenerating test image...")
        image = generator.generate(
            prompt="test pattern, simple colors",
            width=256,
            height=256,
            num_inference_steps=10,
        )
        
        output = "test_output.png"
        image.save(output)
        print(f"✓ Test image saved to: {output}")
        
        print("\nModel ready for upload!")
        print("To upload: python quick_start.py upload")
        print("="*60)
        
    except Exception as e:
        print(f"\n❌ Error: {e}")
        import traceback
        traceback.print_exc()


def main():
    """Main function"""
    print_banner("BYTE DREAM - QUICK START")
    
    print("What would you like to do?")
    print("1. Upload model to Hugging Face")
    print("2. Download model from Hugging Face")
    print("3. Test local model")
    print("4. Exit")
    print()
    
    choice = input("Enter choice (1-4): ").strip()
    
    if choice == "1":
        upload_to_hf()
    elif choice == "2":
        download_from_hf()
    elif choice == "3":
        test_local_model()
    elif choice == "4":
        print("\nGoodbye!")
        return
    else:
        print("❌ Invalid choice!")
        return
    
    print("\nDone!")


if __name__ == "__main__":
    try:
        main()
    except KeyboardInterrupt:
        print("\n\nInterrupted!")
        sys.exit(0)