download_model.py

#!/usr/bin/env python3
"""
Download a sample GGUF model for testing llama.cpp integration
"""

import os
from huggingface_hub import hf_hub_download
from config import get_recommended_model, MODEL_DOWNLOAD_CONFIG

def download_sample_model():
    """Download a recommended small model for testing"""
    model_info = get_recommended_model()
    
    print(f"📥 Downloading {model_info['name']}...")
    print(f"   Repository: {model_info['repo_id']}")
    print(f"   File: {model_info['filename']}")
    print(f"   Size: {model_info['size']}")
    print(f"   Description: {model_info['description']}")
    
    try:
        # Create models directory if it doesn't exist
        os.makedirs(MODEL_DOWNLOAD_CONFIG['cache_dir'], exist_ok=True)
        
        # Download the model
        model_path = hf_hub_download(
            repo_id=model_info['repo_id'],
            filename=model_info['filename'],
            cache_dir=MODEL_DOWNLOAD_CONFIG['cache_dir'],
            resume_download=MODEL_DOWNLOAD_CONFIG['resume_download'],
            token=MODEL_DOWNLOAD_CONFIG['use_auth_token']
        )
        
        print(f"✅ Model downloaded successfully!")
        print(f"   Path: {model_path}")
        
        # Create a symlink in the models directory for easy access
        symlink_path = os.path.join(MODEL_DOWNLOAD_CONFIG['cache_dir'], "model.gguf")
        if os.path.exists(symlink_path):
            os.remove(symlink_path)
        
        try:
            os.symlink(model_path, symlink_path)
            print(f"   Symlink created: {symlink_path}")
        except OSError:
            # Symlinks might not work on all systems, just copy the path
            print(f"   Use this path in your code: {model_path}")
        
        return model_path
        
    except Exception as e:
        print(f"❌ Error downloading model: {e}")
        print("💡 You can manually download a GGUF model and place it in ./models/")
        return None

def list_available_models():
    """List models available in the models directory"""
    models_dir = MODEL_DOWNLOAD_CONFIG['cache_dir']
    
    if not os.path.exists(models_dir):
        print(f"📁 Models directory doesn't exist: {models_dir}")
        return []
    
    model_files = []
    for file in os.listdir(models_dir):
        if file.endswith('.gguf') or file.endswith('.ggml'):
            file_path = os.path.join(models_dir, file)
            file_size = os.path.getsize(file_path)
            model_files.append({
                'name': file,
                'path': file_path,
                'size_mb': file_size / (1024 * 1024)
            })
    
    if model_files:
        print("📋 Available models:")
        for model in model_files:
            print(f"   - {model['name']} ({model['size_mb']:.1f} MB)")
    else:
        print("📭 No GGUF/GGML models found in models directory")
    
    return model_files

if __name__ == "__main__":
    print("🤖 Model Download Utility for llama.cpp")
    print("=" * 50)
    
    # List existing models
    print("\n🔍 Checking for existing models...")
    existing_models = list_available_models()
    
    if not existing_models:
        print("\n📥 No models found. Downloading sample model...")
        download_sample_model()
    else:
        print(f"\n✅ Found {len(existing_models)} existing model(s)")
        
        # Ask if user wants to download another model
        print("\n❓ Download sample model anyway? (y/n): ", end="")
        try:
            response = input().lower().strip()
            if response in ['y', 'yes']:
                download_sample_model()
            else:
                print("👍 Using existing models")
        except (EOFError, KeyboardInterrupt):
            print("\n👍 Using existing models")