simple_package.py

import os
import tarfile
from pathlib import Path

# Define paths
PROCESSED_DATA_DIR = Path("processed_data")
TARGET_PACKAGE = "processed_data.tar.gz"

def simple_package():
    """
    Package the processed data into a single compressed file without validation.
    This creates a tar.gz file containing the entire processed_data directory.
    """
    print("Starting simple packaging process...")
    
    # Check if folder exists
    if not os.path.exists(PROCESSED_DATA_DIR):
        print(f"ERROR: Directory {PROCESSED_DATA_DIR} not found!")
        return
    
    # Create tar.gz file
    print(f"Creating package file: {TARGET_PACKAGE}")
    with tarfile.open(TARGET_PACKAGE, "w:gz") as tar:
        # Add the entire directory
        tar.add(PROCESSED_DATA_DIR, arcname=PROCESSED_DATA_DIR.name)
    
    # Verify the tarfile was created
    if os.path.exists(TARGET_PACKAGE):
        size_mb = os.path.getsize(TARGET_PACKAGE) / (1024 * 1024)
        print(f"Package created successfully: {TARGET_PACKAGE} ({size_mb:.2f} MB)")
        print("\nInstructions:")
        print(f"1. Upload {TARGET_PACKAGE} to your Hugging Face Space")
        print("2. The app will automatically extract it on startup")
    else:
        print("Failed to create package file")

if __name__ == "__main__":
    simple_package()