1. import pandas as pd
  2. import os
  3. import zipfile
  5. def archive_dataset(data, filename, compression='zip'):
  6. """
  7. Archives a pandas DataFrame to a file for backward compatibility.
  9. Args:
  10. data (pd.DataFrame): The DataFrame to archive.
  11. filename (str): The name of the archive file (e.g., 'my_dataset.csv').
  12. compression (str): The compression type ('zip' or None). Defaults to 'zip'.
  13. """
  14. try:
  15. if compression == 'zip':
  16. # Save to a zip file
  17. data.to_csv(filename, index=False, compression='zip')
  18. elif compression is None:
  19. # Save to a csv file
  20. data.to_csv(filename, index=False)
  21. else:
  22. raise ValueError("Invalid compression type. Choose 'zip' or None.")
  24. print(f"Dataset archived to {filename}")
  26. except Exception as e:
  27. print(f"Error archiving dataset: {e}")
  31. if __name__ == '__main__':
  32. # Example Usage
  33. # Create a sample DataFrame
  34. data = {'col1': [1, 2, 3], 'col2': ['a', 'b', 'c']}
  35. df = pd.DataFrame(data)
  37. # Archive to a zip file
  38. archive_dataset(df, 'my_dataset.zip', compression='zip')
  40. # Archive to a csv file
  41. archive_dataset(df, 'my_dataset.csv', compression=None)

Add your comment