Recover original files from `.bak` files. Search in directories recursively.

recover_from_bak_files.py

# A progra conducted file conversion and made a copy of the original file as .bak file
# .bak files have name convention of <original_file_name>.<timestamp>.bak
# This simple program help to reverse such conversion.

import os
import click

def get_sub_files_and_dirs(parent_dir):
  files_in_dir = []
  dirs_in_dir = []
  entities_in_dir = os.listdir(parent_dir)
  for e in entities_in_dir:
      entity_path = os.path.join(parent_dir, e)
      if os.path.isdir(entity_path):
          dirs_in_dir.append(e)
      elif os.path.isfile(entity_path):
          files_in_dir.append(e)
      else:
          print(f'Unknown entity type: {entity_path}')
  return files_in_dir, dirs_in_dir

          
def check_and_recover_file(file_name, parent_dir, files_in_parent_dir):
  if not file_name.endswith('.bak'):
    return
  bak_file_path = os.path.join(parent_dir, file_name)
  ori_file_name = '.'.join(file_name.split('.')[:-2])
  ori_file_path = os.path.join(parent_dir, ori_file_name)
  if ori_file_name in files_in_parent_dir:
    os.remove(ori_file_path)
    print(f'Removed converted file: {ori_file_path}')
  else:
    print(f'[Warning] cannot find converted file of :{bak_file_path}')
  os.rename(bak_file_path, ori_file_path)
  print(f'Recovered converted file: {ori_file_path}')


def entity_recursor(entity_name, parent_dir, files_in_parent_dir):
  entity_path = os.path.join(parent_dir, entity_name)
  if os.path.isfile(entity_path):
    if not files_in_parent_dir:
      files_in_parent_dir = os.listdir(parent_dir)
    check_and_recover_file(entity_name, parent_dir, files_in_parent_dir)
    return
  elif os.path.isdir(entity_path):
    sub_entities = os.listdir(entity_path)
    sub_files = [se for se in sub_entities if os.path.isfile(os.path.join(entity_path, se))]
    for e in sub_entities:
      entity_recursor(e, entity_path, sub_files)
    return
  else:
    print(f'Unsupported entity path: {entity_path}')


@click.command()
@click.argument('entity_path', type=click.Path(exists=True))
def main(entity_path):
  entity_parent, entity_name = os.path.split(entity_path)
  entity_recursor(entity_name, entity_parent, None)
     
    
if __name__ == "__main__":
  main()