Recover LFS file tree from GitLab object directory (hashed path)

main.py

#!/usr/bin/env python3

import sys
import click
import pathlib
import shutil

@click.command()
@click.option('-d', '--delim', default=' - ', help='Delimiter used in the LFS object list')
@click.option('-l', '--level', default=2, type=click.IntRange(0, 31), help='Number of levels in the LFS object tree')
@click.argument("lfs-obj-list", type=click.File('r'))
@click.argument("lfs-obj-dir", type=click.Path(exists=True, file_okay=False))
@click.argument("output-dir", type=click.Path(exists=False, file_okay=False))
def main(delim, level, lfs_obj_list, lfs_obj_dir, output_dir):
    '''
    This script will take a list of LFS objects and a directory containing
    the LFS objects with hashed path, and construct the LFS object tree.

    LFS_OBJ_LIST: generated from `git-lfs ls-files --long`\n
    LFS_OBJ_DIR: directory containing LFS objects with hashed path\n 
    OUTPUT_DIR: directory to store the LFS object tree
    '''
    
    # ensure the output directory exists
    dest = pathlib.Path(output_dir)
    dest.mkdir(parents=True, exist_ok=True)
    lfs_base = pathlib.Path(lfs_obj_dir)

    # Read the LFS object list
    count = 0
    for l in lfs_obj_list.readlines():
        hash, path = l.split(delim)
        hash = hash.strip()
        path = path.strip()
        assert len(hash) == 64, "Invalid hash length"
        # construct the hashed path
        hash_file = lfs_base
        for _ in range(level):
            hash_file = hash_file / hash[:2]
            hash = hash[2:]
        hash_file = hash_file / hash
        if not hash_file.exists():
            print(f"File {path} with hash {hash} does not exist in object directory", file=sys.stderr)
            continue
        # copy the file to the output directory
        dest_file = dest / path
        dest_file.parent.mkdir(parents=True, exist_ok=True)
        shutil.copy2(hash_file, dest_file)
        count += 1

    print(f"Total {count} files copied to {dest}")

if __name__ == '__main__':
    main()