Python script that flattens a directory without overwriting files that share a name. Takes absolute path to directory as input.

flatten_directory.py

#!/usr/local/bin/python3

# Script that flattens a directory without overwriting files that share a name.
# Takes absolute path to directory as input.
#
# All files in any subdirectory of the user-provided directory, regardless of
# depth, are moved to the user-provided directory. When filenames clash,
# special care is taken by adding a counter to the filename (e.g., '_001').
#
# Linux only. Tested only on Python 3.7. Minimum version is 3.6.
#
# This has NOT been tested extensively. Use at your own risk.

import pathlib
import sys


def flatten_directory(path):
    if not _is_valid_path(path):
        return 1

    _rename_files(path)
    return 0


def _is_valid_path(path):
    if not isinstance(path, pathlib.Path):
        print('"path" argument must be instance of "pathlib.Path"')
        return False

    if not path.is_absolute():
        print('Provided path must be an absolute path')
        return False

    if not path.is_dir():
        print('Provided path must be a directory')
        return False

    return True


def _top_dir_files(top_dir):
    return {path for path in top_dir.glob('*') if path.is_file()}


def _nested_dir_files(top_dir):
    return {path for path in top_dir.glob('**/*')
            if path.is_file() and path.parent != top_dir}


def _filename_paths_map(path):
    top_dir_files = _top_dir_files(path)
    nested_dir_files = _nested_dir_files(path)

    filename_paths_map = dict()
    for file in top_dir_files:
        filename_paths_map[file.name] = [file]

    for file in nested_dir_files:
        if file.name in filename_paths_map:
            filename_paths_map[file.name].append(file)
        else:
            filename_paths_map[file.name] = [file]

    return filename_paths_map


# Returns a suffixless filename.
# 'myfile.tar.gz' => 'myfile'
def _suffixless_name(path):
    return path.name.split('.')[0]


# Returns the suffixes combined into a string.
# 'myfile.tar.gz' => '.tar.gz'
def _suffix(path):
    return ''.join(path.suffixes)


# Returns a zero-padded counter suffix.
def _counter_suffix(current, max):
    return '_' + str(current).zfill(len(str(max)))


def _rename_files(top_dir):
    filename_paths_map = _filename_paths_map(top_dir)

    for _, paths in filename_paths_map.items():
        # Filename in top dir does not clash with any other filename.
        # Do nothing.
        if len(paths) == 1 and paths[0].parent == top_dir:
            print(f'{paths[0]} => No action necessary')
            continue

        # Filename in nested dir does not clash with any other filename.
        # Move the file to the top directory.
        if len(paths) == 1 and paths[0].parent != top_dir:
            new_name = top_dir / paths[0].name
            print(f'{paths[0]} => {new_name}')
            paths[0].rename(new_name)
            continue

        # Since we are renaming the top directory file _and_ any files with
        # the same file in any subdirectory, we can just loop over all files
        # and move them to the top directory with a running counter added to
        # their filename to ensure uniqueness.
        for idx, path in enumerate(paths, start=1):
            base_name = _suffixless_name(path)
            suffix = _suffix(path)
            counter = _counter_suffix(idx, len(paths))
            new_name = top_dir / (base_name + counter + suffix)
            print(f'{path} => {new_name}')
            path.rename(new_name)


if __name__ == '__main__':
    if len(sys.argv) != 2:
        print('Need exactly one argument: the absolute path to the directory')
        sys.exit(1)

    path = pathlib.Path(sys.argv[1])
    sys.exit(flatten_directory(path))