Find duplicate files recursively in all subdirectories starting in the directory passed as an argument (or in the current working directory) using checksums (SHA-256)

find_duplicates.py

#!/usr/bin/env python3

import sys
import os
import hashlib


class Directory:
    def __init__(self, path: str) -> None:
        self.path = path
        entries = self.get_directory_entries(path)
        self.directories = self.get_directories(entries)
        self.files = self.get_files(entries)

    def get_directory_entries(self, directory_path: str) -> list[str]:
        return list(map(lambda entry: os.path.join(directory_path, entry),
                        filter(lambda entry: not entry.startswith('.'), os.listdir(directory_path))))

    def get_directories(self, entries: list[str]) -> list['Directory']:
        return list(map(lambda directory: Directory(directory),
                        filter(lambda entry: os.path.isdir(entry), entries)))

    def get_files(self, entries: list[str]) -> list['File']:
        return list(map(lambda file: File(file),
                        filter(lambda entry: not os.path.isdir(entry), entries)))


class File:
    checksums: dict[str, list[str]] = {}

    def __init__(self, path: str) -> None:
        self.path = path
        self.checksum = self.get_file_checksum(path)

        if self.checksum not in self.checksums:
            self.checksums[self.checksum] = [self.path]
        else:
            self.checksums[self.checksum].append(self.path)

    def get_file_checksum(self, file_path: str) -> str:
        with open(file_path, 'rb') as file:
            digest = hashlib.file_digest(file, 'sha256')
        return digest.hexdigest()


def print_directories(directory: Directory, directory_path: str, subdirectory=False) -> None:
    if not subdirectory:
        print('Directory:', '\'' + directory.path + '\'')
    else:
        print('Subdirectory:', '\'.' + directory.path.removeprefix(directory_path) + '\'')
    print_files(list(map(lambda file: os.path.basename(file.path), directory.files)))
    print()
    for subdirectory in directory.directories:
        print_directories(subdirectory, directory_path, True)


def print_files(files: list[str]) -> None:
    if len(files) > 0:
        print('Files: [',
              ',\n'.join(map(lambda file: '  \'' + file + '\'', files)),
              ']', sep='\n')
    else:
        print('No files found in this directory!')


def print_identical_files(directory_path: str) -> None:
    identical_files = dict(filter(lambda checksum: len(checksum[1]) > 1, File.checksums.items()))
    if len(identical_files) > 0:
        print('\nThe identical files are (found using SHA-256):')
        for checksum in identical_files:
            print('\nChecksum:', '\'' + checksum + '\'')
            print_files(list(map(lambda file: '.' + file.removeprefix(directory_path),
                                 File.checksums[checksum])))
    else:
        print('\nNo identical files found (using SHA-256)!')


def main() -> int:
    directory_path: str
    if len(sys.argv) <= 1:
        print('No directory was passed as an argument!')
        print('The current working directory is used instead …\n')
        directory_path = os.getcwd()
    else:
        print('A directory was passed as an argument …\n')
        arguments = sys.argv
        arguments.pop(0)
        directory_path = os.path.abspath(' '.join(arguments))

    try:
        directory = Directory(directory_path)

    except (FileNotFoundError, NotADirectoryError) as error:
        print('But an error occurred:', end=' ')
        if type(error) == FileNotFoundError:
            print('No such directory!')
        elif type(error) == NotADirectoryError:
            print('Not a directory!')
        print('Argument:', '\'' + directory_path + '\'')
        return 1

    else:
        print_directories(directory, directory_path)
        print_identical_files(directory_path)
        return 0


if __name__ == '__main__':
    sys.exit(main())