(WIP - currently not working correctly!) reference: https://stackoverflow.com/a/9758019

normalize_names_of_files_and_folders_in_some_fucking_directory.py

# Normalizes filenames and folders !!

# because fuck you HFS+
# I hate you so
# so much

# Very WIP and probably abandoned because
# I dont feel like fixing the bugs and i'll just 
# consider all this as wasted time and redo my
# rsync command with `--iconv=utf-8-mac,utf-8`,
# bite the fucking bullet and simply re-wait all
# those lost precious hours of slow transfer rate
# god damnit

from rich import print

from pathlib import Path
import unicodedata

base_path = Path(".")
dest_path = base_path / "normalized"


def normalize_name(p: Path) -> Path:
    return Path(unicodedata.ucd_3_2_0.normalize("NFC", str(p)))


def get_dirs(base_path: Path = base_path) -> list[Path]:
    return [p for p in base_path.rglob("*") if p.is_dir()]


def get_files(base_path: Path = base_path) -> list[Path]:
    return [p for p in base_path.rglob("*") if p.is_file()]


def rename_dirs(
    dry_run: bool = True, dest_path: Path = dest_path, base_path: Path = base_path
) -> list[Path]:
    dest_paths = [
        dest_path.absolute() / normalize_name(p)
        for p in get_dirs(base_path)
        if p != dest_path
    ]

    if dry_run:
        return dest_paths

    return [p.parent.mkdir(parents=True, exist_ok=True) for p in dest_paths]


def rename_files(
    dry_run: bool = True, dest_path: Path = dest_path, base_path: Path = base_path
) -> list[Path]:
    dest_paths = [
        (p, dest_path.absolute() / normalize_name(p))
        for p in get_files(base_path)
        if p.parent != dest_path
    ]

    if dry_run:
        return list(map(lambda p: p[1], dest_paths))

    return [source_path.rename(dest_path) for source_path, dest_path in dest_paths]


def normalize(
    dry_run: bool = True, base_path: Path = base_path, dest_path: Path = dest_path
) -> list[Path]:
    return {
        "dirs": rename_dirs(dry_run, dest_path, base_path),
        "files": rename_files(dry_run, dest_path, base_path),
    }


print(normalize())