63 lines
2.3 KiB
Python

import re
from collections.abc import Sequence
from pathlib import Path
from tqdm import tqdm
from karaokatalog.instructions.MoveInstruction import MoveInstruction
from karaokatalog.Song import Song
FORBIDDEN_CHARACTERS = re.compile(r'[\/<>:"\\|?*]')
def _get_canonical_song_dir(song: Song, variant: int = 0) -> Path:
"""
Get the (relative) canonical directory in which this song should be located, that is
the directory f"{song.artist}/{song.title}".
Illegal characters are replaced. If variant > 0, f" ({variant})" is appended to the result.
"""
variant_suffix = f" ({variant})" if variant > 0 else ""
return Path(FORBIDDEN_CHARACTERS.sub("", song.artist)) / Path(
FORBIDDEN_CHARACTERS.sub("", song.title) + variant_suffix
)
def move(songs: Sequence[Song], base_dir: Path) -> Sequence[MoveInstruction]:
"""
Create move instructions to move every song into the proper song directory
within the given base_dir.
"""
song_dir_strs_lower = set(
str(song.dir.relative_to(base_dir)).lower() for song in songs
)
move_instructions = []
for song in tqdm(songs, unit=" songs"):
absolute_song_dir = song.dir
song_dir = absolute_song_dir.relative_to(base_dir)
song_dir_str_lower = str(song_dir).lower()
canonical_song_dir = _get_canonical_song_dir(song)
if song_dir_str_lower not in song_dir_strs_lower:
# A move instruction has already been generated for the dir this song is in
# (which is possible, because some dirs contain multiple (variants of) songs)
continue
if song_dir != canonical_song_dir:
# song_dir is not a good name, we want to replace it, so this path will soon be free
song_dir_strs_lower.remove(song_dir_str_lower)
# Find a canonical song dir variant that does not yet exist
variant = 1
while str(canonical_song_dir).lower() in song_dir_strs_lower:
canonical_song_dir = _get_canonical_song_dir(song, variant)
variant += 1
song_dir_strs_lower.add(str(canonical_song_dir).lower())
move_instructions.append(
MoveInstruction(absolute_song_dir, base_dir / canonical_song_dir)
)
return tuple(move_instructions)