a huge HUGE rewrite that made it faster and less buggy and awful and dependent on linux and BOY this was overdue

This commit is contained in:
Lynne Megido 2020-10-16 23:30:26 +10:00
parent 5eaa942eab
commit 83458f9f30
Signed by: lynnesbian
GPG Key ID: F0A184B5213D9F90
3 changed files with 136 additions and 142 deletions

4
.gitignore vendored Normal file
View File

@ -0,0 +1,4 @@
!bcao.py
!requirements.txt
!.gitignore
*

272
bcao.py
View File

@ -5,45 +5,56 @@
# input: a .zip from bandcamp
# output: it organises it, adds cover art, puts it in the right place...
# TODO: this is written like a hacky bash script and i hate it
import subprocess, argparse, sys, os, re, base64
import argparse
import base64
import os
import re
import subprocess
import tempfile
import shutil
from zipfile import ZipFile, BadZipFile
import mutagen
from mutagen.flac import Picture
from mutagen import id3
from PIL import Image
def log(message: str, importance: int = 0):
if not args.quiet or importance > 0:
print(message)
def die(message: str, code: int = 1):
log(message)
print(message)
exit(code)
try:
from mutagen.oggvorbis import OggVorbis
from mutagen.mp3 import MP3
from mutagen.flac import FLAC
from mutagen.flac import Picture
from mutagen.aac import AAC
except ImportError:
# TODO: requirements.txt
die("Please install python3-mutagen (pip install mutagen)")
def get_tag(m: mutagen.FileType, tag: str):
if tag == "title":
return sanitise(m['title'][0])
elif tag == "track":
return int(m['tracknumber'][0])
elif tag == "album":
return sanitise(m['album'][0])
else:
# may as well try
return sanitise(m[tag])
try:
# TODO: see if there's a decent zip library for python
subprocess.check_output(["unzip", "--help"])
except:
die("Please install unzip (apt install unzip)")
try:
# TODO: don't use which ffs
subprocess.check_output(["which", "convert"])
subprocess.check_output(["which", "identify"])
except:
die("Please install imagemagick, and ensure convert and identify are in your $PATH (apt install imagemagick")
def sanitise(input: str):
if args.sanitise:
return re.sub(r"[?\\/:|*\"<>]", "_", input)
else:
return input
parser = argparse.ArgumentParser(description="Extracts the given zip file downloaded from Bandcamp and organises it.")
parser.add_argument('zip', help='The zip file to use')
parser.add_argument('-d', '--destination', dest='destination', default='/home/lynne/Music/Music/', help="The directory to organise the music into. Default: /home/lynne/Music/Music/")
parser.add_argument('-q', '--quiet', dest='quiet', action='store_true', help='Disable non-error output and assume default artist name.')
parser.add_argument('-t', '--threshold', dest='threshold', nargs=1, default=300, help="Maximum acceptable cover art file size in kilobytes. Default: 300")
parser.add_argument('-d', '--destination', dest='destination', default='/home/lynne/Music/Music/',
help="The directory to organise the music into. Default: /home/lynne/Music/Music/")
parser.add_argument('-q', '--quiet', dest='quiet', action='store_true',
help='Disable non-error output and assume default artist name.')
parser.add_argument('-u', '--unsanitised', dest='sanitise', action='store_false',
help="Don't replace NTFS-unsafe characters with underscores. Not recommended.")
parser.add_argument('-t', '--threshold', dest='threshold', nargs=1, default=300,
help="Maximum acceptable cover art file size in kilobytes. Default: 300")
args = parser.parse_args()
@ -51,129 +62,106 @@ if not os.path.exists(args.zip):
die(f"Couldn't find {args.zip}.", 2)
log("Extracting...")
zipname = os.path.splitext(os.path.basename(args.zip))[0]
tmp = "/tmp/bcao/{0}".format(zipname) # TODO: use OS specific temp dirs
subprocess.check_output(["rm", "-rf", tmp])
subprocess.check_output(['mkdir', '-p', tmp])
subprocess.check_output(["unzip", args.zip, "-d", tmp])
files = []
songs = []
tmp = tempfile.TemporaryDirectory()
cover = None
song_names = []
for root, dirs, filez in os.walk(tmp):
for file in filez: # for every file
files.append(root + os.sep + file)
with ZipFile(args.zip, 'r') as zip_file:
for file in zip_file.namelist():
if re.match(r"^(.+ - ){2}\d{2,} .+\.(ogg|flac|alac|aiff|wav|mp3|opus|m4a|aac|oga)$", file):
# bandcamp zips contains songs with names formatted like "Album - Artist - 01 Song.mp3"
# for example, "King Crimson - In the Wake of Poseidon - 02 Pictures of a City.ogg"
# this regex should match only on those, and cut out (hopefully) all of the bonus material stuff, which shouldn't
# be added to the music folder (since i sync that to my phone, and "making of" videos are cool, but i don't
# have the space for it).
song_names.append(file)
zip_file.extract(file, tmp)
elif cover is None and re.match(r"cover\.(jpe?g|png)", file):
cover = file
zip_file.extract(file, tmp)
# save the format of the songs (ogg, mp3, etc)
# we'll need this to know what metadata format we should write
song_format = os.path.splitext(song_names[0])[1][1:]
log("Resizing album art to embed in songs...")
with Image.open(os.path.join(tmp, cover)) as image:
temp_cover = os.path.join(tmp, "cover-lq.jpg")
image.save(temp_cover, quality=85, optimize=True)
image_smol = image
# keep shrinking the image by 90% until it's less than {args.threshold} kilobytes
while os.path.getsize(temp_cover) / 1024 > args.threshold:
image_smol = image_smol.resize((round(image_smol.size[0] * 0.9), round(image_smol.size[1] * 0.9)))
image_smol.save(temp_cover, quality=85, optimize=True)
if image_smol.size[0] == 10:
die("Failed to resize image")
# read the image file to get the file's raw data
with open(temp_cover, 'r+b') as cover_file:
data = cover_file.read()
with Image.open(temp_cover) as image:
if song_format == "ogg":
# i hate this
cover = Picture()
cover.data = data
cover.type = mutagen.id3.PictureType.COVER_FRONT
cover.mime = "image/jpeg"
cover.width = image.size[0]
cover.height = image.size[1]
cover.depth = image.bits
cover = base64.b64encode(cover.write()).decode("ascii")
else:
log(f"Format {song_format} is not fully supported - cover images will not be modified", 1)
# TODO: redo everything from here to the cover art resizing
cover = ""
artists = []
album = ""
fileExtensionRegex = re.compile(r"[^.]+$")
probablyASongRegex = re.compile(r"^.+ - .+ - \d{2,} .+\.[^.]+$") # matches "artist - album - 01 track.xyz" but not "some weird bonus thing.mp3"
musicExts = ["ogg", "flac", "alac", "aiff", "wav", "mp3", "opus", "m4a", "aac", "oga"]
bannedCharacters = ["?", "\\", "/", ":", "|", "*", "\"", "<", ">"] # characters disallowed in NTFS filenames
trackCount = 0
album = None
songs = {}
zeroes = min(len(song_names), 2)
for song in song_names:
ext = os.path.splitext(song)[1:]
m = mutagen.File(os.path.join(tmp, song))
# add the song's artist to the list, if it hasn't been seen yet
[artists.append(sanitise(artist)) for artist in m['artist'] if artist not in artists]
songs[song] = f"{str(get_tag(m, 'track')).zfill(zeroes)} {get_tag(m, 'title')}.{song_format}"
album = get_tag(m, "album")
# embed cover art
if song_format == "ogg":
m["metadata_block_picture"] = [cover]
m.save()
log("Processing... please wait.")
# use "01 Song.ogg" instead of "1 Song.ogg". Also works for albums with more than 99 tracks if that ever happens
trackNumberLength = len(str(len(files)))
if trackNumberLength < 2:
trackNumberLength = 2
for file in files:
if os.path.basename(file).lower() in ["cover.png", "cover.jpg", "cover.gif"]:
# we've found our cover art
cover = file
ext = re.search(fileExtensionRegex, file.lower()).group(0)
if ext in musicExts:
if re.search(probablyASongRegex, file) != None:
# this is definitely a song and probably not a bonus sound file or whatever
if ext == "ogg":
f = OggVorbis(file)
name = "{0} {1}.{2}".format(f["TRACKNUMBER"][0].zfill(trackNumberLength), f["TITLE"][0], ext)
for bc in bannedCharacters:
if bc in name:
name = name.replace(bc, "-") # replace banned characters with dashes
songs.append(name)
if album == "":
album = f["ALBUM"][0]
if f["ARTIST"][0] not in artists:
artists.append(f["ARTIST"][0])
trackCount = trackCount + 1
subprocess.check_output(["mv", file, os.path.dirname(file) + os.sep + name])
else:
log("UNSUPPORTED FORMAT BECAUSE LYNNE IS A LAZY MORON")
if len(artists) > 1:
if len(artists) > 1 and "Various Artists" not in artists:
artists.append("Various Artists")
if cover == "":
# TODO: HANDLE THIS PROPERLY
die("couldn't find the cover art :)))))")
while os.path.getsize(cover) / 1024 > args.threshold:
if os.path.basename(cover) != "cover-lq.jpg":
nucover = os.path.dirname(cover) + os.sep + "cover-lq.jpg"
subprocess.check_output(["convert", cover, "-quality", "85", "-strip", nucover]) # convert the file to a jpeg
cover = nucover
else:
subprocess.check_output(["convert", cover, "-resize", "90%", cover]) # shrink it slightly
with open(cover, "rb") as cvr:
data = cvr.read()
imginfo = subprocess.check_output(["identify", "-ping", "-format", r"%w,%h,%m,%[bit-depth]", cover]).decode("utf-8").split(",")
# TODO: surely you don't have to do this?? does mutagen not have a higher level way of setting an image?
picture = Picture()
picture.data = data
picture.type = 3
picture.mime = "image/{0}".format(imginfo[2].lower())
picture.width = int(imginfo[0])
picture.height = int(imginfo[1])
picture.depth = int(imginfo[3])
picture_data = picture.write()
encoded_data = base64.b64encode(picture_data)
vcomment_value = encoded_data.decode("ascii")
for song in songs:
f = OggVorbis(tmp + os.sep + song)
f["metadata_block_picture"] = [vcomment_value]
f.save()
artist = artists[0]
artists.append("Custom...")
# log("Please choose the artist name to use when creating the folder.")
choice = 0
while True:
artist = None
while artist is None:
log("Artist directory:")
for i in range(len(artists)):
log("{0}) {1}".format(i + 1, artists[i]))
choice = 1 if args.quiet else input("> ")
try:
choice = artists[int(choice) - 1]
except KeyError:
log(f"Please choose an option from 1 to {len(artists)}.")
continue
break
if choice == "Custom...":
log("Enter the name to use.")
choice = input("> ")
# log("Setting artist to {0}".format(choice))
artist = choice
log(f"{i+1}) {artists[i]}")
log(f"{len(artists) + 1}) Custom...")
choice = "1" if args.quiet else input("> ")
if choice.isdecimal():
if int(choice) == len(artists) + 1:
log("Enter the name to use.")
else:
try:
artist = artists[int(choice) - 1]
except KeyError:
log(f"Please choose a number between 1 and {len(artists) + 1}.")
else:
log(f"Please choose a number between 1 and {len(artists) + 1}")
mPath = os.path.join(args.destination, artist, album)
subprocess.check_output(["mkdir", "-p", mPath])
for root, dirs, filez in os.walk(tmp):
for file in filez: # for every file
# TODO: i'm 99% sure python has a built in move command
# TODO: os.path.join
subprocess.check_output(["mv", root + os.sep + file, mPath + os.sep + file])
log("Deleting {}...".format(tmp))
# TODO: i'm 100% sure python has a built in remove command
subprocess.check_output(["rm", "-rf", tmp])
destination = os.path.join(args.destination, artist, album)
log(f"Moving files to {destination}...")
os.makedirs(destination, exist_ok = True)
for source_name, dest_name in songs.items():
shutil.move(os.path.join(tmp, source_name), os.path.join(destination, dest_name))
shutil.move(os.path.join(tmp, "cover.jpg"), os.path.join(destination, "cover.jpg"))
tmp.cleanup()
log("Done!")

2
requirements.txt Normal file
View File

@ -0,0 +1,2 @@
mutagen~=1.45
Pillow~=8.0