gists/unsync.py

#!/bin/python2

# only using python2 because mutagen
from __future__ import print_function

import os
import os.path
import sys
import mutagen
import mutaext

def G(d, k):
    try:
        return d[k]
    except KeyError:
        return None

def walkfiles(walker):
    for root, _, files in walker:
        for f in files:
            yield os.path.join(root, f)

def filterext(paths, exts):
    for p in paths:
        ext = os.path.splitext(p)[1].lower()
        if ext in exts:
            yield p

def shouldsync(md):
    rating = G(md, 'rating')
    sync = G(md, 'sync')
    if rating:
        rating = rating[0]
    if sync:
        sync = sync[0].lower()
    else:
        sync = u''

    if sync == u'no' or sync == u'space':
        return False
    if sync == u'yes' or sync == u'share':
        return True
    if rating != None and rating >= 3:
        return True
    return False

def fixmetadata(md):
    if 'artist' not in md:
        md['artist'] = "Unknown Artist"
    if 'album' not in md:
        md['album'] = "Unknown Album"
    if 'title' not in md:
        fn = os.path.basename(md.path)
        fn = os.path.splitext(fn)[0]
        # TODO: attempt to infer trackNum/discNum from fn
        md['title'] = fn

def findmatching(haystack, needle):
    # TODO: don't match mismatched lengths (Xing?)
    artist = G(needle, 'artist')
    album = G(needle, 'album')
    title = G(needle, 'title')
    match = None
    for hay in haystack:
        if artist == G(hay, 'artist') \
          and album == G(hay, 'album') \
          and title == G(hay, 'title'):
            match = hay
            if match.seen:
                # TODO: check other tags and filename and such?
                print("Warning: duplicate match found:", file=sys.stderr)
                print("{0} by {1} from {2}".format(artist,album,title), file=sys.stderr)
            else:
                match.seen = True
                break
    return match

def updatemetadata(mdold, mdnew):
    modified = False
    updatabletags = ('rating', 'sync', 'comment')
    # TODO:
    # composer      : probably safe
    # genre         : might need deep equal?
    # trackNum      : TRCK with optional /[total]
    # discNum       : TPOS "    "        "
    # total tracks  : ?
    # total discs   : ?
    # [replaygain]  : TXXX:* in id3
    # albumArtist   : TPE2 in id3
    for tag in updatabletags:
        if tag in mdnew:
            if not tag in mdold or mdnew[tag] != mdold[tag]:
                mdold[tag] = mdnew[tag]
                modified = True
        elif tag in mdold:
            del mdold[tag]
            modified = True
    return modified

def run(args):
    if not len(args) in (2, 3):
        print("I need a path or two!", file=sys.stderr)
        return 1

    inonly = len(args) == 2

    # BUG: doesn't work with my .m4a files?
    goodexts = ('.mp3', '.m4a', '.flac', '.ogg')
    tosync = []
    indir = args[1]
    paths = lambda dir: filterext(walkfiles(os.walk(dir)), goodexts)

    for p in paths(indir):
        md = mutagen.File(p, easy=True)
        if shouldsync(md):
            if inonly:
                print(p)
            else:
                # TODO: don't use custom members on external metadata class
                md.path = p
                md.seen = False
                fixmetadata(md)
                tosync.append(md)

    if inonly:
        return 0
    # TODO: don't print anything
    print("Beginning matching...", file=sys.stderr)

    outdir = args[2]
    for p in paths(outdir):
        md = mutagen.File(p, easy=True)
        match = findmatching(tosync, md)
        if match:
            if updatemetadata(md, match):
                print("UPD", p)
                # save here
        else:
            print("DEL", p)
            # delete files here

    for md in tosync:
        if md.seen:
            continue
        print("ADD", md.path)

    return 0

ret = 0
try:
    ret = run(sys.argv)
except KeyboardInterrupt:
    sys.exit(1)
sys.exit(ret)