id3-picture-tidy.py

#! /usr/bin/env python2
# -*- coding: utf-8 -*-
#
# Traverse a directory tree, picking embedded pictures from .MP3 and .M4A audio
# saving them to a “cover.{png,jpg,bmp}” (as long as the file does *not* exist)
# and then removing the embedded pictures from the audio file.
#
# Requires the Mutagen tag edition library: https://code.google.com/p/mutagen/
# Most distributions ship in a package called “python-mutagen“.
#

from os import path as P
import os, sys
import mutagen

PIC_TAGS = (
    "PIC", u"PIC", u"PIC:",
    "APIC", u"APIC", u"APIC:",
)


def handle_dir(path, files):
    for fpath in (P.join(path, f) for f in files):
        try:
            fileinfo = mutagen.File(fpath)
        except Exception, e:
            print >> sys.stderr, "%s" % e
            continue

        for tag in PIC_TAGS:
            if tag in fileinfo.tags:
                pic_data = fileinfo.tags.get(tag)
                pic_mime = pic_data.mime.lower()
                pic_path = None
                if pic_mime == "image/png":
                    pic_path = P.join(path, "cover.png")
                elif pic_mime == "image/bmp":
                    pic_path = P.join(path, "cover.bmp")
                elif pic_mime == "image/jpeg" or pic_mime == "image/jpg":
                    pic_path = P.join(path, "cover.jpg")
                else:
                    print >> sys.stderr, " - %r unrecognized type %r" % (fpath, pic_mime)

                if pic_path is not None and not P.isfile(pic_path):
                    print "Writing %r" % pic_path
                    with file(pic_path, "wb") as pic_fd:
                        pic_fd.write(pic_data.data)

                print "Removing picture from %r" % fpath
                for tag in PIC_TAGS:
                    fileinfo.tags.delall(tag)
                fileinfo.tags.save()
                fileinfo.save()


def walk_and_handle(path, cb=handle_dir):
    for dirpath, dirnames, filenames in os.walk(path):
        audios = [f for f in filenames
                          if f.lower().endswith(".mp3")
                          or f.lower().endswith(".m4a")]
        if len(audios): cb(dirpath, audios)

if __name__ == "__main__":
    import sys
    if len(sys.argv) == 2:
        walk_and_handle(sys.argv[1])
    else:
        raise SystemExit("Usage: %s <path>" % sys.argv[0])