Most recent podcasts playlist

copy-over-podcasts.py

#!/usr/bin/python2.5

import sys
import os
import stat
import re
from subprocess import call, check_call, Popen, PIPE
from optparse import OptionParser

# I've found this script useful for copying my most recently
# downloaded podcasts to a Sansa Clip MP3 player.  It creates a
# recent-podcasts.m3u playlist in the destination directory which
# lists these podcasts from oldest to newest.
#
# The only subtlety is that if you specify the --mangle option then the
# script will wipe out any "podcast" TCON id3v2 tag in the version of
# the file on the device, since this prevents podcasts appearing in a
# playlist under "MUSIC" on my device.
#
# I'd recommend creating an alias in your .bashrc file to call this
# script with your usual arguments.

usage_message = "Usage: %prog [OPTIONS] [SOURCE_DIRECTORY] [DESTINATION_DIRECTORY]"
parser = OptionParser(usage=usage_message)
parser.add_option('-n', 
                  dest="n",
                  default=40,
                  type="int",
                  help="maximum number of podcasts [default %default]")
parser.add_option('-m', '--mangle',
                  action="store_true",
                  dest="mangle", help="wipe out the TCON tag if it says 'podcast'")
options,args = parser.parse_args()

if len(args) != 2:
    parser.print_help()
    sys.exit(1)

new_podcasts_source, destination_podcasts_folder = args

def get_content_field(mp3_filename):
    command = ["id3v2","-l",mp3_filename]
    result = Popen(command, stdout=PIPE).communicate()[0]
    m = re.search("(?ims)(TCON?) \([^\)]+\): ([^\r\n]*)[\r\n]",result)
    return m and (m.group(1),m.group(2))

def length_in_seconds(mp3_filename):
    command = ["mp3info", "-p", "%S", mp3_filename]
    result = Popen(command, stdout=PIPE, stderr=PIPE).communicate()[0]
    return int(result.strip(),10)

def title_id3v2(mp3_filename):
    command = ["id3v2","-l",mp3_filename]
    result = Popen(command, stdout=PIPE).communicate()[0]
    # The obsolete v2.2 uses TT2 instead of TIT2:
    m = re.search("(?ims)TI?T2 \([^\)]+\): ([^\r\n]*)[\r\n]",result)
    return m and m.group(1)

def title_id3v1(mp3_filename):
    command = ["mp3info", "-p", "%t", mp3_filename]
    result = Popen(command, stdout=PIPE, stderr=PIPE).communicate()[0].strip()
    return len(result) >= 1 and result

def title_from_filename(mp3_filename):
    return re.sub('\.([^\.]*)$','',os.path.basename(mp3_filename))

files = []

audio_file_regexp = re.compile(".mp3$",re.I)

for root, subfolders, basenames in os.walk(new_podcasts_source):
    for file in basenames:
        if audio_file_regexp.search(file):
            files.append(os.path.join(root,file))

def mtime(filename):
    return os.stat(filename)[stat.ST_MTIME]

files.sort(key=mtime)

most_recent_files = files[-options.n:len(files)]

playlist_file = os.path.join(destination_podcasts_folder,"recent-podcasts.m3u")
fp = open(playlist_file,"w")

fp.write("#EXTM3U\r\n")
for file in most_recent_files:
    t1 = title_id3v1(file)
    t2 = title_id3v2(file)
    t_default = title_from_filename(file)
    t = t2 or t1 or t_default
    if not t:
        print "Failed to find a title for: '"+file+"', so using 'Unknown'"
        t = "Unknown"
    l = length_in_seconds(file)
    basename = os.path.basename(file)
    destination = os.path.join(destination_podcasts_folder,basename)
    fp.write("#EXTINF:"+str(l)+","+t+"\r\n")
    fp.write(basename+"\r\n")
    if os.path.exists(destination):
        print "Skipping "+basename+"; plausibly already copied"
        continue
    print "Copying "+basename
    if 0 != call(["cp",
                  file,
                  destination]):
        print "Copy failed: deleting the partial destination file and exiting"
        call(["rm","-f",destination])
        sys.exit(1)
    if options.mangle:
        # Annoyingly, my Sansa Clip won't list files in a playlist if
        # they have a TCON or TCO tag that says that it's a podcast,
        # so if that tag is there, rewrite it to be empty:
        content_tag = get_content_field(destination)
        if content_tag:
            tag, value = content_tag
            if re.search("(?ims)podcast",value):
                print "  Fixing "+tag+" tag for "+basename
                print "  "+tag+" was: '"+value+"'"
                check_call(["id3v2",
                            "--TCON",
                            "",
                            destination])
fp.close()