podcopy

#!/usr/bin/env python

import sys, os
import datetime
import shutil

# Copy podcasts from podget to an mp3 device (or other directory).
def copy_pods(indir, outdir, playlist=None):

    illegal_chars = '?=&;:"\''
    today = datetime.datetime.now().strftime('%y-%m-%d-')

    # podget runs are stored in .m3u files.
    # We need the one with the most recent last-modified date,
    # unless we've been handed one.
    if playlist:
        playlist = os.path.join(indir, playlist)
    else:
        mtime = 0
        for f in os.listdir(indir):
            if not f.endswith('.m3u'):
                continue
            fpath = os.path.join(indir, f)
            statbuf = os.stat(fpath)
            if not playlist or statbuf.st_mtime > mtime:
                mtime = statbuf.st_mtime
                playlist = fpath

    if not playlist:
        print "No m3u files in", indir
        return

    # Get a list of the files already in the output dir.
    # We'll compare names and sizes to try to avoid duplicate copies.
    existing_files = os.listdir(outdir)
    existing_sizes = []
    for f in existing_files:
        statbuf = os.stat(os.path.join(outdir, f))
        existing_sizes.append(statbuf.st_size)

    # Now try to copy the input files.
    fp = open(playlist)
    for f in fp:
        f = f.strip()

        first_bad = -1
        for c in illegal_chars:
            if c in f:
                i = f.find(c)
                if i >= 0 and (first_bad < 0 or i < first_bad):
                    first_bad = i

        #  Save the extension:
        base, ext = os.path.splitext(f)
        f = os.path.join(indir, f)

        if first_bad >= 0:
            # Try just chopping off everything starting with the
            # first bad character.
            base = base[0:first_bad]

        # Split off the local dir path, e.g. Science/filename.mp3
        base = os.path.split(base)[-1]

        # Is the file already there?
        # Some sites, NPR in particular, put the same show
        # out there several times on different dates and with
        # similar but different filenames.
        # E.g. npr_378885682.mp3 and npr_378885682?orgId=1&amp;e=378885682&amp;d=257&amp;ft=pod&amp;f=344098539.mp3
        src_size = os.stat(f).st_size
        try:
            i = existing_sizes.index(src_size)
            print "\n**** Duplicate!", existing_files[i], \
                "is the same size as", f
            # Skip this file. XXX Should probably check for similar filenames.
            continue

        except ValueError:
            pass

        # add today's date.
        base = today + base

        while os.path.exists(os.path.join(outdir, base + ext)):
            # Otherwise, we have a name collision. Add an X.
            base += 'X'

        if base:
            print "cp '" + f + "'", os.path.join(outdir, base + ext)
            shutil.copyfile(f, os.path.join(outdir, base + ext))

if len(sys.argv) > 3:
    playlist = sys.argv[3]
else:
    playlist = None
copy_pods(sys.argv[1], sys.argv[2], playlist)