bandcavall/bandcavall/main.py

from os.path import join
import argparse
from urllib.parse import urlparse
import logging
import json
import multiprocessing
import functools

from splinter import Browser

from .download import download_track
from .visit import visit_track, visit_album, visit_artist

log = logging.getLogger(__name__)


def get_parser():
    p = argparse.ArgumentParser(
        description='Download tracks from bandcamp, sharing is caring!',
        formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    p.add_argument('url')
    p.add_argument('--loglevel', default='INFO', help='Log level',
                   choices=['DEBUG', 'INFO', 'WARNING', 'ERROR'])
    info = p.add_argument_group('fetch track info')
    info.add_argument('--driver', default='firefox',
                      help='Internal splinter option',
                      choices=['firefox', 'chrome'])
    info.add_argument('--mode', default=None,
                      choices=['album', 'track', 'artist'])
    do = p.add_argument_group('what to do with fetched urls')
    do.add_argument('--do', default='print',
                    help='What to do with retrieved informations',
                    choices=['print', 'wgetprint', 'download'])
    do.add_argument('--download-base', default='.',
                    help='Only applies to wget')
    do.add_argument('-j', '--jobs', default=3, type=int,
                    help='Parallel downloads')

    return p


def get_mode(url):
    path = urlparse(url).path
    if path.startswith('/track/'):
        return 'track'
    if path.startswith('/album/'):
        return 'album'
    raise ValueError('unsupported url: "{}"'.format(path))


def main():
    args = get_parser().parse_args()
    logging.basicConfig(level=args.loglevel)
    logging.getLogger('selenium.webdriver').setLevel(logging.WARNING)
    logging.getLogger('requests').setLevel(logging.WARNING)

    if args.mode is None:
        args.mode = get_mode(args.url)
        log.info('Going in %s mode' % args.mode)

    prefs = {
        'dom.audiochannel.mutedByDefault': True
    }
    with Browser(args.driver, profile_preferences=prefs) as br:
        # get track info
        if args.mode == 'track':
            tracks = visit_track(br, args.url)
        elif args.mode == 'album':
            tracks = visit_album(br, args.url)
        else:
            tracks = visit_artist(br, args.url)

        # do something
        if args.do == 'print':
            for track in tracks:
                print(json.dumps(track))
        elif args.do == 'wgetprint':
            for track in tracks:
                cmd = ['wget', '-nv', "'%s'" % track['url'],
                       '-O', "'%s'" %
                       join(args.download_base,
                            '-'.join((track['artist'], track['album'],
                                      track['title']))).replace(' ', '_')
                       ]
                print(' '.join(cmd))
        elif args.do == 'download':
            pool = multiprocessing.Pool(processes=args.jobs)
            for res in pool.imap_unordered(functools.partial(download_track,
                                                             args=args),
                                           tracks):
                print('R', res)
            pool.close()
            pool.join()
        else:
            raise NotImplementedError('action ' + args.do)

if __name__ == '__main__':
    main()