Download and upload to Google Music Regionaljournal Zuerich Schaffhausen podcasts.

podcast.py

#!/usr/bin/env python
'''
Setup:

     $ pip install gmusicapi==4.0.0 requests==2.5.1
     $ mkdir podcasts
     $ python podcast.py
'''
import requests
import bs4
import os.path
import sys
from gmusicapi import Musicmanager

DOWNLOAD_DIR = 'podcasts'
PODCAST_URL = 'http://www.srf.ch/sendungen/regionaljournal-zuerich-schaffhausen'


def retrieve_audio_urls():
    print('Retrieving current list of podcasts...')
    # Retrieve a list of all downloads.
    r = requests.get(PODCAST_URL)
    assert r.ok
    soup = bs4.BeautifulSoup(r.content)
    mp3_urls = [node['href'] for node in soup.find_all('a', class_='download')]

    if not mp3_urls:
        raise Exception('No podcast found on {}.'.format(PODCAST_URL))

    print('  Found {} URLs.'.format(len(mp3_urls)))
    print('')
    return mp3_urls


def download_missing_files(urls, directory):
    print('Downloading new podcasts...'.format(len(urls)))
    # Download those not already downloaded.
    files = []
    for url in urls:
        local_filename = os.path.join(directory, os.path.basename(url))
        # If not already downloaded,
        if not os.path.exists(local_filename):
            # Download
            print('  Downloading {} --> {}'.format(url, local_filename))
            r = requests.get(url, stream=True)
            assert r.ok
            with open(local_filename, 'wb') as f:
                for chunk in r.iter_content(chunk_size=1024):
                    if chunk:  # filter out keep-alive new chunks
                        f.write(chunk)
                        f.flush()
            # Mark to upload
            files.append(local_filename)
    print('')
    return files


def login_music_manager():
    # Login (first time will require to open a page and give a code).
    mm = Musicmanager()
    if not mm.login():
        mm.perform_oauth()
        if not mm.login():
            raise Exception('Could not authenticate.')
    return mm


if __name__ == '__main__':
    # Login first so that it asks credentials if necessary.
    mm = login_music_manager()

    # Retrieve current list of podcasts.
    mp3_urls = retrieve_audio_urls()

    # Download new podcasts
    downloaded_files = download_missing_files(mp3_urls, DOWNLOAD_DIR)
    if not downloaded_files:
        print('No new podcast to upload.')
        sys.exit(0)

    # Upload to Google Music
    print('Uploading {} podcast files...'.format(len(downloaded_files)))
    for filename in downloaded_files:
        print('  Uploading {0}'.format(os.path.basename(filename)))
        mm.upload(filename)