1
0
Fork 0
mirror of https://gitlab.com/dstftw/youtube-dl.git synced 2020-11-16 09:42:26 +00:00
youtube-dl/youtube_dl/extractor/podomatic.py
2014-02-24 17:31:09 +01:00

53 lines
1.6 KiB
Python

from __future__ import unicode_literals
import json
import re
from .common import InfoExtractor
from ..utils import int_or_none
class PodomaticIE(InfoExtractor):
IE_NAME = 'podomatic'
_VALID_URL = r'^(?P<proto>https?)://(?P<channel>[^.]+)\.podomatic\.com/entry/(?P<id>[^?]+)'
_TEST = {
"url": "http://scienceteachingtips.podomatic.com/entry/2009-01-02T16_03_35-08_00",
"file": "2009-01-02T16_03_35-08_00.mp3",
"md5": "84bb855fcf3429e6bf72460e1eed782d",
"info_dict": {
"uploader": "Science Teaching Tips",
"uploader_id": "scienceteachingtips",
"title": "64. When the Moon Hits Your Eye",
"duration": 446,
}
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
channel = mobj.group('channel')
json_url = (('%s://%s.podomatic.com/entry/embed_params/%s' +
'?permalink=true&rtmp=0') %
(mobj.group('proto'), channel, video_id))
data_json = self._download_webpage(
json_url, video_id, note=u'Downloading video info')
data = json.loads(data_json)
video_url = data['downloadLink']
uploader = data['podcast']
title = data['title']
thumbnail = data['imageLocation']
duration = int_or_none(data.get('length'), 1000)
return {
'id': video_id,
'url': video_url,
'title': title,
'uploader': uploader,
'uploader_id': channel,
'thumbnail': thumbnail,
'duration': duration,
}