[cinchcast] Add new extractor (Fixes #4428)
[youtube-dl] / youtube_dl / extractor / cinchcast.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     int_or_none,
7     unified_strdate,
8     xpath_text,
9 )
10
11
12 class CinchcastIE(InfoExtractor):
13     _VALID_URL = r'https?://player\.cinchcast\.com/.*?assetId=(?P<id>[0-9]+)'
14     _TEST = {
15         # Actual test is run in generic, look for undergroundwellness
16         'url': 'http://player.cinchcast.com/?platformId=1&#038;assetType=single&#038;assetId=7141703',
17         'only_matching': True,
18     }
19
20     def _real_extract(self, url):
21         video_id = self._match_id(url)
22         doc = self._download_xml(
23             'http://www.blogtalkradio.com/playerasset/mrss?assetType=single&assetId=%s' % video_id,
24             video_id)
25
26         item = doc.find('.//item')
27         title = xpath_text(item, './title', fatal=True)
28         date_str = xpath_text(
29             item, './{http://developer.longtailvideo.com/trac/}date')
30         upload_date = unified_strdate(date_str, day_first=False)
31         # duration is present but wrong
32         formats = []
33         formats.append({
34             'format_id': 'main',
35             'url': item.find(
36                 './{http://search.yahoo.com/mrss/}content').attrib['url'],
37         })
38         backup_url = xpath_text(
39             item, './{http://developer.longtailvideo.com/trac/}backupContent')
40         if backup_url:
41             formats.append({
42                 'preference': 2,  # seems to be more reliable
43                 'format_id': 'backup',
44                 'url': backup_url,
45             })
46         self._sort_formats(formats)
47
48         return {
49             'id': video_id,
50             'title': title,
51             'upload_date': upload_date,
52             'formats': formats,
53         }