yt_dlp/extractor/swrmediathek.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     parse_duration,
   4     int_or_none,
   5     determine_protocol,
   6 )
   7
   8
   9 class SWRMediathekIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?swrmediathek\.de/(?:content/)?player\.htm\?show=(?P<id>[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})'
  11
  12     _TESTS = [{
  13         'url': 'http://swrmediathek.de/player.htm?show=849790d0-dab8-11e3-a953-0026b975f2e6',
  14         'md5': '8c5f6f0172753368547ca8413a7768ac',
  15         'info_dict': {
  16             'id': '849790d0-dab8-11e3-a953-0026b975f2e6',
  17             'ext': 'mp4',
  18             'title': 'SWR odysso',
  19             'description': 'md5:2012e31baad36162e97ce9eb3f157b8a',
  20             'thumbnail': r're:^http:.*\.jpg$',
  21             'duration': 2602,
  22             'upload_date': '20140515',
  23             'uploader': 'SWR Fernsehen',
  24             'uploader_id': '990030',
  25         },
  26     }, {
  27         'url': 'http://swrmediathek.de/player.htm?show=0e1a8510-ddf2-11e3-9be3-0026b975f2e6',
  28         'md5': 'b10ab854f912eecc5a6b55cd6fc1f545',
  29         'info_dict': {
  30             'id': '0e1a8510-ddf2-11e3-9be3-0026b975f2e6',
  31             'ext': 'mp4',
  32             'title': 'Nachtcafé - Alltagsdroge Alkohol - zwischen Sektempfang und Komasaufen',
  33             'description': 'md5:e0a3adc17e47db2c23aab9ebc36dbee2',
  34             'thumbnail': r're:http://.*\.jpg',
  35             'duration': 5305,
  36             'upload_date': '20140516',
  37             'uploader': 'SWR Fernsehen',
  38             'uploader_id': '990030',
  39         },
  40         'skip': 'redirect to http://swrmediathek.de/index.htm?hinweis=swrlink',
  41     }, {
  42         'url': 'http://swrmediathek.de/player.htm?show=bba23e10-cb93-11e3-bf7f-0026b975f2e6',
  43         'md5': '4382e4ef2c9d7ce6852535fa867a0dd3',
  44         'info_dict': {
  45             'id': 'bba23e10-cb93-11e3-bf7f-0026b975f2e6',
  46             'ext': 'mp3',
  47             'title': 'Saša Stanišic: Vor dem Fest',
  48             'description': 'md5:5b792387dc3fbb171eb709060654e8c9',
  49             'thumbnail': r're:http://.*\.jpg',
  50             'duration': 3366,
  51             'upload_date': '20140520',
  52             'uploader': 'SWR 2',
  53             'uploader_id': '284670',
  54         },
  55         'skip': 'redirect to http://swrmediathek.de/index.htm?hinweis=swrlink',
  56     }]
  57
  58     def _real_extract(self, url):
  59         video_id = self._match_id(url)
  60
  61         video = self._download_json(
  62             'http://swrmediathek.de/AjaxEntry?ekey=%s' % video_id,
  63             video_id, 'Downloading video JSON')
  64
  65         attr = video['attr']
  66         title = attr['entry_title']
  67         media_type = attr.get('entry_etype')
  68
  69         formats = []
  70         for entry in video.get('sub', []):
  71             if entry.get('name') != 'entry_media':
  72                 continue
  73
  74             entry_attr = entry.get('attr', {})
  75             f_url = entry_attr.get('val2')
  76             if not f_url:
  77                 continue
  78             codec = entry_attr.get('val0')
  79             if codec == 'm3u8':
  80                 formats.extend(self._extract_m3u8_formats(
  81                     f_url, video_id, 'mp4', 'm3u8_native',
  82                     m3u8_id='hls', fatal=False))
  83             elif codec == 'f4m':
  84                 formats.extend(self._extract_f4m_formats(
  85                     f_url + '?hdcore=3.7.0', video_id,
  86                     f4m_id='hds', fatal=False))
  87             else:
  88                 formats.append({
  89                     'format_id': determine_protocol({'url': f_url}),
  90                     'url': f_url,
  91                     'quality': int_or_none(entry_attr.get('val1')),
  92                     'vcodec': codec if media_type == 'Video' else 'none',
  93                     'acodec': codec if media_type == 'Audio' else None,
  94                 })
  95
  96         upload_date = None
  97         entry_pdatet = attr.get('entry_pdatet')
  98         if entry_pdatet:
  99             upload_date = entry_pdatet[:-4]
 100
 101         return {
 102             'id': video_id,
 103             'title': title,
 104             'description': attr.get('entry_descl'),
 105             'thumbnail': attr.get('entry_image_16_9'),
 106             'duration': parse_duration(attr.get('entry_durat')),
 107             'upload_date': upload_date,
 108             'uploader': attr.get('channel_title'),
 109             'uploader_id': attr.get('channel_idkey'),
 110             'formats': formats,
 111         }