_ Git - youtube-dl/blob - youtube_dl/extractor/drtv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     ExtractorError,
   7     int_or_none,
   8     float_or_none,
   9     mimetype2ext,
  10     parse_iso8601,
  11     remove_end,
  12 )
  13
  14
  15 class DRTVIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?dr\.dk/(?:tv/se|nyheder)/(?:[^/]+/)*(?P<id>[\da-z-]+)(?:[/#?]|$)'
  17
  18     _TESTS = [{
  19         'url': 'https://www.dr.dk/tv/se/boern/ultra/panisk-paske/panisk-paske-5',
  20         'md5': 'dc515a9ab50577fa14cc4e4b0265168f',
  21         'info_dict': {
  22             'id': 'panisk-paske-5',
  23             'ext': 'mp4',
  24             'title': 'Panisk Påske (5)',
  25             'description': 'md5:ca14173c5ab24cd26b0fcc074dff391c',
  26             'timestamp': 1426984612,
  27             'upload_date': '20150322',
  28             'duration': 1455,
  29         },
  30         'skip': 'Video is no longer available',
  31     }, {
  32         'url': 'https://www.dr.dk/nyheder/indland/live-christianias-rydning-af-pusher-street-er-i-gang',
  33         'md5': '2ada5074f9e79afc0d324a8e9784d850',
  34         'info_dict': {
  35             'id': 'christiania-pusher-street-ryddes-drdkrjpo',
  36             'ext': 'mp4',
  37             'title': 'LIVE Christianias rydning af Pusher Street er i gang',
  38             'description': '- Det er det fedeste, der er sket i 20 år, fortæller christianit til DR Nyheder.',
  39             'timestamp': 1472800279,
  40             'upload_date': '20160902',
  41             'duration': 131.4,
  42         }
  43     }]
  44
  45     def _real_extract(self, url):
  46         video_id = self._match_id(url)
  47
  48         webpage = self._download_webpage(url, video_id)
  49
  50         if '>Programmet er ikke længere tilgængeligt' in webpage:
  51             raise ExtractorError(
  52                 'Video %s is not available' % video_id, expected=True)
  53
  54         video_id = self._search_regex(
  55             (r'data-(?:material-identifier|episode-slug)="([^"]+)"',
  56                 r'data-resource="[^>"]+mu/programcard/expanded/([^"]+)"'),
  57             webpage, 'video id')
  58
  59         programcard = self._download_json(
  60             'http://www.dr.dk/mu/programcard/expanded/%s' % video_id,
  61             video_id, 'Downloading video JSON')
  62         data = programcard['Data'][0]
  63
  64         title = remove_end(self._og_search_title(
  65             webpage, default=None), ' | TV | DR') or data['Title']
  66         description = self._og_search_description(
  67             webpage, default=None) or data.get('Description')
  68
  69         timestamp = parse_iso8601(data.get('CreatedTime'))
  70
  71         thumbnail = None
  72         duration = None
  73
  74         restricted_to_denmark = False
  75
  76         formats = []
  77         subtitles = {}
  78
  79         for asset in data['Assets']:
  80             if asset.get('Kind') == 'Image':
  81                 thumbnail = asset.get('Uri')
  82             elif asset.get('Kind') == 'VideoResource':
  83                 duration = float_or_none(asset.get('DurationInMilliseconds'), 1000)
  84                 restricted_to_denmark = asset.get('RestrictedToDenmark')
  85                 spoken_subtitles = asset.get('Target') == 'SpokenSubtitles'
  86                 for link in asset.get('Links', []):
  87                     uri = link.get('Uri')
  88                     if not uri:
  89                         continue
  90                     target = link.get('Target')
  91                     format_id = target or ''
  92                     preference = None
  93                     if spoken_subtitles:
  94                         preference = -1
  95                         format_id += '-spoken-subtitles'
  96                     if target == 'HDS':
  97                         formats.extend(self._extract_f4m_formats(
  98                             uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43',
  99                             video_id, preference, f4m_id=format_id))
 100                     elif target == 'HLS':
 101                         formats.extend(self._extract_m3u8_formats(
 102                             uri, video_id, 'mp4', entry_protocol='m3u8_native',
 103                             preference=preference, m3u8_id=format_id))
 104                     else:
 105                         bitrate = link.get('Bitrate')
 106                         if bitrate:
 107                             format_id += '-%s' % bitrate
 108                         formats.append({
 109                             'url': uri,
 110                             'format_id': format_id,
 111                             'tbr': int_or_none(bitrate),
 112                             'ext': link.get('FileFormat'),
 113                         })
 114                 subtitles_list = asset.get('SubtitlesList')
 115                 if isinstance(subtitles_list, list):
 116                     LANGS = {
 117                         'Danish': 'da',
 118                     }
 119                     for subs in subtitles_list:
 120                         if not subs.get('Uri'):
 121                             continue
 122                         lang = subs.get('Language') or 'da'
 123                         subtitles.setdefault(LANGS.get(lang, lang), []).append({
 124                             'url': subs['Uri'],
 125                             'ext': mimetype2ext(subs.get('MimeType')) or 'vtt'
 126                         })
 127
 128         if not formats and restricted_to_denmark:
 129             self.raise_geo_restricted(
 130                 'Unfortunately, DR is not allowed to show this program outside Denmark.',
 131                 expected=True)
 132
 133         self._sort_formats(formats)
 134
 135         return {
 136             'id': video_id,
 137             'title': title,
 138             'description': description,
 139             'thumbnail': thumbnail,
 140             'timestamp': timestamp,
 141             'duration': duration,
 142             'formats': formats,
 143             'subtitles': subtitles,
 144         }