yt_dlp/extractor/r7.py

   1 from .common import InfoExtractor
   2 from ..utils import int_or_none
   3
   4
   5 class R7IE(InfoExtractor):
   6     _WORKING = False
   7     _ENABLED = None  # XXX: pass through to GenericIE
   8     _VALID_URL = r'''(?x)
   9                         https?://
  10                         (?:
  11                             (?:[a-zA-Z]+)\.r7\.com(?:/[^/]+)+/idmedia/|
  12                             noticias\.r7\.com(?:/[^/]+)+/[^/]+-|
  13                             player\.r7\.com/video/i/
  14                         )
  15                         (?P<id>[\da-f]{24})
  16                     '''
  17     _TESTS = [{
  18         'url': 'http://videos.r7.com/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-/idmedia/54e7050b0cf2ff57e0279389.html',
  19         'md5': '403c4e393617e8e8ddc748978ee8efde',
  20         'info_dict': {
  21             'id': '54e7050b0cf2ff57e0279389',
  22             'ext': 'mp4',
  23             'title': 'Policiais humilham suspeito à beira da morte: "Morre com dignidade"',
  24             'description': 'md5:01812008664be76a6479aa58ec865b72',
  25             'thumbnail': r're:^https?://.*\.jpg$',
  26             'duration': 98,
  27             'like_count': int,
  28             'view_count': int,
  29         },
  30     }, {
  31         'url': 'http://esportes.r7.com/videos/cigano-manda-recado-aos-fas/idmedia/4e176727b51a048ee6646a1b.html',
  32         'only_matching': True,
  33     }, {
  34         'url': 'http://noticias.r7.com/record-news/video/representante-do-instituto-sou-da-paz-fala-sobre-fim-do-estatuto-do-desarmamento-5480fc580cf2285b117f438d/',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://player.r7.com/video/i/54e7050b0cf2ff57e0279389?play=true&video=http://vsh.r7.com/54e7050b0cf2ff57e0279389/ER7_RE_BG_MORTE_JOVENS_570kbps_2015-02-2009f17818-cc82-4c8f-86dc-89a66934e633-ATOS_copy.mp4&linkCallback=http://videos.r7.com/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-/idmedia/54e7050b0cf2ff57e0279389.html&thumbnail=http://vtb.r7.com/ER7_RE_BG_MORTE_JOVENS_570kbps_2015-02-2009f17818-cc82-4c8f-86dc-89a66934e633-thumb.jpg&idCategory=192&share=true&layout=full&full=true',
  38         'only_matching': True,
  39     }]
  40
  41     def _real_extract(self, url):
  42         video_id = self._match_id(url)
  43
  44         video = self._download_json(
  45             f'http://player-api.r7.com/video/i/{video_id}', video_id)
  46
  47         title = video['title']
  48
  49         formats = []
  50         media_url_hls = video.get('media_url_hls')
  51         if media_url_hls:
  52             formats.extend(self._extract_m3u8_formats(
  53                 media_url_hls, video_id, 'mp4', entry_protocol='m3u8_native',
  54                 m3u8_id='hls', fatal=False))
  55         media_url = video.get('media_url')
  56         if media_url:
  57             f = {
  58                 'url': media_url,
  59                 'format_id': 'http',
  60             }
  61             # m3u8 format always matches the http format, let's copy metadata from
  62             # one to another
  63             m3u8_formats = list(filter(
  64                 lambda f: f.get('vcodec') != 'none', formats))
  65             if len(m3u8_formats) == 1:
  66                 f_copy = m3u8_formats[0].copy()
  67                 f_copy.update(f)
  68                 f_copy['protocol'] = 'http'
  69                 f = f_copy
  70             formats.append(f)
  71
  72         description = video.get('description')
  73         thumbnail = video.get('thumb')
  74         duration = int_or_none(video.get('media_duration'))
  75         like_count = int_or_none(video.get('likes'))
  76         view_count = int_or_none(video.get('views'))
  77
  78         return {
  79             'id': video_id,
  80             'title': title,
  81             'description': description,
  82             'thumbnail': thumbnail,
  83             'duration': duration,
  84             'like_count': like_count,
  85             'view_count': view_count,
  86             'formats': formats,
  87         }
  88
  89
  90 class R7ArticleIE(InfoExtractor):
  91     _WORKING = False
  92     _ENABLED = None  # XXX: pass through to GenericIE
  93     _VALID_URL = r'https?://(?:[a-zA-Z]+)\.r7\.com/(?:[^/]+/)+[^/?#&]+-(?P<id>\d+)'
  94     _TEST = {
  95         'url': 'http://tv.r7.com/record-play/balanco-geral/videos/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-16102015',
  96         'only_matching': True,
  97     }
  98
  99     @classmethod
 100     def suitable(cls, url):
 101         return False if R7IE.suitable(url) else super().suitable(url)
 102
 103     def _real_extract(self, url):
 104         display_id = self._match_id(url)
 105
 106         webpage = self._download_webpage(url, display_id)
 107
 108         video_id = self._search_regex(
 109             r'<div[^>]+(?:id=["\']player-|class=["\']embed["\'][^>]+id=["\'])([\da-f]{24})',
 110             webpage, 'video id')
 111
 112         return self.url_result(f'http://player.r7.com/video/i/{video_id}', R7IE.ie_key())