yt_dlp/extractor/r7.py

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