youtube_dl/extractor/r7.py

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