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 (
   6     js_to_json,
   7     unescapeHTML,
   8     int_or_none,
   9 )
  10
  11
  12 class R7IE(InfoExtractor):
  13     _VALID_URL = r'''(?x)https?://
  14                         (?:
  15                             (?:[a-zA-Z]+)\.r7\.com(?:/[^/]+)+/idmedia/|
  16                             noticias\.r7\.com(?:/[^/]+)+/[^/]+-|
  17                             player\.r7\.com/video/i/
  18                         )
  19                         (?P<id>[\da-f]{24})
  20                         '''
  21     _TESTS = [{
  22         'url': 'http://videos.r7.com/policiais-humilham-suspeito-a-beira-da-morte-morre-com-dignidade-/idmedia/54e7050b0cf2ff57e0279389.html',
  23         'md5': '403c4e393617e8e8ddc748978ee8efde',
  24         'info_dict': {
  25             'id': '54e7050b0cf2ff57e0279389',
  26             'ext': 'mp4',
  27             'title': 'Policiais humilham suspeito à beira da morte: "Morre com dignidade"',
  28             'thumbnail': 're:^https?://.*\.jpg$',
  29             'duration': 98,
  30             'like_count': int,
  31             'view_count': int,
  32         },
  33     }, {
  34         'url': 'http://esportes.r7.com/videos/cigano-manda-recado-aos-fas/idmedia/4e176727b51a048ee6646a1b.html',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://noticias.r7.com/record-news/video/representante-do-instituto-sou-da-paz-fala-sobre-fim-do-estatuto-do-desarmamento-5480fc580cf2285b117f438d/',
  38         'only_matching': True,
  39     }, {
  40         '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',
  41         'only_matching': True,
  42     }]
  43
  44     def _real_extract(self, url):
  45         video_id = self._match_id(url)
  46
  47         webpage = self._download_webpage(
  48             'http://player.r7.com/video/i/%s' % video_id, video_id)
  49
  50         item = self._parse_json(js_to_json(self._search_regex(
  51             r'(?s)var\s+item\s*=\s*({.+?});', webpage, 'player')), video_id)
  52
  53         title = unescapeHTML(item['title'])
  54         thumbnail = item.get('init', {}).get('thumbUri')
  55         duration = None
  56
  57         statistics = item.get('statistics', {})
  58         like_count = int_or_none(statistics.get('likes'))
  59         view_count = int_or_none(statistics.get('views'))
  60
  61         formats = []
  62         for format_key, format_dict in item['playlist'][0].items():
  63             src = format_dict.get('src')
  64             if not src:
  65                 continue
  66             format_id = format_dict.get('format') or format_key
  67             if duration is None:
  68                 duration = format_dict.get('duration')
  69             if '.f4m' in src:
  70                 formats.extend(self._extract_f4m_formats(src, video_id, preference=-1))
  71             elif src.endswith('.m3u8'):
  72                 formats.extend(self._extract_m3u8_formats(src, video_id, 'mp4', preference=-2))
  73             else:
  74                 formats.append({
  75                     'url': src,
  76                     'format_id': format_id,
  77                 })
  78         self._sort_formats(formats)
  79
  80         return {
  81             'id': video_id,
  82             'title': title,
  83             'thumbnail': thumbnail,
  84             'duration': duration,
  85             'like_count': like_count,
  86             'view_count': view_count,
  87             'formats': formats,
  88         }