Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/nowness.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .brightcove import BrightcoveIE
   7 from .common import InfoExtractor
   8 from ..utils import ExtractorError
   9
  10
  11 class NownessIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:(?:www|cn)\.)?nowness\.com/[^?#]*?/(?P<id>[0-9]+)/(?P<slug>[^/]+?)(?:$|[?#])'
  13
  14     _TESTS = [
  15         {
  16             'url': 'http://www.nowness.com/day/2013/6/27/3131/candor--the-art-of-gesticulation',
  17             'md5': '068bc0202558c2e391924cb8cc470676',
  18             'info_dict': {
  19                 'id': '2520295746001',
  20                 'ext': 'mp4',
  21                 'title': 'Candor: The Art of Gesticulation',
  22                 'description': 'Candor: The Art of Gesticulation',
  23                 'thumbnail': 're:^https?://.*\.jpg',
  24                 'uploader': 'Nowness',
  25             }
  26         },
  27         {
  28             'url': 'http://cn.nowness.com/day/2014/8/7/4069/kasper-bj-rke-ft-jaakko-eino-kalevi--tnr',
  29             'md5': 'e79cf125e387216f86b2e0a5b5c63aa3',
  30             'info_dict': {
  31                 'id': '3716354522001',
  32                 'ext': 'mp4',
  33                 'title': 'Kasper Bjørke ft. Jaakko Eino Kalevi: TNR',
  34                 'description': 'Kasper Bjørke ft. Jaakko Eino Kalevi: TNR',
  35                 'thumbnail': 're:^https?://.*\.jpg',
  36                 'uploader': 'Nowness',
  37             }
  38         },
  39     ]
  40
  41     def _real_extract(self, url):
  42         mobj = re.match(self._VALID_URL, url)
  43         video_id = mobj.group('slug')
  44
  45         webpage = self._download_webpage(url, video_id)
  46         player_url = self._search_regex(
  47             r'"([^"]+/content/issue-[0-9.]+.js)"', webpage, 'player URL')
  48         real_id = self._search_regex(
  49             r'\sdata-videoId="([0-9]+)"', webpage, 'internal video ID')
  50
  51         player_code = self._download_webpage(
  52             player_url, video_id,
  53             note='Downloading player JavaScript',
  54             errnote='Player download failed')
  55         player_code = player_code.replace("'+d+'", real_id)
  56
  57         bc_url = BrightcoveIE._extract_brightcove_url(player_code)
  58         if bc_url is None:
  59             raise ExtractorError('Could not find player definition')
  60         return {
  61             '_type': 'url',
  62             'url': bc_url,
  63             'ie_key': 'Brightcove',
  64         }