_ Git - youtube-dl/blob - youtube_dl/extractor/goldenmoustache.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6 )
   7
   8
   9 class GoldenMoustacheIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?goldenmoustache\.com/(?P<display_id>[\w-]+)-(?P<id>\d+)'
  11     _TESTS = [{
  12         'url': 'http://www.goldenmoustache.com/suricate-le-poker-3700/',
  13         'md5': '0f904432fa07da5054d6c8beb5efb51a',
  14         'info_dict': {
  15             'id': '3700',
  16             'ext': 'mp4',
  17             'title': 'Suricate - Le Poker',
  18             'description': 'md5:3d1f242f44f8c8cb0a106f1fd08e5dc9',
  19             'thumbnail': 're:^https?://.*\.jpg$',
  20             'view_count': int,
  21         }
  22     }, {
  23         'url': 'http://www.goldenmoustache.com/le-lab-tout-effacer-mc-fly-et-carlito-55249/',
  24         'md5': '27f0c50fb4dd5f01dc9082fc67cd5700',
  25         'info_dict': {
  26             'id': '55249',
  27             'ext': 'mp4',
  28             'title': 'Le LAB - Tout Effacer (Mc Fly et Carlito)',
  29             'description': 'md5:9b7fbf11023fb2250bd4b185e3de3b2a',
  30             'thumbnail': 're:^https?://.*\.(?:png|jpg)$',
  31             'view_count': int,
  32         }
  33     }]
  34
  35     def _real_extract(self, url):
  36         video_id = self._match_id(url)
  37         webpage = self._download_webpage(url, video_id)
  38
  39         video_url = self._html_search_regex(
  40             r'data-src-type="mp4" data-src="([^"]+)"', webpage, 'video URL')
  41         title = self._html_search_regex(
  42             r'<title>(.*?)(?: - Golden Moustache)?</title>', webpage, 'title')
  43         thumbnail = self._og_search_thumbnail(webpage)
  44         description = self._og_search_description(webpage)
  45         view_count = int_or_none(self._html_search_regex(
  46             r'<strong>([0-9]+)</strong>\s*VUES</span>',
  47             webpage, 'view count', fatal=False))
  48
  49         return {
  50             'id': video_id,
  51             'url': video_url,
  52             'ext': 'mp4',
  53             'title': title,
  54             'description': description,
  55             'thumbnail': thumbnail,
  56             'view_count': view_count,
  57         }