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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6
   7
   8 class NineGagIE(InfoExtractor):
   9     IE_NAME = '9gag'
  10     _VALID_URL = r'^https?://(?:www\.)?9gag\.tv/v/(?P<id>[0-9]+)'
  11
  12     _TEST = {
  13         "url": "http://9gag.tv/v/1912",
  14         "info_dict": {
  15             "id": "1912",
  16             "ext": "mp4",
  17             "description": "This 3-minute video will make you smile and then make you feel untalented and insignificant. Anyway, you should share this awesomeness. (Thanks, Dino!)",
  18             "title": "\"People Are Awesome 2013\" Is Absolutely Awesome",
  19             "view_count": int,
  20             "thumbnail": "re:^https?://",
  21         },
  22         'add_ie': ['Youtube']
  23     }
  24
  25     def _real_extract(self, url):
  26         mobj = re.match(self._VALID_URL, url)
  27         video_id = mobj.group('id')
  28
  29         webpage = self._download_webpage(url, video_id)
  30
  31         youtube_id = self._html_search_regex(
  32             r'(?s)id="jsid-video-post-container".*?data-external-id="([^"]+)"',
  33             webpage, 'video ID')
  34         description = self._html_search_regex(
  35             r'(?s)<div class="video-caption">.*?<p>(.*?)</p>', webpage,
  36             'description', fatal=False)
  37         view_count_str = self._html_search_regex(
  38             r'<p><b>([0-9][0-9,]*)</b> views</p>', webpage, 'view count',
  39             fatal=False)
  40         view_count = (
  41             None if view_count_str is None
  42             else int(view_count_str.replace(',', '')))
  43
  44         return {
  45             '_type': 'url_transparent',
  46             'url': youtube_id,
  47             'ie_key': 'Youtube',
  48             'id': video_id,
  49             'title': self._og_search_title(webpage),
  50             'description': description,
  51             'view_count': view_count,
  52             'thumbnail': self._og_search_thumbnail(webpage),
  53         }