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

   1 from __future__ import unicode_literals
   2
   3 import re
   4 import json
   5
   6 from .common import InfoExtractor
   7 from ..utils import str_to_int
   8
   9
  10 class NineGagIE(InfoExtractor):
  11     IE_NAME = '9gag'
  12     _VALID_URL = r'https?://(?:www\.)?9gag\.com/tv/p/(?P<id>[a-zA-Z0-9]+)/(?P<display_id>[^?#/]+)'
  13
  14     _TESTS = [{
  15         "url": "http://9gag.com/tv/p/Kk2X5/people-are-awesome-2013-is-absolutely-awesome",
  16         "info_dict": {
  17             "id": "Kk2X5",
  18             "ext": "mp4",
  19             "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!)",
  20             "title": "\"People Are Awesome 2013\" Is Absolutely Awesome",
  21             'uploader_id': 'UCdEH6EjDKwtTe-sO2f0_1XA',
  22             'uploader': 'CompilationChannel',
  23             'upload_date': '20131110',
  24             "view_count": int,
  25             "thumbnail": "re:^https?://",
  26         },
  27         'add_ie': ['Youtube']
  28     }, {
  29         'url': 'http://9gag.com/tv/p/KklwM/alternate-banned-opening-scene-of-gravity?ref=fsidebar',
  30         'info_dict': {
  31             'id': 'KklwM',
  32             'ext': 'mp4',
  33             'display_id': 'alternate-banned-opening-scene-of-gravity',
  34             "description": "While Gravity was a pretty awesome movie already, YouTuber Krishna Shenoi came up with a way to improve upon it, introducing a much better solution to Sandra Bullock's seemingly endless tumble in space. The ending is priceless.",
  35             'title': "Banned Opening Scene Of \"Gravity\" That Changes The Whole Movie",
  36             'uploader': 'Krishna Shenoi',
  37             'upload_date': '20140401',
  38             'uploader_id': 'krishnashenoi93',
  39         },
  40         'add_ie': ['Youtube']
  41     }]
  42     _EXTERNAL_VIDEO_PROVIDER = {
  43         '1': {
  44             'url': '%s',
  45             'ie_key': 'Youtube',
  46         },
  47         '2': {
  48             'url': 'http://player.vimeo.com/video/%s',
  49             'ie_key': 'Vimeo',
  50         },
  51         '3': {
  52             'url': 'http://instagram.com/p/%s',
  53             'ie_key': 'Instagram',
  54         },
  55         '4': {
  56             'url': 'http://vine.co/v/%s',
  57             'ie_key': 'Vine',
  58         },
  59     }
  60
  61     def _real_extract(self, url):
  62         mobj = re.match(self._VALID_URL, url)
  63         video_id = mobj.group('id')
  64         display_id = mobj.group('display_id')
  65
  66         webpage = self._download_webpage(url, display_id)
  67
  68         post_view = json.loads(self._html_search_regex(
  69             r'var postView = new app\.PostView\({\s*post:\s*({.+?}),\s*posts:\s*prefetchedCurrentPost', webpage, 'post view'))
  70
  71         ie_key = None
  72         source_url = post_view.get('sourceUrl')
  73         if not source_url:
  74             external_video_id = post_view['videoExternalId']
  75             external_video_provider = post_view['videoExternalProvider']
  76             source_url = self._EXTERNAL_VIDEO_PROVIDER[external_video_provider]['url'] % external_video_id
  77             ie_key = self._EXTERNAL_VIDEO_PROVIDER[external_video_provider]['ie_key']
  78         title = post_view['title']
  79         description = post_view['description']
  80         view_count = str_to_int(post_view['externalView'])
  81         thumbnail = post_view.get('thumbnail_700w') or post_view.get('ogImageUrl') or post_view.get('thumbnail_300w')
  82
  83         return {
  84             '_type': 'url_transparent',
  85             'url': source_url,
  86             'ie_key': ie_key,
  87             'id': video_id,
  88             'display_id': display_id,
  89             'title': title,
  90             'description': description,
  91             'view_count': view_count,
  92             'thumbnail': thumbnail,
  93         }