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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     int_or_none,
   8     parse_iso8601,
   9 )
  10
  11
  12 class IGNIE(InfoExtractor):
  13     """
  14     Extractor for some of the IGN sites, like www.ign.com, es.ign.com de.ign.com.
  15     Some videos of it.ign.com are also supported
  16     """
  17
  18     _VALID_URL = r'https?://.+?\.ign\.com/(?:[^/]+/)?(?P<type>videos|show_videos|articles|feature|(?:[^/]+/\d+/video))(/.+)?/(?P<name_or_id>.+)'
  19     IE_NAME = 'ign.com'
  20
  21     _API_URL_TEMPLATE = 'http://apis.ign.com/video/v3/videos/%s'
  22     _EMBED_RE = r'<iframe[^>]+?["\']((?:https?:)?//.+?\.ign\.com.+?/embed.+?)["\']'
  23
  24     _TESTS = [
  25         {
  26             'url': 'http://www.ign.com/videos/2013/06/05/the-last-of-us-review',
  27             'md5': 'febda82c4bafecd2d44b6e1a18a595f8',
  28             'info_dict': {
  29                 'id': '8f862beef863986b2785559b9e1aa599',
  30                 'ext': 'mp4',
  31                 'title': 'The Last of Us Review',
  32                 'description': 'md5:c8946d4260a4d43a00d5ae8ed998870c',
  33                 'timestamp': 1370440800,
  34                 'upload_date': '20130605',
  35             }
  36         },
  37         {
  38             'url': 'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind',
  39             'info_dict': {
  40                 'id': '100-little-things-in-gta-5-that-will-blow-your-mind',
  41             },
  42             'playlist': [
  43                 {
  44                     'info_dict': {
  45                         'id': '5ebbd138523268b93c9141af17bec937',
  46                         'ext': 'mp4',
  47                         'title': 'GTA 5 Video Review',
  48                         'description': 'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.',
  49                         'timestamp': 1379339880,
  50                         'upload_date': '20130916',
  51                     },
  52                 },
  53                 {
  54                     'info_dict': {
  55                         'id': '638672ee848ae4ff108df2a296418ee2',
  56                         'ext': 'mp4',
  57                         'title': '26 Twisted Moments from GTA 5 in Slow Motion',
  58                         'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
  59                         'timestamp': 1386878820,
  60                         'upload_date': '20131212',
  61                     },
  62                 },
  63             ],
  64             'params': {
  65                 'skip_download': True,
  66             },
  67         },
  68         {
  69             'url': 'http://www.ign.com/articles/2014/08/15/rewind-theater-wild-trailer-gamescom-2014?watch',
  70             'md5': '618fedb9c901fd086f6f093564ef8558',
  71             'info_dict': {
  72                 'id': '078fdd005f6d3c02f63d795faa1b984f',
  73                 'ext': 'mp4',
  74                 'title': 'Rewind Theater - Wild Trailer Gamescom 2014',
  75                 'description': 'Brian and Jared explore Michel Ancel\'s captivating new preview.',
  76                 'timestamp': 1408047180,
  77                 'upload_date': '20140814',
  78             },
  79         },
  80     ]
  81
  82     def _find_video_id(self, webpage):
  83         res_id = [
  84             r'"video_id"\s*:\s*"(.*?)"',
  85             r'class="hero-poster[^"]*?"[^>]*id="(.+?)"',
  86             r'data-video-id="(.+?)"',
  87             r'<object id="vid_(.+?)"',
  88             r'<meta name="og:image" content=".*/(.+?)-(.+?)/.+.jpg"',
  89         ]
  90         return self._search_regex(res_id, webpage, 'video id', default=None)
  91
  92     def _real_extract(self, url):
  93         mobj = re.match(self._VALID_URL, url)
  94         name_or_id = mobj.group('name_or_id')
  95         page_type = mobj.group('type')
  96         webpage = self._download_webpage(url, name_or_id)
  97         if page_type != 'video':
  98             multiple_urls = re.findall(
  99                 '<param name="flashvars"[^>]*value="[^"]*?url=(https?://www\.ign\.com/videos/.*?)["&]',
 100                 webpage)
 101             if multiple_urls:
 102                 entries = [self.url_result(u, ie='IGN') for u in multiple_urls]
 103                 return {
 104                     '_type': 'playlist',
 105                     'id': name_or_id,
 106                     'entries': entries,
 107                 }
 108
 109         video_id = self._find_video_id(webpage)
 110         if not video_id:
 111             return self.url_result(self._search_regex(self._EMBED_RE, webpage, 'embed url'))
 112         return self._get_video_info(video_id)
 113
 114     def _get_video_info(self, video_id):
 115         api_data = self._download_json(self._API_URL_TEMPLATE % video_id, video_id)
 116
 117         formats = []
 118         m3u8_url = api_data['refs'].get('m3uUrl')
 119         if m3u8_url:
 120             formats.extend(self._extract_m3u8_formats(m3u8_url, video_id))
 121         f4m_url = api_data['refs'].get('f4mUrl')
 122         if f4m_url:
 123             formats.extend(self._extract_f4m_formats(f4m_url, video_id))
 124         for asset in api_data['assets']:
 125             formats.append({
 126                 'url': asset['url'],
 127                 'tbr': asset.get('actual_bitrate_kbps'),
 128                 'fps': asset.get('frame_rate'),
 129                 'height': int_or_none(asset.get('height')),
 130                 'width': int_or_none(asset.get('width')),
 131             })
 132         self._sort_formats(formats)
 133
 134         thumbnails = []
 135         for thumbnail in api_data['thumbnails']:
 136             thumbnails.append({'url': thumbnail['url']})
 137
 138         metadata = api_data['metadata']
 139
 140         return {
 141             'id': api_data.get('videoId') or video_id,
 142             'title': metadata.get('longTitle') or metadata.get('name') or metadata.get['title'],
 143             'description': metadata.get('description'),
 144             'timestamp': parse_iso8601(metadata.get('publishDate')),
 145             'duration': int_or_none(metadata.get('duration')),
 146             'display_id': metadata.get('slug') or video_id,
 147             'thumbnails': thumbnails,
 148             'formats': formats,
 149         }
 150
 151
 152 class OneUPIE(IGNIE):
 153     _VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)\.html'
 154     IE_NAME = '1up.com'
 155
 156     _TESTS = [{
 157         'url': 'http://gamevideos.1up.com/video/id/34976.html',
 158         'md5': 'c9cc69e07acb675c31a16719f909e347',
 159         'info_dict': {
 160             'id': '34976',
 161             'ext': 'mp4',
 162             'title': 'Sniper Elite V2 - Trailer',
 163             'description': 'md5:bf0516c5ee32a3217aa703e9b1bc7826',
 164             'timestamp': 1313099220,
 165             'upload_date': '20110811',
 166         }
 167     }]
 168
 169     def _real_extract(self, url):
 170         mobj = re.match(self._VALID_URL, url)
 171         result = super(OneUPIE, self)._real_extract(url)
 172         result['id'] = mobj.group('name_or_id')
 173         return result
 174
 175
 176 class PCMagIE(IGNIE):
 177     _VALID_URL = r'https?://(?:www\.)?pcmag\.com/(?P<type>videos|article2)(/.+)?/(?P<name_or_id>.+)'
 178     IE_NAME = 'pcmag'
 179
 180     _EMBED_RE = r'iframe.setAttribute\("src",\s*__util.objToUrlString\("http://widgets\.ign\.com/video/embed/content.html?[^"]*url=([^"]+)["&]'
 181
 182     _TESTS = [{
 183         'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
 184         'md5': '212d6154fd0361a2781075f1febbe9ad',
 185         'info_dict': {
 186             'id': 'ee10d774b508c9b8ec07e763b9125b91',
 187             'ext': 'mp4',
 188             'title': '010615_What\'s New Now: Is GoGo Snooping on Your Data?',
 189             'description': 'md5:a7071ae64d2f68cc821c729d4ded6bb3',
 190             'timestamp': 1420571160,
 191             'upload_date': '20150106',
 192         }
 193     },{
 194         'url': 'http://www.pcmag.com/article2/0,2817,2470156,00.asp',
 195         'md5': '94130c1ca07ba0adb6088350681f16c1',
 196         'info_dict': {
 197             'id': '042e560ba94823d43afcb12ddf7142ca',
 198             'ext': 'mp4',
 199             'title': 'HTC\'s Weird New Re Camera - What\'s New Now',
 200             'description': 'md5:53433c45df96d2ea5d0fda18be2ca908',
 201             'timestamp': 1412953920,
 202             'upload_date': '20141010',
 203         }
 204     }]