_ 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             'url': 'http://me.ign.com/en/videos/112203/video/how-hitman-aims-to-be-different-than-every-other-s',
  82             'only_matching': True,
  83         },
  84         {
  85             'url': 'http://me.ign.com/ar/angry-birds-2/106533/video/lrd-ldyy-lwl-lfylm-angry-birds',
  86             'only_matching': True,
  87         },
  88     ]
  89
  90     def _find_video_id(self, webpage):
  91         res_id = [
  92             r'"video_id"\s*:\s*"(.*?)"',
  93             r'class="hero-poster[^"]*?"[^>]*id="(.+?)"',
  94             r'data-video-id="(.+?)"',
  95             r'<object id="vid_(.+?)"',
  96             r'<meta name="og:image" content=".*/(.+?)-(.+?)/.+.jpg"',
  97         ]
  98         return self._search_regex(res_id, webpage, 'video id', default=None)
  99
 100     def _real_extract(self, url):
 101         mobj = re.match(self._VALID_URL, url)
 102         name_or_id = mobj.group('name_or_id')
 103         page_type = mobj.group('type')
 104         webpage = self._download_webpage(url, name_or_id)
 105         if page_type != 'video':
 106             multiple_urls = re.findall(
 107                 '<param name="flashvars"[^>]*value="[^"]*?url=(https?://www\.ign\.com/videos/.*?)["&]',
 108                 webpage)
 109             if multiple_urls:
 110                 entries = [self.url_result(u, ie='IGN') for u in multiple_urls]
 111                 return {
 112                     '_type': 'playlist',
 113                     'id': name_or_id,
 114                     'entries': entries,
 115                 }
 116
 117         video_id = self._find_video_id(webpage)
 118         if not video_id:
 119             return self.url_result(self._search_regex(self._EMBED_RE, webpage, 'embed url'))
 120         return self._get_video_info(video_id)
 121
 122     def _get_video_info(self, video_id):
 123         api_data = self._download_json(self._API_URL_TEMPLATE % video_id, video_id)
 124
 125         formats = []
 126         m3u8_url = api_data['refs'].get('m3uUrl')
 127         if m3u8_url:
 128             formats.extend(self._extract_m3u8_formats(m3u8_url, video_id))
 129         f4m_url = api_data['refs'].get('f4mUrl')
 130         if f4m_url:
 131             formats.extend(self._extract_f4m_formats(f4m_url, video_id))
 132         for asset in api_data['assets']:
 133             formats.append({
 134                 'url': asset['url'],
 135                 'tbr': asset.get('actual_bitrate_kbps'),
 136                 'fps': asset.get('frame_rate'),
 137                 'height': int_or_none(asset.get('height')),
 138                 'width': int_or_none(asset.get('width')),
 139             })
 140         self._sort_formats(formats)
 141
 142         thumbnails = []
 143         for thumbnail in api_data['thumbnails']:
 144             thumbnails.append({'url': thumbnail['url']})
 145
 146         metadata = api_data['metadata']
 147
 148         return {
 149             'id': api_data.get('videoId') or video_id,
 150             'title': metadata.get('longTitle') or metadata.get('name') or metadata.get['title'],
 151             'description': metadata.get('description'),
 152             'timestamp': parse_iso8601(metadata.get('publishDate')),
 153             'duration': int_or_none(metadata.get('duration')),
 154             'display_id': metadata.get('slug') or video_id,
 155             'thumbnails': thumbnails,
 156             'formats': formats,
 157         }
 158
 159
 160 class OneUPIE(IGNIE):
 161     _VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)\.html'
 162     IE_NAME = '1up.com'
 163
 164     _TESTS = [{
 165         'url': 'http://gamevideos.1up.com/video/id/34976.html',
 166         'md5': 'c9cc69e07acb675c31a16719f909e347',
 167         'info_dict': {
 168             'id': '34976',
 169             'ext': 'mp4',
 170             'title': 'Sniper Elite V2 - Trailer',
 171             'description': 'md5:bf0516c5ee32a3217aa703e9b1bc7826',
 172             'timestamp': 1313099220,
 173             'upload_date': '20110811',
 174         }
 175     }]
 176
 177     def _real_extract(self, url):
 178         mobj = re.match(self._VALID_URL, url)
 179         result = super(OneUPIE, self)._real_extract(url)
 180         result['id'] = mobj.group('name_or_id')
 181         return result
 182
 183
 184 class PCMagIE(IGNIE):
 185     _VALID_URL = r'https?://(?:www\.)?pcmag\.com/(?P<type>videos|article2)(/.+)?/(?P<name_or_id>.+)'
 186     IE_NAME = 'pcmag'
 187
 188     _EMBED_RE = r'iframe.setAttribute\("src",\s*__util.objToUrlString\("http://widgets\.ign\.com/video/embed/content.html?[^"]*url=([^"]+)["&]'
 189
 190     _TESTS = [{
 191         'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
 192         'md5': '212d6154fd0361a2781075f1febbe9ad',
 193         'info_dict': {
 194             'id': 'ee10d774b508c9b8ec07e763b9125b91',
 195             'ext': 'mp4',
 196             'title': '010615_What\'s New Now: Is GoGo Snooping on Your Data?',
 197             'description': 'md5:a7071ae64d2f68cc821c729d4ded6bb3',
 198             'timestamp': 1420571160,
 199             'upload_date': '20150106',
 200         }
 201     },{
 202         'url': 'http://www.pcmag.com/article2/0,2817,2470156,00.asp',
 203         'md5': '94130c1ca07ba0adb6088350681f16c1',
 204         'info_dict': {
 205             'id': '042e560ba94823d43afcb12ddf7142ca',
 206             'ext': 'mp4',
 207             'title': 'HTC\'s Weird New Re Camera - What\'s New Now',
 208             'description': 'md5:53433c45df96d2ea5d0fda18be2ca908',
 209             'timestamp': 1412953920,
 210             'upload_date': '20141010',
 211         }
 212     }]