Raphaël G. Git Repositories - youtubedl/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                 'uploader_id': 'cberidon@ign.com',
  36             }
  37         },
  38         {
  39             'url': 'http://me.ign.com/en/feature/15775/100-little-things-in-gta-5-that-will-blow-your-mind',
  40             'info_dict': {
  41                 'id': '100-little-things-in-gta-5-that-will-blow-your-mind',
  42             },
  43             'playlist': [
  44                 {
  45                     'info_dict': {
  46                         'id': '5ebbd138523268b93c9141af17bec937',
  47                         'ext': 'mp4',
  48                         'title': 'GTA 5 Video Review',
  49                         'description': 'Rockstar drops the mic on this generation of games. Watch our review of the masterly Grand Theft Auto V.',
  50                         'timestamp': 1379339880,
  51                         'upload_date': '20130916',
  52                         'uploader_id': 'danieljkrupa@gmail.com',
  53                     },
  54                 },
  55                 {
  56                     'info_dict': {
  57                         'id': '638672ee848ae4ff108df2a296418ee2',
  58                         'ext': 'mp4',
  59                         'title': '26 Twisted Moments from GTA 5 in Slow Motion',
  60                         'description': 'The twisted beauty of GTA 5 in stunning slow motion.',
  61                         'timestamp': 1386878820,
  62                         'upload_date': '20131212',
  63                         'uploader_id': 'togilvie@ign.com',
  64                     },
  65                 },
  66             ],
  67             'params': {
  68                 'skip_download': True,
  69             },
  70         },
  71         {
  72             'url': 'http://www.ign.com/articles/2014/08/15/rewind-theater-wild-trailer-gamescom-2014?watch',
  73             'md5': '618fedb9c901fd086f6f093564ef8558',
  74             'info_dict': {
  75                 'id': '078fdd005f6d3c02f63d795faa1b984f',
  76                 'ext': 'mp4',
  77                 'title': 'Rewind Theater - Wild Trailer Gamescom 2014',
  78                 'description': 'Brian and Jared explore Michel Ancel\'s captivating new preview.',
  79                 'timestamp': 1408047180,
  80                 'upload_date': '20140814',
  81                 'uploader_id': 'jamesduggan1990@gmail.com',
  82             },
  83         },
  84         {
  85             'url': 'http://me.ign.com/en/videos/112203/video/how-hitman-aims-to-be-different-than-every-other-s',
  86             'only_matching': True,
  87         },
  88         {
  89             'url': 'http://me.ign.com/ar/angry-birds-2/106533/video/lrd-ldyy-lwl-lfylm-angry-birds',
  90             'only_matching': True,
  91         },
  92         {
  93             # videoId pattern
  94             'url': 'http://www.ign.com/articles/2017/06/08/new-ducktales-short-donalds-birthday-doesnt-go-as-planned',
  95             'only_matching': True,
  96         },
  97     ]
  98
  99     def _find_video_id(self, webpage):
 100         res_id = [
 101             r'"video_id"\s*:\s*"(.*?)"',
 102             r'class="hero-poster[^"]*?"[^>]*id="(.+?)"',
 103             r'data-video-id="(.+?)"',
 104             r'<object id="vid_(.+?)"',
 105             r'<meta name="og:image" content=".*/(.+?)-(.+?)/.+.jpg"',
 106             r'videoId&quot;\s*:\s*&quot;(.+?)&quot;',
 107             r'videoId["\']\s*:\s*["\']([^"\']+?)["\']',
 108         ]
 109         return self._search_regex(res_id, webpage, 'video id', default=None)
 110
 111     def _real_extract(self, url):
 112         mobj = re.match(self._VALID_URL, url)
 113         name_or_id = mobj.group('name_or_id')
 114         page_type = mobj.group('type')
 115         webpage = self._download_webpage(url, name_or_id)
 116         if page_type != 'video':
 117             multiple_urls = re.findall(
 118                 r'<param name="flashvars"[^>]*value="[^"]*?url=(https?://www\.ign\.com/videos/.*?)["&]',
 119                 webpage)
 120             if multiple_urls:
 121                 entries = [self.url_result(u, ie='IGN') for u in multiple_urls]
 122                 return {
 123                     '_type': 'playlist',
 124                     'id': name_or_id,
 125                     'entries': entries,
 126                 }
 127
 128         video_id = self._find_video_id(webpage)
 129         if not video_id:
 130             return self.url_result(self._search_regex(
 131                 self._EMBED_RE, webpage, 'embed url'))
 132         return self._get_video_info(video_id)
 133
 134     def _get_video_info(self, video_id):
 135         api_data = self._download_json(
 136             self._API_URL_TEMPLATE % video_id, video_id)
 137
 138         formats = []
 139         m3u8_url = api_data['refs'].get('m3uUrl')
 140         if m3u8_url:
 141             formats.extend(self._extract_m3u8_formats(
 142                 m3u8_url, video_id, 'mp4', 'm3u8_native',
 143                 m3u8_id='hls', fatal=False))
 144         f4m_url = api_data['refs'].get('f4mUrl')
 145         if f4m_url:
 146             formats.extend(self._extract_f4m_formats(
 147                 f4m_url, video_id, f4m_id='hds', fatal=False))
 148         for asset in api_data['assets']:
 149             formats.append({
 150                 'url': asset['url'],
 151                 'tbr': asset.get('actual_bitrate_kbps'),
 152                 'fps': asset.get('frame_rate'),
 153                 'height': int_or_none(asset.get('height')),
 154                 'width': int_or_none(asset.get('width')),
 155             })
 156         self._sort_formats(formats)
 157
 158         thumbnails = [{
 159             'url': thumbnail['url']
 160         } for thumbnail in api_data.get('thumbnails', [])]
 161
 162         metadata = api_data['metadata']
 163
 164         return {
 165             'id': api_data.get('videoId') or video_id,
 166             'title': metadata.get('longTitle') or metadata.get('name') or metadata.get['title'],
 167             'description': metadata.get('description'),
 168             'timestamp': parse_iso8601(metadata.get('publishDate')),
 169             'duration': int_or_none(metadata.get('duration')),
 170             'display_id': metadata.get('slug') or video_id,
 171             'uploader_id': metadata.get('creator'),
 172             'thumbnails': thumbnails,
 173             'formats': formats,
 174         }
 175
 176
 177 class OneUPIE(IGNIE):
 178     _VALID_URL = r'https?://gamevideos\.1up\.com/(?P<type>video)/id/(?P<name_or_id>.+)\.html'
 179     IE_NAME = '1up.com'
 180
 181     _TESTS = [{
 182         'url': 'http://gamevideos.1up.com/video/id/34976.html',
 183         'md5': 'c9cc69e07acb675c31a16719f909e347',
 184         'info_dict': {
 185             'id': '34976',
 186             'ext': 'mp4',
 187             'title': 'Sniper Elite V2 - Trailer',
 188             'description': 'md5:bf0516c5ee32a3217aa703e9b1bc7826',
 189             'timestamp': 1313099220,
 190             'upload_date': '20110811',
 191             'uploader_id': 'IGN',
 192         }
 193     }]
 194
 195     def _real_extract(self, url):
 196         mobj = re.match(self._VALID_URL, url)
 197         result = super(OneUPIE, self)._real_extract(url)
 198         result['id'] = mobj.group('name_or_id')
 199         return result
 200
 201
 202 class PCMagIE(IGNIE):
 203     _VALID_URL = r'https?://(?:www\.)?pcmag\.com/(?P<type>videos|article2)(/.+)?/(?P<name_or_id>.+)'
 204     IE_NAME = 'pcmag'
 205
 206     _EMBED_RE = r'iframe.setAttribute\("src",\s*__util.objToUrlString\("http://widgets\.ign\.com/video/embed/content.html?[^"]*url=([^"]+)["&]'
 207
 208     _TESTS = [{
 209         'url': 'http://www.pcmag.com/videos/2015/01/06/010615-whats-new-now-is-gogo-snooping-on-your-data',
 210         'md5': '212d6154fd0361a2781075f1febbe9ad',
 211         'info_dict': {
 212             'id': 'ee10d774b508c9b8ec07e763b9125b91',
 213             'ext': 'mp4',
 214             'title': '010615_What\'s New Now: Is GoGo Snooping on Your Data?',
 215             'description': 'md5:a7071ae64d2f68cc821c729d4ded6bb3',
 216             'timestamp': 1420571160,
 217             'upload_date': '20150106',
 218             'uploader_id': 'cozzipix@gmail.com',
 219         }
 220     }, {
 221         'url': 'http://www.pcmag.com/article2/0,2817,2470156,00.asp',
 222         'md5': '94130c1ca07ba0adb6088350681f16c1',
 223         'info_dict': {
 224             'id': '042e560ba94823d43afcb12ddf7142ca',
 225             'ext': 'mp4',
 226             'title': 'HTC\'s Weird New Re Camera - What\'s New Now',
 227             'description': 'md5:53433c45df96d2ea5d0fda18be2ca908',
 228             'timestamp': 1412953920,
 229             'upload_date': '20141010',
 230             'uploader_id': 'chris_snyder@pcmag.com',
 231         }
 232     }]