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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import itertools
   5 import re
   6 import random
   7
   8 from .common import InfoExtractor
   9 from ..compat import (
  10     compat_str,
  11     compat_urllib_parse,
  12     compat_urllib_request,
  13 )
  14 from ..utils import (
  15     ExtractorError,
  16     parse_iso8601,
  17 )
  18
  19
  20 class TwitchBaseIE(InfoExtractor):
  21     _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv'
  22
  23     _API_BASE = 'https://api.twitch.tv'
  24     _USHER_BASE = 'http://usher.twitch.tv'
  25     _LOGIN_URL = 'https://secure.twitch.tv/user/login'
  26     _NETRC_MACHINE = 'twitch'
  27
  28     def _handle_error(self, response):
  29         if not isinstance(response, dict):
  30             return
  31         error = response.get('error')
  32         if error:
  33             raise ExtractorError(
  34                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  35                 expected=True)
  36
  37     def _download_json(self, url, video_id, note='Downloading JSON metadata'):
  38         headers = {
  39             'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2',
  40             'X-Requested-With': 'XMLHttpRequest',
  41         }
  42         for cookie in self._downloader.cookiejar:
  43             if cookie.name == 'api_token':
  44                 headers['Twitch-Api-Token'] = cookie.value
  45         request = compat_urllib_request.Request(url, headers=headers)
  46         response = super(TwitchBaseIE, self)._download_json(request, video_id, note)
  47         self._handle_error(response)
  48         return response
  49
  50     def _real_initialize(self):
  51         self._login()
  52
  53     def _login(self):
  54         (username, password) = self._get_login_info()
  55         if username is None:
  56             return
  57
  58         login_page = self._download_webpage(
  59             self._LOGIN_URL, None, 'Downloading login page')
  60
  61         authenticity_token = self._search_regex(
  62             r'<input name="authenticity_token" type="hidden" value="([^"]+)"',
  63             login_page, 'authenticity token')
  64
  65         login_form = {
  66             'utf8': '✓'.encode('utf-8'),
  67             'authenticity_token': authenticity_token,
  68             'redirect_on_login': '',
  69             'embed_form': 'false',
  70             'mp_source_action': '',
  71             'follow': '',
  72             'user[login]': username,
  73             'user[password]': password,
  74         }
  75
  76         request = compat_urllib_request.Request(
  77             self._LOGIN_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8'))
  78         request.add_header('Referer', self._LOGIN_URL)
  79         response = self._download_webpage(
  80             request, None, 'Logging in as %s' % username)
  81
  82         m = re.search(
  83             r"id=([\"'])login_error_message\1[^>]*>(?P<msg>[^<]+)", response)
  84         if m:
  85             raise ExtractorError(
  86                 'Unable to login: %s' % m.group('msg').strip(), expected=True)
  87
  88
  89 class TwitchItemBaseIE(TwitchBaseIE):
  90     def _download_info(self, item, item_id):
  91         return self._extract_info(self._download_json(
  92             '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
  93             'Downloading %s info JSON' % self._ITEM_TYPE))
  94
  95     def _extract_media(self, item_id):
  96         info = self._download_info(self._ITEM_SHORTCUT, item_id)
  97         response = self._download_json(
  98             '%s/api/videos/%s%s' % (self._API_BASE, self._ITEM_SHORTCUT, item_id), item_id,
  99             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 100         entries = []
 101         chunks = response['chunks']
 102         qualities = list(chunks.keys())
 103         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 104             formats = []
 105             for fmt_num, fragment_fmt in enumerate(fragment):
 106                 format_id = qualities[fmt_num]
 107                 fmt = {
 108                     'url': fragment_fmt['url'],
 109                     'format_id': format_id,
 110                     'quality': 1 if format_id == 'live' else 0,
 111                 }
 112                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 113                 if m:
 114                     fmt['height'] = int(m.group('height'))
 115                 formats.append(fmt)
 116             self._sort_formats(formats)
 117             entry = dict(info)
 118             entry['id'] = '%s_%d' % (entry['id'], num)
 119             entry['title'] = '%s part %d' % (entry['title'], num)
 120             entry['formats'] = formats
 121             entries.append(entry)
 122         return self.playlist_result(entries, info['id'], info['title'])
 123
 124     def _extract_info(self, info):
 125         return {
 126             'id': info['_id'],
 127             'title': info['title'],
 128             'description': info['description'],
 129             'duration': info['length'],
 130             'thumbnail': info['preview'],
 131             'uploader': info['channel']['display_name'],
 132             'uploader_id': info['channel']['name'],
 133             'timestamp': parse_iso8601(info['recorded_at']),
 134             'view_count': info['views'],
 135         }
 136
 137     def _real_extract(self, url):
 138         return self._extract_media(self._match_id(url))
 139
 140
 141 class TwitchVideoIE(TwitchItemBaseIE):
 142     IE_NAME = 'twitch:video'
 143     _VALID_URL = r'%s/[^/]+/b/(?P<id>[^/]+)' % TwitchBaseIE._VALID_URL_BASE
 144     _ITEM_TYPE = 'video'
 145     _ITEM_SHORTCUT = 'a'
 146
 147     _TEST = {
 148         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 149         'info_dict': {
 150             'id': 'a577357806',
 151             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 152         },
 153         'playlist_mincount': 12,
 154     }
 155
 156
 157 class TwitchChapterIE(TwitchItemBaseIE):
 158     IE_NAME = 'twitch:chapter'
 159     _VALID_URL = r'%s/[^/]+/c/(?P<id>[^/]+)' % TwitchBaseIE._VALID_URL_BASE
 160     _ITEM_TYPE = 'chapter'
 161     _ITEM_SHORTCUT = 'c'
 162
 163     _TESTS = [{
 164         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 165         'info_dict': {
 166             'id': 'c5285812',
 167             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 168         },
 169         'playlist_mincount': 3,
 170     }, {
 171         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 172         'only_matching': True,
 173     }]
 174
 175
 176 class TwitchVodIE(TwitchItemBaseIE):
 177     IE_NAME = 'twitch:vod'
 178     _VALID_URL = r'%s/[^/]+/v/(?P<id>[^/]+)' % TwitchBaseIE._VALID_URL_BASE
 179     _ITEM_TYPE = 'vod'
 180     _ITEM_SHORTCUT = 'v'
 181
 182     _TEST = {
 183         'url': 'http://www.twitch.tv/ksptv/v/3622000',
 184         'info_dict': {
 185             'id': 'v3622000',
 186             'ext': 'mp4',
 187             'title': '''KSPTV: Squadcast: "Everyone's on vacation so here's Dahud" Edition!''',
 188             'thumbnail': 're:^https?://.*\.jpg$',
 189             'duration': 6951,
 190             'timestamp': 1419028564,
 191             'upload_date': '20141219',
 192             'uploader': 'KSPTV',
 193             'uploader_id': 'ksptv',
 194             'view_count': int,
 195         },
 196         'params': {
 197             # m3u8 download
 198             'skip_download': True,
 199         },
 200     }
 201
 202     def _real_extract(self, url):
 203         item_id = self._match_id(url)
 204         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 205         access_token = self._download_json(
 206             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
 207             'Downloading %s access token' % self._ITEM_TYPE)
 208         formats = self._extract_m3u8_formats(
 209             '%s/vod/%s?nauth=%s&nauthsig=%s'
 210             % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']),
 211             item_id, 'mp4')
 212         info['formats'] = formats
 213         return info
 214
 215
 216 class TwitchPlaylistBaseIE(TwitchBaseIE):
 217     _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 218     _PAGE_LIMIT = 100
 219
 220     def _extract_playlist(self, channel_id):
 221         info = self._download_json(
 222             '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 223             channel_id, 'Downloading channel info JSON')
 224         channel_name = info.get('display_name') or info.get('name')
 225         entries = []
 226         offset = 0
 227         limit = self._PAGE_LIMIT
 228         for counter in itertools.count(1):
 229             response = self._download_json(
 230                 self._PLAYLIST_URL % (channel_id, offset, limit),
 231                 channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter))
 232             page_entries = self._extract_playlist_page(response)
 233             if not page_entries:
 234                 break
 235             entries.extend(page_entries)
 236             offset += limit
 237         return self.playlist_result(
 238             [self.url_result(entry) for entry in set(entries)],
 239             channel_id, channel_name)
 240
 241     def _extract_playlist_page(self, response):
 242         videos = response.get('videos')
 243         return [video['url'] for video in videos] if videos else []
 244
 245     def _real_extract(self, url):
 246         return self._extract_playlist(self._match_id(url))
 247
 248
 249 class TwitchProfileIE(TwitchPlaylistBaseIE):
 250     IE_NAME = 'twitch:profile'
 251     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 252     _PLAYLIST_TYPE = 'profile'
 253
 254     _TEST = {
 255         'url': 'http://www.twitch.tv/vanillatv/profile',
 256         'info_dict': {
 257             'id': 'vanillatv',
 258             'title': 'VanillaTV',
 259         },
 260         'playlist_mincount': 412,
 261     }
 262
 263
 264 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 265     IE_NAME = 'twitch:past_broadcasts'
 266     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 267     _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
 268     _PLAYLIST_TYPE = 'past broadcasts'
 269
 270     _TEST = {
 271         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 272         'info_dict': {
 273             'id': 'spamfish',
 274             'title': 'Spamfish',
 275         },
 276         'playlist_mincount': 54,
 277     }
 278
 279
 280 class TwitchBookmarksIE(TwitchPlaylistBaseIE):
 281     IE_NAME = 'twitch:bookmarks'
 282     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 283     _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 284     _PLAYLIST_TYPE = 'bookmarks'
 285
 286     _TEST = {
 287         'url': 'http://www.twitch.tv/ognos/profile/bookmarks',
 288         'info_dict': {
 289             'id': 'ognos',
 290             'title': 'Ognos',
 291         },
 292         'playlist_mincount': 3,
 293     }
 294
 295     def _extract_playlist_page(self, response):
 296         entries = []
 297         for bookmark in response.get('bookmarks', []):
 298             video = bookmark.get('video')
 299             if not video:
 300                 continue
 301             entries.append(video['url'])
 302         return entries
 303
 304
 305 class TwitchStreamIE(TwitchBaseIE):
 306     IE_NAME = 'twitch:stream'
 307     _VALID_URL = r'%s/(?P<id>[^/]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 308
 309     _TEST = {
 310         'url': 'http://www.twitch.tv/shroomztv',
 311         'info_dict': {
 312             'id': '12772022048',
 313             'display_id': 'shroomztv',
 314             'ext': 'mp4',
 315             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 316             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 317             'is_live': True,
 318             'timestamp': 1421928037,
 319             'upload_date': '20150122',
 320             'uploader': 'ShroomzTV',
 321             'uploader_id': 'shroomztv',
 322             'view_count': int,
 323         },
 324         'params': {
 325             # m3u8 download
 326             'skip_download': True,
 327         },
 328     }
 329
 330     def _real_extract(self, url):
 331         channel_id = self._match_id(url)
 332
 333         stream = self._download_json(
 334             '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id,
 335             'Downloading stream JSON').get('stream')
 336
 337         # Fallback on profile extraction if stream is offline
 338         if not stream:
 339             return self.url_result(
 340                 'http://www.twitch.tv/%s/profile' % channel_id,
 341                 'TwitchProfile', channel_id)
 342
 343         access_token = self._download_json(
 344             '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id,
 345             'Downloading channel access token')
 346
 347         query = {
 348             'allow_source': 'true',
 349             'p': random.randint(1000000, 10000000),
 350             'player': 'twitchweb',
 351             'segment_preference': '4',
 352             'sig': access_token['sig'],
 353             'token': access_token['token'],
 354         }
 355
 356         formats = self._extract_m3u8_formats(
 357             '%s/api/channel/hls/%s.m3u8?%s'
 358             % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query).encode('utf-8')),
 359             channel_id, 'mp4')
 360
 361         # prefer the 'source' stream, the others are limited to 30 fps
 362         def _sort_source(f):
 363             if f.get('m3u8_media') is not None and f['m3u8_media'].get('NAME') == 'Source':
 364                 return 1
 365             return 0
 366         formats = sorted(formats, key=_sort_source)
 367
 368         view_count = stream.get('viewers')
 369         timestamp = parse_iso8601(stream.get('created_at'))
 370
 371         channel = stream['channel']
 372         title = self._live_title(channel.get('display_name') or channel.get('name'))
 373         description = channel.get('status')
 374
 375         thumbnails = []
 376         for thumbnail_key, thumbnail_url in stream['preview'].items():
 377             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 378             if not m:
 379                 continue
 380             thumbnails.append({
 381                 'url': thumbnail_url,
 382                 'width': int(m.group('width')),
 383                 'height': int(m.group('height')),
 384             })
 385
 386         return {
 387             'id': compat_str(stream['_id']),
 388             'display_id': channel_id,
 389             'title': title,
 390             'description': description,
 391             'thumbnails': thumbnails,
 392             'uploader': channel.get('display_name'),
 393             'uploader_id': channel.get('name'),
 394             'timestamp': timestamp,
 395             'view_count': view_count,
 396             'formats': formats,
 397             'is_live': True,
 398         }