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_parse_qs,
  11     compat_str,
  12     compat_urllib_parse,
  13     compat_urllib_parse_urlparse,
  14     compat_urllib_request,
  15 )
  16 from ..utils import (
  17     ExtractorError,
  18     parse_duration,
  19     parse_iso8601,
  20 )
  21
  22
  23 class TwitchBaseIE(InfoExtractor):
  24     _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv'
  25
  26     _API_BASE = 'https://api.twitch.tv'
  27     _USHER_BASE = 'http://usher.twitch.tv'
  28     _LOGIN_URL = 'https://secure.twitch.tv/login'
  29     _LOGIN_POST_URL = 'https://passport.twitch.tv/authorize'
  30     _NETRC_MACHINE = 'twitch'
  31
  32     def _handle_error(self, response):
  33         if not isinstance(response, dict):
  34             return
  35         error = response.get('error')
  36         if error:
  37             raise ExtractorError(
  38                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  39                 expected=True)
  40
  41     def _download_json(self, url, video_id, note='Downloading JSON metadata'):
  42         headers = {
  43             'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2',
  44             'X-Requested-With': 'XMLHttpRequest',
  45         }
  46         for cookie in self._downloader.cookiejar:
  47             if cookie.name == 'api_token':
  48                 headers['Twitch-Api-Token'] = cookie.value
  49         request = compat_urllib_request.Request(url, headers=headers)
  50         response = super(TwitchBaseIE, self)._download_json(request, video_id, note)
  51         self._handle_error(response)
  52         return response
  53
  54     def _real_initialize(self):
  55         self._login()
  56
  57     def _login(self):
  58         (username, password) = self._get_login_info()
  59         if username is None:
  60             return
  61
  62         login_page = self._download_webpage(
  63             self._LOGIN_URL, None, 'Downloading login page')
  64
  65         login_form = self._hidden_inputs(login_page)
  66
  67         login_form.update({
  68             'login': username.encode('utf-8'),
  69             'password': password.encode('utf-8'),
  70         })
  71
  72         request = compat_urllib_request.Request(
  73             self._LOGIN_POST_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8'))
  74         request.add_header('Referer', self._LOGIN_URL)
  75         response = self._download_webpage(
  76             request, None, 'Logging in as %s' % username)
  77
  78         error_message = self._search_regex(
  79             r'<div[^>]+class="subwindow_notice"[^>]*>([^<]+)</div>',
  80             response, 'error message', default=None)
  81         if error_message:
  82             raise ExtractorError(
  83                 'Unable to login. Twitch said: %s' % error_message, expected=True)
  84
  85         if '>Reset your password<' in response:
  86             self.report_warning('Twitch asks you to reset your password, go to https://secure.twitch.tv/reset/submit')
  87
  88     def _prefer_source(self, formats):
  89         try:
  90             source = next(f for f in formats if f['format_id'] == 'Source')
  91             source['preference'] = 10
  92         except StopIteration:
  93             pass  # No Source stream present
  94         self._sort_formats(formats)
  95
  96
  97 class TwitchItemBaseIE(TwitchBaseIE):
  98     def _download_info(self, item, item_id):
  99         return self._extract_info(self._download_json(
 100             '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
 101             'Downloading %s info JSON' % self._ITEM_TYPE))
 102
 103     def _extract_media(self, item_id):
 104         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 105         response = self._download_json(
 106             '%s/api/videos/%s%s' % (self._API_BASE, self._ITEM_SHORTCUT, item_id), item_id,
 107             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 108         entries = []
 109         chunks = response['chunks']
 110         qualities = list(chunks.keys())
 111         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 112             formats = []
 113             for fmt_num, fragment_fmt in enumerate(fragment):
 114                 format_id = qualities[fmt_num]
 115                 fmt = {
 116                     'url': fragment_fmt['url'],
 117                     'format_id': format_id,
 118                     'quality': 1 if format_id == 'live' else 0,
 119                 }
 120                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 121                 if m:
 122                     fmt['height'] = int(m.group('height'))
 123                 formats.append(fmt)
 124             self._sort_formats(formats)
 125             entry = dict(info)
 126             entry['id'] = '%s_%d' % (entry['id'], num)
 127             entry['title'] = '%s part %d' % (entry['title'], num)
 128             entry['formats'] = formats
 129             entries.append(entry)
 130         return self.playlist_result(entries, info['id'], info['title'])
 131
 132     def _extract_info(self, info):
 133         return {
 134             'id': info['_id'],
 135             'title': info['title'],
 136             'description': info['description'],
 137             'duration': info['length'],
 138             'thumbnail': info['preview'],
 139             'uploader': info['channel']['display_name'],
 140             'uploader_id': info['channel']['name'],
 141             'timestamp': parse_iso8601(info['recorded_at']),
 142             'view_count': info['views'],
 143         }
 144
 145     def _real_extract(self, url):
 146         return self._extract_media(self._match_id(url))
 147
 148
 149 class TwitchVideoIE(TwitchItemBaseIE):
 150     IE_NAME = 'twitch:video'
 151     _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 152     _ITEM_TYPE = 'video'
 153     _ITEM_SHORTCUT = 'a'
 154
 155     _TEST = {
 156         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 157         'info_dict': {
 158             'id': 'a577357806',
 159             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 160         },
 161         'playlist_mincount': 12,
 162     }
 163
 164
 165 class TwitchChapterIE(TwitchItemBaseIE):
 166     IE_NAME = 'twitch:chapter'
 167     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 168     _ITEM_TYPE = 'chapter'
 169     _ITEM_SHORTCUT = 'c'
 170
 171     _TESTS = [{
 172         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 173         'info_dict': {
 174             'id': 'c5285812',
 175             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 176         },
 177         'playlist_mincount': 3,
 178     }, {
 179         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 180         'only_matching': True,
 181     }]
 182
 183
 184 class TwitchVodIE(TwitchItemBaseIE):
 185     IE_NAME = 'twitch:vod'
 186     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 187     _ITEM_TYPE = 'vod'
 188     _ITEM_SHORTCUT = 'v'
 189
 190     _TEST = {
 191         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 192         'info_dict': {
 193             'id': 'v6528877',
 194             'ext': 'mp4',
 195             'title': 'LCK Summer Split - Week 6 Day 1',
 196             'thumbnail': 're:^https?://.*\.jpg$',
 197             'duration': 17208,
 198             'timestamp': 1435131709,
 199             'upload_date': '20150624',
 200             'uploader': 'Riot Games',
 201             'uploader_id': 'riotgames',
 202             'view_count': int,
 203             'start_time': 310,
 204         },
 205         'params': {
 206             # m3u8 download
 207             'skip_download': True,
 208         },
 209     }
 210
 211     def _real_extract(self, url):
 212         item_id = self._match_id(url)
 213         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 214         access_token = self._download_json(
 215             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
 216             'Downloading %s access token' % self._ITEM_TYPE)
 217         formats = self._extract_m3u8_formats(
 218             '%s/vod/%s?nauth=%s&nauthsig=%s&allow_source=true'
 219             % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']),
 220             item_id, 'mp4')
 221         self._prefer_source(formats)
 222         info['formats'] = formats
 223
 224         parsed_url = compat_urllib_parse_urlparse(url)
 225         query = compat_parse_qs(parsed_url.query)
 226         if 't' in query:
 227             info['start_time'] = parse_duration(query['t'][0])
 228
 229         return info
 230
 231
 232 class TwitchPlaylistBaseIE(TwitchBaseIE):
 233     _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 234     _PAGE_LIMIT = 100
 235
 236     def _extract_playlist(self, channel_id):
 237         info = self._download_json(
 238             '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 239             channel_id, 'Downloading channel info JSON')
 240         channel_name = info.get('display_name') or info.get('name')
 241         entries = []
 242         offset = 0
 243         limit = self._PAGE_LIMIT
 244         for counter in itertools.count(1):
 245             response = self._download_json(
 246                 self._PLAYLIST_URL % (channel_id, offset, limit),
 247                 channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter))
 248             page_entries = self._extract_playlist_page(response)
 249             if not page_entries:
 250                 break
 251             entries.extend(page_entries)
 252             offset += limit
 253         return self.playlist_result(
 254             [self.url_result(entry) for entry in set(entries)],
 255             channel_id, channel_name)
 256
 257     def _extract_playlist_page(self, response):
 258         videos = response.get('videos')
 259         return [video['url'] for video in videos] if videos else []
 260
 261     def _real_extract(self, url):
 262         return self._extract_playlist(self._match_id(url))
 263
 264
 265 class TwitchProfileIE(TwitchPlaylistBaseIE):
 266     IE_NAME = 'twitch:profile'
 267     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 268     _PLAYLIST_TYPE = 'profile'
 269
 270     _TEST = {
 271         'url': 'http://www.twitch.tv/vanillatv/profile',
 272         'info_dict': {
 273             'id': 'vanillatv',
 274             'title': 'VanillaTV',
 275         },
 276         'playlist_mincount': 412,
 277     }
 278
 279
 280 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 281     IE_NAME = 'twitch:past_broadcasts'
 282     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 283     _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
 284     _PLAYLIST_TYPE = 'past broadcasts'
 285
 286     _TEST = {
 287         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 288         'info_dict': {
 289             'id': 'spamfish',
 290             'title': 'Spamfish',
 291         },
 292         'playlist_mincount': 54,
 293     }
 294
 295
 296 class TwitchBookmarksIE(TwitchPlaylistBaseIE):
 297     IE_NAME = 'twitch:bookmarks'
 298     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 299     _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 300     _PLAYLIST_TYPE = 'bookmarks'
 301
 302     _TEST = {
 303         'url': 'http://www.twitch.tv/ognos/profile/bookmarks',
 304         'info_dict': {
 305             'id': 'ognos',
 306             'title': 'Ognos',
 307         },
 308         'playlist_mincount': 3,
 309     }
 310
 311     def _extract_playlist_page(self, response):
 312         entries = []
 313         for bookmark in response.get('bookmarks', []):
 314             video = bookmark.get('video')
 315             if not video:
 316                 continue
 317             entries.append(video['url'])
 318         return entries
 319
 320
 321 class TwitchStreamIE(TwitchBaseIE):
 322     IE_NAME = 'twitch:stream'
 323     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 324
 325     _TESTS = [{
 326         'url': 'http://www.twitch.tv/shroomztv',
 327         'info_dict': {
 328             'id': '12772022048',
 329             'display_id': 'shroomztv',
 330             'ext': 'mp4',
 331             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 332             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 333             'is_live': True,
 334             'timestamp': 1421928037,
 335             'upload_date': '20150122',
 336             'uploader': 'ShroomzTV',
 337             'uploader_id': 'shroomztv',
 338             'view_count': int,
 339         },
 340         'params': {
 341             # m3u8 download
 342             'skip_download': True,
 343         },
 344     }, {
 345         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 346         'only_matching': True,
 347     }]
 348
 349     def _real_extract(self, url):
 350         channel_id = self._match_id(url)
 351
 352         stream = self._download_json(
 353             '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id,
 354             'Downloading stream JSON').get('stream')
 355
 356         # Fallback on profile extraction if stream is offline
 357         if not stream:
 358             return self.url_result(
 359                 'http://www.twitch.tv/%s/profile' % channel_id,
 360                 'TwitchProfile', channel_id)
 361
 362         # Channel name may be typed if different case than the original channel name
 363         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 364         # an invalid m3u8 URL. Working around by use of original channel name from stream
 365         # JSON and fallback to lowercase if it's not available.
 366         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 367
 368         access_token = self._download_json(
 369             '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id,
 370             'Downloading channel access token')
 371
 372         query = {
 373             'allow_source': 'true',
 374             'p': random.randint(1000000, 10000000),
 375             'player': 'twitchweb',
 376             'segment_preference': '4',
 377             'sig': access_token['sig'].encode('utf-8'),
 378             'token': access_token['token'].encode('utf-8'),
 379         }
 380         formats = self._extract_m3u8_formats(
 381             '%s/api/channel/hls/%s.m3u8?%s'
 382             % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query)),
 383             channel_id, 'mp4')
 384         self._prefer_source(formats)
 385
 386         view_count = stream.get('viewers')
 387         timestamp = parse_iso8601(stream.get('created_at'))
 388
 389         channel = stream['channel']
 390         title = self._live_title(channel.get('display_name') or channel.get('name'))
 391         description = channel.get('status')
 392
 393         thumbnails = []
 394         for thumbnail_key, thumbnail_url in stream['preview'].items():
 395             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 396             if not m:
 397                 continue
 398             thumbnails.append({
 399                 'url': thumbnail_url,
 400                 'width': int(m.group('width')),
 401                 'height': int(m.group('height')),
 402             })
 403
 404         return {
 405             'id': compat_str(stream['_id']),
 406             'display_id': channel_id,
 407             'title': title,
 408             'description': description,
 409             'thumbnails': thumbnails,
 410             'uploader': channel.get('display_name'),
 411             'uploader_id': channel.get('name'),
 412             'timestamp': timestamp,
 413             'view_count': view_count,
 414             'formats': formats,
 415             'is_live': True,
 416         }