_ 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_parse_qs,
  11     compat_str,
  12     compat_urllib_parse,
  13     compat_urllib_parse_urlparse,
  14     compat_urllib_request,
  15     compat_urlparse,
  16 )
  17 from ..utils import (
  18     ExtractorError,
  19     int_or_none,
  20     parse_duration,
  21     parse_iso8601,
  22 )
  23
  24
  25 class TwitchBaseIE(InfoExtractor):
  26     _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv'
  27
  28     _API_BASE = 'https://api.twitch.tv'
  29     _USHER_BASE = 'http://usher.twitch.tv'
  30     _LOGIN_URL = 'https://secure.twitch.tv/login'
  31     _LOGIN_POST_URL = 'https://passport.twitch.tv/authentications/new'
  32     _NETRC_MACHINE = 'twitch'
  33
  34     def _handle_error(self, response):
  35         if not isinstance(response, dict):
  36             return
  37         error = response.get('error')
  38         if error:
  39             raise ExtractorError(
  40                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  41                 expected=True)
  42
  43     def _download_json(self, url, video_id, note='Downloading JSON metadata'):
  44         headers = {
  45             'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2',
  46             'X-Requested-With': 'XMLHttpRequest',
  47         }
  48         for cookie in self._downloader.cookiejar:
  49             if cookie.name == 'api_token':
  50                 headers['Twitch-Api-Token'] = cookie.value
  51         request = compat_urllib_request.Request(url, headers=headers)
  52         response = super(TwitchBaseIE, self)._download_json(request, video_id, note)
  53         self._handle_error(response)
  54         return response
  55
  56     def _real_initialize(self):
  57         self._login()
  58
  59     def _login(self):
  60         (username, password) = self._get_login_info()
  61         if username is None:
  62             return
  63
  64         login_page = self._download_webpage(
  65             self._LOGIN_URL, None, 'Downloading login page')
  66
  67         login_form = self._hidden_inputs(login_page)
  68
  69         login_form.update({
  70             'login': username.encode('utf-8'),
  71             'password': password.encode('utf-8'),
  72         })
  73
  74         post_url = self._search_regex(
  75             r'<form[^>]+action=(["\'])(?P<url>.+?)\1', login_page,
  76             'post url', default=self._LOGIN_POST_URL, group='url')
  77
  78         if not post_url.startswith('http'):
  79             post_url = compat_urlparse.urljoin(self._LOGIN_URL, post_url)
  80
  81         request = compat_urllib_request.Request(
  82             post_url, compat_urllib_parse.urlencode(login_form).encode('utf-8'))
  83         request.add_header('Referer', self._LOGIN_URL)
  84         response = self._download_webpage(
  85             request, None, 'Logging in as %s' % username)
  86
  87         error_message = self._search_regex(
  88             r'<div[^>]+class="subwindow_notice"[^>]*>([^<]+)</div>',
  89             response, 'error message', default=None)
  90         if error_message:
  91             raise ExtractorError(
  92                 'Unable to login. Twitch said: %s' % error_message, expected=True)
  93
  94         if '>Reset your password<' in response:
  95             self.report_warning('Twitch asks you to reset your password, go to https://secure.twitch.tv/reset/submit')
  96
  97     def _prefer_source(self, formats):
  98         try:
  99             source = next(f for f in formats if f['format_id'] == 'Source')
 100             source['preference'] = 10
 101         except StopIteration:
 102             pass  # No Source stream present
 103         self._sort_formats(formats)
 104
 105
 106 class TwitchItemBaseIE(TwitchBaseIE):
 107     def _download_info(self, item, item_id):
 108         return self._extract_info(self._download_json(
 109             '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
 110             'Downloading %s info JSON' % self._ITEM_TYPE))
 111
 112     def _extract_media(self, item_id):
 113         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 114         response = self._download_json(
 115             '%s/api/videos/%s%s' % (self._API_BASE, self._ITEM_SHORTCUT, item_id), item_id,
 116             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 117         entries = []
 118         chunks = response['chunks']
 119         qualities = list(chunks.keys())
 120         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 121             formats = []
 122             for fmt_num, fragment_fmt in enumerate(fragment):
 123                 format_id = qualities[fmt_num]
 124                 fmt = {
 125                     'url': fragment_fmt['url'],
 126                     'format_id': format_id,
 127                     'quality': 1 if format_id == 'live' else 0,
 128                 }
 129                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 130                 if m:
 131                     fmt['height'] = int(m.group('height'))
 132                 formats.append(fmt)
 133             self._sort_formats(formats)
 134             entry = dict(info)
 135             entry['id'] = '%s_%d' % (entry['id'], num)
 136             entry['title'] = '%s part %d' % (entry['title'], num)
 137             entry['formats'] = formats
 138             entries.append(entry)
 139         return self.playlist_result(entries, info['id'], info['title'])
 140
 141     def _extract_info(self, info):
 142         return {
 143             'id': info['_id'],
 144             'title': info.get('title') or 'Untitled Broadcast',
 145             'description': info.get('description'),
 146             'duration': int_or_none(info.get('length')),
 147             'thumbnail': info.get('preview'),
 148             'uploader': info.get('channel', {}).get('display_name'),
 149             'uploader_id': info.get('channel', {}).get('name'),
 150             'timestamp': parse_iso8601(info.get('recorded_at')),
 151             'view_count': int_or_none(info.get('views')),
 152         }
 153
 154     def _real_extract(self, url):
 155         return self._extract_media(self._match_id(url))
 156
 157
 158 class TwitchVideoIE(TwitchItemBaseIE):
 159     IE_NAME = 'twitch:video'
 160     _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 161     _ITEM_TYPE = 'video'
 162     _ITEM_SHORTCUT = 'a'
 163
 164     _TEST = {
 165         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 166         'info_dict': {
 167             'id': 'a577357806',
 168             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 169         },
 170         'playlist_mincount': 12,
 171     }
 172
 173
 174 class TwitchChapterIE(TwitchItemBaseIE):
 175     IE_NAME = 'twitch:chapter'
 176     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 177     _ITEM_TYPE = 'chapter'
 178     _ITEM_SHORTCUT = 'c'
 179
 180     _TESTS = [{
 181         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 182         'info_dict': {
 183             'id': 'c5285812',
 184             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 185         },
 186         'playlist_mincount': 3,
 187     }, {
 188         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 189         'only_matching': True,
 190     }]
 191
 192
 193 class TwitchVodIE(TwitchItemBaseIE):
 194     IE_NAME = 'twitch:vod'
 195     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 196     _ITEM_TYPE = 'vod'
 197     _ITEM_SHORTCUT = 'v'
 198
 199     _TESTS = [{
 200         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 201         'info_dict': {
 202             'id': 'v6528877',
 203             'ext': 'mp4',
 204             'title': 'LCK Summer Split - Week 6 Day 1',
 205             'thumbnail': 're:^https?://.*\.jpg$',
 206             'duration': 17208,
 207             'timestamp': 1435131709,
 208             'upload_date': '20150624',
 209             'uploader': 'Riot Games',
 210             'uploader_id': 'riotgames',
 211             'view_count': int,
 212             'start_time': 310,
 213         },
 214         'params': {
 215             # m3u8 download
 216             'skip_download': True,
 217         },
 218     }, {
 219         # Untitled broadcast (title is None)
 220         'url': 'http://www.twitch.tv/belkao_o/v/11230755',
 221         'info_dict': {
 222             'id': 'v11230755',
 223             'ext': 'mp4',
 224             'title': 'Untitled Broadcast',
 225             'thumbnail': 're:^https?://.*\.jpg$',
 226             'duration': 1638,
 227             'timestamp': 1439746708,
 228             'upload_date': '20150816',
 229             'uploader': 'BelkAO_o',
 230             'uploader_id': 'belkao_o',
 231             'view_count': int,
 232         },
 233         'params': {
 234             # m3u8 download
 235             'skip_download': True,
 236         },
 237     }]
 238
 239     def _real_extract(self, url):
 240         item_id = self._match_id(url)
 241         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 242         access_token = self._download_json(
 243             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
 244             'Downloading %s access token' % self._ITEM_TYPE)
 245         formats = self._extract_m3u8_formats(
 246             '%s/vod/%s?nauth=%s&nauthsig=%s&allow_source=true'
 247             % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']),
 248             item_id, 'mp4')
 249         self._prefer_source(formats)
 250         info['formats'] = formats
 251
 252         parsed_url = compat_urllib_parse_urlparse(url)
 253         query = compat_parse_qs(parsed_url.query)
 254         if 't' in query:
 255             info['start_time'] = parse_duration(query['t'][0])
 256
 257         return info
 258
 259
 260 class TwitchPlaylistBaseIE(TwitchBaseIE):
 261     _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 262     _PAGE_LIMIT = 100
 263
 264     def _extract_playlist(self, channel_id):
 265         info = self._download_json(
 266             '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 267             channel_id, 'Downloading channel info JSON')
 268         channel_name = info.get('display_name') or info.get('name')
 269         entries = []
 270         offset = 0
 271         limit = self._PAGE_LIMIT
 272         for counter in itertools.count(1):
 273             response = self._download_json(
 274                 self._PLAYLIST_URL % (channel_id, offset, limit),
 275                 channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter))
 276             page_entries = self._extract_playlist_page(response)
 277             if not page_entries:
 278                 break
 279             entries.extend(page_entries)
 280             offset += limit
 281         return self.playlist_result(
 282             [self.url_result(entry) for entry in set(entries)],
 283             channel_id, channel_name)
 284
 285     def _extract_playlist_page(self, response):
 286         videos = response.get('videos')
 287         return [video['url'] for video in videos] if videos else []
 288
 289     def _real_extract(self, url):
 290         return self._extract_playlist(self._match_id(url))
 291
 292
 293 class TwitchProfileIE(TwitchPlaylistBaseIE):
 294     IE_NAME = 'twitch:profile'
 295     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 296     _PLAYLIST_TYPE = 'profile'
 297
 298     _TEST = {
 299         'url': 'http://www.twitch.tv/vanillatv/profile',
 300         'info_dict': {
 301             'id': 'vanillatv',
 302             'title': 'VanillaTV',
 303         },
 304         'playlist_mincount': 412,
 305     }
 306
 307
 308 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 309     IE_NAME = 'twitch:past_broadcasts'
 310     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 311     _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
 312     _PLAYLIST_TYPE = 'past broadcasts'
 313
 314     _TEST = {
 315         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 316         'info_dict': {
 317             'id': 'spamfish',
 318             'title': 'Spamfish',
 319         },
 320         'playlist_mincount': 54,
 321     }
 322
 323
 324 class TwitchBookmarksIE(TwitchPlaylistBaseIE):
 325     IE_NAME = 'twitch:bookmarks'
 326     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 327     _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 328     _PLAYLIST_TYPE = 'bookmarks'
 329
 330     _TEST = {
 331         'url': 'http://www.twitch.tv/ognos/profile/bookmarks',
 332         'info_dict': {
 333             'id': 'ognos',
 334             'title': 'Ognos',
 335         },
 336         'playlist_mincount': 3,
 337     }
 338
 339     def _extract_playlist_page(self, response):
 340         entries = []
 341         for bookmark in response.get('bookmarks', []):
 342             video = bookmark.get('video')
 343             if not video:
 344                 continue
 345             entries.append(video['url'])
 346         return entries
 347
 348
 349 class TwitchStreamIE(TwitchBaseIE):
 350     IE_NAME = 'twitch:stream'
 351     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 352
 353     _TESTS = [{
 354         'url': 'http://www.twitch.tv/shroomztv',
 355         'info_dict': {
 356             'id': '12772022048',
 357             'display_id': 'shroomztv',
 358             'ext': 'mp4',
 359             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 360             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 361             'is_live': True,
 362             'timestamp': 1421928037,
 363             'upload_date': '20150122',
 364             'uploader': 'ShroomzTV',
 365             'uploader_id': 'shroomztv',
 366             'view_count': int,
 367         },
 368         'params': {
 369             # m3u8 download
 370             'skip_download': True,
 371         },
 372     }, {
 373         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 374         'only_matching': True,
 375     }]
 376
 377     def _real_extract(self, url):
 378         channel_id = self._match_id(url)
 379
 380         stream = self._download_json(
 381             '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id,
 382             'Downloading stream JSON').get('stream')
 383
 384         # Fallback on profile extraction if stream is offline
 385         if not stream:
 386             return self.url_result(
 387                 'http://www.twitch.tv/%s/profile' % channel_id,
 388                 'TwitchProfile', channel_id)
 389
 390         # Channel name may be typed if different case than the original channel name
 391         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 392         # an invalid m3u8 URL. Working around by use of original channel name from stream
 393         # JSON and fallback to lowercase if it's not available.
 394         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 395
 396         access_token = self._download_json(
 397             '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id,
 398             'Downloading channel access token')
 399
 400         query = {
 401             'allow_source': 'true',
 402             'p': random.randint(1000000, 10000000),
 403             'player': 'twitchweb',
 404             'segment_preference': '4',
 405             'sig': access_token['sig'].encode('utf-8'),
 406             'token': access_token['token'].encode('utf-8'),
 407         }
 408         formats = self._extract_m3u8_formats(
 409             '%s/api/channel/hls/%s.m3u8?%s'
 410             % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query)),
 411             channel_id, 'mp4')
 412         self._prefer_source(formats)
 413
 414         view_count = stream.get('viewers')
 415         timestamp = parse_iso8601(stream.get('created_at'))
 416
 417         channel = stream['channel']
 418         title = self._live_title(channel.get('display_name') or channel.get('name'))
 419         description = channel.get('status')
 420
 421         thumbnails = []
 422         for thumbnail_key, thumbnail_url in stream['preview'].items():
 423             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 424             if not m:
 425                 continue
 426             thumbnails.append({
 427                 'url': thumbnail_url,
 428                 'width': int(m.group('width')),
 429                 'height': int(m.group('height')),
 430             })
 431
 432         return {
 433             'id': compat_str(stream['_id']),
 434             'display_id': channel_id,
 435             'title': title,
 436             'description': description,
 437             'thumbnails': thumbnails,
 438             'uploader': channel.get('display_name'),
 439             'uploader_id': channel.get('name'),
 440             'timestamp': timestamp,
 441             'view_count': view_count,
 442             'formats': formats,
 443             'is_live': True,
 444         }