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_urlparse,
  15 )
  16 from ..utils import (
  17     encode_dict,
  18     ExtractorError,
  19     int_or_none,
  20     orderedSet,
  21     parse_duration,
  22     parse_iso8601,
  23     sanitized_Request,
  24 )
  25
  26
  27 class TwitchBaseIE(InfoExtractor):
  28     _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv'
  29
  30     _API_BASE = 'https://api.twitch.tv'
  31     _USHER_BASE = 'http://usher.twitch.tv'
  32     _LOGIN_URL = 'http://www.twitch.tv/login'
  33     _NETRC_MACHINE = 'twitch'
  34
  35     def _handle_error(self, response):
  36         if not isinstance(response, dict):
  37             return
  38         error = response.get('error')
  39         if error:
  40             raise ExtractorError(
  41                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  42                 expected=True)
  43
  44     def _download_json(self, url, video_id, note='Downloading JSON metadata'):
  45         headers = {
  46             'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2',
  47             'X-Requested-With': 'XMLHttpRequest',
  48         }
  49         for cookie in self._downloader.cookiejar:
  50             if cookie.name == 'api_token':
  51                 headers['Twitch-Api-Token'] = cookie.value
  52         request = sanitized_Request(url, headers=headers)
  53         response = super(TwitchBaseIE, self)._download_json(request, video_id, note)
  54         self._handle_error(response)
  55         return response
  56
  57     def _real_initialize(self):
  58         self._login()
  59
  60     def _login(self):
  61         (username, password) = self._get_login_info()
  62         if username is None:
  63             return
  64
  65         login_page, handle = self._download_webpage_handle(
  66             self._LOGIN_URL, None, 'Downloading login page')
  67
  68         login_form = self._hidden_inputs(login_page)
  69
  70         login_form.update({
  71             'username': username,
  72             'password': password,
  73         })
  74
  75         redirect_url = handle.geturl()
  76
  77         post_url = self._search_regex(
  78             r'<form[^>]+action=(["\'])(?P<url>.+?)\1', login_page,
  79             'post url', default=redirect_url, group='url')
  80
  81         if not post_url.startswith('http'):
  82             post_url = compat_urlparse.urljoin(redirect_url, post_url)
  83
  84         request = sanitized_Request(
  85             post_url, compat_urllib_parse.urlencode(encode_dict(login_form)).encode('utf-8'))
  86         request.add_header('Referer', redirect_url)
  87         response = self._download_webpage(
  88             request, None, 'Logging in as %s' % username)
  89
  90         error_message = self._search_regex(
  91             r'<div[^>]+class="subwindow_notice"[^>]*>([^<]+)</div>',
  92             response, 'error message', default=None)
  93         if error_message:
  94             raise ExtractorError(
  95                 'Unable to login. Twitch said: %s' % error_message, expected=True)
  96
  97         if '>Reset your password<' in response:
  98             self.report_warning('Twitch asks you to reset your password, go to https://secure.twitch.tv/reset/submit')
  99
 100     def _prefer_source(self, formats):
 101         try:
 102             source = next(f for f in formats if f['format_id'] == 'Source')
 103             source['preference'] = 10
 104         except StopIteration:
 105             pass  # No Source stream present
 106         self._sort_formats(formats)
 107
 108
 109 class TwitchItemBaseIE(TwitchBaseIE):
 110     def _download_info(self, item, item_id):
 111         return self._extract_info(self._download_json(
 112             '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
 113             'Downloading %s info JSON' % self._ITEM_TYPE))
 114
 115     def _extract_media(self, item_id):
 116         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 117         response = self._download_json(
 118             '%s/api/videos/%s%s' % (self._API_BASE, self._ITEM_SHORTCUT, item_id), item_id,
 119             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 120         entries = []
 121         chunks = response['chunks']
 122         qualities = list(chunks.keys())
 123         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 124             formats = []
 125             for fmt_num, fragment_fmt in enumerate(fragment):
 126                 format_id = qualities[fmt_num]
 127                 fmt = {
 128                     'url': fragment_fmt['url'],
 129                     'format_id': format_id,
 130                     'quality': 1 if format_id == 'live' else 0,
 131                 }
 132                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 133                 if m:
 134                     fmt['height'] = int(m.group('height'))
 135                 formats.append(fmt)
 136             self._sort_formats(formats)
 137             entry = dict(info)
 138             entry['id'] = '%s_%d' % (entry['id'], num)
 139             entry['title'] = '%s part %d' % (entry['title'], num)
 140             entry['formats'] = formats
 141             entries.append(entry)
 142         return self.playlist_result(entries, info['id'], info['title'])
 143
 144     def _extract_info(self, info):
 145         return {
 146             'id': info['_id'],
 147             'title': info.get('title') or 'Untitled Broadcast',
 148             'description': info.get('description'),
 149             'duration': int_or_none(info.get('length')),
 150             'thumbnail': info.get('preview'),
 151             'uploader': info.get('channel', {}).get('display_name'),
 152             'uploader_id': info.get('channel', {}).get('name'),
 153             'timestamp': parse_iso8601(info.get('recorded_at')),
 154             'view_count': int_or_none(info.get('views')),
 155         }
 156
 157     def _real_extract(self, url):
 158         return self._extract_media(self._match_id(url))
 159
 160
 161 class TwitchVideoIE(TwitchItemBaseIE):
 162     IE_NAME = 'twitch:video'
 163     _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 164     _ITEM_TYPE = 'video'
 165     _ITEM_SHORTCUT = 'a'
 166
 167     _TEST = {
 168         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 169         'info_dict': {
 170             'id': 'a577357806',
 171             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 172         },
 173         'playlist_mincount': 12,
 174     }
 175
 176
 177 class TwitchChapterIE(TwitchItemBaseIE):
 178     IE_NAME = 'twitch:chapter'
 179     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 180     _ITEM_TYPE = 'chapter'
 181     _ITEM_SHORTCUT = 'c'
 182
 183     _TESTS = [{
 184         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 185         'info_dict': {
 186             'id': 'c5285812',
 187             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 188         },
 189         'playlist_mincount': 3,
 190     }, {
 191         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 192         'only_matching': True,
 193     }]
 194
 195
 196 class TwitchVodIE(TwitchItemBaseIE):
 197     IE_NAME = 'twitch:vod'
 198     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 199     _ITEM_TYPE = 'vod'
 200     _ITEM_SHORTCUT = 'v'
 201
 202     _TESTS = [{
 203         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 204         'info_dict': {
 205             'id': 'v6528877',
 206             'ext': 'mp4',
 207             'title': 'LCK Summer Split - Week 6 Day 1',
 208             'thumbnail': 're:^https?://.*\.jpg$',
 209             'duration': 17208,
 210             'timestamp': 1435131709,
 211             'upload_date': '20150624',
 212             'uploader': 'Riot Games',
 213             'uploader_id': 'riotgames',
 214             'view_count': int,
 215             'start_time': 310,
 216         },
 217         'params': {
 218             # m3u8 download
 219             'skip_download': True,
 220         },
 221     }, {
 222         # Untitled broadcast (title is None)
 223         'url': 'http://www.twitch.tv/belkao_o/v/11230755',
 224         'info_dict': {
 225             'id': 'v11230755',
 226             'ext': 'mp4',
 227             'title': 'Untitled Broadcast',
 228             'thumbnail': 're:^https?://.*\.jpg$',
 229             'duration': 1638,
 230             'timestamp': 1439746708,
 231             'upload_date': '20150816',
 232             'uploader': 'BelkAO_o',
 233             'uploader_id': 'belkao_o',
 234             'view_count': int,
 235         },
 236         'params': {
 237             # m3u8 download
 238             'skip_download': True,
 239         },
 240     }]
 241
 242     def _real_extract(self, url):
 243         item_id = self._match_id(url)
 244
 245         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 246         access_token = self._download_json(
 247             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
 248             'Downloading %s access token' % self._ITEM_TYPE)
 249
 250         formats = self._extract_m3u8_formats(
 251             '%s/vod/%s?%s' % (
 252                 self._USHER_BASE, item_id,
 253                 compat_urllib_parse.urlencode({
 254                     'allow_source': 'true',
 255                     'allow_audio_only': 'true',
 256                     'allow_spectre': 'true',
 257                     'player': 'twitchweb',
 258                     'nauth': access_token['token'],
 259                     'nauthsig': access_token['sig'],
 260                 })),
 261             item_id, 'mp4')
 262
 263         self._prefer_source(formats)
 264         info['formats'] = formats
 265
 266         parsed_url = compat_urllib_parse_urlparse(url)
 267         query = compat_parse_qs(parsed_url.query)
 268         if 't' in query:
 269             info['start_time'] = parse_duration(query['t'][0])
 270
 271         return info
 272
 273
 274 class TwitchPlaylistBaseIE(TwitchBaseIE):
 275     _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 276     _PAGE_LIMIT = 100
 277
 278     def _extract_playlist(self, channel_id):
 279         info = self._download_json(
 280             '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 281             channel_id, 'Downloading channel info JSON')
 282         channel_name = info.get('display_name') or info.get('name')
 283         entries = []
 284         offset = 0
 285         limit = self._PAGE_LIMIT
 286         broken_paging_detected = False
 287         counter_override = None
 288         for counter in itertools.count(1):
 289             response = self._download_json(
 290                 self._PLAYLIST_URL % (channel_id, offset, limit),
 291                 channel_id,
 292                 'Downloading %s videos JSON page %s'
 293                 % (self._PLAYLIST_TYPE, counter_override or counter))
 294             page_entries = self._extract_playlist_page(response)
 295             if not page_entries:
 296                 break
 297             total = int_or_none(response.get('_total'))
 298             # Since the beginning of March 2016 twitch's paging mechanism
 299             # is completely broken on the twitch side. It simply ignores
 300             # a limit and returns the whole offset number of videos.
 301             # Working around by just requesting all videos at once.
 302             if not broken_paging_detected and total and len(page_entries) > limit:
 303                 self.report_warning(
 304                     'Twitch paging is broken on twitch side, requesting all videos at once',
 305                     channel_id)
 306                 broken_paging_detected = True
 307                 offset = total
 308                 counter_override = '(all at once)'
 309                 continue
 310             entries.extend(page_entries)
 311             if broken_paging_detected or total and len(page_entries) >= total:
 312                 break
 313             offset += limit
 314         return self.playlist_result(
 315             [self.url_result(entry) for entry in orderedSet(entries)],
 316             channel_id, channel_name)
 317
 318     def _extract_playlist_page(self, response):
 319         videos = response.get('videos')
 320         return [video['url'] for video in videos] if videos else []
 321
 322     def _real_extract(self, url):
 323         return self._extract_playlist(self._match_id(url))
 324
 325
 326 class TwitchProfileIE(TwitchPlaylistBaseIE):
 327     IE_NAME = 'twitch:profile'
 328     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 329     _PLAYLIST_TYPE = 'profile'
 330
 331     _TEST = {
 332         'url': 'http://www.twitch.tv/vanillatv/profile',
 333         'info_dict': {
 334             'id': 'vanillatv',
 335             'title': 'VanillaTV',
 336         },
 337         'playlist_mincount': 412,
 338     }
 339
 340
 341 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 342     IE_NAME = 'twitch:past_broadcasts'
 343     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 344     _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
 345     _PLAYLIST_TYPE = 'past broadcasts'
 346
 347     _TEST = {
 348         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 349         'info_dict': {
 350             'id': 'spamfish',
 351             'title': 'Spamfish',
 352         },
 353         'playlist_mincount': 54,
 354     }
 355
 356
 357 class TwitchBookmarksIE(TwitchPlaylistBaseIE):
 358     IE_NAME = 'twitch:bookmarks'
 359     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 360     _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 361     _PLAYLIST_TYPE = 'bookmarks'
 362
 363     _TEST = {
 364         'url': 'http://www.twitch.tv/ognos/profile/bookmarks',
 365         'info_dict': {
 366             'id': 'ognos',
 367             'title': 'Ognos',
 368         },
 369         'playlist_mincount': 3,
 370     }
 371
 372     def _extract_playlist_page(self, response):
 373         entries = []
 374         for bookmark in response.get('bookmarks', []):
 375             video = bookmark.get('video')
 376             if not video:
 377                 continue
 378             entries.append(video['url'])
 379         return entries
 380
 381
 382 class TwitchStreamIE(TwitchBaseIE):
 383     IE_NAME = 'twitch:stream'
 384     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 385
 386     _TESTS = [{
 387         'url': 'http://www.twitch.tv/shroomztv',
 388         'info_dict': {
 389             'id': '12772022048',
 390             'display_id': 'shroomztv',
 391             'ext': 'mp4',
 392             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 393             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 394             'is_live': True,
 395             'timestamp': 1421928037,
 396             'upload_date': '20150122',
 397             'uploader': 'ShroomzTV',
 398             'uploader_id': 'shroomztv',
 399             'view_count': int,
 400         },
 401         'params': {
 402             # m3u8 download
 403             'skip_download': True,
 404         },
 405     }, {
 406         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 407         'only_matching': True,
 408     }]
 409
 410     def _real_extract(self, url):
 411         channel_id = self._match_id(url)
 412
 413         stream = self._download_json(
 414             '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id,
 415             'Downloading stream JSON').get('stream')
 416
 417         # Fallback on profile extraction if stream is offline
 418         if not stream:
 419             return self.url_result(
 420                 'http://www.twitch.tv/%s/profile' % channel_id,
 421                 'TwitchProfile', channel_id)
 422
 423         # Channel name may be typed if different case than the original channel name
 424         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 425         # an invalid m3u8 URL. Working around by use of original channel name from stream
 426         # JSON and fallback to lowercase if it's not available.
 427         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 428
 429         access_token = self._download_json(
 430             '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id,
 431             'Downloading channel access token')
 432
 433         query = {
 434             'allow_source': 'true',
 435             'allow_audio_only': 'true',
 436             'p': random.randint(1000000, 10000000),
 437             'player': 'twitchweb',
 438             'segment_preference': '4',
 439             'sig': access_token['sig'].encode('utf-8'),
 440             'token': access_token['token'].encode('utf-8'),
 441         }
 442         formats = self._extract_m3u8_formats(
 443             '%s/api/channel/hls/%s.m3u8?%s'
 444             % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query)),
 445             channel_id, 'mp4')
 446         self._prefer_source(formats)
 447
 448         view_count = stream.get('viewers')
 449         timestamp = parse_iso8601(stream.get('created_at'))
 450
 451         channel = stream['channel']
 452         title = self._live_title(channel.get('display_name') or channel.get('name'))
 453         description = channel.get('status')
 454
 455         thumbnails = []
 456         for thumbnail_key, thumbnail_url in stream['preview'].items():
 457             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 458             if not m:
 459                 continue
 460             thumbnails.append({
 461                 'url': thumbnail_url,
 462                 'width': int(m.group('width')),
 463                 'height': int(m.group('height')),
 464             })
 465
 466         return {
 467             'id': compat_str(stream['_id']),
 468             'display_id': channel_id,
 469             'title': title,
 470             'description': description,
 471             'thumbnails': thumbnails,
 472             'uploader': channel.get('display_name'),
 473             'uploader_id': channel.get('name'),
 474             'timestamp': timestamp,
 475             'view_count': view_count,
 476             'formats': formats,
 477             'is_live': True,
 478         }