_ 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_HTTPError,
  11     compat_parse_qs,
  12     compat_str,
  13     compat_urllib_parse_urlencode,
  14     compat_urllib_parse_urlparse,
  15     compat_urlparse,
  16 )
  17 from ..utils import (
  18     clean_html,
  19     ExtractorError,
  20     int_or_none,
  21     js_to_json,
  22     orderedSet,
  23     parse_duration,
  24     parse_iso8601,
  25     urlencode_postdata,
  26 )
  27
  28
  29 class TwitchBaseIE(InfoExtractor):
  30     _VALID_URL_BASE = r'https?://(?:www\.)?twitch\.tv'
  31
  32     _API_BASE = 'https://api.twitch.tv'
  33     _USHER_BASE = 'https://usher.ttvnw.net'
  34     _LOGIN_URL = 'http://www.twitch.tv/login'
  35     _NETRC_MACHINE = 'twitch'
  36
  37     def _handle_error(self, response):
  38         if not isinstance(response, dict):
  39             return
  40         error = response.get('error')
  41         if error:
  42             raise ExtractorError(
  43                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  44                 expected=True)
  45
  46     def _call_api(self, path, item_id, note):
  47         headers = {
  48             'Referer': 'http://api.twitch.tv/crossdomain/receiver.html?v=2',
  49             'X-Requested-With': 'XMLHttpRequest',
  50         }
  51         for cookie in self._downloader.cookiejar:
  52             if cookie.name == 'api_token':
  53                 headers['Twitch-Api-Token'] = cookie.value
  54         response = self._download_json(
  55             '%s/%s' % (self._API_BASE, path), item_id, note)
  56         self._handle_error(response)
  57         return response
  58
  59     def _real_initialize(self):
  60         self._login()
  61
  62     def _login(self):
  63         (username, password) = self._get_login_info()
  64         if username is None:
  65             return
  66
  67         def fail(message):
  68             raise ExtractorError(
  69                 'Unable to login. Twitch said: %s' % message, expected=True)
  70
  71         login_page, handle = self._download_webpage_handle(
  72             self._LOGIN_URL, None, 'Downloading login page')
  73
  74         # Some TOR nodes and public proxies are blocked completely
  75         if 'blacklist_message' in login_page:
  76             fail(clean_html(login_page))
  77
  78         login_form = self._hidden_inputs(login_page)
  79
  80         login_form.update({
  81             'username': username,
  82             'password': password,
  83         })
  84
  85         redirect_url = handle.geturl()
  86
  87         post_url = self._search_regex(
  88             r'<form[^>]+action=(["\'])(?P<url>.+?)\1', login_page,
  89             'post url', default=redirect_url, group='url')
  90
  91         if not post_url.startswith('http'):
  92             post_url = compat_urlparse.urljoin(redirect_url, post_url)
  93
  94         headers = {'Referer': redirect_url}
  95
  96         try:
  97             response = self._download_json(
  98                 post_url, None, 'Logging in as %s' % username,
  99                 data=urlencode_postdata(login_form),
 100                 headers=headers)
 101         except ExtractorError as e:
 102             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400:
 103                 response = self._parse_json(
 104                     e.cause.read().decode('utf-8'), None)
 105                 fail(response['message'])
 106             raise
 107
 108         if response.get('redirect'):
 109             self._download_webpage(
 110                 response['redirect'], None, 'Downloading login redirect page',
 111                 headers=headers)
 112
 113     def _prefer_source(self, formats):
 114         try:
 115             source = next(f for f in formats if f['format_id'] == 'Source')
 116             source['preference'] = 10
 117         except StopIteration:
 118             pass  # No Source stream present
 119         self._sort_formats(formats)
 120
 121
 122 class TwitchItemBaseIE(TwitchBaseIE):
 123     def _download_info(self, item, item_id):
 124         return self._extract_info(self._call_api(
 125             'kraken/videos/%s%s' % (item, item_id), item_id,
 126             'Downloading %s info JSON' % self._ITEM_TYPE))
 127
 128     def _extract_media(self, item_id):
 129         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 130         response = self._call_api(
 131             'api/videos/%s%s' % (self._ITEM_SHORTCUT, item_id), item_id,
 132             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 133         entries = []
 134         chunks = response['chunks']
 135         qualities = list(chunks.keys())
 136         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 137             formats = []
 138             for fmt_num, fragment_fmt in enumerate(fragment):
 139                 format_id = qualities[fmt_num]
 140                 fmt = {
 141                     'url': fragment_fmt['url'],
 142                     'format_id': format_id,
 143                     'quality': 1 if format_id == 'live' else 0,
 144                 }
 145                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 146                 if m:
 147                     fmt['height'] = int(m.group('height'))
 148                 formats.append(fmt)
 149             self._sort_formats(formats)
 150             entry = dict(info)
 151             entry['id'] = '%s_%d' % (entry['id'], num)
 152             entry['title'] = '%s part %d' % (entry['title'], num)
 153             entry['formats'] = formats
 154             entries.append(entry)
 155         return self.playlist_result(entries, info['id'], info['title'])
 156
 157     def _extract_info(self, info):
 158         return {
 159             'id': info['_id'],
 160             'title': info.get('title') or 'Untitled Broadcast',
 161             'description': info.get('description'),
 162             'duration': int_or_none(info.get('length')),
 163             'thumbnail': info.get('preview'),
 164             'uploader': info.get('channel', {}).get('display_name'),
 165             'uploader_id': info.get('channel', {}).get('name'),
 166             'timestamp': parse_iso8601(info.get('recorded_at')),
 167             'view_count': int_or_none(info.get('views')),
 168         }
 169
 170     def _real_extract(self, url):
 171         return self._extract_media(self._match_id(url))
 172
 173
 174 class TwitchVideoIE(TwitchItemBaseIE):
 175     IE_NAME = 'twitch:video'
 176     _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 177     _ITEM_TYPE = 'video'
 178     _ITEM_SHORTCUT = 'a'
 179
 180     _TEST = {
 181         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 182         'info_dict': {
 183             'id': 'a577357806',
 184             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 185         },
 186         'playlist_mincount': 12,
 187         'skip': 'HTTP Error 404: Not Found',
 188     }
 189
 190
 191 class TwitchChapterIE(TwitchItemBaseIE):
 192     IE_NAME = 'twitch:chapter'
 193     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 194     _ITEM_TYPE = 'chapter'
 195     _ITEM_SHORTCUT = 'c'
 196
 197     _TESTS = [{
 198         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 199         'info_dict': {
 200             'id': 'c5285812',
 201             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 202         },
 203         'playlist_mincount': 3,
 204         'skip': 'HTTP Error 404: Not Found',
 205     }, {
 206         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 207         'only_matching': True,
 208     }]
 209
 210
 211 class TwitchVodIE(TwitchItemBaseIE):
 212     IE_NAME = 'twitch:vod'
 213     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 214     _ITEM_TYPE = 'vod'
 215     _ITEM_SHORTCUT = 'v'
 216
 217     _TESTS = [{
 218         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 219         'info_dict': {
 220             'id': 'v6528877',
 221             'ext': 'mp4',
 222             'title': 'LCK Summer Split - Week 6 Day 1',
 223             'thumbnail': 're:^https?://.*\.jpg$',
 224             'duration': 17208,
 225             'timestamp': 1435131709,
 226             'upload_date': '20150624',
 227             'uploader': 'Riot Games',
 228             'uploader_id': 'riotgames',
 229             'view_count': int,
 230             'start_time': 310,
 231         },
 232         'params': {
 233             # m3u8 download
 234             'skip_download': True,
 235         },
 236     }, {
 237         # Untitled broadcast (title is None)
 238         'url': 'http://www.twitch.tv/belkao_o/v/11230755',
 239         'info_dict': {
 240             'id': 'v11230755',
 241             'ext': 'mp4',
 242             'title': 'Untitled Broadcast',
 243             'thumbnail': 're:^https?://.*\.jpg$',
 244             'duration': 1638,
 245             'timestamp': 1439746708,
 246             'upload_date': '20150816',
 247             'uploader': 'BelkAO_o',
 248             'uploader_id': 'belkao_o',
 249             'view_count': int,
 250         },
 251         'params': {
 252             # m3u8 download
 253             'skip_download': True,
 254         },
 255     }]
 256
 257     def _real_extract(self, url):
 258         item_id = self._match_id(url)
 259
 260         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 261         access_token = self._call_api(
 262             'api/vods/%s/access_token' % item_id, item_id,
 263             'Downloading %s access token' % self._ITEM_TYPE)
 264
 265         formats = self._extract_m3u8_formats(
 266             '%s/vod/%s?%s' % (
 267                 self._USHER_BASE, item_id,
 268                 compat_urllib_parse_urlencode({
 269                     'allow_source': 'true',
 270                     'allow_audio_only': 'true',
 271                     'allow_spectre': 'true',
 272                     'player': 'twitchweb',
 273                     'nauth': access_token['token'],
 274                     'nauthsig': access_token['sig'],
 275                 })),
 276             item_id, 'mp4', entry_protocol='m3u8_native')
 277
 278         self._prefer_source(formats)
 279         info['formats'] = formats
 280
 281         parsed_url = compat_urllib_parse_urlparse(url)
 282         query = compat_parse_qs(parsed_url.query)
 283         if 't' in query:
 284             info['start_time'] = parse_duration(query['t'][0])
 285
 286         return info
 287
 288
 289 class TwitchPlaylistBaseIE(TwitchBaseIE):
 290     _PLAYLIST_PATH = 'kraken/channels/%s/videos/?offset=%d&limit=%d'
 291     _PAGE_LIMIT = 100
 292
 293     def _extract_playlist(self, channel_id):
 294         info = self._call_api(
 295             'kraken/channels/%s' % channel_id,
 296             channel_id, 'Downloading channel info JSON')
 297         channel_name = info.get('display_name') or info.get('name')
 298         entries = []
 299         offset = 0
 300         limit = self._PAGE_LIMIT
 301         broken_paging_detected = False
 302         counter_override = None
 303         for counter in itertools.count(1):
 304             response = self._call_api(
 305                 self._PLAYLIST_PATH % (channel_id, offset, limit),
 306                 channel_id,
 307                 'Downloading %s videos JSON page %s'
 308                 % (self._PLAYLIST_TYPE, counter_override or counter))
 309             page_entries = self._extract_playlist_page(response)
 310             if not page_entries:
 311                 break
 312             total = int_or_none(response.get('_total'))
 313             # Since the beginning of March 2016 twitch's paging mechanism
 314             # is completely broken on the twitch side. It simply ignores
 315             # a limit and returns the whole offset number of videos.
 316             # Working around by just requesting all videos at once.
 317             # Upd: pagination bug was fixed by twitch on 15.03.2016.
 318             if not broken_paging_detected and total and len(page_entries) > limit:
 319                 self.report_warning(
 320                     'Twitch pagination is broken on twitch side, requesting all videos at once',
 321                     channel_id)
 322                 broken_paging_detected = True
 323                 offset = total
 324                 counter_override = '(all at once)'
 325                 continue
 326             entries.extend(page_entries)
 327             if broken_paging_detected or total and len(page_entries) >= total:
 328                 break
 329             offset += limit
 330         return self.playlist_result(
 331             [self.url_result(entry) for entry in orderedSet(entries)],
 332             channel_id, channel_name)
 333
 334     def _extract_playlist_page(self, response):
 335         videos = response.get('videos')
 336         return [video['url'] for video in videos] if videos else []
 337
 338     def _real_extract(self, url):
 339         return self._extract_playlist(self._match_id(url))
 340
 341
 342 class TwitchProfileIE(TwitchPlaylistBaseIE):
 343     IE_NAME = 'twitch:profile'
 344     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 345     _PLAYLIST_TYPE = 'profile'
 346
 347     _TEST = {
 348         'url': 'http://www.twitch.tv/vanillatv/profile',
 349         'info_dict': {
 350             'id': 'vanillatv',
 351             'title': 'VanillaTV',
 352         },
 353         'playlist_mincount': 412,
 354     }
 355
 356
 357 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 358     IE_NAME = 'twitch:past_broadcasts'
 359     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 360     _PLAYLIST_PATH = TwitchPlaylistBaseIE._PLAYLIST_PATH + '&broadcasts=true'
 361     _PLAYLIST_TYPE = 'past broadcasts'
 362
 363     _TEST = {
 364         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 365         'info_dict': {
 366             'id': 'spamfish',
 367             'title': 'Spamfish',
 368         },
 369         'playlist_mincount': 54,
 370     }
 371
 372
 373 class TwitchStreamIE(TwitchBaseIE):
 374     IE_NAME = 'twitch:stream'
 375     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 376
 377     _TESTS = [{
 378         'url': 'http://www.twitch.tv/shroomztv',
 379         'info_dict': {
 380             'id': '12772022048',
 381             'display_id': 'shroomztv',
 382             'ext': 'mp4',
 383             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 384             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 385             'is_live': True,
 386             'timestamp': 1421928037,
 387             'upload_date': '20150122',
 388             'uploader': 'ShroomzTV',
 389             'uploader_id': 'shroomztv',
 390             'view_count': int,
 391         },
 392         'params': {
 393             # m3u8 download
 394             'skip_download': True,
 395         },
 396     }, {
 397         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 398         'only_matching': True,
 399     }]
 400
 401     def _real_extract(self, url):
 402         channel_id = self._match_id(url)
 403
 404         stream = self._call_api(
 405             'kraken/streams/%s' % channel_id, channel_id,
 406             'Downloading stream JSON').get('stream')
 407
 408         # Fallback on profile extraction if stream is offline
 409         if not stream:
 410             return self.url_result(
 411                 'http://www.twitch.tv/%s/profile' % channel_id,
 412                 'TwitchProfile', channel_id)
 413
 414         # Channel name may be typed if different case than the original channel name
 415         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 416         # an invalid m3u8 URL. Working around by use of original channel name from stream
 417         # JSON and fallback to lowercase if it's not available.
 418         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 419
 420         access_token = self._call_api(
 421             'api/channels/%s/access_token' % channel_id, channel_id,
 422             'Downloading channel access token')
 423
 424         query = {
 425             'allow_source': 'true',
 426             'allow_audio_only': 'true',
 427             'p': random.randint(1000000, 10000000),
 428             'player': 'twitchweb',
 429             'segment_preference': '4',
 430             'sig': access_token['sig'].encode('utf-8'),
 431             'token': access_token['token'].encode('utf-8'),
 432         }
 433         formats = self._extract_m3u8_formats(
 434             '%s/api/channel/hls/%s.m3u8?%s'
 435             % (self._USHER_BASE, channel_id, compat_urllib_parse_urlencode(query)),
 436             channel_id, 'mp4')
 437         self._prefer_source(formats)
 438
 439         view_count = stream.get('viewers')
 440         timestamp = parse_iso8601(stream.get('created_at'))
 441
 442         channel = stream['channel']
 443         title = self._live_title(channel.get('display_name') or channel.get('name'))
 444         description = channel.get('status')
 445
 446         thumbnails = []
 447         for thumbnail_key, thumbnail_url in stream['preview'].items():
 448             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 449             if not m:
 450                 continue
 451             thumbnails.append({
 452                 'url': thumbnail_url,
 453                 'width': int(m.group('width')),
 454                 'height': int(m.group('height')),
 455             })
 456
 457         return {
 458             'id': compat_str(stream['_id']),
 459             'display_id': channel_id,
 460             'title': title,
 461             'description': description,
 462             'thumbnails': thumbnails,
 463             'uploader': channel.get('display_name'),
 464             'uploader_id': channel.get('name'),
 465             'timestamp': timestamp,
 466             'view_count': view_count,
 467             'formats': formats,
 468             'is_live': True,
 469         }
 470
 471
 472 class TwitchClipsIE(InfoExtractor):
 473     IE_NAME = 'twitch:clips'
 474     _VALID_URL = r'https?://clips\.twitch\.tv/(?:[^/]+/)*(?P<id>[^/?#&]+)'
 475
 476     _TESTS = [{
 477         'url': 'https://clips.twitch.tv/ea/AggressiveCobraPoooound',
 478         'md5': '761769e1eafce0ffebfb4089cb3847cd',
 479         'info_dict': {
 480             'id': 'AggressiveCobraPoooound',
 481             'ext': 'mp4',
 482             'title': 'EA Play 2016 Live from the Novo Theatre',
 483             'thumbnail': 're:^https?://.*\.jpg',
 484             'creator': 'EA',
 485             'uploader': 'stereotype_',
 486             'uploader_id': 'stereotype_',
 487         },
 488     }, {
 489         # multiple formats
 490         'url': 'https://clips.twitch.tv/rflegendary/UninterestedBeeDAESuppy',
 491         'only_matching': True,
 492     }]
 493
 494     def _real_extract(self, url):
 495         video_id = self._match_id(url)
 496
 497         webpage = self._download_webpage(url, video_id)
 498
 499         clip = self._parse_json(
 500             self._search_regex(
 501                 r'(?s)clipInfo\s*=\s*({.+?});', webpage, 'clip info'),
 502             video_id, transform_source=js_to_json)
 503
 504         title = clip.get('channel_title') or self._og_search_title(webpage)
 505
 506         formats = [{
 507             'url': option['source'],
 508             'format_id': option.get('quality'),
 509             'height': int_or_none(option.get('quality')),
 510         } for option in clip.get('quality_options', []) if option.get('source')]
 511
 512         if not formats:
 513             formats = [{
 514                 'url': clip['clip_video_url'],
 515             }]
 516
 517         self._sort_formats(formats)
 518
 519         return {
 520             'id': video_id,
 521             'title': title,
 522             'thumbnail': self._og_search_thumbnail(webpage),
 523             'creator': clip.get('broadcaster_display_name') or clip.get('broadcaster_login'),
 524             'uploader': clip.get('curator_login'),
 525             'uploader_id': clip.get('curator_display_name'),
 526             'formats': formats,
 527         }