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