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     orderedSet,
  20     parse_duration,
  21     parse_iso8601,
  22     sanitized_Request,
  23     urlencode_postdata,
  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, urlencode_postdata(login_form))
  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         'skip': 'HTTP Error 404: Not Found',
 175     }
 176
 177
 178 class TwitchChapterIE(TwitchItemBaseIE):
 179     IE_NAME = 'twitch:chapter'
 180     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 181     _ITEM_TYPE = 'chapter'
 182     _ITEM_SHORTCUT = 'c'
 183
 184     _TESTS = [{
 185         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 186         'info_dict': {
 187             'id': 'c5285812',
 188             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 189         },
 190         'playlist_mincount': 3,
 191         'skip': 'HTTP Error 404: Not Found',
 192     }, {
 193         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 194         'only_matching': True,
 195     }]
 196
 197
 198 class TwitchVodIE(TwitchItemBaseIE):
 199     IE_NAME = 'twitch:vod'
 200     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 201     _ITEM_TYPE = 'vod'
 202     _ITEM_SHORTCUT = 'v'
 203
 204     _TESTS = [{
 205         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 206         'info_dict': {
 207             'id': 'v6528877',
 208             'ext': 'mp4',
 209             'title': 'LCK Summer Split - Week 6 Day 1',
 210             'thumbnail': 're:^https?://.*\.jpg$',
 211             'duration': 17208,
 212             'timestamp': 1435131709,
 213             'upload_date': '20150624',
 214             'uploader': 'Riot Games',
 215             'uploader_id': 'riotgames',
 216             'view_count': int,
 217             'start_time': 310,
 218         },
 219         'params': {
 220             # m3u8 download
 221             'skip_download': True,
 222         },
 223     }, {
 224         # Untitled broadcast (title is None)
 225         'url': 'http://www.twitch.tv/belkao_o/v/11230755',
 226         'info_dict': {
 227             'id': 'v11230755',
 228             'ext': 'mp4',
 229             'title': 'Untitled Broadcast',
 230             'thumbnail': 're:^https?://.*\.jpg$',
 231             'duration': 1638,
 232             'timestamp': 1439746708,
 233             'upload_date': '20150816',
 234             'uploader': 'BelkAO_o',
 235             'uploader_id': 'belkao_o',
 236             'view_count': int,
 237         },
 238         'params': {
 239             # m3u8 download
 240             'skip_download': True,
 241         },
 242     }]
 243
 244     def _real_extract(self, url):
 245         item_id = self._match_id(url)
 246
 247         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 248         access_token = self._download_json(
 249             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
 250             'Downloading %s access token' % self._ITEM_TYPE)
 251
 252         formats = self._extract_m3u8_formats(
 253             '%s/vod/%s?%s' % (
 254                 self._USHER_BASE, item_id,
 255                 compat_urllib_parse_urlencode({
 256                     'allow_source': 'true',
 257                     'allow_audio_only': 'true',
 258                     'allow_spectre': 'true',
 259                     'player': 'twitchweb',
 260                     'nauth': access_token['token'],
 261                     'nauthsig': access_token['sig'],
 262                 })),
 263             item_id, 'mp4', entry_protocol='m3u8_native')
 264
 265         self._prefer_source(formats)
 266         info['formats'] = formats
 267
 268         parsed_url = compat_urllib_parse_urlparse(url)
 269         query = compat_parse_qs(parsed_url.query)
 270         if 't' in query:
 271             info['start_time'] = parse_duration(query['t'][0])
 272
 273         return info
 274
 275
 276 class TwitchPlaylistBaseIE(TwitchBaseIE):
 277     _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
 278     _PAGE_LIMIT = 100
 279
 280     def _extract_playlist(self, channel_id):
 281         info = self._download_json(
 282             '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 283             channel_id, 'Downloading channel info JSON')
 284         channel_name = info.get('display_name') or info.get('name')
 285         entries = []
 286         offset = 0
 287         limit = self._PAGE_LIMIT
 288         broken_paging_detected = False
 289         counter_override = None
 290         for counter in itertools.count(1):
 291             response = self._download_json(
 292                 self._PLAYLIST_URL % (channel_id, offset, limit),
 293                 channel_id,
 294                 'Downloading %s videos JSON page %s'
 295                 % (self._PLAYLIST_TYPE, counter_override or counter))
 296             page_entries = self._extract_playlist_page(response)
 297             if not page_entries:
 298                 break
 299             total = int_or_none(response.get('_total'))
 300             # Since the beginning of March 2016 twitch's paging mechanism
 301             # is completely broken on the twitch side. It simply ignores
 302             # a limit and returns the whole offset number of videos.
 303             # Working around by just requesting all videos at once.
 304             # Upd: pagination bug was fixed by twitch on 15.03.2016.
 305             if not broken_paging_detected and total and len(page_entries) > limit:
 306                 self.report_warning(
 307                     'Twitch pagination is broken on twitch side, requesting all videos at once',
 308                     channel_id)
 309                 broken_paging_detected = True
 310                 offset = total
 311                 counter_override = '(all at once)'
 312                 continue
 313             entries.extend(page_entries)
 314             if broken_paging_detected or total and len(page_entries) >= total:
 315                 break
 316             offset += limit
 317         return self.playlist_result(
 318             [self.url_result(entry) for entry in orderedSet(entries)],
 319             channel_id, channel_name)
 320
 321     def _extract_playlist_page(self, response):
 322         videos = response.get('videos')
 323         return [video['url'] for video in videos] if videos else []
 324
 325     def _real_extract(self, url):
 326         return self._extract_playlist(self._match_id(url))
 327
 328
 329 class TwitchProfileIE(TwitchPlaylistBaseIE):
 330     IE_NAME = 'twitch:profile'
 331     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 332     _PLAYLIST_TYPE = 'profile'
 333
 334     _TEST = {
 335         'url': 'http://www.twitch.tv/vanillatv/profile',
 336         'info_dict': {
 337             'id': 'vanillatv',
 338             'title': 'VanillaTV',
 339         },
 340         'playlist_mincount': 412,
 341     }
 342
 343
 344 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 345     IE_NAME = 'twitch:past_broadcasts'
 346     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 347     _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
 348     _PLAYLIST_TYPE = 'past broadcasts'
 349
 350     _TEST = {
 351         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 352         'info_dict': {
 353             'id': 'spamfish',
 354             'title': 'Spamfish',
 355         },
 356         'playlist_mincount': 54,
 357     }
 358
 359
 360 class TwitchStreamIE(TwitchBaseIE):
 361     IE_NAME = 'twitch:stream'
 362     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 363
 364     _TESTS = [{
 365         'url': 'http://www.twitch.tv/shroomztv',
 366         'info_dict': {
 367             'id': '12772022048',
 368             'display_id': 'shroomztv',
 369             'ext': 'mp4',
 370             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 371             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 372             'is_live': True,
 373             'timestamp': 1421928037,
 374             'upload_date': '20150122',
 375             'uploader': 'ShroomzTV',
 376             'uploader_id': 'shroomztv',
 377             'view_count': int,
 378         },
 379         'params': {
 380             # m3u8 download
 381             'skip_download': True,
 382         },
 383     }, {
 384         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 385         'only_matching': True,
 386     }]
 387
 388     def _real_extract(self, url):
 389         channel_id = self._match_id(url)
 390
 391         stream = self._download_json(
 392             '%s/kraken/streams/%s' % (self._API_BASE, channel_id), channel_id,
 393             'Downloading stream JSON').get('stream')
 394
 395         # Fallback on profile extraction if stream is offline
 396         if not stream:
 397             return self.url_result(
 398                 'http://www.twitch.tv/%s/profile' % channel_id,
 399                 'TwitchProfile', channel_id)
 400
 401         # Channel name may be typed if different case than the original channel name
 402         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 403         # an invalid m3u8 URL. Working around by use of original channel name from stream
 404         # JSON and fallback to lowercase if it's not available.
 405         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 406
 407         access_token = self._download_json(
 408             '%s/api/channels/%s/access_token' % (self._API_BASE, channel_id), channel_id,
 409             'Downloading channel access token')
 410
 411         query = {
 412             'allow_source': 'true',
 413             'allow_audio_only': 'true',
 414             'p': random.randint(1000000, 10000000),
 415             'player': 'twitchweb',
 416             'segment_preference': '4',
 417             'sig': access_token['sig'].encode('utf-8'),
 418             'token': access_token['token'].encode('utf-8'),
 419         }
 420         formats = self._extract_m3u8_formats(
 421             '%s/api/channel/hls/%s.m3u8?%s'
 422             % (self._USHER_BASE, channel_id, compat_urllib_parse_urlencode(query)),
 423             channel_id, 'mp4')
 424         self._prefer_source(formats)
 425
 426         view_count = stream.get('viewers')
 427         timestamp = parse_iso8601(stream.get('created_at'))
 428
 429         channel = stream['channel']
 430         title = self._live_title(channel.get('display_name') or channel.get('name'))
 431         description = channel.get('status')
 432
 433         thumbnails = []
 434         for thumbnail_key, thumbnail_url in stream['preview'].items():
 435             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 436             if not m:
 437                 continue
 438             thumbnails.append({
 439                 'url': thumbnail_url,
 440                 'width': int(m.group('width')),
 441                 'height': int(m.group('height')),
 442             })
 443
 444         return {
 445             'id': compat_str(stream['_id']),
 446             'display_id': channel_id,
 447             'title': title,
 448             'description': description,
 449             'thumbnails': thumbnails,
 450             'uploader': channel.get('display_name'),
 451             'uploader_id': channel.get('name'),
 452             'timestamp': timestamp,
 453             'view_count': view_count,
 454             'formats': formats,
 455             'is_live': True,
 456         }