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