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     _CLIENT_ID = 'jzkbprff40iqj646a697cyrvl0zt2m6'
  36     _NETRC_MACHINE = 'twitch'
  37
  38     def _handle_error(self, response):
  39         if not isinstance(response, dict):
  40             return
  41         error = response.get('error')
  42         if error:
  43             raise ExtractorError(
  44                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  45                 expected=True)
  46
  47     def _call_api(self, path, item_id, note):
  48         response = self._download_json(
  49             '%s/%s' % (self._API_BASE, path), item_id, note,
  50             headers={'Client-ID': self._CLIENT_ID})
  51         self._handle_error(response)
  52         return response
  53
  54     def _real_initialize(self):
  55         self._login()
  56
  57     def _login(self):
  58         (username, password) = self._get_login_info()
  59         if username is None:
  60             return
  61
  62         def fail(message):
  63             raise ExtractorError(
  64                 'Unable to login. Twitch said: %s' % message, expected=True)
  65
  66         login_page, handle = self._download_webpage_handle(
  67             self._LOGIN_URL, None, 'Downloading login page')
  68
  69         # Some TOR nodes and public proxies are blocked completely
  70         if 'blacklist_message' in login_page:
  71             fail(clean_html(login_page))
  72
  73         login_form = self._hidden_inputs(login_page)
  74
  75         login_form.update({
  76             'username': username,
  77             'password': password,
  78         })
  79
  80         redirect_url = handle.geturl()
  81
  82         post_url = self._search_regex(
  83             r'<form[^>]+action=(["\'])(?P<url>.+?)\1', login_page,
  84             'post url', default=redirect_url, group='url')
  85
  86         if not post_url.startswith('http'):
  87             post_url = compat_urlparse.urljoin(redirect_url, post_url)
  88
  89         headers = {'Referer': redirect_url}
  90
  91         try:
  92             response = self._download_json(
  93                 post_url, None, 'Logging in as %s' % username,
  94                 data=urlencode_postdata(login_form),
  95                 headers=headers)
  96         except ExtractorError as e:
  97             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400:
  98                 response = self._parse_json(
  99                     e.cause.read().decode('utf-8'), None)
 100                 fail(response['message'])
 101             raise
 102
 103         if response.get('redirect'):
 104             self._download_webpage(
 105                 response['redirect'], None, 'Downloading login redirect page',
 106                 headers=headers)
 107
 108     def _prefer_source(self, formats):
 109         try:
 110             source = next(f for f in formats if f['format_id'] == 'Source')
 111             source['preference'] = 10
 112         except StopIteration:
 113             pass  # No Source stream present
 114         self._sort_formats(formats)
 115
 116
 117 class TwitchItemBaseIE(TwitchBaseIE):
 118     def _download_info(self, item, item_id):
 119         return self._extract_info(self._call_api(
 120             'kraken/videos/%s%s' % (item, item_id), item_id,
 121             'Downloading %s info JSON' % self._ITEM_TYPE))
 122
 123     def _extract_media(self, item_id):
 124         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 125         response = self._call_api(
 126             'api/videos/%s%s' % (self._ITEM_SHORTCUT, item_id), item_id,
 127             'Downloading %s playlist JSON' % self._ITEM_TYPE)
 128         entries = []
 129         chunks = response['chunks']
 130         qualities = list(chunks.keys())
 131         for num, fragment in enumerate(zip(*chunks.values()), start=1):
 132             formats = []
 133             for fmt_num, fragment_fmt in enumerate(fragment):
 134                 format_id = qualities[fmt_num]
 135                 fmt = {
 136                     'url': fragment_fmt['url'],
 137                     'format_id': format_id,
 138                     'quality': 1 if format_id == 'live' else 0,
 139                 }
 140                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 141                 if m:
 142                     fmt['height'] = int(m.group('height'))
 143                 formats.append(fmt)
 144             self._sort_formats(formats)
 145             entry = dict(info)
 146             entry['id'] = '%s_%d' % (entry['id'], num)
 147             entry['title'] = '%s part %d' % (entry['title'], num)
 148             entry['formats'] = formats
 149             entries.append(entry)
 150         return self.playlist_result(entries, info['id'], info['title'])
 151
 152     def _extract_info(self, info):
 153         return {
 154             'id': info['_id'],
 155             'title': info.get('title') or 'Untitled Broadcast',
 156             'description': info.get('description'),
 157             'duration': int_or_none(info.get('length')),
 158             'thumbnail': info.get('preview'),
 159             'uploader': info.get('channel', {}).get('display_name'),
 160             'uploader_id': info.get('channel', {}).get('name'),
 161             'timestamp': parse_iso8601(info.get('recorded_at')),
 162             'view_count': int_or_none(info.get('views')),
 163         }
 164
 165     def _real_extract(self, url):
 166         return self._extract_media(self._match_id(url))
 167
 168
 169 class TwitchVideoIE(TwitchItemBaseIE):
 170     IE_NAME = 'twitch:video'
 171     _VALID_URL = r'%s/[^/]+/b/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 172     _ITEM_TYPE = 'video'
 173     _ITEM_SHORTCUT = 'a'
 174
 175     _TEST = {
 176         'url': 'http://www.twitch.tv/riotgames/b/577357806',
 177         'info_dict': {
 178             'id': 'a577357806',
 179             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
 180         },
 181         'playlist_mincount': 12,
 182         'skip': 'HTTP Error 404: Not Found',
 183     }
 184
 185
 186 class TwitchChapterIE(TwitchItemBaseIE):
 187     IE_NAME = 'twitch:chapter'
 188     _VALID_URL = r'%s/[^/]+/c/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 189     _ITEM_TYPE = 'chapter'
 190     _ITEM_SHORTCUT = 'c'
 191
 192     _TESTS = [{
 193         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
 194         'info_dict': {
 195             'id': 'c5285812',
 196             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
 197         },
 198         'playlist_mincount': 3,
 199         'skip': 'HTTP Error 404: Not Found',
 200     }, {
 201         'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
 202         'only_matching': True,
 203     }]
 204
 205
 206 class TwitchVodIE(TwitchItemBaseIE):
 207     IE_NAME = 'twitch:vod'
 208     _VALID_URL = r'%s/[^/]+/v/(?P<id>\d+)' % TwitchBaseIE._VALID_URL_BASE
 209     _ITEM_TYPE = 'vod'
 210     _ITEM_SHORTCUT = 'v'
 211
 212     _TESTS = [{
 213         'url': 'http://www.twitch.tv/riotgames/v/6528877?t=5m10s',
 214         'info_dict': {
 215             'id': 'v6528877',
 216             'ext': 'mp4',
 217             'title': 'LCK Summer Split - Week 6 Day 1',
 218             'thumbnail': 're:^https?://.*\.jpg$',
 219             'duration': 17208,
 220             'timestamp': 1435131709,
 221             'upload_date': '20150624',
 222             'uploader': 'Riot Games',
 223             'uploader_id': 'riotgames',
 224             'view_count': int,
 225             'start_time': 310,
 226         },
 227         'params': {
 228             # m3u8 download
 229             'skip_download': True,
 230         },
 231     }, {
 232         # Untitled broadcast (title is None)
 233         'url': 'http://www.twitch.tv/belkao_o/v/11230755',
 234         'info_dict': {
 235             'id': 'v11230755',
 236             'ext': 'mp4',
 237             'title': 'Untitled Broadcast',
 238             'thumbnail': 're:^https?://.*\.jpg$',
 239             'duration': 1638,
 240             'timestamp': 1439746708,
 241             'upload_date': '20150816',
 242             'uploader': 'BelkAO_o',
 243             'uploader_id': 'belkao_o',
 244             'view_count': int,
 245         },
 246         'params': {
 247             # m3u8 download
 248             'skip_download': True,
 249         },
 250     }]
 251
 252     def _real_extract(self, url):
 253         item_id = self._match_id(url)
 254
 255         info = self._download_info(self._ITEM_SHORTCUT, item_id)
 256         access_token = self._call_api(
 257             'api/vods/%s/access_token' % item_id, item_id,
 258             'Downloading %s access token' % self._ITEM_TYPE)
 259
 260         formats = self._extract_m3u8_formats(
 261             '%s/vod/%s?%s' % (
 262                 self._USHER_BASE, item_id,
 263                 compat_urllib_parse_urlencode({
 264                     'allow_source': 'true',
 265                     'allow_audio_only': 'true',
 266                     'allow_spectre': 'true',
 267                     'player': 'twitchweb',
 268                     'nauth': access_token['token'],
 269                     'nauthsig': access_token['sig'],
 270                 })),
 271             item_id, 'mp4', entry_protocol='m3u8_native')
 272
 273         self._prefer_source(formats)
 274         info['formats'] = formats
 275
 276         parsed_url = compat_urllib_parse_urlparse(url)
 277         query = compat_parse_qs(parsed_url.query)
 278         if 't' in query:
 279             info['start_time'] = parse_duration(query['t'][0])
 280
 281         return info
 282
 283
 284 class TwitchPlaylistBaseIE(TwitchBaseIE):
 285     _PLAYLIST_PATH = 'kraken/channels/%s/videos/?offset=%d&limit=%d'
 286     _PAGE_LIMIT = 100
 287
 288     def _extract_playlist(self, channel_id):
 289         info = self._call_api(
 290             'kraken/channels/%s' % channel_id,
 291             channel_id, 'Downloading channel info JSON')
 292         channel_name = info.get('display_name') or info.get('name')
 293         entries = []
 294         offset = 0
 295         limit = self._PAGE_LIMIT
 296         broken_paging_detected = False
 297         counter_override = None
 298         for counter in itertools.count(1):
 299             response = self._call_api(
 300                 self._PLAYLIST_PATH % (channel_id, offset, limit),
 301                 channel_id,
 302                 'Downloading %s videos JSON page %s'
 303                 % (self._PLAYLIST_TYPE, counter_override or counter))
 304             page_entries = self._extract_playlist_page(response)
 305             if not page_entries:
 306                 break
 307             total = int_or_none(response.get('_total'))
 308             # Since the beginning of March 2016 twitch's paging mechanism
 309             # is completely broken on the twitch side. It simply ignores
 310             # a limit and returns the whole offset number of videos.
 311             # Working around by just requesting all videos at once.
 312             # Upd: pagination bug was fixed by twitch on 15.03.2016.
 313             if not broken_paging_detected and total and len(page_entries) > limit:
 314                 self.report_warning(
 315                     'Twitch pagination is broken on twitch side, requesting all videos at once',
 316                     channel_id)
 317                 broken_paging_detected = True
 318                 offset = total
 319                 counter_override = '(all at once)'
 320                 continue
 321             entries.extend(page_entries)
 322             if broken_paging_detected or total and len(page_entries) >= total:
 323                 break
 324             offset += limit
 325         return self.playlist_result(
 326             [self.url_result(entry) for entry in orderedSet(entries)],
 327             channel_id, channel_name)
 328
 329     def _extract_playlist_page(self, response):
 330         videos = response.get('videos')
 331         return [video['url'] for video in videos] if videos else []
 332
 333     def _real_extract(self, url):
 334         return self._extract_playlist(self._match_id(url))
 335
 336
 337 class TwitchProfileIE(TwitchPlaylistBaseIE):
 338     IE_NAME = 'twitch:profile'
 339     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 340     _PLAYLIST_TYPE = 'profile'
 341
 342     _TEST = {
 343         'url': 'http://www.twitch.tv/vanillatv/profile',
 344         'info_dict': {
 345             'id': 'vanillatv',
 346             'title': 'VanillaTV',
 347         },
 348         'playlist_mincount': 412,
 349     }
 350
 351
 352 class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
 353     IE_NAME = 'twitch:past_broadcasts'
 354     _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 355     _PLAYLIST_PATH = TwitchPlaylistBaseIE._PLAYLIST_PATH + '&broadcasts=true'
 356     _PLAYLIST_TYPE = 'past broadcasts'
 357
 358     _TEST = {
 359         'url': 'http://www.twitch.tv/spamfish/profile/past_broadcasts',
 360         'info_dict': {
 361             'id': 'spamfish',
 362             'title': 'Spamfish',
 363         },
 364         'playlist_mincount': 54,
 365     }
 366
 367
 368 class TwitchStreamIE(TwitchBaseIE):
 369     IE_NAME = 'twitch:stream'
 370     _VALID_URL = r'%s/(?P<id>[^/#?]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
 371
 372     _TESTS = [{
 373         'url': 'http://www.twitch.tv/shroomztv',
 374         'info_dict': {
 375             'id': '12772022048',
 376             'display_id': 'shroomztv',
 377             'ext': 'mp4',
 378             'title': 're:^ShroomzTV [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 379             'description': 'H1Z1 - lonewolfing with ShroomzTV | A3 Battle Royale later - @ShroomzTV',
 380             'is_live': True,
 381             'timestamp': 1421928037,
 382             'upload_date': '20150122',
 383             'uploader': 'ShroomzTV',
 384             'uploader_id': 'shroomztv',
 385             'view_count': int,
 386         },
 387         'params': {
 388             # m3u8 download
 389             'skip_download': True,
 390         },
 391     }, {
 392         'url': 'http://www.twitch.tv/miracle_doto#profile-0',
 393         'only_matching': True,
 394     }]
 395
 396     def _real_extract(self, url):
 397         channel_id = self._match_id(url)
 398
 399         stream = self._call_api(
 400             'kraken/streams/%s' % channel_id, channel_id,
 401             'Downloading stream JSON').get('stream')
 402
 403         # Fallback on profile extraction if stream is offline
 404         if not stream:
 405             return self.url_result(
 406                 'http://www.twitch.tv/%s/profile' % channel_id,
 407                 'TwitchProfile', channel_id)
 408
 409         # Channel name may be typed if different case than the original channel name
 410         # (e.g. http://www.twitch.tv/TWITCHPLAYSPOKEMON) that will lead to constructing
 411         # an invalid m3u8 URL. Working around by use of original channel name from stream
 412         # JSON and fallback to lowercase if it's not available.
 413         channel_id = stream.get('channel', {}).get('name') or channel_id.lower()
 414
 415         access_token = self._call_api(
 416             'api/channels/%s/access_token' % channel_id, channel_id,
 417             'Downloading channel access token')
 418
 419         query = {
 420             'allow_source': 'true',
 421             'allow_audio_only': 'true',
 422             'p': random.randint(1000000, 10000000),
 423             'player': 'twitchweb',
 424             'segment_preference': '4',
 425             'sig': access_token['sig'].encode('utf-8'),
 426             'token': access_token['token'].encode('utf-8'),
 427         }
 428         formats = self._extract_m3u8_formats(
 429             '%s/api/channel/hls/%s.m3u8?%s'
 430             % (self._USHER_BASE, channel_id, compat_urllib_parse_urlencode(query)),
 431             channel_id, 'mp4')
 432         self._prefer_source(formats)
 433
 434         view_count = stream.get('viewers')
 435         timestamp = parse_iso8601(stream.get('created_at'))
 436
 437         channel = stream['channel']
 438         title = self._live_title(channel.get('display_name') or channel.get('name'))
 439         description = channel.get('status')
 440
 441         thumbnails = []
 442         for thumbnail_key, thumbnail_url in stream['preview'].items():
 443             m = re.search(r'(?P<width>\d+)x(?P<height>\d+)\.jpg$', thumbnail_key)
 444             if not m:
 445                 continue
 446             thumbnails.append({
 447                 'url': thumbnail_url,
 448                 'width': int(m.group('width')),
 449                 'height': int(m.group('height')),
 450             })
 451
 452         return {
 453             'id': compat_str(stream['_id']),
 454             'display_id': channel_id,
 455             'title': title,
 456             'description': description,
 457             'thumbnails': thumbnails,
 458             'uploader': channel.get('display_name'),
 459             'uploader_id': channel.get('name'),
 460             'timestamp': timestamp,
 461             'view_count': view_count,
 462             'formats': formats,
 463             'is_live': True,
 464         }
 465
 466
 467 class TwitchClipsIE(InfoExtractor):
 468     IE_NAME = 'twitch:clips'
 469     _VALID_URL = r'https?://clips\.twitch\.tv/(?:[^/]+/)*(?P<id>[^/?#&]+)'
 470
 471     _TESTS = [{
 472         'url': 'https://clips.twitch.tv/ea/AggressiveCobraPoooound',
 473         'md5': '761769e1eafce0ffebfb4089cb3847cd',
 474         'info_dict': {
 475             'id': 'AggressiveCobraPoooound',
 476             'ext': 'mp4',
 477             'title': 'EA Play 2016 Live from the Novo Theatre',
 478             'thumbnail': 're:^https?://.*\.jpg',
 479             'creator': 'EA',
 480             'uploader': 'stereotype_',
 481             'uploader_id': 'stereotype_',
 482         },
 483     }, {
 484         # multiple formats
 485         'url': 'https://clips.twitch.tv/rflegendary/UninterestedBeeDAESuppy',
 486         'only_matching': True,
 487     }]
 488
 489     def _real_extract(self, url):
 490         video_id = self._match_id(url)
 491
 492         webpage = self._download_webpage(url, video_id)
 493
 494         clip = self._parse_json(
 495             self._search_regex(
 496                 r'(?s)clipInfo\s*=\s*({.+?});', webpage, 'clip info'),
 497             video_id, transform_source=js_to_json)
 498
 499         title = clip.get('channel_title') or self._og_search_title(webpage)
 500
 501         formats = [{
 502             'url': option['source'],
 503             'format_id': option.get('quality'),
 504             'height': int_or_none(option.get('quality')),
 505         } for option in clip.get('quality_options', []) if option.get('source')]
 506
 507         if not formats:
 508             formats = [{
 509                 'url': clip['clip_video_url'],
 510             }]
 511
 512         self._sort_formats(formats)
 513
 514         return {
 515             'id': video_id,
 516             'title': title,
 517             'thumbnail': self._og_search_thumbnail(webpage),
 518             'creator': clip.get('broadcaster_display_name') or clip.get('broadcaster_login'),
 519             'uploader': clip.get('curator_login'),
 520             'uploader_id': clip.get('curator_display_name'),
 521             'formats': formats,
 522         }