yt_dlp/extractor/twitcasting.py

   1 import base64
   2 import itertools
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..dependencies import websockets
   7 from ..utils import (
   8     ExtractorError,
   9     UserNotLive,
  10     clean_html,
  11     float_or_none,
  12     get_element_by_class,
  13     get_element_by_id,
  14     int_or_none,
  15     parse_duration,
  16     qualities,
  17     str_to_int,
  18     traverse_obj,
  19     try_get,
  20     unified_timestamp,
  21     urlencode_postdata,
  22     urljoin,
  23 )
  24
  25
  26 class TwitCastingIE(InfoExtractor):
  27     _VALID_URL = r'https?://(?:[^/?#]+\.)?twitcasting\.tv/(?P<uploader_id>[^/?#]+)/(?:movie|twplayer)/(?P<id>\d+)'
  28     _M3U8_HEADERS = {
  29         'Origin': 'https://twitcasting.tv',
  30         'Referer': 'https://twitcasting.tv/',
  31     }
  32     _TESTS = [{
  33         'url': 'https://twitcasting.tv/ivetesangalo/movie/2357609',
  34         'md5': '745243cad58c4681dc752490f7540d7f',
  35         'info_dict': {
  36             'id': '2357609',
  37             'ext': 'mp4',
  38             'title': 'Live #2357609',
  39             'uploader_id': 'ivetesangalo',
  40             'description': 'Twitter Oficial da cantora brasileira Ivete Sangalo.',
  41             'thumbnail': r're:^https?://.*\.jpg$',
  42             'upload_date': '20110822',
  43             'timestamp': 1313978424,
  44             'duration': 32,
  45             'view_count': int,
  46         },
  47         'params': {
  48             'skip_download': True,
  49         },
  50     }, {
  51         'url': 'https://twitcasting.tv/mttbernardini/movie/3689740',
  52         'info_dict': {
  53             'id': '3689740',
  54             'ext': 'mp4',
  55             'title': 'Live playing something #3689740',
  56             'uploader_id': 'mttbernardini',
  57             'description': 'md5:1dc7efa2f1ab932fcd119265cebeec69',
  58             'thumbnail': r're:^https?://.*\.jpg$',
  59             'upload_date': '20120211',
  60             'timestamp': 1328995624,
  61             'duration': 681,
  62             'view_count': int,
  63         },
  64         'params': {
  65             'skip_download': True,
  66             'videopassword': 'abc',
  67         },
  68     }, {
  69         'url': 'https://twitcasting.tv/loft_heaven/movie/685979292',
  70         'info_dict': {
  71             'id': '685979292',
  72             'ext': 'mp4',
  73             'title': '【無料配信】南波一海のhear/here “ナタリー望月哲さんに聞く編集と「渋谷系狂騒曲」”',
  74             'uploader_id': 'loft_heaven',
  75             'description': 'md5:3a0c7b53019df987ce545c935538bacf',
  76             'upload_date': '20210604',
  77             'timestamp': 1622802114,
  78             'thumbnail': r're:^https?://.*\.jpg$',
  79             'duration': 6964,
  80             'view_count': int,
  81         },
  82         'params': {
  83             'skip_download': True,
  84         },
  85     }]
  86
  87     def _parse_data_movie_playlist(self, dmp, video_id):
  88         # attempt 1: parse as JSON directly
  89         try:
  90             return self._parse_json(dmp, video_id)
  91         except ExtractorError:
  92             pass
  93         # attempt 2: decode reversed base64
  94         decoded = base64.b64decode(dmp[::-1])
  95         return self._parse_json(decoded, video_id)
  96
  97     def _real_extract(self, url):
  98         uploader_id, video_id = self._match_valid_url(url).groups()
  99
 100         webpage, urlh = self._download_webpage_handle(url, video_id)
 101         video_password = self.get_param('videopassword')
 102         request_data = None
 103         if video_password:
 104             request_data = urlencode_postdata({
 105                 'password': video_password,
 106                 **self._hidden_inputs(webpage),
 107             }, encoding='utf-8')
 108             webpage, urlh = self._download_webpage_handle(
 109                 url, video_id, data=request_data,
 110                 headers={'Origin': 'https://twitcasting.tv'},
 111                 note='Trying video password')
 112         if urlh.url != url and request_data:
 113             webpage = self._download_webpage(
 114                 urlh.url, video_id, data=request_data,
 115                 headers={'Origin': 'https://twitcasting.tv'},
 116                 note='Retrying authentication')
 117         # has to check here as the first request can contain password input form even if the password is correct
 118         if re.search(r'<form\s+method="POST">\s*<input\s+[^>]+?name="password"', webpage):
 119             raise ExtractorError('This video is protected by a password, use the --video-password option', expected=True)
 120
 121         title = (clean_html(get_element_by_id('movietitle', webpage))
 122                  or self._html_search_meta(['og:title', 'twitter:title'], webpage, fatal=True))
 123
 124         video_js_data = try_get(
 125             webpage,
 126             lambda x: self._parse_data_movie_playlist(self._search_regex(
 127                 r'data-movie-playlist=\'([^\']+?)\'',
 128                 x, 'movie playlist', default=None), video_id)['2'], list)
 129
 130         thumbnail = traverse_obj(video_js_data, (0, 'thumbnailUrl')) or self._og_search_thumbnail(webpage)
 131         description = clean_html(get_element_by_id(
 132             'authorcomment', webpage)) or self._html_search_meta(
 133             ['description', 'og:description', 'twitter:description'], webpage)
 134         duration = (try_get(video_js_data, lambda x: sum(float_or_none(y.get('duration')) for y in x) / 1000)
 135                     or parse_duration(clean_html(get_element_by_class('tw-player-duration-time', webpage))))
 136         view_count = str_to_int(self._search_regex(
 137             (r'Total\s*:\s*Views\s*([\d,]+)', r'総視聴者\s*:\s*([\d,]+)\s*</'), webpage, 'views', None))
 138         timestamp = unified_timestamp(self._search_regex(
 139             r'data-toggle="true"[^>]+datetime="([^"]+)"',
 140             webpage, 'datetime', None))
 141
 142         stream_server_data = self._download_json(
 143             'https://twitcasting.tv/streamserver.php?target=%s&mode=client' % uploader_id, video_id,
 144             'Downloading live info', fatal=False)
 145
 146         is_live = any(f'data-{x}' in webpage for x in ['is-onlive="true"', 'live-type="live"', 'status="online"'])
 147         if not traverse_obj(stream_server_data, 'llfmp4') and is_live:
 148             self.raise_login_required(method='cookies')
 149
 150         base_dict = {
 151             'title': title,
 152             'description': description,
 153             'thumbnail': thumbnail,
 154             'timestamp': timestamp,
 155             'uploader_id': uploader_id,
 156             'duration': duration,
 157             'view_count': view_count,
 158             'is_live': is_live,
 159         }
 160
 161         def find_dmu(x):
 162             data_movie_url = self._search_regex(
 163                 r'data-movie-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
 164                 x, 'm3u8 url', group='url', default=None)
 165             if data_movie_url:
 166                 return [data_movie_url]
 167
 168         m3u8_urls = (try_get(webpage, find_dmu, list)
 169                      or traverse_obj(video_js_data, (..., 'source', 'url'))
 170                      or ([f'https://twitcasting.tv/{uploader_id}/metastream.m3u8'] if is_live else None))
 171         if not m3u8_urls:
 172             raise ExtractorError('Failed to get m3u8 playlist')
 173
 174         if is_live:
 175             m3u8_url = m3u8_urls[0]
 176             formats = self._extract_m3u8_formats(
 177                 m3u8_url, video_id, ext='mp4', m3u8_id='hls',
 178                 live=True, headers=self._M3U8_HEADERS)
 179
 180             if traverse_obj(stream_server_data, ('hls', 'source')):
 181                 formats.extend(self._extract_m3u8_formats(
 182                     m3u8_url, video_id, ext='mp4', m3u8_id='source',
 183                     live=True, query={'mode': 'source'},
 184                     note='Downloading source quality m3u8',
 185                     headers=self._M3U8_HEADERS, fatal=False))
 186
 187             if websockets:
 188                 qq = qualities(['base', 'mobilesource', 'main'])
 189                 streams = traverse_obj(stream_server_data, ('llfmp4', 'streams')) or {}
 190                 for mode, ws_url in streams.items():
 191                     formats.append({
 192                         'url': ws_url,
 193                         'format_id': 'ws-%s' % mode,
 194                         'ext': 'mp4',
 195                         'quality': qq(mode),
 196                         'source_preference': -10,
 197                         # TwitCasting simply sends moof atom directly over WS
 198                         'protocol': 'websocket_frag',
 199                     })
 200
 201             infodict = {
 202                 'formats': formats,
 203                 '_format_sort_fields': ('source', ),
 204             }
 205         elif len(m3u8_urls) == 1:
 206             formats = self._extract_m3u8_formats(
 207                 m3u8_urls[0], video_id, 'mp4', headers=self._M3U8_HEADERS)
 208             infodict = {
 209                 # No problem here since there's only one manifest
 210                 'formats': formats,
 211                 'http_headers': self._M3U8_HEADERS,
 212             }
 213         else:
 214             infodict = {
 215                 '_type': 'multi_video',
 216                 'entries': [{
 217                     'id': f'{video_id}-{num}',
 218                     'url': m3u8_url,
 219                     'ext': 'mp4',
 220                     # Requesting the manifests here will cause download to fail.
 221                     # So use ffmpeg instead. See: https://github.com/yt-dlp/yt-dlp/issues/382
 222                     'protocol': 'm3u8',
 223                     'http_headers': self._M3U8_HEADERS,
 224                     **base_dict,
 225                 } for (num, m3u8_url) in enumerate(m3u8_urls)],
 226             }
 227
 228         return {
 229             'id': video_id,
 230             **base_dict,
 231             **infodict,
 232         }
 233
 234
 235 class TwitCastingLiveIE(InfoExtractor):
 236     _VALID_URL = r'https?://(?:[^/?#]+\.)?twitcasting\.tv/(?P<id>[^/?#]+)/?(?:[#?]|$)'
 237     _TESTS = [{
 238         'url': 'https://twitcasting.tv/ivetesangalo',
 239         'only_matching': True,
 240     }, {
 241         'url': 'https://twitcasting.tv/c:unusedlive',
 242         'expected_exception': 'UserNotLive',
 243     }]
 244
 245     _PROTECTED_LIVE_RE = r'(?s)(<span\s*class="tw-movie-thumbnail2-badge"\s*data-status="live">\s*LIVE)'
 246
 247     def _real_extract(self, url):
 248         uploader_id = self._match_id(url)
 249         self.to_screen(
 250             'Downloading live video of user {0}. '
 251             'Pass "https://twitcasting.tv/{0}/show" to download the history'.format(uploader_id))
 252
 253         webpage = self._download_webpage(url, uploader_id)
 254         is_live = self._search_regex(  # first pattern is for public live
 255             (r'(data-is-onlive="true")', self._PROTECTED_LIVE_RE), webpage, 'is live?', default=None)
 256         current_live = int_or_none(self._search_regex(
 257             (r'data-type="movie" data-id="(\d+)">',  # not available?
 258              r'tw-sound-flag-open-link" data-id="(\d+)" style=',  # not available?
 259              r'data-movie-id="(\d+)"'),  # if not currently live, value may be 0
 260             webpage, 'current live ID', default=None))
 261         if is_live and not current_live:
 262             # fetch unfiltered /show to find running livestreams; we can't get ID of the password-protected livestream above
 263             webpage = self._download_webpage(
 264                 f'https://twitcasting.tv/{uploader_id}/show/', uploader_id,
 265                 note='Downloading live history')
 266             is_live = self._search_regex(self._PROTECTED_LIVE_RE, webpage, 'is live?', default=None)
 267             if is_live:
 268                 # get the first live; running live is always at the first
 269                 current_live = self._search_regex(
 270                     r'(?s)<a\s+class="tw-movie-thumbnail2"\s*href="/[^/]+/movie/(?P<video_id>\d+)"\s*>.+?</a>',
 271                     webpage, 'current live ID 2', default=None, group='video_id')
 272         if not current_live:
 273             raise UserNotLive(video_id=uploader_id)
 274         return self.url_result(f'https://twitcasting.tv/{uploader_id}/movie/{current_live}', TwitCastingIE)
 275
 276
 277 class TwitCastingUserIE(InfoExtractor):
 278     _VALID_URL = r'https?://(?:[^/?#]+\.)?twitcasting\.tv/(?P<id>[^/?#]+)/(:?show|archive)/?(?:[#?]|$)'
 279     _TESTS = [{
 280         'url': 'https://twitcasting.tv/natsuiromatsuri/archive/',
 281         'info_dict': {
 282             'id': 'natsuiromatsuri',
 283             'title': 'natsuiromatsuri - Live History',
 284         },
 285         'playlist_mincount': 235,
 286     }, {
 287         'url': 'https://twitcasting.tv/noriyukicas/show',
 288         'only_matching': True,
 289     }]
 290
 291     def _entries(self, uploader_id):
 292         base_url = next_url = 'https://twitcasting.tv/%s/show' % uploader_id
 293         for page_num in itertools.count(1):
 294             webpage = self._download_webpage(
 295                 next_url, uploader_id, query={'filter': 'watchable'}, note='Downloading page %d' % page_num)
 296             matches = re.finditer(
 297                 r'''(?isx)<a\s+class="tw-movie-thumbnail"\s*href="(?P<url>/[^/]+/movie/\d+)"\s*>.+?</a>''',
 298                 webpage)
 299             for mobj in matches:
 300                 yield self.url_result(urljoin(base_url, mobj.group('url')))
 301
 302             next_url = self._search_regex(
 303                 r'<a href="(/%s/show/%d-\d+)[?"]' % (re.escape(uploader_id), page_num),
 304                 webpage, 'next url', default=None)
 305             next_url = urljoin(base_url, next_url)
 306             if not next_url:
 307                 return
 308
 309     def _real_extract(self, url):
 310         uploader_id = self._match_id(url)
 311         return self.playlist_result(
 312             self._entries(uploader_id), uploader_id, '%s - Live History' % uploader_id)