diff options
Diffstat (limited to 'hypervideo_dl/extractor/twitcasting.py')
| -rw-r--r-- | hypervideo_dl/extractor/twitcasting.py | 166 |
1 files changed, 120 insertions, 46 deletions
diff --git a/hypervideo_dl/extractor/twitcasting.py b/hypervideo_dl/extractor/twitcasting.py index 3acf1b1..5c4d26c 100644 --- a/hypervideo_dl/extractor/twitcasting.py +++ b/hypervideo_dl/extractor/twitcasting.py @@ -8,22 +8,27 @@ from .common import InfoExtractor from ..downloader.websocket import has_websockets from ..utils import ( clean_html, + ExtractorError, float_or_none, get_element_by_class, get_element_by_id, parse_duration, qualities, str_to_int, + traverse_obj, try_get, unified_timestamp, urlencode_postdata, urljoin, - ExtractorError, ) class TwitCastingIE(InfoExtractor): _VALID_URL = r'https?://(?:[^/]+\.)?twitcasting\.tv/(?P<uploader_id>[^/]+)/(?:movie|twplayer)/(?P<id>\d+)' + _M3U8_HEADERS = { + 'Origin': 'https://twitcasting.tv', + 'Referer': 'https://twitcasting.tv/', + } _TESTS = [{ 'url': 'https://twitcasting.tv/ivetesangalo/movie/2357609', 'md5': '745243cad58c4681dc752490f7540d7f', @@ -60,6 +65,16 @@ class TwitCastingIE(InfoExtractor): 'skip_download': True, 'videopassword': 'abc', }, + }, { + 'note': 'archive is split in 2 parts', + 'url': 'https://twitcasting.tv/loft_heaven/movie/685979292', + 'info_dict': { + 'id': '685979292', + 'ext': 'mp4', + 'title': '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」”', + 'duration': 6964.599334, + }, + 'playlist_mincount': 2, }] def _real_extract(self, url): @@ -70,66 +85,49 @@ class TwitCastingIE(InfoExtractor): if video_password: request_data = urlencode_postdata({ 'password': video_password, - }) - webpage = self._download_webpage( + }, encoding='utf-8') + webpage, urlh = self._download_webpage_handle( url, video_id, data=request_data, headers={'Origin': 'https://twitcasting.tv'}) + if urlh.geturl() != url and request_data: + webpage = self._download_webpage( + urlh.geturl(), video_id, data=request_data, + headers={'Origin': 'https://twitcasting.tv'}, + note='Retrying authentication') + # has to check here as the first request can contain password input form even if the password is correct + if re.search(r'<form\s+method="POST">\s*<input\s+[^>]+?name="password"', webpage): + raise ExtractorError('This video is protected by a password, use the --video-password option', expected=True) title = (clean_html(get_element_by_id('movietitle', webpage)) or self._html_search_meta(['og:title', 'twitter:title'], webpage, fatal=True)) - video_js_data = {} - m3u8_url = self._search_regex( - r'data-movie-url=(["\'])(?P<url>(?:(?!\1).)+)\1', - webpage, 'm3u8 url', group='url', default=None) - if not m3u8_url: - video_js_data = self._parse_json(self._search_regex( - r'data-movie-playlist=(["\'])(?P<url>(?:(?!\1).)+)', - webpage, 'movie playlist', group='url', default='[{}]'), video_id) - if isinstance(video_js_data, dict): - video_js_data = list(video_js_data.values())[0] - video_js_data = video_js_data[0] - m3u8_url = try_get(video_js_data, lambda x: x['source']['url']) - - stream_server_data = self._download_json( - 'https://twitcasting.tv/streamserver.php?target=%s&mode=client' % uploader_id, video_id, - 'Downloading live info', fatal=False) - - is_live = 'data-status="online"' in webpage - formats = [] - if is_live and not m3u8_url: - m3u8_url = 'https://twitcasting.tv/%s/metastream.m3u8' % uploader_id - if is_live and has_websockets and stream_server_data: - qq = qualities(['base', 'mobilesource', 'main']) - for mode, ws_url in stream_server_data['llfmp4']['streams'].items(): - formats.append({ - 'url': ws_url, - 'format_id': 'ws-%s' % mode, - 'ext': 'mp4', - 'quality': qq(mode), - 'protocol': 'websocket_frag', # TwitCasting simply sends moof atom directly over WS - }) + video_js_data = try_get( + webpage, + lambda x: self._parse_json(self._search_regex( + r'data-movie-playlist=\'([^\']+?)\'', + x, 'movie playlist', default=None), video_id)['2'], list) - thumbnail = video_js_data.get('thumbnailUrl') or self._og_search_thumbnail(webpage) + thumbnail = traverse_obj(video_js_data, (0, 'thumbnailUrl')) or self._og_search_thumbnail(webpage) description = clean_html(get_element_by_id( 'authorcomment', webpage)) or self._html_search_meta( ['description', 'og:description', 'twitter:description'], webpage) - duration = float_or_none(video_js_data.get( - 'duration'), 1000) or parse_duration(clean_html( - get_element_by_class('tw-player-duration-time', webpage))) + duration = (try_get(video_js_data, lambda x: sum(float_or_none(y.get('duration')) for y in x) / 1000) + or parse_duration(clean_html(get_element_by_class('tw-player-duration-time', webpage)))) view_count = str_to_int(self._search_regex( - r'Total\s*:\s*([\d,]+)\s*Views', webpage, 'views', None)) + (r'Total\s*:\s*([\d,]+)\s*Views', r'総視聴者\s*:\s*([\d,]+)\s*</'), webpage, 'views', None)) timestamp = unified_timestamp(self._search_regex( r'data-toggle="true"[^>]+datetime="([^"]+)"', webpage, 'datetime', None)) - if m3u8_url: - formats.extend(self._extract_m3u8_formats( - m3u8_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', live=is_live)) - self._sort_formats(formats) + stream_server_data = self._download_json( + 'https://twitcasting.tv/streamserver.php?target=%s&mode=client' % uploader_id, video_id, + 'Downloading live info', fatal=False) - return { - 'id': video_id, + is_live = 'data-status="online"' in webpage + if not traverse_obj(stream_server_data, 'llfmp4') and is_live: + self.raise_login_required(method='cookies') + + base_dict = { 'title': title, 'description': description, 'thumbnail': thumbnail, @@ -137,10 +135,75 @@ class TwitCastingIE(InfoExtractor): 'uploader_id': uploader_id, 'duration': duration, 'view_count': view_count, - 'formats': formats, 'is_live': is_live, } + def find_dmu(x): + data_movie_url = self._search_regex( + r'data-movie-url=(["\'])(?P<url>(?:(?!\1).)+)\1', + x, 'm3u8 url', group='url', default=None) + if data_movie_url: + return [data_movie_url] + + m3u8_urls = (try_get(webpage, find_dmu, list) + or traverse_obj(video_js_data, (..., 'source', 'url')) + or ([f'https://twitcasting.tv/{uploader_id}/metastream.m3u8'] if is_live else None)) + if not m3u8_urls: + raise ExtractorError('Failed to get m3u8 playlist') + + if is_live: + m3u8_url = m3u8_urls[0] + formats = self._extract_m3u8_formats( + m3u8_url, video_id, ext='mp4', m3u8_id='hls', + live=True, headers=self._M3U8_HEADERS) + + if traverse_obj(stream_server_data, ('hls', 'source')): + formats.extend(self._extract_m3u8_formats( + m3u8_url, video_id, ext='mp4', m3u8_id='source', + live=True, query={'mode': 'source'}, + note='Downloading source quality m3u8', + headers=self._M3U8_HEADERS, fatal=False)) + + if has_websockets: + qq = qualities(['base', 'mobilesource', 'main']) + streams = traverse_obj(stream_server_data, ('llfmp4', 'streams')) or {} + for mode, ws_url in streams.items(): + formats.append({ + 'url': ws_url, + 'format_id': 'ws-%s' % mode, + 'ext': 'mp4', + 'quality': qq(mode), + 'source_preference': -10, + # TwitCasting simply sends moof atom directly over WS + 'protocol': 'websocket_frag', + }) + + self._sort_formats(formats, ('source',)) + + infodict = { + 'formats': formats + } + else: + infodict = { + '_type': 'multi_video', + 'entries': [{ + 'id': f'{video_id}-{num}', + 'url': m3u8_url, + 'ext': 'mp4', + # Requesting the manifests here will cause download to fail. + # So use ffmpeg instead. See: https://github.com/hypervideo/hypervideo/issues/382 + 'protocol': 'm3u8', + 'http_headers': self._M3U8_HEADERS, + **base_dict, + } for (num, m3u8_url) in enumerate(m3u8_urls)], + } + + return { + 'id': video_id, + **base_dict, + **infodict, + } + class TwitCastingLiveIE(InfoExtractor): _VALID_URL = r'https?://(?:[^/]+\.)?twitcasting\.tv/(?P<id>[^/]+)/?(?:[#?]|$)' @@ -161,6 +224,17 @@ class TwitCastingLiveIE(InfoExtractor): r'tw-sound-flag-open-link" data-id="(\d+)" style=',), webpage, 'current live ID', default=None) if not current_live: + # fetch unfiltered /show to find running livestreams; we can't get ID of the password-protected livestream above + webpage = self._download_webpage( + f'https://twitcasting.tv/{uploader_id}/show/', uploader_id, + note='Downloading live history') + is_live = self._search_regex(r'(?s)(<span\s*class="tw-movie-thumbnail-badge"\s*data-status="live">\s*LIVE)', webpage, 'is live?', default=None) + if is_live: + # get the first live; running live is always at the first + current_live = self._search_regex( + r'(?s)<a\s+class="tw-movie-thumbnail"\s*href="/[^/]+/movie/(?P<video_id>\d+)"\s*>.+?</a>', + webpage, 'current live ID 2', default=None, group='video_id') + if not current_live: raise ExtractorError('The user is not currently live') return self.url_result('https://twitcasting.tv/%s/movie/%s' % (uploader_id, current_live)) |