[twitcasting] Refactor extractor (#2310)

Co-authored-by: Ashish Gupta <39122144+Ashish0804@users.noreply.github.com>
Co-authored-by: pukkandan <pukkandan.ytdlp@gmail.com>

Authored by: Lesmiscore

yt_dlp/extractor/twitcasting.py

@@ -8,6 +8,7 @@
 from ..downloader.websocket import has_websockets
 from ..utils import (
     clean_html,
+    ExtractorError,
     float_or_none,
     get_element_by_class,
     get_element_by_id,
@@ -19,12 +20,15 @@
     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',
@@ -61,6 +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):
@@ -79,63 +93,33 @@ def _real_extract(self, url):
         title = (clean_html(get_element_by_id('movietitle', webpage))
                  or self._html_search_meta(['og:title', 'twitter:title'], webpage, fatal=True))
 
-        video_js_data = {}
+        video_js_data = try_get(
-        m3u8_url = self._search_regex(
+            webpage,
-            r'data-movie-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
+            lambda x: self._parse_json(self._search_regex(
-            webpage, 'm3u8 url', group='url', default=None)
+                r'data-movie-playlist=\'([^\']+?)\'',
-        if not m3u8_url:
+                x, 'movie playlist', default=None), video_id)['2'], list)
-            video_js_data = self._parse_json(self._search_regex(
+
-                r'data-movie-playlist=(["\'])(?P<url>(?:(?!\1).)+)',
+        thumbnail = traverse_obj(video_js_data, (0, 'thumbnailUrl')) or self._og_search_thumbnail(webpage)
-                webpage, 'movie playlist', group='url', default='[{}]'), video_id)
+        description = clean_html(get_element_by_id(
-            if isinstance(video_js_data, dict):
+            'authorcomment', webpage)) or self._html_search_meta(
-                video_js_data = list(video_js_data.values())[0]
+            ['description', 'og:description', 'twitter:description'], webpage)
-            video_js_data = video_js_data[0]
+        duration = (try_get(video_js_data, lambda x: sum(float_or_none(y.get('duration')) for y in x) / 1000)
-            m3u8_url = try_get(video_js_data, lambda x: x['source']['url'])
+                    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', 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))
 
         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
-
         if not traverse_obj(stream_server_data, 'llfmp4') and is_live:
             self.raise_login_required(method='cookies')
 
-        formats = []
+        base_dict = {
-        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'])
-            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),
-                    'protocol': 'websocket_frag',  # TwitCasting simply sends moof atom directly over WS
-                })
-
-        thumbnail = video_js_data.get('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)))
-        view_count = str_to_int(self._search_regex(
-            r'Total\s*:\s*([\d,]+)\s*Views', 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)
-
-        return {
-            'id': video_id,
             'title': title,
             'description': description,
             'thumbnail': thumbnail,
@@ -143,10 +127,73 @@ def _real_extract(self, url):
             '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)
+
+            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),
+                        # TwitCasting simply sends moof atom directly over WS
+                        'protocol': 'websocket_frag',
+                    })
+
+            self._sort_formats(formats)
+
+            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/yt-dlp/yt-dlp/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>[^/]+)/?(?:[#?]|$)'