[ie/crunchyroll] Fix stream extraction (#10005)

[yt-dlp.git] / yt_dlp / extractor / twitcasting.py

diff --git a/yt_dlp/extractor/twitcasting.py b/yt_dlp/extractor/twitcasting.py
index 32a38c86efb8221fc6e9e932d880ee1988653766..28ea16cc21a406a8f1a3047c7b47cf1ac7226715 100644 (file)
--- a/yt_dlp/extractor/twitcasting.py
+++ b/yt_dlp/extractor/twitcasting.py
@@ -247,25 +247,26 @@ def _real_extract(self, url):
             'Downloading live video of user {0}. '
             'Pass "https://twitcasting.tv/{0}/show" to download the history'.format(uploader_id))
 
-        webpage = self._download_webpage(url, uploader_id)
+        is_live = traverse_obj(self._download_json(
+            f'https://frontendapi.twitcasting.tv/watch/user/{uploader_id}',
+            uploader_id, 'Checking live status', data=b'', fatal=False), ('is_live', {bool}))
+        if is_live is False:  # only raise here if API response was as expected
+            raise UserNotLive(video_id=uploader_id)
+
+        # Use /show/ page so that password-protected and members-only livestreams can be found
+        webpage = self._download_webpage(
+            f'https://twitcasting.tv/{uploader_id}/show/', uploader_id, 'Downloading live history')
+        is_live = is_live or self._search_regex(
+            r'(?s)(<span\s*class="tw-movie-thumbnail2-badge"\s*data-status="live">\s*LIVE)',
+            webpage, 'is live?', default=False)
+        # Current live is always the first match
         current_live = self._search_regex(
-            (r'data-type="movie" data-id="(\d+)">',
-             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:
+            r'(?s)<a\s+class="tw-movie-thumbnail2"\s+href="/[^/"]+/movie/(?P<video_id>\d+)"',
+            webpage, 'current live ID', default=None, group='video_id')
+        if not is_live or not current_live:
             raise UserNotLive(video_id=uploader_id)
-        return self.url_result('https://twitcasting.tv/%s/movie/%s' % (uploader_id, current_live))
+
+        return self.url_result(f'https://twitcasting.tv/{uploader_id}/movie/{current_live}', TwitCastingIE)
 
 
 class TwitCastingUserIE(InfoExtractor):
@@ -288,8 +289,7 @@ def _entries(self, uploader_id):
             webpage = self._download_webpage(
                 next_url, uploader_id, query={'filter': 'watchable'}, note='Downloading page %d' % page_num)
             matches = re.finditer(
-                r'''(?isx)<a\s+class="tw-movie-thumbnail"\s*href="(?P<url>/[^/]+/movie/\d+)"\s*>.+?</a>''',
-                webpage)
+                r'(?s)<a\s+class="tw-movie-thumbnail2"\s+href="(?P<url>/[^/"]+/movie/\d+)"', webpage)
             for mobj in matches:
                 yield self.url_result(urljoin(base_url, mobj.group('url')))