diff --git a/yt_dlp/extractor/_extractors.py b/yt_dlp/extractor/_extractors.py index 9445270858..3eea0cdf6b 100644 --- a/yt_dlp/extractor/_extractors.py +++ b/yt_dlp/extractor/_extractors.py @@ -2235,6 +2235,7 @@ from .tvplay import ( from .tvplayer import TVPlayerIE from .tvw import ( TvwIE, + TvwNewsIE, TvwTvChannelsIE, ) from .tweakers import TweakersIE diff --git a/yt_dlp/extractor/tvw.py b/yt_dlp/extractor/tvw.py index 0ab926dbdd..74d9b6424b 100644 --- a/yt_dlp/extractor/tvw.py +++ b/yt_dlp/extractor/tvw.py @@ -10,12 +10,15 @@ from ..utils import ( unified_timestamp, url_or_none, ) -from ..utils.traversal import find_element, traverse_obj +from ..utils.traversal import find_element, find_elements, traverse_obj class TvwIE(InfoExtractor): IE_NAME = 'tvw' - _VALID_URL = r'https?://(?:www\.)?tvw\.org/video/(?P[^/?#]+)' + _VALID_URL = [ + r'https?://(?:www\.)?tvw\.org/video/(?P[^/?#]+)', + r'https?://(?:www\.)?tvw\.org/watch/?\?(?:[^#]+&)?eventID=(?P\d+)', + ] _TESTS = [{ 'url': 'https://tvw.org/video/billy-frank-jr-statue-maquette-unveiling-ceremony-2024011211/', 'md5': '9ceb94fe2bb7fd726f74f16356825703', @@ -75,6 +78,20 @@ class TvwIE(InfoExtractor): 'display_id': 'washington-to-washington-a-new-space-race-2022041111', 'categories': ['Washington to Washington', 'General Interest'], }, + }, { + 'url': 'https://tvw.org/watch?eventID=2025041235', + 'md5': '7d697c02f110b37d6a47622ea608ca90', + 'info_dict': { + 'id': '2025041235', + 'ext': 'mp4', + 'title': 'Legislative Review - Medicaid Postpartum Bill Sparks Debate & Senate Approves Automatic Voter Registration', + 'thumbnail': r're:^https?://.*\.(?:jpe?g|png)$', + 'description': 'md5:37d0f3a9187ae520aac261b3959eaee6', + 'timestamp': 1745006400, + 'upload_date': '20250418', + 'location': 'Hayner Media Center', + 'categories': ['Legislative Review'], + }, }] def _real_extract(self, url): @@ -125,6 +142,41 @@ class TvwIE(InfoExtractor): } +class TvwNewsIE(InfoExtractor): + IE_NAME = 'tvw:news' + _VALID_URL = r'https?://(?:www\.)?tvw\.org/\d{4}/\d{2}/(?P[^/?#]+)' + _TESTS = [{ + 'url': 'https://tvw.org/2024/01/the-impact-issues-to-watch-in-the-2024-legislative-session/', + 'info_dict': { + 'id': 'the-impact-issues-to-watch-in-the-2024-legislative-session', + 'title': 'The Impact - Issues to Watch in the 2024 Legislative Session', + 'description': 'md5:65f0b33ec8f18ff1cd401c5547aa5441', + }, + 'playlist_count': 6, + }, { + 'url': 'https://tvw.org/2024/06/the-impact-water-rights-and-the-skookumchuck-dam-debate/', + 'info_dict': { + 'id': 'the-impact-water-rights-and-the-skookumchuck-dam-debate', + 'title': 'The Impact - Water Rights and the Skookumchuck Dam Debate', + 'description': 'md5:185f3a2350ef81e3fa159ac3e040a94b', + }, + 'playlist_count': 1, + }] + + def _real_extract(self, url): + playlist_id = self._match_id(url) + webpage = self._download_webpage(url, playlist_id) + + video_ids = traverse_obj(webpage, ( + {find_elements(cls='invintus-player', html=True)}, ..., {extract_attributes}, 'data-eventid')) + + return self.playlist_from_matches( + video_ids, playlist_id, + playlist_title=remove_end(self._og_search_title(webpage, default=None), ' - TVW'), + playlist_description=self._og_search_description(webpage, default=None), + getter=lambda x: f'https://tvw.org/watch?eventID={x}', ie=TvwIE) + + class TvwTvChannelsIE(InfoExtractor): IE_NAME = 'tvw:tvchannels' _VALID_URL = r'https?://(?:www\.)?tvw\.org/tvchannels/(?P[^/?#]+)'