From 785ab1af7f131e73444634ad57b39478651a43d3 Mon Sep 17 00:00:00 2001 From: Xpl0itU <24777100+Xpl0itU@users.noreply.github.com> Date: Sun, 10 Mar 2024 00:03:18 +0100 Subject: [PATCH] [ie/crtvg] Fix `_VALID_URL` (#9404) Authored by: Xpl0itU --- yt_dlp/extractor/crtvg.py | 25 ++++++++++++++++++++++--- 1 file changed, 22 insertions(+), 3 deletions(-) diff --git a/yt_dlp/extractor/crtvg.py b/yt_dlp/extractor/crtvg.py index 1aa8d77055..21325e331d 100644 --- a/yt_dlp/extractor/crtvg.py +++ b/yt_dlp/extractor/crtvg.py @@ -1,18 +1,32 @@ +import re + from .common import InfoExtractor -from ..utils import remove_end +from ..utils import make_archive_id, remove_end class CrtvgIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?crtvg\.es/tvg/a-carta/[^/#?]+-(?P\d+)' + _VALID_URL = r'https?://(?:www\.)?crtvg\.es/tvg/a-carta/(?P[^/#?]+)' _TESTS = [{ 'url': 'https://www.crtvg.es/tvg/a-carta/os-caimans-do-tea-5839623', 'md5': 'c0958d9ff90e4503a75544358758921d', 'info_dict': { - 'id': '5839623', + 'id': 'os-caimans-do-tea-5839623', 'title': 'Os caimáns do Tea', 'ext': 'mp4', 'description': 'md5:f71cfba21ae564f0a6f415b31de1f842', 'thumbnail': r're:^https?://.*\.(?:jpg|png)', + '_old_archive_ids': ['crtvg 5839623'], + }, + 'params': {'skip_download': 'm3u8'} + }, { + 'url': 'https://www.crtvg.es/tvg/a-carta/a-parabolica-love-story', + 'md5': '9a47b95a1749db7b7eb3214904624584', + 'info_dict': { + 'id': 'a-parabolica-love-story', + 'title': 'A parabólica / Trabuco, o can mordedor / Love Story', + 'ext': 'mp4', + 'description': 'md5:f71cfba21ae564f0a6f415b31de1f842', + 'thumbnail': r're:^https?://.*\.(?:jpg|png)', }, 'params': {'skip_download': 'm3u8'} }] @@ -24,8 +38,13 @@ class CrtvgIE(InfoExtractor): formats = self._extract_m3u8_formats(video_url + '/playlist.m3u8', video_id, fatal=False) formats.extend(self._extract_mpd_formats(video_url + '/manifest.mpd', video_id, fatal=False)) + old_video_id = None + if mobj := re.fullmatch(r'[^/#?]+-(?P\d{7})', video_id): + old_video_id = [make_archive_id(self, mobj.group('old_id'))] + return { 'id': video_id, + '_old_archive_ids': old_video_id, 'formats': formats, 'title': remove_end(self._html_search_meta( ['og:title', 'twitter:title'], webpage, 'title', default=None), ' | CRTVG'),