From c1ea7f5a242dbaad3a4646d73bb1cae22439a788 Mon Sep 17 00:00:00 2001
From: dirkf <fieldhouse@gmx.net>
Date: Wed, 26 Mar 2025 12:17:49 +0000
Subject: [PATCH] [ITV] Mark ITVX not working * update old shim * correct
[_]IE_DESC
---
youtube_dl/extractor/itv.py | 17 +++++------------
1 file changed, 5 insertions(+), 12 deletions(-)
diff --git a/youtube_dl/extractor/itv.py b/youtube_dl/extractor/itv.py
index c64af3be6..2510ad887 100644
--- a/youtube_dl/extractor/itv.py
+++ b/youtube_dl/extractor/itv.py
@@ -35,15 +35,6 @@ from ..utils import (
class ITVBaseIE(InfoExtractor):
- def _search_nextjs_data(self, webpage, video_id, **kw):
- transform_source = kw.pop('transform_source', None)
- fatal = kw.pop('fatal', True)
- return self._parse_json(
- self._search_regex(
- r'''<script\b[^>]+\bid=('|")__NEXT_DATA__\1[^>]*>(?P<js>[^<]+)</script>''',
- webpage, 'next.js data', group='js', fatal=fatal, **kw),
- video_id, transform_source=transform_source, fatal=fatal)
-
def __handle_request_webpage_error(self, err, video_id=None, errnote=None, fatal=True):
if errnote is False:
return False
@@ -109,7 +100,9 @@ class ITVBaseIE(InfoExtractor):
class ITVIE(ITVBaseIE):
_VALID_URL = r'https?://(?:www\.)?itv\.com/(?:(?P<w>watch)|hub)/[^/]+/(?(w)[\w-]+/)(?P<id>\w+)'
- _IE_DESC = 'ITVX'
+ IE_DESC = 'ITVX'
+ _WORKING = False
+
_TESTS = [{
'note': 'Hub URLs redirect to ITVX',
'url': 'https://www.itv.com/hub/liar/2a4547a0012',
@@ -270,7 +263,7 @@ class ITVIE(ITVBaseIE):
'ext': determine_ext(href, 'vtt'),
})
- next_data = self._search_nextjs_data(webpage, video_id, fatal=False, default='{}')
+ next_data = self._search_nextjs_data(webpage, video_id, fatal=False, default={})
video_data.update(traverse_obj(next_data, ('props', 'pageProps', ('title', 'episode')), expected_type=dict)[0] or {})
title = traverse_obj(video_data, 'headerTitle', 'episodeTitle')
info = self._og_extract(webpage, require_title=not title)
@@ -323,7 +316,7 @@ class ITVIE(ITVBaseIE):
class ITVBTCCIE(ITVBaseIE):
_VALID_URL = r'https?://(?:www\.)?itv\.com/(?!(?:watch|hub)/)(?:[^/]+/)+(?P<id>[^/?#&]+)'
- _IE_DESC = 'ITV articles: News, British Touring Car Championship'
+ IE_DESC = 'ITV articles: News, British Touring Car Championship'
_TESTS = [{
'note': 'British Touring Car Championship',
'url': 'https://www.itv.com/btcc/articles/btcc-2018-all-the-action-from-brands-hatch',