From a13c91a722dd4ff6b95556292601cd4175e6dd44 Mon Sep 17 00:00:00 2001 From: Alexander Seiler Date: Fri, 7 Dec 2018 05:02:11 +0100 Subject: [PATCH] [regiotv] Fix broken extractor (closes #15348) --- youtube_dl/extractor/regiotv.py | 47 +++++++++------------------------ 1 file changed, 13 insertions(+), 34 deletions(-) diff --git a/youtube_dl/extractor/regiotv.py b/youtube_dl/extractor/regiotv.py index e250a52f0..a3a732e21 100644 --- a/youtube_dl/extractor/regiotv.py +++ b/youtube_dl/extractor/regiotv.py @@ -3,59 +3,38 @@ from __future__ import unicode_literals from .common import InfoExtractor -from ..utils import ( - sanitized_Request, - xpath_text, - xpath_with_ns, -) - class RegioTVIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?regio-tv\.de/video/(?P[0-9]+)' - _TESTS = [{ - 'url': 'http://www.regio-tv.de/video/395808.html', + _VALID_URL = r'https?://(?:www\.)?regio-tv\.de/.*vidid,(?P[0-9]+)' + _TEST = { + 'url': 'https://www.regio-tv.de/video_video,-nikolaus-besucht-sindelfinger-markt-_vidid,151732.html', 'info_dict': { - 'id': '395808', + 'id': '151732', 'ext': 'mp4', - 'title': 'Wir in Ludwigsburg', - 'description': 'Mit unseren zuckersüßen Adventskindern, außerdem besuchen wir die Abendsterne!', + 'title': 'Nikolaus besucht Sindelfinger Markt', + 'description': 'md5:205f8c1aa68fcc110c54a4e82b4ce43b', } - }, { - 'url': 'http://www.regio-tv.de/video/395808', - 'only_matching': True, - }] + } def _real_extract(self, url): video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) key = self._search_regex( - r'key\s*:\s*(["\'])(?P.+?)\1', webpage, 'key', group='key') + r'key\s*=\s*(?P[0-9a-f]+)', webpage, 'key', group='key') title = self._og_search_title(webpage) - SOAP_TEMPLATE = '<{0} xmlns="http://v.telvi.de/">{1}' - - request = sanitized_Request( - 'http://v.telvi.de/', - SOAP_TEMPLATE.format('GetHTML5VideoData', key).encode('utf-8')) - video_data = self._download_xml(request, video_id, 'Downloading video XML') - - NS_MAP = { - 'xsi': 'http://www.w3.org/2001/XMLSchema-instance', - 'soap': 'http://schemas.xmlsoap.org/soap/envelope/', - } - - video_url = xpath_text( - video_data, xpath_with_ns('.//video', NS_MAP), 'video url', fatal=True) - thumbnail = xpath_text( - video_data, xpath_with_ns('.//image', NS_MAP), 'thumbnail') + video_url = 'https://vimp.schwaebische.de/getMedia.php?key=%s&type=mp4' % key description = self._og_search_description( webpage) or self._html_search_meta('description', webpage) + thumbnail = self._og_search_thumbnail(webpage) + if thumbnail and thumbnail.startswith('/'): + thumbnail = 'https://regio-tv.de' + thumbnail return { 'id': video_id, 'url': video_url, + 'ext': 'mp4', 'title': title, 'description': description, 'thumbnail': thumbnail,