From c7bab7c39fd51c0812f70020172766303191bc01 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Sun, 5 May 2024 19:37:42 +0200 Subject: Adding upstream version 2023.03.04. Signed-off-by: Daniel Baumann --- yt_dlp/extractor/unistra.py | 64 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 64 insertions(+) create mode 100644 yt_dlp/extractor/unistra.py (limited to 'yt_dlp/extractor/unistra.py') diff --git a/yt_dlp/extractor/unistra.py b/yt_dlp/extractor/unistra.py new file mode 100644 index 0000000..6e872cd --- /dev/null +++ b/yt_dlp/extractor/unistra.py @@ -0,0 +1,64 @@ +import re + +from .common import InfoExtractor +from ..utils import qualities + + +class UnistraIE(InfoExtractor): + _VALID_URL = r'https?://utv\.unistra\.fr/(?:index|video)\.php\?id_video\=(?P\d+)' + + _TESTS = [ + { + 'url': 'http://utv.unistra.fr/video.php?id_video=154', + 'md5': '736f605cfdc96724d55bb543ab3ced24', + 'info_dict': { + 'id': '154', + 'ext': 'mp4', + 'title': 'M!ss Yella', + 'description': 'md5:104892c71bd48e55d70b902736b81bbf', + }, + }, + { + 'url': 'http://utv.unistra.fr/index.php?id_video=437', + 'md5': '1ddddd6cccaae76f622ce29b8779636d', + 'info_dict': { + 'id': '437', + 'ext': 'mp4', + 'title': 'Prix Louise Weiss 2014', + 'description': 'md5:cc3a8735f079f4fb6b0b570fc10c135a', + }, + } + ] + + def _real_extract(self, url): + mobj = self._match_valid_url(url) + video_id = mobj.group('id') + + webpage = self._download_webpage(url, video_id) + + files = set(re.findall(r'file\s*:\s*"(/[^"]+)"', webpage)) + + quality = qualities(['SD', 'HD']) + formats = [] + for file_path in files: + format_id = 'HD' if file_path.endswith('-HD.mp4') else 'SD' + formats.append({ + 'url': 'http://vod-flash.u-strasbg.fr:8080%s' % file_path, + 'format_id': format_id, + 'quality': quality(format_id) + }) + + title = self._html_search_regex( + r'UTV - (.*?)</', webpage, 'title') + description = self._html_search_regex( + r'<meta name="Description" content="(.*?)"', webpage, 'description', flags=re.DOTALL) + thumbnail = self._search_regex( + r'image: "(.*?)"', webpage, 'thumbnail') + + return { + 'id': video_id, + 'title': title, + 'description': description, + 'thumbnail': thumbnail, + 'formats': formats + } -- cgit v1.2.3