From 41a241ca6ffb95b3d9aaf4f42106ca8cba9af1a6 Mon Sep 17 00:00:00 2001 From: sepro Date: Sat, 14 Sep 2024 02:16:34 +0200 Subject: [PATCH] [ie/Sen] Add extractor (#10952) Closes #10951 Authored by: seproDev --- yt_dlp/extractor/_extractors.py | 1 + yt_dlp/extractor/sen.py | 36 +++++++++++++++++++++++++++++++++ 2 files changed, 37 insertions(+) create mode 100644 yt_dlp/extractor/sen.py diff --git a/yt_dlp/extractor/_extractors.py b/yt_dlp/extractor/_extractors.py index 597876197..d8abf0b5d 100644 --- a/yt_dlp/extractor/_extractors.py +++ b/yt_dlp/extractor/_extractors.py @@ -1823,6 +1823,7 @@ SCTECourseIE, ) from .sejmpl import SejmIE +from .sen import SenIE from .senalcolombia import SenalColombiaLiveIE from .senategov import ( SenateGovIE, diff --git a/yt_dlp/extractor/sen.py b/yt_dlp/extractor/sen.py new file mode 100644 index 000000000..d8f14ecdc --- /dev/null +++ b/yt_dlp/extractor/sen.py @@ -0,0 +1,36 @@ +from .common import InfoExtractor +from ..utils import url_or_none +from ..utils.traversal import traverse_obj + + +class SenIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?sen\.com/video/(?P[0-9a-f-]+)' + _TEST = { + 'url': 'https://www.sen.com/video/eef46eb1-4d79-4e28-be9d-bd937767f8c4', + 'md5': 'ff615aca9691053c94f8f10d96cd7884', + 'info_dict': { + 'id': 'eef46eb1-4d79-4e28-be9d-bd937767f8c4', + 'ext': 'mp4', + 'description': 'Florida, 28 Sep 2022', + 'title': 'Hurricane Ian', + 'tags': ['North America', 'Storm', 'Weather'], + }, + } + + def _real_extract(self, url): + video_id = self._match_id(url) + + api_data = self._download_json(f'https://api.sen.com/content/public/video/{video_id}', video_id) + m3u8_url = (traverse_obj(api_data, ( + 'data', 'nodes', lambda _, v: v['id'] == 'player', 'video', 'url', {url_or_none}, any)) + or f'https://vod.sen.com/videos/{video_id}/manifest.m3u8') + + return { + 'id': video_id, + 'formats': self._extract_m3u8_formats(m3u8_url, video_id, 'mp4'), + **traverse_obj(api_data, ('data', 'nodes', lambda _, v: v['id'] == 'details', any, 'content', { + 'title': ('title', 'text', {str}), + 'description': ('descriptions', 0, 'text', {str}), + 'tags': ('badges', ..., 'text', {str}), + })), + }