From f3411af12e209bc5624e1ac31271b8aabe2d3c90 Mon Sep 17 00:00:00 2001 From: megumin <34505936+megumintyan@users.noreply.github.com> Date: Tue, 25 Jun 2024 01:49:09 +0300 Subject: [PATCH] [ie/matchtv] Fix extractor (#10190) Authored by: megumintyan --- yt_dlp/extractor/matchtv.py | 40 +++++++++++-------------------------- 1 file changed, 12 insertions(+), 28 deletions(-) diff --git a/yt_dlp/extractor/matchtv.py b/yt_dlp/extractor/matchtv.py index a67fa9fe4..93799fe85 100644 --- a/yt_dlp/extractor/matchtv.py +++ b/yt_dlp/extractor/matchtv.py @@ -1,51 +1,35 @@ -import random - from .common import InfoExtractor -from ..utils import xpath_text class MatchTVIE(InfoExtractor): - _VALID_URL = r'https?://matchtv\.ru(?:/on-air|/?#live-player)' + _VALID_URL = [ + r'https?://matchtv\.ru/on-air/?(?:$|[?#])', + r'https?://video\.matchtv\.ru/iframe/channel/106/?(?:$|[?#])', + ] _TESTS = [{ - 'url': 'http://matchtv.ru/#live-player', + 'url': 'http://matchtv.ru/on-air/', 'info_dict': { 'id': 'matchtv-live', - 'ext': 'flv', + 'ext': 'mp4', 'title': r're:^Матч ТВ - Прямой эфир \d{4}-\d{2}-\d{2} \d{2}:\d{2}$', - 'is_live': True, + 'live_status': 'is_live', }, 'params': { 'skip_download': True, }, }, { - 'url': 'http://matchtv.ru/on-air/', + 'url': 'https://video.matchtv.ru/iframe/channel/106', 'only_matching': True, }] def _real_extract(self, url): video_id = 'matchtv-live' - video_url = self._download_json( - 'http://player.matchtv.ntvplus.tv/player/smil', video_id, - query={ - 'ts': '', - 'quality': 'SD', - 'contentId': '561d2c0df7159b37178b4567', - 'sign': '', - 'includeHighlights': '0', - 'userId': '', - 'sessionId': random.randint(1, 1000000000), - 'contentType': 'channel', - 'timeShift': '0', - 'platform': 'portal', - }, - headers={ - 'Referer': 'http://player.matchtv.ntvplus.tv/embed-player/NTVEmbedPlayer.swf', - })['data']['videoUrl'] - f4m_url = xpath_text(self._download_xml(video_url, video_id), './to') - formats = self._extract_f4m_formats(f4m_url, video_id) + webpage = self._download_webpage('https://video.matchtv.ru/iframe/channel/106', video_id) + video_url = self._html_search_regex( + r'data-config="config=(https?://[^?"]+)[?"]', webpage, 'video URL').replace('/feed/', '/media/') + '.m3u8' return { 'id': video_id, 'title': 'Матч ТВ - Прямой эфир', 'is_live': True, - 'formats': formats, + 'formats': self._extract_m3u8_formats(video_url, video_id, 'mp4', live=True), }