commit: bcf150e43514e2308355a94a513020d99d594f2a
parent 240f26229d2ef655a6d16e067094a24ab77bce29
Author: Sergey M․ <dstftw@gmail.com>
Date: Mon, 5 Feb 2018 23:43:07 +0700
[telebruxelles] Fix extraction (closes #15504)
Diffstat:
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/youtube_dl/extractor/telebruxelles.py b/youtube_dl/extractor/telebruxelles.py
@@ -38,13 +38,14 @@ class TeleBruxellesIE(InfoExtractor):
webpage = self._download_webpage(url, display_id)
article_id = self._html_search_regex(
- r"<article id=\"post-(\d+)\"", webpage, 'article ID', default=None)
+ r'<article[^>]+\bid=["\']post-(\d+)', webpage, 'article ID', default=None)
title = self._html_search_regex(
- r'<h1 class=\"entry-title\">(.*?)</h1>', webpage, 'title')
+ r'<h1[^>]*>(.+?)</h1>', webpage, 'title',
+ default=None) or self._og_search_title(webpage)
description = self._og_search_description(webpage, default=None)
rtmp_url = self._html_search_regex(
- r'file\s*:\s*"(rtmp://[^/]+/vod/mp4:"\s*\+\s*"[^"]+"\s*\+\s*".mp4)"',
+ r'file\s*:\s*"(rtmps?://[^/]+/vod/mp4:"\s*\+\s*"[^"]+"\s*\+\s*".mp4)"',
webpage, 'RTMP url')
rtmp_url = re.sub(r'"\s*\+\s*"', '', rtmp_url)
formats = self._extract_wowza_formats(rtmp_url, article_id or display_id)