diff options
| author | Alexander Seiler <seileralex@gmail.com> | 2018-12-06 21:41:02 +0100 |
|---|---|---|
| committer | Sergey M <dstftw@gmail.com> | 2018-12-07 03:41:02 +0700 |
| commit | 8c5879715f4d979b83c49d44a9094307247097ba (patch) | |
| tree | 0d8c57f4a268ec5f3e44f9bda790c102db3c9293 | |
| parent | ebb0449049c198f04103502c95a13171b854d1c7 (diff) | |
[ard:mediathek] Fix title and description extraction (closes #18349)
| -rw-r--r-- | youtube_dl/extractor/ard.py | 9 |
1 files changed, 7 insertions, 2 deletions
diff --git a/youtube_dl/extractor/ard.py b/youtube_dl/extractor/ard.py index 6bf8f61eb..84e96f769 100644 --- a/youtube_dl/extractor/ard.py +++ b/youtube_dl/extractor/ard.py @@ -173,13 +173,18 @@ class ARDMediathekIE(InfoExtractor): title = self._html_search_regex( [r'<h1(?:\s+class="boxTopHeadline")?>(.*?)</h1>', r'<meta name="dcterms\.title" content="(.*?)"/>', - r'<h4 class="headline">(.*?)</h4>'], + r'<h4 class="headline">(.*?)</h4>', + r'<title[^>]*>(.*?)</title>'], webpage, 'title') description = self._html_search_meta( 'dcterms.abstract', webpage, 'description', default=None) if description is None: description = self._html_search_meta( - 'description', webpage, 'meta description') + 'description', webpage, 'meta description', default=None) + if description is None: + description = self._html_search_regex( + r'<p\s+class="teasertext">(.+?)</p>', + webpage, 'teaser text', default=None) # Thumbnail is sometimes not present. # It is in the mobile version, but that seems to use a different URL |