diff options
author | Sergey M․ <dstftw@gmail.com> | 2015-03-18 00:05:40 +0600 |
---|---|---|
committer | Sergey M․ <dstftw@gmail.com> | 2015-03-18 00:05:40 +0600 |
commit | ed9a25dd612fb06d9cf007a6491ac9982535a8f9 (patch) | |
tree | 07dc954c0604bb089d3046c7ebd75e5606fd1cc5 | |
parent | 84f810160667f9129e6a8d841fae2d0e2e1fec86 (diff) |
[generic] Generalize redirect regex
-rw-r--r-- | youtube_dl/extractor/generic.py | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 0b8d96c27..dc5755d12 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -1268,16 +1268,16 @@ class GenericIE(InfoExtractor): # HTML5 video found = re.findall(r'(?s)<video[^<]*(?:>.*?<source[^>]*)?\s+src=["\'](.*?)["\']', webpage) if not found: + REDIRECT_REGEX = r'[0-9]{,2};\s*(?:URL|url)=\'?([^\'"]+)' found = re.search( r'(?i)<meta\s+(?=(?:[a-z-]+="[^"]+"\s+)*http-equiv="refresh")' - r'(?:[a-z-]+="[^"]+"\s+)*?content="[0-9]{,2};\s*(?:URL|url)=\'?([^\'"]+)', + r'(?:[a-z-]+="[^"]+"\s+)*?content="%s' % REDIRECT_REGEX, webpage) if not found: # Look also in Refresh HTTP header refresh_header = head_response.headers.get('Refresh') if refresh_header: - found = re.search( - r'[0-9]{,2};\s*(?:URL|url)=(.+)', refresh_header) + found = re.search(REDIRECT_REGEX, refresh_header) if found: new_url = found.group(1) self.report_following_redirect(new_url) |