diff options
| author | Philipp Hagemeister <phihag@phihag.de> | 2012-12-20 21:28:32 +0100 |
|---|---|---|
| committer | Philipp Hagemeister <phihag@phihag.de> | 2012-12-20 21:28:32 +0100 |
| commit | a7c0f8602e91cc96962c7eade10860b61afc3728 (patch) | |
| tree | 01182f741a7b474d91e565e4beb91cc577e726d8 /youtube_dl/utils.py | |
| parent | 21a9c6aaac074b3ad85adc55818573e7971f8043 (diff) | |
| parent | 162e3c52616a72f4ddc11b0e5de0f2425e512896 (diff) | |
Merge branch 'master' of github.com:rg3/youtube-dl
Diffstat (limited to 'youtube_dl/utils.py')
| -rw-r--r-- | youtube_dl/utils.py | 11 |
1 files changed, 2 insertions, 9 deletions
diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index a5196b0ae..463804e18 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -298,7 +298,8 @@ def clean_html(html): """Clean an HTML snippet into a readable string""" # Newline vs <br /> html = html.replace('\n', ' ') - html = re.sub('\s*<\s*br\s*/?\s*>\s*', '\n', html) + html = re.sub(r'\s*<\s*br\s*/?\s*>\s*', '\n', html) + html = re.sub(r'<\s*/\s*p\s*>\s*<\s*p[^>]*>', '\n', html) # Strip html tags html = re.sub('<.*?>', '', html) # Replace html entities @@ -465,14 +466,6 @@ class ContentTooShortError(Exception): self.downloaded = downloaded self.expected = expected - -class Trouble(Exception): - """Trouble helper exception - - This is an exception to be handled with - FileDownloader.trouble - """ - class YoutubeDLHandler(compat_urllib_request.HTTPHandler): """Handler for HTTP requests and responses. |