diff options
| author | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2013-12-26 15:00:09 +0100 |
|---|---|---|
| committer | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2013-12-26 15:00:09 +0100 |
| commit | db3b0dbc41a814f9185eca81cddc7caf47b9f04e (patch) | |
| tree | ff31d9ab801d3a22d87ecd62207bce8ca68d3ea2 /pyaggr3g470r | |
| parent | Get the 'real' url. (diff) | |
| download | newspipe-db3b0dbc41a814f9185eca81cddc7caf47b9f04e.tar.gz newspipe-db3b0dbc41a814f9185eca81cddc7caf47b9f04e.tar.bz2 newspipe-db3b0dbc41a814f9185eca81cddc7caf47b9f04e.zip | |
If it is not possible to get the real URL, uses the URL provided by feedparser.
Diffstat (limited to 'pyaggr3g470r')
| -rw-r--r-- | pyaggr3g470r/feedgetter.py | 9 |
1 files changed, 6 insertions, 3 deletions
diff --git a/pyaggr3g470r/feedgetter.py b/pyaggr3g470r/feedgetter.py index c0d93c50..d3195fb0 100644 --- a/pyaggr3g470r/feedgetter.py +++ b/pyaggr3g470r/feedgetter.py @@ -95,9 +95,12 @@ class FeedGetter(object): articles = [] for article in a_feed['entries']: - r = requests.get(article.link) - parsed_url = urlparse(r.url) - real_url = parsed_url.scheme + '://' + parsed_url.netloc + parsed_url.path + try: + r = requests.get(article.link) + parsed_url = urlparse(r.url) + real_url = parsed_url.scheme + '://' + parsed_url.netloc + parsed_url.path + except: + real_url = article.link if models.Article.objects(link=real_url).first() != None: # if article already in the database continue with the next article |