diff options
author | Cédric Bonhomme <kimble.mandel@gmail.com> | 2014-01-21 07:51:49 +0100 |
---|---|---|
committer | Cédric Bonhomme <kimble.mandel@gmail.com> | 2014-01-21 07:51:49 +0100 |
commit | 918790b5e51fb21343ec001a24b770ab188203f9 (patch) | |
tree | 36ab99d4413c16a6bfd25bbbad97a1754e703397 /pyaggr3g470r/feedgetter.py | |
parent | Updated default configuration file and README. (diff) | |
download | newspipe-918790b5e51fb21343ec001a24b770ab188203f9.tar.gz newspipe-918790b5e51fb21343ec001a24b770ab188203f9.tar.bz2 newspipe-918790b5e51fb21343ec001a24b770ab188203f9.zip |
Added an option to choose if URL of articles behind proxies should be resolved.
Diffstat (limited to 'pyaggr3g470r/feedgetter.py')
-rw-r--r-- | pyaggr3g470r/feedgetter.py | 21 |
1 files changed, 11 insertions, 10 deletions
diff --git a/pyaggr3g470r/feedgetter.py b/pyaggr3g470r/feedgetter.py index 3b77509f..d29d3c9e 100644 --- a/pyaggr3g470r/feedgetter.py +++ b/pyaggr3g470r/feedgetter.py @@ -110,16 +110,17 @@ class FeedGetter(object): for article in a_feed['entries']: nice_url = article.link.encode("utf-8") - try: - # resolves URL behind proxies (like feedproxy.google.com) - r = requests.get(article.link, timeout=5.0, proxies=self.proxies) - nice_url = r.url.encode("utf-8") - except Timeout: - pyaggr3g470r_log.warning("Timeout when getting the real URL of %s." % (article.link,)) - continue - except Exception as e: - pyaggr3g470r_log.warning("Unable to get the real URL of %s. Error: %s" % (article.link, str(e))) - continue + if conf.RESOLVE_ARTICLE_URL: + try: + # resolves URL behind proxies (like feedproxy.google.com) + r = requests.get(article.link, timeout=5.0, proxies=self.proxies) + nice_url = r.url.encode("utf-8") + except Timeout: + pyaggr3g470r_log.warning("Timeout when getting the real URL of %s." % (article.link,)) + continue + except Exception as e: + pyaggr3g470r_log.warning("Unable to get the real URL of %s. Error: %s" % (article.link, str(e))) + continue # remove utm_* parameters parsed_url = urlparse(nice_url) qd = parse_qs(parsed_url.query, keep_blank_values=True) |