New limits are possible: time limit, max number of item fetched, and max number of item taken from cache. Fill third argument is now Fast=True, which is self-explicit. (Complexity of the changes made separate commits impossible).
The SERVER variable is no longer needed. RSS .xml file is now cached for a very short time, so as to make loading faster, and hopefully reduce bann a little. Use a more common User-Agent to try to cut down bann. Added ability to test whether a key is in the Cache.
No longer writes everytime a value is added, since it could cause some issues if two instances of the script were run at the same time. Now it only writes when the Cache object is no longer in use (ie. garbage colllected).
Makes the whole "xpath rules" things useless. Almost any feed is now supported. CSS liferea stylesheets are also uneeded now, since readability cleans up html code a more efficient way. README was updated.
Now uses OOP where it fits. Atom feeds are supported, but no real tests were made. Unix globbing is now possible for urls. Caching is done a cleaner way. Feedburner links are also replaced. HTML is cleaned a more efficient way. Code is now much cleaner, using lxml.objectify and a small wrapper to access Atom feeds as if they were RSS feeds (and much faster than feedparser). README has been updated.
This is a huge commit. The whole code is ported to Object-Oritented Programming. This makes the code cleaner, which became required to deal with all the different cases, for example with encoding detection. Encoding detection now works better, and uses 3 different methods. HTML pages with an xml declaration are now supported. Feed urls with parameters (eg. "index.php?option=par") are also supported. Cache is now smarter, since it no longer grows indefinitely, since only in-use pages are kept in the cache. Caching is now mandatory. urllib (not urllib2) is no longer needed. Solved a possible crash with log function (when passing list of str with non-unicode encoging).
README is also updated.
pictuga