Willow for Internet content filtering

Description

Currently testing this software for content filtering. The nice thing about Willow is: it uses a Bayesian filter (well known for filtering out SPAM). In short: when a page is requested it determines on the fly if the page is “good” or “bad”. To determine this it uses a set of good and bad pages (the set which is included in the package seems to work well). No need for crappy white/black lists! Personally I couldn’t find any porn website which slipped through the filter.

Notes

• It depends on an multiverse package python2.4-profiles.

• The webinterface works by using the user configured in willow.conf and the password hardcoded on line 1607 of willow.py
• The documentation is a bit scarce. You’ll need your linux experience to set it up.
• The page which appears if you visit a blocked site is hardcoded in several python scripts. Not a big deal but not really convienant.

Future

Ogra was considering it for Edubuntu inclusion but obviously the multiverse dependancy is a problem. If included there will probably be some GTK interface instead of the current webinterface.

Installation

Just get it from the website. And untar it. In configuration example we untarred it into /opt/willow.

Configuration

example willow.conf

syspath = ['/opt/willow']
port = 8080
filters = ['urlfilter','domainfilter','contentfilter']
domainfilterpath = '/opt/willow/filters/domain'
urlfilterpath = '/opt/willow/filters/url'
contentfilterpath = '/opt/willow/filters/content'

Run it

/opt/willow/willow.py --config=/opt/willow/willow.conf

Sources

• Blog: http://users.lichtsnel.nl/~spacey/?p=129

• Willow homepage: http://www.digitallumber.net/software/willow/

Discussion