Language Filtering

How to filter out specific phrases like URL's

You may be able to get the profanity filter to do what you want, depending upon exactly what it is you want to do. The first thing you need to keep in mind is that the profanity filter works in three steps (it doesn't actually, but it acts as if it does):

1) Separation: The text being sent is separated into works along the stops. The space character is always a stop. Other characters can be configured to be stops or not.

2) Comparison: Each word from step 1 is compared against the list of disallowed words. Characters that map equivalentally are considered to match.

3) Action: On a match, the profanity filter takes the selected action or compinations configured for that that category of text (drop, filter, and/or warn).

If you want to stop URLs like 'http://xxx.xxx.xxx', you could map ':' to itself (it's a stop by default) and make sure '/' is a stop. Then add 'http:' to the filtered words list. If you only wanted to stop 'http://www.xxx.xxx', you could map ':' and '/' to themselves and make sure '.' is a stop. Then you could add 'http://www' to the filtered words list.

If only specific sites are a problem, map '.' to itself, make '/' a stop character and filter out the specific sites. For example 'www.foo.com' would stop 'http://www.foo.com' since the '/' is a stop character and '.' is not.

There's really no good way to catch things like "Come see my web site at www.foo.com" (such that the filter would catch any web site). There's just too many ways to disguise it.

If you have a repeat problem with specific sites, you may find that it's helpful to complain to the administrators of those sites and (in some cases) the abuse contacts for their uplinks. For extreme cases, where the site themself is sponsoring or encouraging this, you can take action to get the site blacklisted until they clean up their act.