I still think the best way to eliminate this kind of business (I brought this up in another post) is to replace any sentence that contains any character other than a letter or a puctuation mark (read: numerals or symbols) or punctuation marks that are not immediately followed by a space or a quote with \" mumbles unintelligibly.\" This way, we don\'t have to worry about alternate spelling of leetspeak words used to circumvent any filter.
None of those aformentioned characters are actually needed to convey any ideas, they can all be replaced by real words, in any language:
& = and
% = percent
$ = dollar
# = pound or number
@ = at
2 = two
5 = five
etc.
I think this is a better solution than a translator because it can get rid of most of the \"offending\" speech patterns without requiring an extensive list of the unwanted \"words\" (which can be circumvented anyway) and without causing problems for people who are not typing in English.
This restriction should not be placed on /tells or /helps, for obvious reasons.
Granted, it WON\'T get rid of some of the other patterns we don\'t like to see like lol, rofl, wtf, omg, etc. but there may be a solution to that as well - user translations! Here\'s how they might work:
Chat: /defs omg
Chat: Jimminy Christmas!
Now whenever I type \"omg\" by itself in the chat window, the server would see that I have a macro set for it and would
/say Jimminy Christmas!
Likewise:
Chat: /defm rofl
Chat: is lauging so hard he can\'t breathe!
so a /defs defines a /say macro, and a /defm defines a /me macro. These are the only message types that can be /def\'ed. And messages sent via a /def\'ed macro have to abide by the same symbol/numeral rule as mentioned above.
Better yet, /def\'s could be the responsibility of the client, relieving server resources, and nobody really cares what word you use to prompt one of these responses anyway.
--
Of course, the numeral/symbol restriction (we already have such a rule in place for character names!) doesn\'t preclude the use of a list of \"words\" to further control messages, as in a profanity filter -- the two concepts can be used together -- but this simple restriction would drastically reduce the complexity of any attempt to parse for content, since many of the \"offending\" words which would constitute a large portion of the list would be eliminated by the character restriction!
