WatchBot

[Lorenzo J. Lucchini]

I have introduced a new bot into the main channels (#ubuntu, #kubuntu, 
#ubuntu+1, #ubuntu-offtopic, as well as some in the #ubuntu-it hierarchy).

The bot is a fork of FloodBot, and is intended to provide a relatively wide 
range of features.

I'm not simply adding features to FloodBot (actually, I've moved some of the 
less vital FloodBot features into WatchBot), and I've made a new bot instead, 
because FloodBot needs to stay very responsive, while WatchBot performs some 
CPU and network intensive tasks.


The only feature that is currently working, in beta phase if you like, is URL 
screening. I'll describe it now, while I'll describe other features later 
when I've actually got them implemented.


WatchBot looks for URLs posted into the channels, and does the following 
things in this order:
- consults an internal whitelist, and if the site is listed, nothing is done 
about it
- consults an internal blacklist, and if the site is listed, !ops is called 
and a warning to the poster is sent on the channel
- consults http://domain.opendns.com/<hostname>, and if the site matches 
a "bad" category (and does not match a "good" category), then again !ops is 
called and a warning is sent
- looks at the HTML code on the URL's page for some "bad" keyphrases, and 
calls !ops and warns if one is found


"Bad" and "good" categories, as well as "bad" keyphrases, can be defined on a 
channel-by-channel basis (while the whitelist and blacklist are currently 
global).

At the time being, "bad" categories for #ubuntu, #kubuntu and #ubuntu+1 are:
Malware, Dating, Drugs, Gambling, Hate/Discrimination, Nudity, Phishing, 
Politics, Pornography, Religious, Sexuality, Tasteless, Adult Themes, 
Television
There are no "good" categories.

"Bad" categories for #ubuntu-offtopic are:
Dating, Drugs, Gambling, Hate/Discrimination, Nudity, Phishing, Politics, 
Pornography, Religious, Sexuality, Tasteless, Adult Themes, Television
"Good" categories are:
Top site, Photo sharing

The only "bad" keyphrase currently present is "sudo passwd root".


I've sent this email both to clarify the purpose of the new bot and to ask for 
feedback with regards to choice of categories and keyphrases (new keyphrases 
are especially welcome, it the use of keyphrase matching is deemed 
appropriate to begin with).


Note that currently, !ops is not really called, but only "simulated" in 
#ubuntu-ops-monitor, to avoid triggering annoying false positives during the 
testing phrase.


by LjL
ljl at ubuntu.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <https://lists.ubuntu.com/archives/ubuntu-irc/attachments/20090323/67ee865d/attachment.pgp>