MishaBot is a combination IRC bot and web interface that acts as a link scraper. The bot component connects to IRC and silently logs web links to a database. The web interface grabs links from the database and presents them to the user.
The best way to explain what it does is by demonstration, so run on over to http://mishabot.taricorp.net/ to see the web interface.
The IRC bot portion of the code is implemented in Python, on top of python-irclib. It communicates with an SQL database to store links.
I think this highly decoupled approach is better than the traditional approach taken by web applications of having the server generate pages from templates, since it forces all features to have accessible hooks, thus allowing alternate clients to be easily designed, and it is easier to experiment with changes to the presentation.
As long as this project continues to hold my interest, I’ll be adding additional features. Some highlights of what I want to add:
- Clean bot reloading. The database layer in use (SQLObject) doesn’t handle reloading modules that use it gracefully, so simply reloading modules isn’t an option for applying code changes to a running instance of the bot. To allow new code to be loaded without disconnecting the bot from IRC, a new instance can be started that receives state information from the currently running isntance. To avoid dropping the open network connections, they can be passed to the new instance via a (magical) UNIX domain socket (AF_UNIX).
- Link filtering in the interface. The current version has some rudimentary filter capabilities (select a single channel to show), but there’s room for it to be much more capable.
- Allow users on IRC to “subscribe” to the bot, to have it announce link information to them.
As with most of my projects, MishaBot is open-source, provided under the zlib/libpng license. The repository is over on Bitbucket.