blackskad / mollommw Goto Github PK

Right now, all Mollom functionality is in one class. Before the reseller api can be implemented. The core functionality (updateing server list, managing mollom keys, the doCall-function) should be in a separate core class.
Proposal:

• Mollom: core functionality
• MollomClient extends Mollom: implementation of the client api
• MollomReseller extends Mollom: implementation of the reseller api

To decide which group can get less restrictive permissions (mollommw-no-check and mollommw-no-captcha), it would be interesting to get some detailed stats for which user posts what type of content.

This would need some database support to store information about the calls made to Mollom. Useful information that would need to be stored:

• timestamp
• user id
• user ip, as reported to Mollom
• Mollom sessionid
• Mollom action: (check) content or captcha
• Mollom response: 'ham/spam/unsure' or 'true/false'

Posts of users with special rights (like sysops) should not be checked for spam, and those users certainly should not see captcha's.
We should maintain a matrix with permissions:

      | check    | captcha
---------------------------
sysop |          | 
auth  |     x    |
anon  |     x    |    x

The xmlrpc core uses an experimental php module (which is experimental for almost 10 years). It's not in the default install and some people might not be able to install it (company policy, no shell access,...) It would be nice if there is an alternative for these people. Perhaps we can include the drupal xmlrpc functions?

Right now, the plugin does a text analysis first, then shows a captcha when the analysis wasn't conclusive. This might not be the preferred action.
Usecases:

• You might want the plugin to report spam, not detect it. So no analysis or captcha's are needed.
• You might want to just analyse the text and rely on the correctness of Mollom, so you don't have to bother the visitors with captcha's.

Configuration options: do nothing, analysis, analysis+captcha

Mollom has a function to detected languages. It would be nice if the plugin could be configured to only allow edits in a specific language. You might not want english text on a french wiki.

Mollom has a function to send feedback about edits. This teaches Mollom things about the content: the quality, profanity, spamminess, (un)wanted,... It would be nice to send feedback to mollom when an edit is reverted.

When the Mollom servers are offline, the plugin will accept anything it receives. While this is sensible for servers with a lot of legitimate users, which rely on their comments and activity, this isn't the right option for sites with very little legitimate users that doen't post very often. It would imply some work to clean up the spam, while only very little legitimate edits were made.
So it would be nice to have an option to let the administrators decide about it.

Mollom has an api for resellers. Would be nice to have it available