fhocutt / matchbot Goto Github PK

On test-wiki; needs a nonexistent profile page.

When the same thing is posted soon after, we want it not to post. When the same topic is posted because of a recategorization of the page (new category posted), we want it to post (as a new topic? Perhaps.)

API calls may not use maximum limits, and it is theoretically possible for not all results to be returned if there are too many of them.

This is going to mean writing more in mbapi, not matchbot.

Should cover:

• purpose of the script
• how to run the script
• how to run it on cron on Labs?
• assumptions baked in
• how to edit the config (oh json)/what the categories mean/correspond to
• log formats and information, where logs are stored

Someone should be automatically notified if the bot is broken.

Log format is neither documented nor obvious, nor is it consistent across types of log.

Figure out how to log the stack trace for errors.

Log errors both to the console and the log file (so that there's a notification on cron? figure out how that works.)

Consider a rotating set of debugging logs with a bunch of information.

Consider separating learner-related data and methods into a Learner class.

Errors are currently mostly not handled. This should change.

This probably involves swapping out the engine in SQLAlchemy.

Rotating run logs currently only log a few lines. They should be able to get longer (week of data? month of data?).

Redirects? Does the push notification go to the right place?

Things that should not be hard-coded:

• category names
• greeting/topic
• namespace/title of Co-op (?)
• site, protocol, user-agent, username, password
• base of categories ("editing", "technical editing", etc.)

Log a unique run ID every time matchbot.py is run. Suggestions: epoch time; the process ID, followed by a dot, followed by the current unix epoch time.