Bernard J. Ortcutt is a Reddit moderation bot. Moderators send it commands by reporting posts or comments. The bot reads these commands and automates moderation, based on a customizable configuration.
You can install the latest development version with:
pip install https://github.com/leviroth/bernard/archive/master.zip
Your subreddit must be configured for toolbox's usernotes module, or else the built-in commands will fail.
The database should be set up by reading create_tables.sql with SQLite3, as
in:
sqlite3 bernard-database.sq3 ".read create_tables.sql"
The bot requires access, config, posts, and wiki permissions. While not
strictly necessary, mail permissions keep replies to ban messages from going
to your bot's inbox. That makes it easier to isolate the messages sent to your
bot, which would be missed otherwise and which are very occasionally useful.
Login information should be configured with a praw.ini file.
python -m bernard [command_config_file] [database]
Configuration is not currently well-documented, but much can be gleaned from the
examples/ directory. Separate files are used for the reddit API login
information (JSON) and the command configuration (YAML).
Please note that the bot performs no validation of these configuration files.
You should not try to load configurations from untrusted sources, and even with
trusted sources you'll want to monitor startup to make sure that nothing has
crashed. Some basic validation can be performed manually via schema.yaml
and pyKwalify.
The per-subreddit default_post_actions and default_comment_actions keys
allow automatic loading of the following defaults:
-
For each subreddit rule affecting posts, moderator reports using that rule as the report reason will result in the post being removed, the rule being posted as a notification, and a usernote being added for the user. These actions are also triggered by the commands
[i]andrule [i], where[i]is the number of the rule as it appears on the rules page. -
For the ith comment-specific rule, the command
n [i]ornuke [i]removes the comment and all its replies, adds a warning to follow the rule in question, and adds a usernote.
Bernard understands the following basic actions, which can be composed into more complex actions via a configuration file. A target can be a comment or a submission.
Set via the remove configuration key. Please note that the target is approved
otherwise; this is necessary to prevent repeat actions.
A submission that is removed is locked.
Via a comment, distinguished as a moderator. This comment is stickied if the target is a submission.
Adds the author's username to a list the AutoModerator configuration. This requires that the list include a placeholder name. I put exclamation marks in my placeholders to ensure that they can't match a valid username.
Example uses: Automatically report a problem user's comments for review; automatically flair posts by bots or other specific-purpose accounts.
Bans the author from the subreddit. Can set a duration, but only on a fixed per-command basis. That is, it isn't currently possible to make a report-command take duration as a parameter. If no duration is set, the ban is permanent.
Removes all descendants from the comment tree. Does not affect the target; use
the remove key if it should be removed. Does not affect distinguished comments.
Currently only works for comment targets.
Adds a usernote for the author.
Bernard logs its actions to a database. bernard-logs is a simple Flask app providing a web-based view for these logs. reddit's OAuth is used for login: a user can view the logs of a specific subreddit if and only if she is a moderator of that subreddit.
Bernard J. Ortcutt is a character who appears in W. V. Quine's classic paper "Quantifiers and Propositional Attitudes." This doesn't have anything to do with what the bot does; I was just reading the paper back when I began this project, and I needed a name.
MIT