Skip to content
The Python Bot Framework for Mastodon
Branch: master
Clone or download
chr-1x Merge pull request #17 from StarlitGhost/reconnect
Auto-reconnect if the user stream dies (Fixes #2)
Latest commit e4625a3 Oct 30, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information.
ananas auto-reconnect if the user stream dies Oct 11, 2018
.gitignore Update .gitignore Aug 31, 2017
LICENSE Add license Aug 30, 2017 Content type of readme and minor fix Jul 31, 2018


What is Ananas?

Ananas allows you to write simple (or complicated!) mastodon bots without having to rewrite config file loading, interval-based posting, scheduled posting, auto-replying, and so on.

Some bots are as simple as a configuration file:

class = tracery.TraceryBot
access_token = ....
grammar_file = "bepis.json"

But it's easy to write one with customized behavior:

class MyBot(ananas.PineappleBot):
    def start(self):
        with open('trivia.txt', 'r') as trivia_file:
           self.trivia = trivia_file.lines()

    def post_trivia(self):

    def respond_trivia(self, status, user):
        self.mastodon.toot("@{}: {}".format(user["acct"], random.choice(self.trivia)))

Run multiple bots on multiple instances out of a single config file:

class = custom.JortsBot
domain =
access_token = ....
line = 632

class = roll.DiceBot
domain =
access_token = ....

And use the DEFAULT section to share common configuration options between them:

domain =
client_id = ....
client_secret = ....

Getting started

pip install ananas

The ananas pip package comes with a script to help you manage your bots.

Simply give it a config file and it'll load your bots and close them safely when it receives a keyboard interrupt, SIGINT, SIGTERM, or SIGKILL.

ananas config.cfg

If you haven't specified a client id/secret or access token, the script will exit unless you run it with the --interactive flag, which allows it to prompt you for the instance login information. (The only part of the input you enter here that's stored in the config file is the instance name -- the email and password are only used to generate the access token).


The following fields are interpreted by the PineappleBot base classs and will work for every bot:

class: the fully-specified python class that the runner script should instantiate to start your bot. e.g. "ananas.default.TraceryBot"

domain\ ¹: the domain of the instance to run the bot on. Must support https connections. Only include the domain, no protocol or slashes. e.g. ""

client_id\ ¹, client_secret\ ¹: the tokens that the instance uses to identify what client this bot is posting from/as. Will be used to determine what's displayed underneath all the posts made by this bot.

access_token\ ¹: the access token used to authenticate API requests with the instance. Make sure this is secret, don't distribute config files with this field filled out or people will be able to post under the account this token was created with.

admin: the full username (without leading @) of the user to DM error reports to. Can be left unspecified, but is useful for keeping an eye on the health of the bot without constantly monitoring the script logs. e.g.

¹: Filled out automatically if the bot is run in interactive mode.

Additional fields are specific to the type of bot, refer to the documentation for the bot's class for more information about the fields it expects.

Writing Bots

Custom bot classes should be subclasses of ananas.PineappleBot. If you override __init__, be sure to call the base class's __init__.


In order for the bot to do anything, you should add a method decorated with at least one of the following decorators:

@ananas.reply: Calls the decorated function when the bot is mentioned by any other user. Decorator takes no parameters, but should only be called on functions matching this signature: def reply_fn(self, mention, user). mention will be the dictionary corresponding to the status containing the mention (as returned by the mastodon API, user will be the dictionary corresponding to the user that mentioned the bot.

@ananas.interval(secs): Calls the decorated function every secs seconds, starting when the bot is initialized. For intervals longer than ~an hour, you may want to use @schedule instead. e.g. @ananas.interval(60)

@ananas.schedule(**kwargs): Allows you to schedule, cron-style, the decorated function. Accepted keywords are "second", "minute", "hour", "day_of_week" or "day_of_month" (but not both), "month", and "year". If any of these keywords are not specified, they will be treated like cron treats an *, that is, as long as the time matches the other values, any value will be accepted. Speaking of which, the cron-like syntax "*" as well as "*/3" are both accepted, and will expand to the expected thing: for example, schedule(hour="*/2", minute="*/10") will post every 10 minutes during hours which are multiples of 2.

@ananas.hourly(minute=0), @ananas.daily(hour=0, minute=0): Shortcuts for @ananas.schedule() that call the decorated function once an hour at the specified minute or once a day at the specified hour and minute. If parameters are omitted they'll post at the top of the hour or midnight (UTC).

@ananas.error_reporter: specifies custom behavior for reporting errors. The decorated function should match this signature: def err(self, error) where error is a string representation of the error.

Overrideable Functions

You can also define the following functions and they will be called at the relevant points in the bot's lifecycle:

init(self): called before the configuration file has been loaded, so that you can set default values for config fields in case the config file doesn't specify them.

start(self): called after all of the internal PineappleBot initialization is complete and the mastodon API is ready to use. A good place to load files specified in the config, post a startup notice, or otherwise do bot-specific setup.

stop(self): called when the bot has received a shutdown signal and needs to stop. The config file will be saved after this, so if you need to make any last minute changes to the config, do that here.

Configuration Fields

All of the configuration fields for the current bot are available through the self.config object, which exposes them with both field-accessor syntax and dictionary-accessor syntax, for example:

foo =
bar = self.config["bar"]

These can be read (to get the user's configuration data) or written to (to affect the config file on next save) or deleted (to remove that field from the config file).

You can call self.config.load() to get the latest values from the config file. load takes an optional parameter name, which is the name of the section to load in the config file in case you want to load a different one than the bot was started with.

You can also call to write any changes made since the last load back to the config file.

Note that if you call self.config.load() during bot operation, without first calling, you will discard any changes made to the configuration since the last load.

Distributing Bots

You can distribute bots however you want; as long as the class is available in some module in python's sys.path or a module accessible from the current directory, the runner script will be able to load it.

If you think your bot might be generally useful to other people, feel free to create a pull request on this repository to get it added to the collection of default bots.

Questions? Ping me on Mastodon at or shoot me an email at and I'll answer as best I can!

You can’t perform that action at this time.