XMPP bots for humans
Go to file
decentral1se 6788e6a4f1
fix: allow template loading without the web server
See https://git.vvvvvvaria.org/decentral1se/xbotlib/issues/12
2022-02-21 11:01:27 +01:00
.drone.yml Attempt to integrate codecov 2021-01-26 23:16:50 +01:00
.gitignore Rig up first-class file system support 2021-01-23 23:31:00 +01:00
CHANGELOG.md Bump slixmpp 2021-02-02 22:40:29 +01:00
LICENSE First working prototype 2021-01-10 14:10:39 +01:00
README.md fix: allow template loading without the web server 2022-02-21 11:01:27 +01:00
poetry.lock Bump slixmpp 2021-02-02 22:40:29 +01:00
pyproject.toml Bump version 2021-02-02 22:43:11 +01:00
test_xbotlib.py Remove unused import 2021-02-02 22:30:57 +01:00
xbotlib.py fix: allow template loading without the web server 2022-02-21 11:01:27 +01:00



PyPI version Build Status

XMPP bots for humans

status: experimental

A friendly lightweight wrapper around slixmpp for writing XMPP bots in Python. The goal is to make writing and running XMPP bots easy and fun. xbotlib is a single file implementation which can easily be understood and extended. The xbotlib source code and ideas are largely borrowed from the XMPP bot experiments going on in Varia.


$ pip install xbotlib

Getting Started

Put the following in a echo.py file. This bot echoes back whatever message you send it in both direct messages and group messages. In group chats, you need to message the bot directly (e.g. echobot: hi) (see the commands section for more).

from xbotlib import Bot

class EchoBot(Bot):
    def direct(self, message):
        self.reply(message.text, to=message.sender)

    def group(self, message):
        self.reply(message.content, room=message.room)


And then python echo.py. You will be asked a few questions in order to load the account details that your bot will be using. This will generate an echobot.conf file in the same working directory for further use. See the configuration section for more.

That's it! If you want to go further, continue reading here.

Working with your bot

The following sections try to cover all the ways you can configure and extend your bot using xbotlib. If anything is unclear, please let us know on the chat.


Add a help = "my help" to your Bot class like so.

class MyBot(Bot):
    help = """This is my cool help.

    I can list some commands too:

      @foo - some command

Your bot will then respond to mybot: @help invocations. This string can be a multi-line string with indentation. xbotlib will try to format this sensibly for showing in chat clients.

See more in the commands section for more.


Using @<command> in direct messages and <nick>, @<command> (the , is optional, anything will be accepted here and there doesn't seem to be a consensus on what is most common way to "at" another user in XMPP) in group chats, here are the supported commands.

  • @uptime: how long the bot has been running
  • @help: the help text for what the bot does

There are also more general status commands which all bots respond to.

  • @bots: status check on who is a bot in the group chat

These commands will be detected in any part of the message sent to the bot. So you can write echobot, can we see your @uptime, or I'd love to know which @bots are here.


By default, xbotlib will look for an avatar.png (so far tested with .png but other file types may work) file alongside your Python script which contains your bot implementation. You can also specify another path using the --avatar option on the command-line interface. The images should ideally have a height of 64 and a width of 64 pixels each.


All the ways you can pass configuration details to your bot. There are three ways to configure your bot, the configuration file, command-line interface and the environment. Use whichever one suits you best. The values are loaded in the following order: command-line > configuration file > environment. This means you can override everything from the command-line easily.

Using the .conf configuration file

If you run simply run your Python script which contains the bot then xbotlib will generate a configuration for you by asking a few questions. This is the simplest way to run your bot locally.

  • account: the account of the bot
  • password: the password of the bot account
  • nick: the nickname of the bot
  • avatar: the avatar of the bot (default: avatar.png)
  • redis_url: the Redis connection URL
  • rooms: a list of rooms to automatically join (comma separated)
  • no_auto_join: disable auto-join when invited (default: False)
  • template: the port to serve from (default: index.html.j2)
  • serve: turn on the web server (default: False)
  • port: the port to serve from (default: 8080)
  • storage: storage back-end (default: file)
  • output: path to the output directory (default: ./)

Using the command-line interface

Every bot accepts a number of comand-line arguments to load configuration. You can use the --help option to see what is available (e.g. python bot.py --help).

  • -h, --help: show this help message and exit
  • -d, --debug: enable verbose debug logs
  • -a ACCOUNT, --account ACCOUNT: account for the bot account
  • -p PASSWORD, --password PASSWORD: password for the bot account
  • -n NICK, --nick NICK: nickname for the bot account
  • -av AVATAR, --avatar AVATAR: avatar for the bot account (default: avatar.png)
  • -ru REDIS_URL, --redis-url REDIS_URL: redis storage connection URL
  • -r ROOMS [ROOMS ...], --rooms ROOMS [ROOMS ...]: Rooms to automatically join
  • -naj, --no-auto-join: disable automatically joining rooms when invited (default: False)
  • -pt PORT, --port PORT: the port to serve from (default: 8080)
  • -t TEMPLATE, --template TEMPLATE: the template to render (default: index.html.j2)
  • -s, --serve: turn on the web server (default: False)
  • -st {file,redis}, --storage {file,redis}: choice of storage back-end (default: file)
  • -o OUTPUT, --output OUTPUT: path to the output directory (default: ./)

Using the environment

xbotlib will try to read the following configuration values from the environment if it cannot read them from a configuration file or the command-line interface. This can be useful when doing remote server deployments.

  • XBOT_ACCOUNT: The bot account
  • XBOT_PASSWORD: The bot password
  • XBOT_NICK: The bot nickname
  • XBOT_AVATAR: The bot avatar icon (default: avatar.png)
  • XBOT_REDIS_URL: Redis key store connection URL
  • XBOT_ROOMS: The rooms to automatically join
  • XBOT_NO_AUTO_JOIN: Disable auto-joining on invite (default: False)
  • XBOT_TEMPLATE: the template to render (default: index.html.j2)
  • XBOT_SERVE: Turn on the web server (default: False)
  • XBOT_PORT: The port to serve from (default: 8080)
  • XBOT_STORAGE: choice of storage back-end (default: file)
  • XBOT_OUTPUT: path to the output directory (default: ./)

Storage back-end

In order to store data you can make use of the self.db attribute of the Bot class. It is a Python dictionary which will be saved to disk automatically for you as a <nick>.json in your current working directory. The name and path to this file can be configured using the output option (e.g. python bot.py --output /var/www/html)

def group(self, message):
    if not message.room in self.db.keys():
        self.db[message.room] = "visited"

If you want to inspect the database when the bot is not running, you can look in the file directly.

$ cat <nick>.json

For more advanced use cases, xbotlib also supports Redis as a storage back-end. You'll need to configure this (e.g. --storage redis) as the default uses the filesystem approach mentioned above. The same self.db will then be passed as a Redis connection object. You will also need to install additional dependencies using pip install xbotlib[redis].

Loading Plugins

You can specify a plugins = [...] on your bot definition and they will be automatically loaded when you start your bot.

class MyBot(Bot):
    plugins = ["xep_0066"]

See here for the list of supported plugins.

Serving HTTP

Firstly, you'll need to install additional dependencies.

$ pip install xbotlib[web]

Your bot will run a web server if you configure it to do so. Use the --serve option on the command-line, the serve = True configuration option or the XBOT_SERVE=True environment variable.

If you're running your bot locally, just visit to see. The default response is just some placeholder text. You can write your own responses using the Bot.serve function.

xbotlib provides a small wrapper API for Jinja2 which allows you to easily template and generate HTML. The web server is provided by aiohttp.

The default template search path is index.html.j2 in the current working directory. This can be configured through the usual configuration entrypoints. It is possible to use the template without the built-in server too!

Here's a small example that renders a random ASCII letter and uses a stylesheet.


  <style> h1 { color: red; } </style>
  <h1>{{ letter }}</h1>


from string import ascii_letters

async def serve(self, request):
    letter = choice(ascii_letters)
    rendered = self.template.render(letter=letter)
    return self.respond(rendered)

Please note the use of the return keyword here. The serve function must return a response that will be passed to the web server. Also, the async keyword. This function can handle asynchronous operations and must be marked as such. You can return any content type that you might find on the web (e.g. HTML, XML, JSON, audio and maybe even video) but you must specify the content_type=... keyword argument for respond.

See the list of available content types for more.

If you want to pass data from your direct/group functions to the serve function, you'll need to make use of some type of persistent storage. Your serve function can read from the storage back-end and then respond. This is usually as simple as accessing the self.db attribute.


As long as the --no-auto-join option is not set (via the configuration file or environment also), then your bot will automatically join any room it is invited to. Rooms that your bot has been invited to will be stored in the .xbotlib/data.json file. If your bot is turned off or fails for some reason then it will read this file when turned back on to see what rooms it should re-join automatically. The data.json file can be edited freely by hand.

API Reference

When writing your own bot, you always sub-class the Bot class provided from xbotlib. Then if you want to respond to a direct message, you write a direct function. If you want to respond to a group chat message, you write a group function. That's it for the basics.


Respond to direct messages.


  • message: received message (see SimpleMessage below for available attributes)


Respond to a message in a group chat.


  • message: received message (see SimpleMessage below for available attributes)


Serve requests via the built-in web server.

See this section for more.


  • request: the web request


Register additional routes for web serving.

See this section for more.

This is an advanced feature. Use self.web.add_routes.

from aiohttp.web import get

def routes(self):
    self.web.add_routes([get("/foo", self.foo)])


Run some setup logic before starting your bot.


A simple message interface.


  • text: the entire text of the message
  • content: the text of the message after the nick
  • sender: the user the message came from
  • room: the room the message came from
  • receiver: the receiver of the message
  • nick: the nickname of the sender
  • type: the type of message
  • url: The URL of a sent file


Bot.reply(message, to=None, room=None)

Send a reply back.


  • message: the message that is sent
  • to: the user to send the reply to
  • room: the room to send the reply to

Bot.respond(response, content_type="text/html")

Return a response via the web server.


  • response: the text of the response
  • content_type: the type of response

Other useful attributes on the Bot class are:


We're lurking in xbotlibtest@muc.vvvvvvaria.org if you want to chat or just invite your bots for testing.

More Examples

See more examples on git.vvvvvvaria.org.

Deploy your bots

Documentation coming soon.


See the issue tracker.


See the CHANGELOG.md.


See the LICENSE.


Any and all contributions most welcome! Happy to hear how you use the library or what could be improved from a usability perspective.

To test, install Tox (pip install tox) and run tox to run the test suite locally.