A python-based IRC bot using Twisted. Original inspiration came from olga. Why re-implement another bot? Because olga is written in perl, and I wanted something a bit more sane to look at.
All requirements for helga are listed in requirements.txt
.
However, there is a single external requirement, and that is MongoDB.
You don't need it, per se, but many of the included plugins use MongoDB
for storing consistent state between restarts.
Start by creating a virtualenv where helga will reside:
$ virtualenv helga
$ cd helga
$ source bin/activate
Then grab the latest copy and install requirements:
$ git clone https://github.com/shaunduncan/helga src/helga
$ cd src/helga
$ python setup.py develop
$ pip install -r requirements.txt
Once you have performed the above steps, there will be a helga
executable
placed in the bin
dir of your virtualenv. Run helga by calling this:
$ /path/to/venv/bin/helga
Note that this uses the default settings file, helga.settings
to start.
You can, and should, use your own custom setttings. This file must a be an
importable python file on $PYTHONPATH. To run helga with your custom settings
file, set an environment variable HELGA_SETTINGS
to be a python import path:
$ export HELGA_SETTINGS=path.to.mysettings
This will preserve any defaults in helga.settings
, but you can override at will.
The included Vagrantfile will let you spin up a VM to run both MongoDB and an IRC server
for local development. Once you've followed the previous instructions for installing helga,
simply vagrant up
. This will forward host ports 6667 (irc) and 27017 (mongo) to the guest.
At this point, simply runing helga
from the command line will connect to this VM.
Helga supports plugins outside of the core source code. Plugins have a minimal API, but there
are some basic rules that should be followed. All core plugin implementations can be found
in helga.plugins.core
. The basic requirement for plugins is that they have a process
attribute that is a callable and determines if the plugin should handle a message, and
a run
method that actually performs the legwork of what the plugin should do. By convention,
the process
method should accept four arguments:
- client: an instance of
helga.comm.Client
- channel: the channel on which the message was received
- nick: the current nick of the message sender
- message: the message string itself
The run
is a bit different as it is up to the plugin implementation itself to decide what
arguments are necessary to generate a response. This method should be called by process
and
should return one of:
- None or empty string, if no response is to be sent over IRC
- Non-empty string for a single line response
- List of strings for multiline responses
Really, as long as you follow the above conventions, you can write plugins however you wish.
However, you should try to keep plugins simple and use the included decorators command
,
match
, and preprocessor
(explained later). However, if you prefer writing a plugin
as a class, you can subclass the included Plugin
base class, provided you have followed
the above rules. Here is a simple example:
import time
from helga.plugins.core import Plugin
class MyPlugin(Plugin):
def run(self, channel, nick, message):
return 'Current timestamp: {0}'.format(time.time())
def process(self, channel, nick, message):
if message.startswith('!time'):
return self.run(channel, nick, message)
NOTE the previous example is not the preferred way. You should use the included decorators instead (shown below).
For the most part, there are two main types of plugins: commands and matches. Commands are plugins
that require a user to specifically ask for helga to perform some action. For example,
helga haiku
or helga google something to search
. Matches are on the other hand are
intended to be autoresponders that give some extra meaning or context to what a user has said.
For example, if helga matches for a string "foo":
<sduncan> i'm talking about foo in this message <helga> sduncan is talking about foo
For the sake of simplicity, there are two convenient decorators for authoring these types of plugins (which is usually the case). For example:
from helga.plugins import command, match
@command('foo', aliases=['foobar'], help="The foo command")
def foo(client, channel, nick, message, cmd, args):
# This is run on "helga foo" or "helga foobar"
return "Running the foo command"
@match(r'bar')
def bar(client, channel, nick, message, matches):
# This will run whenever a user mentions the word 'bar'
return "{0} said bar!".format(nick)
You may notice in the above example that each decorated function accepts different arguments.
For commands, there are two additional arguments cmd
and args
. The former is the parsed
command that was used to run the method (which could be "foo" in the above case, or the alias
"foobar"). The latter is a list of whitespace delimited strings that follow the parsed commend.
For example helga foo a b c
would mean the args param would be ['a', 'b', 'c']
.
For the match plugin, the single additional argument is matches
which is for the most part,
the result of re.findall
. However, the @match
decorator accepts a callable in place of
a regex string. This callable should accept one argument: the message being processed. It should
return a value that can be evaluated for truthiness and will be passed to the decorated function
as the matches
parameter.
Plugins can also be message preprocessors. These are callables that may perform some modification
on an incoming message prior to that message being delivered to any plugins. Preprocessors should
accept arguments (in order) for client
, channel
, nick
, and message
and should
return a three-tuple consisting of (in order) channel
, nick
, and message
. To declare
a function as a preprocessor, a convenient decorator can be used:
from helga.plugins import preprocessor
@preprocessor
def blank_message(client, channel, nick, message):
return channel, nick, ''
Some plugins do both matching and act as a command. For this reason, plugin decorators are chainable.
However, remember that different plugin types expect decorated functions to accept different arguments.
It is best to accept *args
for these:
from helga.plugins import command, match, preprocessor
@preprocessor
@match(r'bar')
@command('foo')
def complex(client, channel, nick, message, *args):
# len(args) == 0 for preprocessors
# len(args) == 1 for matches
# len(args) == 2 for commands
You can control the priority in which a plugin is run. Note though, that preprocessors will always
run first. A priority value should be an integer value. There are no limits or bounds for this value,
but know that a higher value will mean a higher priority. If you are writing Plugin
subclass
style plugins, you will need to set a priority
attribute of your object. This is done automatically
if you call super(MyClass, self).__init__(priority=some_value)
in your class's __init__
.
However, if you are using the preferred decorator style for writing plugins, you can supply a priority
keyword argument to the decorator:
from helga import command, match, preprocessor
@preprocessor(priority=10)
def foo_preprocess(*args):
pass
@command('foo', priority=20)
def foo_command(*args):
pass
@match(r'foo', priority=30)
def foo_match(*args):
pass
For convenience, there are constants that can be used for setting priorities:
- PRIORITY_LOW = 25
- PRIORITY_NORMAL = 50
- PRIORITY_HIGH = 75
Also, each decorator/plugin type has its own default value for priority:
- Preprocessors have default priority of
PRIORITY_NORMAL
- Commands have default priority of
PRIORITY_NORMAL
- Matches have default priority of
PRIORITY_LOW
Helga uses setuptools entry points for plugin loading. Once you've written a plugin you wish to use,
you will need to make sure your python package's setup.py contains an entry_point under the group
name helga_plugins
. For example:
entry_points = {
'helga_plugins': [
'plugin_name = mylib.mymodule:MyPluginClass',
],
},
Note that if you are using decorated function for a plugin, you will want to specify the method name
for your entry point, i.e. mylib.mymodule:myfn
.
As of helga version 1.3, there is an included plugin for exposing an HTTP server to support webhooks.
This might be useful if you need to have a public facing HTTP service that you would like to use to
perform some sort of announcement on a particular channel. This is also very extensible and should allow
you to create new webhooks in a very similar way plugins are created. This plugin is enabled by default
and requires two settings: WEBHOOKS_PORT
and WEBHOOKS_CREDENTIALS
. The former is of course the
port on which to run this service. The latter should be a list of tuples in the form of (username, password).
These are used to perform HTTP basic authentication on any webhook that requires it.
Webhook plugins work by declaring routes. This will not only feel similar to helga's decorator style
plugins, but it will also feel very similar to anyone who has used something like Flask. There are two
primary decorators you will need to get started: route
, which declares a function as a route endpoint,
and authenticated
, which ensures that the route function cannot be called without proper HTTP basic
authentication. Both of these can be imported from helga.plugins.webhooks
. For example:
from helga.plugins.webhooks import authenticated, route
@route(r'/foo/(?P<id>[0-9]+)')
@authenticated
def foo(request, irc_client, id):
# This will require auth
pass
@route('/bar', methods=['POST'])
def bar(request, irc_client):
# This will not require auth, and will only accept POST
pass
NOTE: For authenticated routes, you MUST specify @authenticated
as the first decorator. This may be
changed in the future.
The route decorator accepts two arguments: 1) a path regular expression and 2) an optional list of HTTP methods to accept. If you do not specify a list of HTTP methods, only GET requests will be served. All regex paths must be named groups and they will be passed as keyword arguments.
To register a new webhook plugin, you must declare an entry_point much in the same way normal plugins
are done. However, the entry_point group name is helga_webhooks
. For example:
entry_points = {
'helga_webhooks': [
'name = mylib.mymodule:myhook',
],
},
The webhook plugin itself has some commands for IRC interaction: start/stop to control the running HTTP listener, and routes, which will show all the route paths and the HTTP methods they accept.
Here are some plugins that have been written that you can use:
Plugin | Description | Link |
---|---|---|
excuses | Generate a response from http://developerexcuses.com | https://github.com/alfredodeza/helga-excuses |
haskell | Evaluate Haskell expressions. | https://github.com/carymrobbins/helga-haskell |
isup | Check downforeveryoneorjustme.com | https://github.com/shaunduncan/helga-isup |
karma | Dish out karma points to other people | https://github.com/coddingtonbear/helga-karma |
norris | Generate Chuck Norris facts for users | https://github.com/alfredodeza/helga-norris |
updates | List and record IRC channel updates. | https://github.com/cobbdb/helga-contrib-updates |
zen | The Zen of Python | https://github.com/shaunduncan/helga-zen |
Written a plugin? Send a pull request to be listed in the above table!
All tests are written to be run via tox
. To run the test suite, inside your virtualenv:
$ cd src/helga
$ tox
Contributions are welcomed, as well as any bug reports! Please note that any pull request will be denied if tests run via tox do not pass
Copyright (c) 2013 Shaun Duncan