EarwigBot is a Python robot that edits Wikipedia and interacts with people over IRC.
EarwigBot is a Python robot that edits Wikipedia and interacts with people over IRC. This file provides a basic overview of how to install and setup the bot; more detailed information is located in the docs/ directory (available online at PyPI).
Development began, based on the Pywikipedia framework, in early 2009. Approval for its fist task, a copyright violation detector, was carried out in May, and the bot has been running consistently ever since (with the exception of Jan/Feb 2011). It currently handles several ongoing tasks ranging from statistics generation to category cleanup, and on-demand tasks such as WikiProject template tagging. Since it started running, the bot has made over 50,000 edits.
A project to rewrite it from scratch began in early April 2011, thus moving away from the Pywikipedia framework and allowing for less overall code, better integration between bot parts, and easier maintenance.
This package contains the core earwigbot, abstracted enough that it should be usable and customizable by anyone running a bot on a MediaWiki site. Since it is component-based, the IRC components can be disabled if desired. IRC commands and bot tasks specific to my instance of EarwigBot that I don’t feel the average user will need are available from the repository earwigbot-plugins.
It’s recommended to run the bot’s unit tests before installing. Run python setup.py test from the project’s root directory. Note that some tests require an internet connection, and others may take a while to run. Coverage is currently rather incomplete.
If you get an error while pip is installing dependencies, you may be missing some header files. For example, on Ubuntu, see this StackOverflow post.
You can also install it from source  directly:
curl -Lo earwigbot.tgz https://github.com/earwig/earwigbot/tarball/v0.2 tar -xf earwigbot.tgz cd earwig-earwigbot-* python setup.py install cd .. rm -r earwigbot.tgz earwig-earwigbot-*
You can install the development version of the bot from git by using setuptools’s develop command , probably on the develop branch which contains (usually) working code. master contains the latest release. EarwigBot uses git flow, so you’re free to browse by tags or by new features (feature/* branches):
git clone git://github.com/earwig/earwigbot.git earwigbot cd earwigbot python setup.py develop
The bot stores its data in a “working directory”, including its config file and databases. This is also the location where you will place custom IRC commands and bot tasks, which will be explained later. It doesn’t matter where this directory is, as long as the bot can write to it.
Start the bot with earwigbot path/to/working/dir, or just earwigbot if the working directory is the current directory. It will notice that no config.yml file exists and take you through the setup process.
There is currently no way to edit the config.yml file from within the bot after it has been created, but YAML is a very straightforward format, so you should be able to make any necessary changes yourself. Check out the explanation of YAML on Wikipedia for help.
After setup, the bot will start. This means it will connect to the IRC servers it has been configured for, schedule bot tasks to run at specific times, and then wait for instructions (as commands on IRC). For a list of commands, say “!help” (commands are messages prefixed with an exclamation mark).
You can stop the bot at any time with Control+C, same as you stop a normal Python program, and it will try to exit safely. You can also use the “!quit” command on IRC.
The bot’s working directory contains a commands subdirectory and a tasks subdirectory. Custom IRC commands can be placed in the former, whereas custom wiki bot tasks go into the latter. Developing custom modules is explained below, and in more detail through the bot’s documentation on PyPI (or in the docs/ dir).
Note that custom commands will override built-in commands and tasks with the same name.
earwigbot.bot.Bot is EarwigBot’s main class. You don’t have to instantiate this yourself, but it’s good to be familiar with its attributes and methods, because it is the main way to communicate with other parts of the bot. A Bot object is accessible as an attribute of commands and tasks (i.e., self.bot).
earwigbot.config.BotConfig stores configuration information for the bot. Its docstring explains what each attribute is used for, but essentially each “node” (one of config.components, wiki, irc, commands, tasks, and metadata) maps to a section of the bot’s config.yml file. For example, if config.yml includes something like:
irc: frontend: nick: MyAwesomeBot channels: - "##earwigbot" - "#channel" - "#other-channel"
…then config.irc["frontend"]["nick"] will be "MyAwesomeBot" and config.irc["frontend"]["channels"] will be ["##earwigbot", "#channel", "#other-channel"].
Custom commands are subclasses of earwigbot.commands.Command that override Command’s process() (and optionally check(), setup(), or unload()) methods.
EarwigBot’s answer to the Pywikipedia framework is the Wiki Toolset (earwigbot.wiki), which you will mainly access through bot.wiki.
bot.wiki provides three methods for the management of Sites - get_site(), add_site(), and remove_site(). Sites are objects that simply represent a MediaWiki site. A single instance of EarwigBot (i.e. a single working directory) is expected to relate to a single site or group of sites using the same login info (like all WMF wikis with CentralAuth).
Load your default site (the one that you picked during setup) with site = bot.wiki.get_site().
Not all aspects of the toolset are covered in the docs. Explore its code and docstrings to learn how to use it in a more hands-on fashion. For reference, bot.wiki is an instance of earwigbot.wiki.SitesDB tied to the sites.db file in the bot’s working directory.