Skip to content

Latest commit

 

History

History
162 lines (118 loc) · 6.12 KB

README.md

File metadata and controls

162 lines (118 loc) · 6.12 KB
Raw

Reports bot maintains reports and other useful things for WikiProjects.

Setup

First, clone the bot:

git clone [email protected]:harej/reports_bot.git
cd reports_bot

Python 3.4+ is required. You should set up and activate a virtual environment in the venv directory, though this is not required. The following workaround may be necessary on Debian systems:

python3 -m venv --without-pip venv
source venv/bin/activate
curl https://bootstrap.pypa.io/get-pip.py | python

Next, install these dependencies:

pip install mwoauth requests PyYAML oursql3 mwparserfromhell BTrees \
mediawiki-utilities numpy scikit-learn

You also need Pywikibot. You can pip install pywikibot, but I recommend installing it from source to get the latest updates:

cd venv
git clone https://gerrit.wikimedia.org/r/pywikibot/core.git pywikibot
cd pywikibot
git submodule update --init
python setup.py develop

If you set up a virtualenv, run the following command to ensure the bot's task runner always uses it:

sed -e "1s|.*|#! $PWD/venv/bin/python|" -i "" ./run

Unprivileged user

Depending on your setup, you may wish to create a separate, unprivileged user for the bot. The recommended method is:

sudo adduser --system --home /path/to/reportsbot reportsbot

If so, make sure to create the bot's config and logs directories with the appropriate ownership:

mkdir config logs && sudo chown reportsbot config logs

Database

Reports bot uses a MySQL database to store its on-wiki config, the WikiProject page indices, and some other information. Create this database using the schema described in schema.sql. You may change the database name—for example, on Labs, it should probably be something like sXXXXX__wpx. Make note of this for the next step.

Config

The bot requires a config/config.yml file for itself and config/user-config.py file for Pywikibot. The easiest way to create these is with the configuration assistant:

./run -q configure

(As described below, you should use sudo ./run -q configure if the bot is running under an unprivileged user.)

Afterwards, you may edit these files manually whenever necessary.

Usage

Reports bot's standard tasks are located in the tasks/ directory. A ./run script is provided to make things simple, assuming you've followed the standard setup procedure above.

To run a task located at tasks/foobar.py:

./run foobar

For a full description of the command-line interface:

./run --help

If you're using a separate user for the bot, be aware that the ./run script tries to ensure that it is running under the account that owns the config directory. You can run jobs from reportsbot's crontab using the plain ./run syntax, but manual jobs under your own account should be initiated with sudo ./run.

If you prefer to use reportsbot as a regular Python package and execute task files at arbitrary locations, you can use this syntax, which supports the same arguments as ./run:

python3 -m reportsbot.cli full/path/to/task.py
python3 -m reportsbot.cli --help

Logs

The bot stores logs in the logs/ directory unless -t (--traceless) is passed to ./run. A few different kinds of logs are kept:

  • all.log stores non-DEBUG level logs for all tasks. It automatically rotates when it grows large.
  • all.err stores WARNING-level logs and above for all tasks. It automatically rotates when it grows large.
  • <sitename>/<taskname>.log stores non-DEBUG level logs for the specified task running on the specified site. It automatically rotates nightly.
  • <sitename>/<taskname>.err stores WARNING-level logs and above. It automatically rotates when it grows large.
  • <sitename>/<taskname>.log.verbose stores full logs for the last run of the task. It is cleared at the start of each run.

The bot also prints all logs (including DEBUG-level) to standard error unless -q (--quiet) is passed to ./run, in which case only ERROR-level and higher are printed. This option can be useful for cron jobs; if you set up cron to email you the output of ./run -q, you will be notified immediately when problems occur.

Tasks

A number of tasks are provided. Advice on developing your own is given at the end of this section.

  • load_project_config: Loads WikiProject-specific configuration from the wiki and stores it in the bot's database.
  • metrics: Updates monthly metrics on the number of articles in a project.
  • new_discussions: Provides a list of new discussions within a WikiProject's scope.
  • update_members: Updates WikiProject membership lists based on WikiProjectCard transclusions.
  • update_project_index: Updates the index of articles associated with each project.

Developing

To create new tasks, you can follow the skeleton in tasks/example.py. Important things to keep in mind:

  • The name of the task class doesn't matter (the bot searches by filename), but it should be descriptive.
  • The run method is the only method called by the task runner, other than __init__, which should only do inexpensive setup if you override it.
  • There should only be one Task subclass per module. __all__ is used to identify which class to run in case multiple exist in the module namespace, like if you import other Task subclasses to use their methods.

The task has access to two important attributes:

  • self._bot is the Bot instance, which provides the following functionality:
    • self._bot.site: the Pywikibot site instance
    • self._bot.wikidb: a database connection to the wiki replica
    • self._bot.localdb: a connection to the bot's local database
    • self._bot.wikidata: an interface to Wikidata Some methods are available for working with WikiProjects in a structured manner. See the reportsbot.bot.Bot class documentation for details.
  • self._logger is the Logger instance that you should use for all log messages. print and writing to stdout directly should be avoided.