How to Write a Plugin

Plugins are the building blocks of dōteki. They are standalone Python modules that provide data to the main application.

What kind of data? Anything you'd like to showcase on your GitHub profile. Take a look at the available plugins to see some examples.

tip

Before you start developing a plugin, set up your development environment.

General recommendations

• Set sane defaults for settings that allow it. For example, the current_date plugin uses the ISO standard ("%Y-%m-%d") as the default format.
• Think of other users; aim for flexibility. For example, the lastfm plugin can display top artists, albums, tracks, or tags, even though the original use case was to display top artists.
• Follow the coding guidelines.

Plugin structure

TL;DR: Plugins must have a run function that returns a single item or a list of items. In case of an error, the plugin should log an appropriate message and return None.

The run function

Plugins must have a run function. This is the entry point for the plugin and is called by the main application. The run function should adhere to the following signature:

from typing import Any

def run(settings: dict[str, Any]) -> Any:
    # Plugin logic.

You can be more specific about the return types. For example, if your plugin only returns a string, a list of strings, you can use:

from typing import Any

def run(settings: dict[str, Any]) -> str | list[str] | None:

We include None in the return type for error handling.

Generally, you'll want to follow this second signature to have more control over the output.

Remember: GitHub readmes support Markdown, so the plugin can return Markdown-formatted strings for bold text, links, etc. See the GitHub docs to learn more about the supported Markdown syntax.

Arguments

The run function accepts a single argument: a dictionary containing the settings for the plugin.

Return values

The run function can return a single item or a list of items. These items will be rendered as strings. The main application will format the list.

In case of an error, the plugin must return None so dōteki skips the section.

tip

To render an empty section, return an empty string.

Other functions

Plugins may have other functions, but they will not be called by the main application.

For example, it's a good idea to have a validate_settings function to validate user settings. See the code of the current_date plugin for a simple example.

Error handling

In case of an error, the plugin should log an appropriate message using logging and return None:

important_setting = settings.get("url")
if not important_setting:
    logging.error("No 'important_setting' provided for the Foo plugin")
    return None

You don't need to configure the logger; the main application takes care of that.

Environment variables

If your plugin needs to access sensitive information, such as an API key, it must do so through environment variables. This is to avoid leaking the information in the configuration file.

The environment variable name must be DOTEKI_<PLUGIN_NAME>_<SETTING_NAME>. For example, the lastfm plugin uses DOTEKI_LASTFM_API_KEY.

Make sure you mask the environment variable in your logs. For example:

try:
    response = requests.get(url)
    response.raise_for_status()
except requests.RequestException as e:
    safe_error_message = str(e).replace(api_key, "HIDDEN_API_KEY")
    logging.error(f"Error fetching data: {safe_error_message}")
    return None

Testing

Each plugin must have a corresponding test file in the tests directory. The filename must be test_<plugin_name>.py. For example, the current_date plugin has its tests in tests/test_current_date.py.

Write clear and concise tests, considering all edge cases you can think of.

To run all tests:

poetry run pytest

If you use VSCode, you might want to install the Python Test Explorer extension.

Code coverage

Plugins must have 80%+ code coverage. Aim for 100%, but don't let this stop you from contributing; do your best and reach out for help if needed. Do keep in mind functionality coverage (testing all features and edge cases) is more important than code coverage.

To generate a code coverage report, run:

poetry run pytest --cov=doteki

For VSCode users, the Coverage Gutters extension can highlight lines not covered by tests. This extension needs a a coverage report; generate it with:

poetry run pytest --cov=doteki --cov-report xml

Adding dependencies

dōteki uses Poetry to manage dependencies. The pyproject.toml file contains all the dependencies, including development dependencies.

All plugin dependencies must be marked as optional.

If your plugin adds a dependency, you need to:

1. Add it to the pyproject.toml file by using poetry:

    poetry add <dependency_name>

2. Edit pyproject.toml to mark it as optional. For example, if you just added pandas:

   [tool.poetry.dependencies]
   pandas = {version = "^2.1.4", optional = true}

3. Add your plugin name with its dependencies to the [tool.poetry.extras] section. Example:

   [tool.poetry.extras]
   plugin_name = ["pandas"]

4. Update the all section in [tool.poetry.extras] to add the new dependencies, if needed:

   [tool.poetry.extras]
   all = ["requests", "feedparser", "pandas"]

5. Finally, update poetry.lock:

   poetry update <dependency_name>

Documentation

Each plugin must have a Markdown file in the website/docs/plugins directory. The filename must match the name of the plugin. For example, the documentation for the current_date plugin can be found in website/docs/plugins/current_date.md.

This Markdown file must include:

• A description of the plugin's purpose.
• A list of dependencies.
• An explanation of all configuration options.
• Usage example(s).

Check out the Feed Plugin Documentation for an example.

Read more about the website and documentation here.

Example

The Pull Request to add the Current Date plugin would need to include:

• The plugin code in doteki/plugins/current_date.py.
• The tests in tests/plugins/test_current_date.py.
• The documentation in website/docs/plugins/current_date.md. (See the the page live here.)
• Any plugin dependency changes in pyproject.toml. (No changes in this case.)

Questions?

If something is not clear, or you have any questions, don't hesitate to reach out via the issue tracker, discussions, or email.