# lint_roller - A plugin specification for linters `lint_roller` is an itty-bitty plugin API for code analysis tools like linters and formatters. It provides plugins for those tools to load extensions and specify custom rulesets. (As of this release, only [Standard Ruby](https://github.com/standardrb/standard) supports `lint_roller` plugins, but we think [RuboCop](https://github.com/rubocop/rubocop) could add support and most plugins would be compatible with both `rubocop` and `standardrb`. Additionally, there's nothing that would prevent other tools like [rufo](https://github.com/ruby-formatter/rufo) from adopting it.) ## How to make a plugin If you want to make a plugin, the first thing you should do is extend [LintRoller::Plugin](/lib/lint_roller/plugin.rb) with a custom class. Let's take this example plugin for banana-related static analysis: ```ruby module BananaRoller class Plugin < LintRoller::Plugin # `config' is a Hash of options passed to the plugin by the user def initialize(config = {}) @alternative = config["alternative"] ||= "chocolate" end def about LintRoller::About.new( name: "banana_roller", version: "1.0", # or, in a gem, probably BananaRoller::VERSION homepage: "https://github.com/example/banana_roller", description: "Configuration of banana-related code" ) end # `context' is an instance of LintRoller::Context provided by the runner def supported?(context) context.engine == :rubocop end # `context' is an instance of LintRoller::Context provided by the runner def rules(context) LintRoller::Rules.new( type: :path, config_format: :rubocop, value: Pathname.new(__dir__).join("../../config/default.yml") ) end end end ``` And that's pretty much it. Just a declarative way to identify your plugin, detect whether it supports the given [LintRoller::Context](/lib/lint_roller_context.rb) (e.g. the current `runner` and `engine` and their respective `_version`s), for which the plugin will ultimately its configuration as a [LintRoller::Rules](/lib/lint_roller/rules.rb) object. ## Packaging a plugin in a gem In order for a formatter to use your plugin, it needs to know what path to require as well as the name of the plugin class to instantiate and invoke. To make this work seamlessly for your users without additional configuration of their own, all you need to do is specify a metadata attribute called `default_lint_roller_plugin` in your gemspec. Taking [standard-custom](https://github.com/standardrb/standard-custom) as an example, its gemspec contains: ```ruby Gem::Specification.new do |spec| # … spec.metadata["default_lint_roller_plugin"] = "Standard::Custom::Plugin" # … end ``` Because gem specs are loaded by RubyGems and Bundler very early, remember to specify the plugin as a string representation of the constant, as load order usually matters, and most tools will need to be loaded before any custom extensions. Hence, `"Standard::Custom::Plugin"` instead of `Standard::Custom::Plugin`. ## Using your plugin Once you've made your plugin, here's how it's configured from a Standard Ruby `.standard.yml` file. ### If your plugin is packaged as a gem Packaging your plugin in a gem is the golden path, both because distributing code via [RubyGems.org](https://rubygems.org) is very neat, but also because it makes the least work for your users. If your gem name is `banana_roller` and you've set `spec.metadata["default_lint_roller_plugin"]` to `"BananaRoller::Plugin"`, then your users could just add this to their `.standard.yml` file: ```yaml plugins: - banana_roller ``` And that's it! During initialization, `standardrb` will `require "banana_roller"` and know to call `BananaRoller::Plugin.new(config)` to instantiate it. ### If your plugin ISN'T in a gem If you're developing a plugin for internal use or in conjunction with a single project, you may want it to live in the same repo as opposed to packaging it in a gem. To do this, then—in lieu of a gem name—provide the path you want to be required as its name, and (since there is no `spec.metadata` to learn of your plugin's class name), specify it as an option on the plugin: ```yaml plugins: - lib/banana_roller/plugin: plugin_class_name: BananaRoller::Plugin ``` (Be careful with the indentation here! Any configuration under a plugin must be indented in order for it to be parsed as a hash under the `"lib/banana_roller/plugin"` key.) Additionally, if you want the plugin's name to make more sense, you can give it whatever name you like in the configuration and specify the `require_path` explicitly: ```yaml plugins: - banana_roller: require_path: lib/banana_roller/plugin plugin_class_name: BananaRoller::Plugin ``` ### Passing user configuration to the plugin When a `LintRoller::Plugin` is instantiated, users can pass a configuration hash that tells your plugin how to behave. To illustrate how this works in Standard Ruby, anything passed as a hash beneath a plugin will be passed to the class's `initialize` method: ```yaml plugins: - apple_roller - banana_roller: require_path: lib/banana_roller/plugin plugin_class_name: BananaRoller::Plugin alternative: "apples" - orange_roller: rind: false ``` In the above case, `apple_roller`'s plugin class will be instantiated with `new({})`, `banana_roller` will get all 3 of those parameters passed `BananaRoller::Plugin.new({require_path…})`, and `orange_roller`'s class will be called with `new({rind: false})`. ## Code of Conduct This project follows Test Double's [code of conduct](https://testdouble.com/code-of-conduct) for all community interactions, including (but not limited to) one-on-one communications, public posts/comments, code reviews, pull requests, and GitHub issues. If violations occur, Test Double will take any action they deem appropriate for the infraction, up to and including blocking a user from the organization's repositories.