README.md

renamer.nvim tests

This directory holds the tests for renamer.nvim. In here you will find the all the information needed to write, run and understand those tests.

Frameworks and dependencies

Tests are possible through the busted and luassert implementation that plenary.nvim provides. It uses the busted framework to structure and run the tests and luassert for mocking the Neovim API or renamer methods which should not be validated by the respective unit test.

Running tests

To run the tests you can either use the make test command (from the project root) or the more complex:

nvim --headless --noplugin -u scripts/minimal_init.vim -c "PlenaryBustedDirectory lua/tests/ { minimal_init = './scripts/minimal_init.vim' }"

In order to properly run, the tests require a minimal Neovim setup to be made before they are actually ran, which is defined in scripts/minimal_init.vim. plenary also needs to be aware of where tests are found, which is done by specifying the PlenaryBustedDirectory.

Style

All tests must follow the AAA (Arrange, Act, Assert) pattern. The structure of unit tests should be:

describe(<component_name>, function()
    describe(<subcomponent_name>, function()
        it(<tested_behaviour_description>, function()
            -- arrange
            -- ...

            -- act
            -- ...

            -- assert
            -- ...
        end)
    end)

    -- or

    it(<tested_behaviour_description>, function()
        -- arrange
        -- ...

        -- act
        -- ...

        -- assert
        -- ...
    end)
end)

An example of this style would be:

describe('renamer', function()
    describe('on_close', function()
        it('should delete valid buffer', function()
            -- arrange
            -- ...

            -- act
            -- ...

            -- assert
            -- ...
        end)
    end)
end)

Tests should not cover the same behaviour twice or more, if possible. For example, if a tests validates the behaviour when a window ID is invalid, a test validating the behaviour when a buffer ID is invalid should not check the former again.