napari-plugin-devtools

A repo with tools and services for napari plugin developers

Installation

This tool can be installed via pip: pip install napari-plugin-devtools, you can also add it to your requirements.txt to be installed with other dependencies automatically.

Validation tool

The validation tools provides automated tests for plugin developers, these checks verify that the plugin is available for users of napari to install, and would register entry points with napari.

There are two parts to the validation tool. One part is a command line interface and the other is a pytest fixture.

Command Line Interface (CLI) Usage

The CLI can be used by continuous integration (CI) pipelines to perform a quick verification of a plugins setup without any specific input required. It serves as a quick "sanity check". (It is also accessible from python in validation.py)

The tool can be run as npd <cmd>. where currently we support cmd validate:

-a|--all: validation is then run on everything, including package checks and all hooks checks. In this mode the command has exit code 1 if no hook of any type is found, or any package failed to be validated.

-p|--packages [PACKAGES [PACKAGES ...]]: validation runs on packages built, for example python setup.py sdist bdist_wheel, it checks packages with given paths, and if no path is given, it runs on all files under dist folder to checks that all packages built are correctly tagged with classifier "Framework :: napari". This is recommended for most plugins unless you do not want your plugin to appear in the napari built-in plugin installation tool. Can be used with -s|--skip-repackaging when not rebuilding dist folder.

-k|--hooks {reader,writer,function,widget} [{reader,writer,function,widget} ...] validation runs on provided list of hook types that after the plugin is installed to current python environment, typically done through pip install -e ., there is at least one hook under corresponding category registered by the plugin, and there are no conflicts in name or registration error, when used with verbose mode, the function signatures are also printed, for example:

    [{
        "plugin name": "napari-demo",
        "function name": "image arithmetic",
        "args": ["layerA", "operation", "layerB"],
        "annotations": {
            "return": "napari.types.ImageData",
            "layerA": "napari.types.ImageData",
            "operation": "<enum 'Operation'>",
            "layerB": "napari.types.ImageData"
        },
        "defaults": null
    }]

-i|--include-plugin INCLUDE_PLUGIN [INCLUDE_PLUGIN ...] run hook checks only on listed plugins, this is useful to filter out other plugins on a complicated python environment.

-e|--exclude-plugin EXCLUDE_PLUGIN [EXCLUDE_PLUGIN ...] do not run hook checks on listed plugins, this is useful to filter out other plugins on a complicated python environment.

-s|--skip-repackaging when specified, skipping the packaging step before package checks. This is useful when the dist folder has correct packages to validate. Otherwise we would rebuild dist folder to make sure the result is correctly reflecting latest code status.

-v|--verbose enable verbose mode, gives slightly more information on the underlying findings of validation process.

Pytest fixture usage

devtools provides a pytest fixture: napari_plugin_tester in plugin_tester.py, it extends a plugin manager used by napari and have additional assertion modes ready:

def test_pm(napari_plugin_tester):
    napari_plugin_tester.assert_plugin_name_registered("test-plugin")
    napari_plugin_tester.assert_module_registered(_test)
    napari_plugin_tester.assert_implementations_registered(
        "test-plugin", "napari_get_reader"
    )

where you can check more specifically on a module or function being registered under napari annotations, see hook specifications to find what other annotations are avaiable in addition to napari_get_reader