pylsp-mypy 0.8.0

Mypy linter for the Python LSP Server

Mypy plugin for PYLSP

This is a plugin for the Python LSP Server, to make mypy output available directly in your IDE. pylsp-mypy uses mypy to help you write statically typed Python code. It, like mypy, requires Python 3.10 or newer.

Installation

Install into the same virtualenv as python-lsp-server itself.

pip install pylsp-mypy

Configuration

pylsp-mypy supports the use of pyproject.toml for configuration. This is the recommended way.

For more advanced usecases, it can also be configuered using configs provided directly to the LSP server. To use this prepend pylsp.plugins.pylsp_mypy. to any configuration key. For example strict becomes pylsp.plugins.pylsp_mypy.strict

The configuration keys are listed in the following.

pyproject.toml key	Type	Description	Default
strict	bool	Refers to the strict option of mypy. This option often is too strict to be useful.	false
live_mode	bool	Provides type checking as you type. This writes to a tempfile every time a check is done. Turning off live_mode means you must save your changes for mypy diagnostics to update correctly.	true
hover	bool	Displays types as inferred by mypy on hover. This requires dmypy to work.	false
dmypy	bool	Executes via dmypy run rather than mypy. This uses the dmypy daemon and may dramatically improve the responsiveness of the pylsp server, however this currently does not work in live_mode. Enabling this disables live_mode, even for conflicting configs.	false
report_progress	bool	Report basic progress to the LSP client. With this option, pylsp-mypy will report when mypy is running, given your editor supports LSP progress reporting. For small files this might produce annoying flashing in your editor, especially in live_mode. For large projects, enabling this can be helpful to assure yourself whether mypy is still running.	false
exclude	list of str items	A list of regular expressions which should be ignored. The mypy runner wil not be invoked when a document path is matched by one of the expressions. Note that this differs from the exclude directive of a mypy config which is only used for recursively discovering files when mypy is invoked on a whole directory. For both windows or unixoid platforms you should use forward slashes (/) to indicate paths.	[]
follow-imports	"normal", "silent", "skip" or "error"	mypy parameter follow-imports. In mypy this is normal by default. We set it silent, to sort out unwanted results. This can cause cache invalidation if you also run mypy in other ways. Setting this to normal avoids this at the cost of a small performance penalty.	"silent"
config_sub_paths	list of str items	Specifies sub paths under which the mypy configuration file may be found. For each directory searched for the mypy config file, this also searches the sub paths specified here.	[]
mypy_command	list of str items	The command to run mypy. This is useful if you want to run mypy in a specific virtual environment. Requires env variable PYLSP_MYPY_ALLOW_DANGEROUS_CODE_EXECUTION to be set.	[]
dmypy_command	list of str items	The command to run dmypy. This is useful if you want to run dmypy in a specific virtual environment. Requires env variable PYLSP_MYPY_ALLOW_DANGEROUS_CODE_EXECUTION to be set.	[]
dmypy_status_file	str	Specifies which status file dmypy should use. This modifies the --status-file option passed to dmypy given dmypy is active.	.dmypy.json
overrides	list of (str items or true)	A list of alternate or supplemental command-line options. This modifies the options passed to mypy or the mypy-specific ones passed to dmypy run. When present, the special boolean member true is replaced with the command-line options that would've been passed had overrides not been specified.	[true]

Both mypy_command and dmypy_command could be used by a malicious repo to execute arbitrary code by looking at its source with this plugin active. Still users want this feature. For security reasons this is disabled by default. If you really want it and accept the risks, set the environment variable PYLSP_MYPY_ALLOW_DANGEROUS_CODE_EXECUTION in order to activate it.

Using a pyproject.toml for configuration, which is in fact the preferred way, your configuration could look like this:

[tool.pylsp-mypy]
enabled = true
live_mode = true
strict = true
exclude = ["tests/*"]

It should be roughly like this for a standard configuration:

[tool.pylsp-mypy]
enabled = true
live_mode = true
strict = false
exclude = ["tests/*"]

With dmypy enabled your config should look like this:

[tool.pylsp-mypy]
enabled = true
live_mode = false
dmypy = true
strict = false

With overrides specified (for example to tell mypy to use a different python than the currently active venv), your config could look like this:

[tool.pylsp-mypy]
enabled = true
overrides = ["--python-executable", "/home/me/bin/python", true]

With dmypy_status_file your config could look like this:

[tool.pylsp-mypy]
enabled = true
live_mode = false
dmypy = true
strict = false
dmypy_status_file = ".custom_dmypy_status_file.json"

With config_sub_paths your config could look like this:

[tool.pylsp-mypy]
enabled = true
config_sub_paths = [".config"]

With report_progress your config could look like this:

[tool.pylsp-mypy]
enabled = true
report_progress = true

With mypy_command your config could look like this:

[tool.pylsp-mypy]
enabled = true
mypy_command = ["poetry", "run", "mypy"]

With dmypy_command your config could look like this:

[tool.pylsp-mypy]
enabled = true
live_mode = false
dmypy = true
dmypy_command = ["/path/to/venv/bin/dmypy"]

Developing

Install development dependencies with (you might want to create a virtualenv first):

pip install -r requirements.txt

The project is formatted with black. You can either configure your IDE to automatically format code with it, run it manually (black .) or rely on pre-commit (see below) to format files on git commit.

The project is formatted with isort. You can either configure your IDE to automatically sort imports with it, run it manually (isort .) or rely on pre-commit (see below) to sort files on git commit.

The project uses a markdown test in order to assure uploadability to pypi: markdownlint as a pre-commit hook and in a GitHub workflow. This does not catch all errors.

This project uses pre-commit to enforce code-quality. After cloning the repository install the pre-commit hooks with:

pre-commit install

After that pre-commit will run all defined hooks on every git commit and keep you from committing if there are any errors.