spacemacs/layers/+lang/python

Python layer

• Description
  • Features:
• Install
  • Layer
  • Choosing a backend
• Backends
  • Anaconda
  • Language Server Protocol
    • python-language-server
    • Microsoft python language server
• Additional tools
  • Syntax checking
  • Test runner
  • Buffer formatting
  • Automatic buffer formatting on save
  • Automatic save of buffer when testing
  • autoflake
  • pylookup
  • dap-mode debugger (only for lsp backend)
• Configuration
  • Fill column
  • Sort imports
  • Importmagic
• Management of Python versions and virtual environments
  • Manage virtual environments with pyvenv
  • Manage multiple Python versions with pyenv
    • Automatic activation of local pyenv version
• Key bindings
  • Inferior REPL process
  • Running Python Script in shell
  • Testing
  • Refactoring
  • Pip package management
  • Live coding
  • Other Python commands
  • Debugger

Description

This layer adds support for the Python language.

Features:

• Support for the following backends:

  • anaconda (default),

  • Language Server Protocol (experimental - 2 implementations),

    • python-language-server
    • Microsoft python language server
• Auto-completion
• Code Navigation
• Documentation Lookup using anaconda-mode and pylookup
• Test Runners using nose.el or pytest
• Virtual Environment using pyvenv and pyenv
• semantic mode is enabled
• PEP8 compliant formatting via YAPF or black
• PEP8 checks with flake8 or pylint
• Suppression of unused import with autoflake
• Use the % key to jump between blocks with evil-matchit
• Sort imports with isort
• Fix a missing import statement with importmagic
• Pip package manager with pippel

Install

Layer

To use this configuration layer, add it to your ~/.spacemacs. You will need to add python to the existing dotspacemacs-configuration-layers list in this file.

Choosing a backend

To choose a default backend set the layer variable python-backend:

  (python :variables python-backend 'anaconda)

Backend can be chosen on a per project basis using directory local variables (files named .dir-locals.el at the root of a project), an example to use the lsp backend:

  ;;; Directory Local Variables
  ;;; For more information see (info "(emacs) Directory Variables")

  ((python-mode (python-backend . lsp)))

Note: you can easily add a directory local variable with SPC f v d.

The available options are:

symbol	description
'anaconda	Default
'lsp	python-language-server package

Backends

Anaconda

anaconda-mode tries to install the dependencies itself but sometimes it does not work and you may encounter the following message when opening a python buffer:

  Blocking call to accept-process-output with quit inhibited!!

To fix this, install the anaconda-mode anaconda-deps by hand:

  pip install --upgrade "jedi>=0.9.0" "json-rpc>=1.8.1" "service_factory>=0.1.5"

If you encounter problems with Jedi 1.0 consider downgrading to 0.9.0. See this issue for details.

Source: https://github.com/proofit404/anaconda-mode#issues

If you are facing errors such as "Unable to run anaconda-mode server", try setting your PYTHONPATH as explained at https://github.com/proofit404/anaconda-mode#pythonpath

Language Server Protocol

The lsp backend can use either of the following language server implementations:

symbol	description
'pyls	python-language-server package (default)
'mspyls	Microsoft python language server

pyls is used by default - to use the Microsoft server, set the python-lsp-server layer variable as follows:

  (python :variables python-backend 'lsp python-lsp-server 'mspyls)

python-language-server

You just have to install python language server package:

  pip install python-language-server

Additionally you can install the following other packages:

  # for import sorting
  pip install pyls-isort
  # for mypy checking (python 3.4+ is needed)
  pip install pyls-mypy

If you've installed the language server and related packages as development dependencies in a pipenv environment, you'll want to set the python-pipenv-activate config variable to t. This activates your pipenv before enabling the lsp backend.

Microsoft python language server

Paraphrasing the instructions provided by the author of the lsp-python-ms package:

  git clone https://github.com/Microsoft/python-language-server.git
  cd python-language-server/src/LanguageServer/Impl
  dotnet build -c Release
  dotnet publish -c Release -r <RUNTIME>

where <RUNTIME> is one of the runtime IDs supported by dotnet core. One of linux-x64, osx-x64, win10-x64 should cover most use cases.

The default package configuration assumes the executable is located in a folder included in your system path. To use the latest built version in a cloned git repo, use the python-lsp-git-root config variable, e.g.:

  (setq-default dotspacemacs-configuration-layers
    '((python :variables
              python-backend 'lsp
              python-lsp-server 'mspyls
              python-lsp-git-root "~/dev/python/python-language-server")))

N.B. If you're using Arch linux or a derivative distribution, you can install the microsoft-python-language-server package from the AUR.

Additional tools

Syntax checking

Syntax checking uses flake8 package:

  pip install flake8

Test runner

Both nose and pytest are supported. By default nose is used. To choose your test runner set the layer variable python-test-runner to either nose or pytest.

  (setq-default dotspacemacs-configuration-layers
    '((python :variables python-test-runner 'pytest)))

If you need both then you can set python-test-runner to a list like this:

  (setq-default dotspacemacs-configuration-layers
    '((python :variables python-test-runner '(pytest nose))))

This means that pytest is your primary test runner. To use the secondary test runner you can call the test functions with a prefix argument e.g. SPC u SPC m t t to run one test with nose.

To set project specific test runners you can set python-test-runner in a directory local variable in your project root. SPC f v d in Spacemacs. See the official documentation for more information.

The root of the project is detected with a .git directory or a setup.cfg file.

Buffer formatting

One of YAPF (the default) or black may be selected as the formatter, via python-formatter, as 'yapf or 'black respectively.

  (setq-default dotspacemacs-configuration-layers '(
    (python :variables python-formatter 'yapf)))

The key binding SPC m = invokes the selected formatter on the current buffer when in python mode.

Note that YAPF and black may also be invoked unconditionally via yapfify-buffer and blacken-buffer, respectively, provided that they are installed.

Automatic buffer formatting on save

To enable automatic buffer formatting on save set the variable python-format-on-save to t. The formatter specified by python-formatter will be used.

  (setq-default dotspacemacs-configuration-layers '(
    (python :variables python-format-on-save t)))

Automatic save of buffer when testing

By default a buffer is automatically saved before tests are executed upon it, you can disable this feature by setting python-save-before-test to nil.

  (setq-default dotspacemacs-configuration-layers '(
    (python :variables python-save-before-test nil)))

autoflake

To be able to suppress unused imports easily, install autoflake:

  pip install autoflake

pylookup

To use pylookup on SPC m h H, make sure you update the database first, using SPC SPC pylookup-update.

dap-mode debugger (only for lsp backend)

To use dap-mode for debugging do:

  pip install "ptvsd>=4.2"

Configuration

Fill column

If you want to customize the fill column value, use something like this inside the user-init function in your .spacemacs:

  (setq-default dotspacemacs-configuration-layers '(
      (python :variables python-fill-column 99)))

Sort imports

If you want imports to be automatically sorted when you save a file (using isort), set the python-sort-imports-on-save variable in the python layer config section:

  (setq-default dotspacemacs-configuration-layers
    '((python :variables python-sort-imports-on-save t)))

or as a directory-local variable (for per-project settings).

Importmagic

Install importmagic and epc for importmagic functionality.

  pip install importmagic epc

Management of Python versions and virtual environments

Manage virtual environments with pyvenv

A virtual environment provides isolation of your Python package versions. For a general overview see this site. Virtualenvwrapper which is also explained in the previous link, is a program which manages your virtual environments in a central location set by the WORKON_HOME environment variable.

Spacemacs integration of virtual environments and virtualenvwrapper is provided by the pyvenv package. It provides the following key bindings:

Key binding	Description
SPC m V a	activate a virtual environment in any directory
SPC m V d	deactivate active virtual environment
SPC m V w	work on virtual environment in WORKON_HOME

Manage multiple Python versions with pyenv

If you need multiple Python versions (e.g. Python 2 and Python 3) then take a look at pyenv. It enables the installation and management of multiple Python versions. This blogpost gives a good overview on how to use the tool. Spacemacs integration is provided by pyenv mode which has the following key bindings.

Key binding	Description
SPC m v s	set a pyenv environment with pyenv
SPC m v u	unset a pyenv environment with pyenv

Pyenv can also manage virtual environments for each of the Python versions it has installed. Those will be listed alongside your Python versions.

Automatic activation of local pyenv version

A project-specific pyenv version may be written to a file called .python-version using the pyenv local command.

Spacemacs can search in parent directories for this file, and automatically set the pyenv version. The behavior can be set with the variable python-auto-set-local-pyenv-version to:

• on-visit (default) set the version when you visit a python buffer,
• on-project-switch set the version when you switch projects,
• nil to disable.

The same is also possible on pyvenv with a file called .venv. The behavior can be set with the variable python-auto-set-local-pyvenv-virtualenv to:

• on-visit (default) set the virtualenv when you visit a python buffer,
• on-project-switch set the virtualenv when you switch projects,
• nil to disable.

Key bindings

Inferior REPL process

Start a Python or iPython inferior REPL process with SPC m s i. If ipython is available in system executable search paths, ipython will be used to launch python shell; otherwise, default python interpreter will be used. You may change your system executable search path by activating a virtual environment.

Send code to inferior process commands:

Key binding	Description
SPC m s b	send buffer and keep code buffer focused
SPC m s B	send buffer and switch to REPL in insert mode
SPC m s f	send function and keep code buffer focused
SPC m s F	send function and switch to REPL in insert mode
SPC m s i	start inferior REPL process
SPC m s r	send region and keep code buffer focused
SPC m s R	send region and switch to REPL in insert mode
CTRL+j	next item in REPL history
CTRL+k	previous item in REPL history

Running Python Script in shell

To run a Python script like you would in the shell press SPC m c c to start the Python script in comint mode. This is useful when working with multiple Python files since the REPL does not reload changes made in other modules.

Key binding	Description
SPC m c c	Execute current file in a comint shell
SPC m c C	Execute current file in a comint shell and switch to it in insert state

Note: With the universal argument SPC u you can enter a new compilation command.

Testing

Test commands start with m t. To use the secondary test runner call the function with a prefix argument, for example SPC u SPC m t a.

No Debug	Description
SPC m t a	launch all tests of the project
SPC m t b	launch all tests of the current buffer (same as module)
SPC m t l	launch last tests
SPC m t m	launch all tests of the current module
SPC m t s	launch all tests of the current suite (only with nose)
SPC m t t	launch the current test (function)

Debug	Description
SPC m t A	launch all tests of the project in debug mode
SPC m t B	launch all tests of the current buffer (module) in debug mode
SPC m t M	launch all tests of the current module in debug mode
SPC m t S	launch all tests of the current suite in debug mode (only with nose)
SPC m t T	launch the current test (function) in debug mode

Refactoring

Key binding	Description
SPC m r f	fix a missing import statement with importmagic
SPC m r i	remove unused imports with autoflake
SPC m r I	sort imports with isort

Pip package management

In python buffer type SPC m P to open buffer listing all installed pip packages in the currently activated virtual environment.

Note: To open this menu from outside a python buffer type SPC SPC pippel-list-packages RET.

In the package list buffer:

Key binding	Description
RET	follow link (pippel-menu-visit-homepage)
d	mark for deletion (pippel-menu-mark-delete)
i	prompt user for packages (pippel-install-package)
m	remove mark (pippel-menu-mark-unmark)
r	refresh package list (pippel-list-packages)
U	mark all upgradable (pippel-menu-mark-all-upgrades)
u	mark for upgrade (pippel-menu-mark-upgrade)
x	perform marked package menu actions (pippel-menu-execute)

Live coding

Live coding is provided by the live-py-plugin.

Key binding	Description
SPC m l	Toggle live-py-mode

Other Python commands

Key binding	Description
SPC m =	reformat the buffer using formatter specified in python-formatter
SPC m d b	toggle a breakpoint using wdb, ipdb, pudb, pdb or python3.7 (and above)
SPC m g a	go to assignment using anaconda-mode-find-assignments (C-o to jump back)
SPC m g b	jump back
SPC m g g	go to definition using anaconda-mode-find-definitions (C-o to jump back)
SPC m g u	navigate between usages with anaconda-mode-find-references
SPC m h d	look for documentation using helm-pydoc
SPC m h h	quick documentation using anaconda
SPC m h H	open documentation in firefox using pylookup
SPC m v a	activate a virtual environment in any directory
SPC m v d	deactivate active virtual environment
SPC m v s	set a pyenv environment with pyenv
SPC m v u	unset a pyenv environment with pyenv
SPC m v w	work on virtual environment in WORKON_HOME
SPC m v p a	activate pipenv in current project
SPC m v p d	deactivate pipenv in current project
SPC m v p i	install module into pipenv environment
SPC m v p o	open pipenv module in buffer
SPC m v p s	launch pipenv shell in current project
SPC m v p u	uninstall module from pipenv environment

Debugger

Key binding	Description
SPC m d d d	start debugging
SPC m d d l	debug last configuration
SPC m d d r	debug recent configuration
SPC m d c	continue
SPC m d i	step in
SPC m d o	step out
SPC m d s	next step
SPC m d v	inspect value at point
SPC m d r	restart frame
SPC m d .	debug transient state
SPC m d a	abandon current session
SPC m d A	abandon all process
SPC m d e e	eval
SPC m d e r	eval region
SPC m d e t	eval value at point
SPC m d S s	switch session
SPC m d S t	switch thread
SPC m d S f	switch frame
SPC m d I i	inspect
SPC m d I r	inspect region
SPC m d I t	inspect value at point
SPC m d b b	toggle a breakpoint
SPC m d b c	change breakpoint condition
SPC m d b l	change breakpoint log condition
SPC m d b h	change breakpoint hit count
SPC m d b a	add a breakpoint
SPC m d b d	delete a breakpoint
SPC m d b D	clear all breakpoints
SPC m d '_	Run debug REPL
SPC m d w l	list local variables
SPC m d w o	goto output buffer if present
SPC m d w s	list sessions
SPC m d w b	list breakpoints