An extension for nvim-dap providing default configurations for python and methods to debug individual test methods or classes.
- Requires Neovim >= 0.5
- Requires nvim-dap
- Requires debugpy
- Install like any other neovim plugin:
- If using vim-plug:
Plug 'mfussenegger/nvim-dap-python'
- If using packer.nvim:
use 'mfussenegger/nvim-dap-python'
- If using vim-plug:
If you want to use the test runner functionality, it additionally requires a tree sitter parser for Python.
It is recommended to install debugpy into a dedicated virtualenv. To do so:
mkdir .virtualenvs
cd .virtualenvs
python -m venv debugpy
debugpy/bin/python -m pip install debugpy
The debugger will automatically pick-up another virtual environment if it is activated before neovim is started.
Install either:
- Via
:TSInstall python
of nvim-treesitter - Compile the parser from tree-sitter-python and copy it into
.config/nvim/parser/
:git clone https://github.com/tree-sitter/tree-sitter-python.git
cd tree-sitter-python
cc -O2 -o ~/.config/nvim/parser/python}.so -I./src src/parser.c src/scanner.cc -shared -Os -lstdc++ -fPIC
- Call
setup
in yourinit.vim
to register the adapter and configurations:
lua require('dap-python').setup('~/.virtualenvs/debugpy/bin/python')
The argument to setup
is the path to the python installation which contains the debugpy
module.
- Use nvim-dap as usual.
- Call
:lua require('dap').continue()
to start debugging. - See
:help dap-mappings
and:help dap-api
. - Use
:lua require('dap-python').test_method()
to debug the closest method above the cursor.
Supported test frameworks are unittest
, pytest
and django
. By default it
tries to detect the runner by probing for pytest.ini
and manage.py
, if
neither are present it defaults to unittest
.
To configure a different runner, change the test_runner
variable. For example
to configure pytest
set the test runner like this in vimL
:
lua require('dap-python').test_runner = 'pytest'
You can also add custom runners. An example in Lua
:
local test_runners = require('dap-python').test_runners
-- `test_runners` is a table. The keys are the runner names like `unittest` or `pytest`.
-- The value is a function that takes three arguments:
-- The classname, a methodname and the opts
-- (The `opts` are coming passed through from either `test_method` or `test_class`)
-- The function must return a module name and the arguments passed to the module as list.
test_runners.your_runner = function(classname, methodname, opts)
local args = {classname, methodname}
return 'modulename', args
end
See :help dap-python
nnoremap <silent> <leader>dn :lua require('dap-python').test_method()<CR>
nnoremap <silent> <leader>df :lua require('dap-python').test_class()<CR>
vnoremap <silent> <leader>ds <ESC>:lua require('dap-python').debug_selection()<CR>
If you call the require('dap-python').setup
method it will create a few nvim-dap
configuration entries. These configurations are general purpose configurations suitable for many use cases, but you may need to customize the configurations - for example if you want to use Docker containers.
To add your own entries, you can extend the dap.configurations.python
list after calling the setup
function:
lua << EOF
require('dap-python').setup('/path/to/python')
table.insert(require('dap').configurations.python, {
type = 'python',
request = 'launch',
name = 'My custom launch configuration',
program = '${file}',
-- ... more options, see https://github.com/microsoft/debugpy/wiki/Debug-configuration-settings
})
EOF
An alternative is to use project specific .vscode/launch.json
files, see :help dap-launch.json
.
The Debugpy Wiki contains a list of all supported configuration options.
nvim-dap-python
by default looks for VIRTUAL_ENV
and CONDA_PREFIX
environment variables and if present uses these environments to execute your
application or tests.
If you're using another way to manage virtual environments, you can set a
custom resolve_python
function:
require('dap-python').resolve_python = function()
return '/absolute/path/to/python'
end
A test runner building upon vim-test with nvim-dap support. Aims to work for all python runners.