A simple tool that auto-inserts import statements.
Sin.titulo3.mov
It matches with:
- Pre-defined lookup tables
- Existing imports in the project (implemented in Python:
python-import
cli)- It finds the imports in the project and shows a list of them to choose from.
- pyright/basedpyright LSP completion
- Just
import <word>
:P
This plugin doesn't detect duplicated imports. Use ruff
to sort imports.
Path
->from pathlib import Path
np
->import numpy as np
torch
->import torch
logging
->import logging
logger
->
import logging
logger = logging.getLogger(__name__)
Most will use the current word to find the import statement, but the treesitter node can be used to find the import statement more accurately. (e.g. class SomeDataset(torch.utils.data.DataLoader)
-> import torch.utils.data
)
Warning
This work-in-progress plugin is not yet ready for the public. There aren't many customisation options (e.g. how to locate the import place) and the behaviour and build/configuration instructions change rapidly. If you don't want breaking changes, please fix the version of this plugin.
- 💻 Neovim >= 0.10
- pipx or uv (or any other way to install
python-import
cli in PATH) - ripgrep (
brew install ripgrep
orcargo install ripgrep
) - fd (
brew install fd
,cargo install fd-find
ornpm install -g fd-find
)
{
"kiyoon/python-import.nvim",
build = "pipx install . --force",
keys = {
{
"<M-CR>",
function()
require("python_import.api").add_import_current_word_and_notify()
end,
mode = { "i", "n" },
silent = true,
desc = "Add python import",
ft = "python",
},
{
"<M-CR>",
function()
require("python_import.api").add_import_current_selection_and_notify()
end,
mode = "x",
silent = true,
desc = "Add python import",
ft = "python",
},
{
"<space>i",
function()
require("python_import.api").add_import_current_word_and_move_cursor()
end,
mode = "n",
silent = true,
desc = "Add python import and move cursor",
ft = "python",
},
{
"<space>i",
function()
require("python_import.api").add_import_current_selection_and_move_cursor()
end,
mode = "x",
silent = true,
desc = "Add python import and move cursor",
ft = "python",
},
{
"<space>tr",
function()
require("python_import.api").add_rich_traceback()
end,
silent = true,
desc = "Add rich traceback",
ft = "python",
},
},
opts = {
-- Example 1:
-- Default behaviour for `tqdm` is `from tqdm.auto import tqdm`.
-- If you want to change it to `import tqdm`, you can set `import = {"tqdm"}` and `import_from = {tqdm = nil}` here.
-- If you want to change it to `from tqdm import tqdm`, you can set `import_from = {tqdm = "tqdm"}` here.
-- Example 2:
-- Default behaviour for `logger` is `import logging`, ``, `logger = logging.getLogger(__name__)`.
-- If you want to change it to `import my_custom_logger`, ``, `logger = my_custom_logger.get_logger()`,
-- you can set `statement_after_imports = {logger = {"import my_custom_logger", "", "logger = my_custom_logger.get_logger()"}}` here.
extend_lookup_table = {
---@type string[]
import = {
-- "tqdm",
},
---@type table<string, string>
import_as = {
-- These are the default values. Here for demonstration.
-- np = "numpy",
-- pd = "pandas",
},
---@type table<string, string>
import_from = {
-- tqdm = nil,
-- tqdm = "tqdm",
},
---@type table<string, string[]>
statement_after_imports = {
-- logger = { "import my_custom_logger", "", "logger = my_custom_logger.get_logger()" },
},
},
---Return nil to indicate no match is found and continue with the default lookup
---Return a table to stop the lookup and use the returned table as the result
---Return an empty table to stop the lookup. This is useful when you want to add to wherever you need to.
---@type fun(winnr: integer, word: string, ts_node: TSNode?): string[]?
custom_function = function(winnr, word, ts_node)
-- if vim.endswith(word, "_DIR") then
-- return { "from my_module import " .. word }
-- end
end,
},
},
"rcarriga/nvim-notify", -- optional
pipx
is easy to configure, but uv
is much faster. Make sure to install uv >= 0.2 and change the build script as follows:
With lazy.nvim:
{
"kiyoon/python-import.nvim",
build = "bash scripts/build_with_uv.sh ~/.virtualenvs/python-import",
-- Other configurations ...
},
This simply creates a virtual environment in ~/.virtualenvs/python-import
and installs the cli there with uv pip install .
. Then it sym-links the binary to ~/.local/bin/python-import
.
Run :checkhealth python_import
and see if python-import.nvim is installed correctly.
You need to disable lazy loading or run any commands in a python file to activate the plugin first.
==============================================================================
python_import: require("python_import.health").check()
python-import ~
- OK Using Neovim >= 0.10.0
- OK `rg` is installed
- OK `fd` is installed
- OK `python-import` is installed
- ERROR python-import cli (0.1.0+41.g9513418.dirty) and nvim plugin version (0.1.0+42.g9f74e1d.dirty) mismatch.
- Add more tests
- Command to add imports in TYPE_CHECKING
- Command to add lazy imports in methods
- Command to add imports in jupytext cell
- Search in ipynb files
- VSCode-neovim integration (use
vim.ui.select
for vscode UI support. However, this complicates stuff because it runs async)- Currently, if you open the vscode-neovim output terminal, it is quite usable. The notification works as well. LSP doesn't work yet.