jedi-vim is a VIM binding to the autocompletion library Jedi.
Here are some pictures:
Completion for almost anything (Ctrl+Space).
Display of function/class bodies, docstrings.
Documentation (Pydoc) support (with highlighting, Shift+k).
There is also support for goto and renaming.
Get the latest from github.
Documentation is available in your vim: :help jedi-vim
. You can also look
it up on github.
You can read the Jedi library documentation here.
If you want to report issues, just use the github issue tracker. In case of
questions about the software, please use stackoverflow and tag your question with jedi-vim
.
We love Pull Requests! Read the instructions in CONTRIBUTING.md
.
The Jedi library understands most of Python's core features. From decorators to generators, there is broad support.
Apart from that, jedi-vim supports the following commands
- Completion
<C-Space>
- Goto assignment
<leader>g
(typical goto function) - Goto definition
<leader>d
(follow identifier as far as possible, includes imports and statements) - Goto (typing) stub
<leader>s
- Show Documentation/Pydoc
K
(shows a popup with assignments) - Renaming
<leader>r
- Usages
<leader>n
(shows all the usages of a name) - Open module, e.g.
:Pyimport os
(opens theos
module)
You need a VIM version that was compiled with Python 2.7 or later
(+python
or +python3
). You can check this from within VIM using
:python3 import sys; print(sys.version)
(use :python
for Python 2).
You might want to use pathogen or Vundle to install jedi-vim.
The first thing you need after that is an up-to-date version of Jedi. Install
git submodule update --init --recursive
in your jedi-vim repository.
Example installation command using Pathogen:
git clone --recursive https://github.com/davidhalter/jedi-vim.git ~/.vim/bundle/jedi-vim
Example installation using Vundle:
Add the following line in your ~/.vimrc
Plugin 'davidhalter/jedi-vim'
For installing Jedi, pip install jedi
will also work, but you might run
into issues when working in virtual environments. Please use git submodules.
On Arch Linux, you can also install jedi-vim from official repositories as vim-jedi. It is also available on Debian (≥8) and Ubuntu (≥14.04) as vim-python-jedi. On Fedora Linux, it is available as vim-jedi.
Please note that this version might be quite old compared to using jedi-vim from Git.
Note that the python-mode VIM plugin seems to conflict with jedi-vim, therefore you should disable it before enabling jedi-vim.
To enjoy the full features of jedi-vim, you should have VIM >= 7.3, compiled with
+conceal
(which is not the case on some platforms, including OS X). If your VIM
does not meet these requirements, the parameter recommendation list may not appear
when you type an open bracket after a function name. Please read
the documentation
for details.
Jedi is by default automatically initialized. If you don't want that I suggest
you disable the auto-initialization in your .vimrc
:
let g:jedi#auto_initialization = 0
There are also some VIM options (like completeopt
and key defaults) which
are automatically initialized, but you can skip this:
let g:jedi#auto_vim_configuration = 0
You can make jedi-vim use tabs when going to a definition etc:
let g:jedi#use_tabs_not_buffers = 1
If you are a person who likes to use VIM-splits, you might want to put this in your .vimrc
:
let g:jedi#use_splits_not_buffers = "left"
This options could be "left", "right", "top", "bottom" or "winwidth". It will decide the direction where the split open.
Jedi automatically starts the completion, if you type a dot, e.g. str.
, if
you don't want this:
let g:jedi#popup_on_dot = 0
Jedi selects the first line of the completion menu: for a better typing-flow and usually saves one keypress.
let g:jedi#popup_select_first = 0
Jedi displays function call signatures in insert mode in real-time, highlighting the current argument. The call signatures can be displayed as a pop-up in the buffer (set to 1 by default (with the conceal feature), 2 otherwise), which has the advantage of being easier to refer to (but is a hack with many drawbacks since it changes the buffer's contents), or in Vim's command line aligned with the function call (set to 2), which can improve the integrity of Vim's undo history.
let g:jedi#show_call_signatures = "1"
Here are a few more defaults for actions, read the docs (:help jedi-vim
) to
get more information. If you set them to ""
, they are not assigned.
NOTE: subject to change!
let g:jedi#goto_command = "<leader>d"
let g:jedi#goto_assignments_command = "<leader>g"
let g:jedi#goto_stubs_command = "<leader>s"
let g:jedi#goto_definitions_command = ""
let g:jedi#documentation_command = "K"
let g:jedi#usages_command = "<leader>n"
let g:jedi#completions_command = "<C-Space>"
let g:jedi#rename_command = "<leader>r"
An example for setting up your project:
let g:jedi#environment_path = "/usr/bin/python3.9"
jedi-vim tries its best to guess your virtual env. If you want to work with a specific virtual environment however, you can point jedi-vim towards it:
let g:jedi#environment_path = "venv"
Finally, if you don't want completion, but all the other features, use:
let g:jedi#completions_enabled = 0
This depends on the completeopt
option. Jedi initializes it in its
ftplugin
. Add the following line to your .vimrc
to disable it:
autocmd FileType python setlocal completeopt-=preview
Don't even think about changing the Jedi command to <Tab>
,
use supertab!
Completion of complex libraries (like Numpy) should only be slow the first time you complete them. After that the results should be cached and very fast.
If it is still slow after the initial completion and you have installed the python-mode Vim plugin, try disabling its rope mode:
let g:pymode_rope = 0
See issue #163.
You can also use deoplete-jedi for completions, which uses Jedi, but does completions asynchronously (requires Neovim). It makes sense to use both jedi-vim and deoplete-jedi, but you should disable jedi-vim's completions then:
let g:jedi#completions_enabled = 0
jedi-vim is being tested with a combination of vspec and py.test.
The tests are in the test
subdirectory, you can run them calling:
py.test
The tests are automatically run with travis.