Jedi - an awesome autocompletion, static analysis and refactoring library for Python
Jedi is a static analysis tool for Python that is typically used in
IDEs/editors plugins. Jedi has a focus on autocompletion and goto
functionality. Other features include refactoring, code search and
finding references.
Jedi has a simple API to work with. There is a reference implementation
as a VIM-Plugin.
Autocompletion in your REPL is also possible, IPython uses it natively
and for the CPython REPL you can install it. Jedi is well tested and
bugs should be rare.
Jedi can currently be used with the following editors/projects:
- Vim (jedi-vim,
YouCompleteMe,
deoplete-jedi,
completor.vim) - Visual Studio Code (via Python
Extension) - Emacs (Jedi.el,
company-mode,
elpy,
anaconda-mode,
ycmd) - Sublime Text (SublimeJEDI
[ST2 + ST3], anaconda
[only ST3]) - TextMate
(Not sure if it's actually working) - Kate version 4.13+ supports it natively,
you have to enable it, though.
[see] - Atom
(autocomplete-python-jedi) - GNOME Builder (with support
for GObject Introspection) - Gedit (gedi)
- wdb - Web Debugger
- Eric IDE
- IPython
6.0.0+ - xonsh shell has jedi
extension
and many more!
There are a few language servers that use Jedi:
- jedi-language-server
- python-language-server
(currently unmaintained) - python-lsp-server
(fork from python-language-server) - anakin-language-server
Here are some pictures taken from
jedi-vim:
Completion for almost anything:
Documentation:
Get the latest version from
github (master branch should
always be kind of stable/working).
Docs are available at https://jedi.readthedocs.org/en/latest/. Pull
requests with enhancements and/or fixes are awesome and most welcome.
Jedi uses semantic versioning.
If you want to stay up-to-date with releases, please subscribe
to this mailing list: https://groups.google.com/g/jedi-announce. To
subscribe you can simply send an empty email to
jedi-announce+subscribe@googlegroups.com
.
Issues & Questions
You can file issues and questions in the [issue tracker
<https://github.com/davidhalter/jedi/>]{.title-ref}. Alternatively you
can also ask on Stack
Overflow with
the label python-jedi
.
Installation
Features and Limitations
Jedi's features are listed here:
Features.
You can run Jedi on Python 3.6+ but it should also understand code that
is older than those versions. Additionally you should be able to use
Virtualenvs
very well.
Tips on how to use Jedi efficiently can be found
here.
API
You can find a comprehensive documentation for the API
here.
Autocompletion / Goto / Documentation
There are the following commands:
jedi.Script.goto
jedi.Script.infer
jedi.Script.help
jedi.Script.complete
jedi.Script.get_references
jedi.Script.get_signatures
jedi.Script.get_context
The returned objects are very powerful and are really all you might
need.
Autocompletion in your REPL (IPython, etc.)
Jedi is a dependency of IPython. Autocompletion in IPython with Jedi is
therefore possible without additional configuration.
Here is an example video how REPL
completion can look like. For the python
shell you can enable tab
completion in a
REPL.
Static Analysis
For a lot of forms of static analysis, you can try to use
jedi.Script(...).get_names
. It will return a list of names that you
can then filter and work with. There is also a way to list the syntax
errors in a file: jedi.Script.get_syntax_errors
.
Refactoring
Jedi supports the following refactorings:
jedi.Script.inline
jedi.Script.rename
jedi.Script.extract_function
jedi.Script.extract_variable
Code Search
There is support for module search with jedi.Script.search
, and
project search for jedi.Project.search
. The way to search is either by
providing a name like foo
or by using dotted syntax like foo.bar
.
Additionally you can provide the API type like class foo.bar.Bar
.
There are also the functions jedi.Script.complete_search
and
jedi.Project.complete_search
.
Development
There's a pretty good and extensive development
documentation.
Testing
The test suite uses pytest
:
pip install pytest
If you want to test only a specific Python version (e.g. Python 3.8), it
is as easy as:
python3.8 -m pytest
For more detailed information visit the testing
documentation.
Acknowledgements
Thanks a lot to all the
contributors!