Code navigation, documentation lookup and completion for Python.
Anaconda mode provides the following features
- context-sensitive code completion
- jump to definitions
- find references
- view documentation
- virtual environment
- eldoc mode
- all this stuff inside vagrant and remote hosts
2.6, 2.7, 3.3, 3.4
To use this package you need to install pip
.
All you need to do is install the package from Melpa:
M-x package-install RET anaconda-mode RET
Clone this repository somewhere and add this directory to your load-path
.
anaconda-mode
is included in the Emacs Prelude distribution. You can use it as well. Look at the prelude-python
module to see more details.
anaconda-mode
is included in the Spacemacs distribution. You can use it as well. Look at the python
language layer to see more details.
You can automatically enable anaconda-mode
in all python buffers with following code in your configuration:
(add-hook 'python-mode-hook 'anaconda-mode)
anaconda-eldoc-mode
provide document function to eldoc-mode
. All you need is to enable anaconda-eldoc-mode
in addition to the previous setup.
(add-hook 'python-mode-hook 'anaconda-eldoc-mode)
To start a completion press C-M-i
. This is the standard emacs binding for complete-at-point
function. You can use company-mode with company-anaconda backend to get more intelligent ui. Or auto-complete-mode with ac-anaconda as a last try.
Here is a list of interactive commands available with anaconda-mode
Keybinding | Description |
---|---|
C-M-i | anaconda-mode-complete |
M-. | anaconda-mode-find-definitions |
M-, | anaconda-mode-find-assignments |
M-r | anaconda-mode-find-references |
M-* | anaconda-mode-go-back |
M-? | anaconda-mode-show-doc |
If multiple candidates are found for definitions, assignments or usages, you'll see an advanced anaconda navigator buffer.
You can add your project to the Emacs PYTHONPATH
. If you store project dependencies somewhere on your machine, you can add them as well.
(add-to-list 'python-shell-extra-pythonpaths "/path/to/the/project")
(add-to-list 'python-shell-extra-pythonpaths "/path/to/the/dependency")
Use a virtual environment to isolate your project dependencies from others on the system. You can additionally install your project in an editable mode in the virtual environment. This will improve search functionality. Then activate this virtual environment inside Emacs.
M-x pythonic-activate RET /path/to/virtualenv RET
Also you can use pyenv-mode or similar package to hold virtual environment in actual state.
Each action above applies to anaconda-mode
immediately. The next anaconda-mode
command you call will use this environment for search completion candidates.
It's possible to use anaconda-mode on a remote server when you connect to it using tramp. Anaconda-mode can search for completion candidates and all other stuff on remote server while you running Emacs locally. First of all open interesting remote file.
C-x C-f /ssh:remote_host:project/__init__.py RET
After tramp successfully connects and you see actual buffer content, activate the remote virtual environment.
M-x pythoninc-activate RET /ssh:remote_host:/home/user/venv RET
Now any anaconda-mode command will use /home/user/venv/bin/python
interpreter running on remote_host
over ssh. If you don't use the virtual environment remotely then you have an option to specify the remote interpreter directly.
(setq python-shell-interpreter "/ssh:remote_host:/usr/bin/python")
It is important to remember that remote_host
must be a real host name or an IP address. SSH aliases not allowed to be used with anaconda-mode. Also port 9000 on the remote host should be open to incoming connections from your local machine. A final note about project scope: all kinds of searching from inside the virtual environment are available from any buffer. However searching inside your project is available only if you open it on the same machine as the interpreter.
You can get all the intelligent features of anaconda-mode with virtual environments deployed on your vagrant box. Add a port forwarding line to your Vagrantfile.
config.vm.network "forwarded_port", guest: 9000, host: 9000
Fire up vagrant machine as usual and open your project inside vagrant box.
C-x C-f /ssh:vagrant@localhost#2222:/vagrant/polls/views.py
Then activate your project environment installed inside vagrant.
M-x pythonic-activate RET /ssh:vagrant@localhost#2222:/vagrant/polls/venv RET
Remember that the standard password for vagrant user is vagrant
. It is too annoying to type this password each time you want to connect. I use ssh-copy-id
to upload my public ssh key the box.
ssh-copy-id vagrant@localhost -p 2222
If you have random connection errors during interaction with running server - try to replace host name with IP address. For example localhost
with 127.0.0.1
.
Now you are ready to go.
Anaconda mode comes with anaconda_mode.py
server. This server allows you to use the jedi python library over jsonrpc api. The server chooses first available port starting from 9000. Anaconda mode will run this server automatically on first call of any anaconda-mode command.
This means that completion results and reference searches depend on your project installation.
Please attach *anaconda-mode*
buffer content to every created issue.
DistutilsOptionError: must supply either home or prefix/exec-prefix -- not both
This occurs due to distutils bug when pip -t
option conflict with distutils.cfg
prefix
option. If you install pip
with homebrew
you are on fire. There are few options to avoid this issue.
- install
anaconda-mode
dependencies manually - remove
prefix
option fromdistutils.cfg
These kinds of problems were reported with jedi 0.9 version. You can try to downgrade jedi version down to 0.8.
M-: (dired (anaconda-mode-server-directory)) RET
M-! rm -rf jedi* RET
M-! pip install "jedi<0.9" -t . RET
Are very welcome. But any significant change has to be accompanied with tests, both for Emacs Lisp and Python code. To run the test suite, call:
tox
- Dmitry Gutov @dgutov
- Bo Lin @sadboy
- Vasilij Schneidermann @wasamasa
- Fredrik Bergroth @fbergroth
- Fabio Corneti @fabiocorneti
- Tom Davis @tdavis
- Sviridov Alexander @sviridov
- Mario Rodas @marsam