autocomplete-python-pulsar

Python IDE features for Pulsar, powered by Jedi: completions, go-to-definition, find usages, project-wide rename, method override and argument snippets.

A rewrite of the original Atom autocomplete-python for Pulsar. The Kite integration and its telemetry are gone, the CoffeeScript is now TypeScript, and the Jedi backend targets the modern API. See CHANGELOG.md for what changed and docs/MIGRATION.md if you are coming from 1.x.

Published as autocomplete-python-pulsar. The original autocomplete-python name still points at the unmaintained Atom package.

• Pulsar 1.100 or newer
• Python 3.10 or newer
• Jedi 0.19 or newer, installed for the interpreter you want completions from:

python3 -m pip install --upgrade "jedi>=0.19"

The package tells you which interpreter it picked, and says so plainly if Jedi is missing from it.

ppm install autocomplete-python-pulsar

• Completions for modules, classes, functions, methods and variables, with docstrings and signatures.
• Function argument completion: type ( and get a snippet of the parameters, tab-navigable. Off by default; see Autocomplete Function Parameters.
• Go to definition: Ctrl+Alt+G (Alt+Cmd+G on macOS), or Ctrl/Cmd-click with the hyperclick package installed.
• Show usages of the symbol under the cursor, across the project.
• Rename a symbol across every file in the project.
• Override method: pick an inherited method and get a correct super()-calling stub.
• Tooltips showing the docstring of the symbol under the cursor. Off by default.
• Semantic highlighting: recolor identifiers by what Jedi knows them to be - function, class, parameter, builtin, constant, module - layered over the grammar. Off by default; see Semantic Highlighting.
• Run button in the status bar: run the current file with the interpreter the package discovered, with output in a dock pane. F5, or click the button again to stop. The pane takes stdin, so scripts that call input() work.
• Interpreter picker in the status bar, listing every environment found.

Most of the time there is nothing to configure. The package looks for interpreters the way the VS Code Python extension does, in this order:

1. The interpreter you picked with Autocomplete Python Pulsar: Select Interpreter
2. Python Executable Paths from the settings
3. VIRTUAL_ENV / CONDA_PREFIX: the environment Pulsar was launched from
4. Virtual environments inside your project (.venv, venv, env, or any directory with a pyvenv.cfg)
5. Poetry, Pipenv, pyenv (honouring .python-version), Conda, virtualenvwrapper
6. PATH
7. Well-known system locations

Whichever it picked is shown in the status bar. Click it to choose another, or run Autocomplete Python Pulsar: Show Environment to see the full list and the Jedi version in use.

Full details in docs/CONFIGURATION.md.

Start with Autocomplete Python Pulsar: Show Environment: it reports the interpreter and the Jedi version actually in use, which answers most reports. docs/TROUBLESHOOTING.md covers the rest.

CONTRIBUTING.md has the setup, and docs/ARCHITECTURE.md explains how the pieces fit together.

npm install          # TypeScript toolchain
npm run build        # src/*.ts -> dist/
npm test             # typecheck + TypeScript tests + Python tests

If this saves you time, you can buy me a coffee.

Originally written by Dmitry Sadovnychyi and contributors, as autocomplete-python for Atom. Completions come from Jedi by David Halter.

GPL-3.0-or-later. See LICENSE.