Skip to main content

Identify unused dependencies and avoid a bloated virtual environment.

Project description

creosote

check test

Identify unused dependencies and avoid a bloated virtual environment.

⚡️ Quickstart

Install creosote in separate virtual environment (using e.g. pipx):

pipx install creosote

Scan virtual environment for unused dependencies (PEP-621 example below, but Poetry, Pipenv and requirements.txt files are also supported, see this table):

$ creosote
Found dependencies in pyproject.toml: distlib, dotty-dict, loguru, pip-requirements-parser, requests, toml
Oh no, bloated venv! 🤢 🪣
Unused dependencies found: requests

And after having removed/uninstalled requests:

$ creosote
Found dependencies in pyproject.toml: distlib, dotty-dict, loguru, pip-requirements-parser, toml
No unused dependencies found! ✨

Get help:

creosote --help

🤔 How this works

Some data is required as input:

Argument Default value Description
--venv .venv The path to your virtual environment.
--paths src The path to your source code, one or more files/folders.
--deps-file pyproject.toml The path to the file specifying your dependencies, like pyproject.toml, requirements_*.txt | .in.
--sections project.dependencies One or more toml sections to parse, e.g. project.dependencies.

The creosote tool will first scan the given python file(s) for all its imports. Then it fetches all dependency names (from the dependencies spec file). Finally, all imports are associated with their corresponding dependency name (requires the virtual environment for resolving). If a dependency does not have any imports associated, it is considered unused.

😤 Known limitations

  • importlib imports are not detected by the AST parser (a great first contribution for anyone inclined 😄, reach out or start here).

🥧 History and ambition

The idea of a project like this was hatched from having security vulnerability reports about production dependencies (shipped into production) which turned out to not not even be in use.

The goal of this project is to run the creosote tool in CI, which will catch cases where the developer forgets to remove unused dependencies. An example of such a case could be when doing refactorings.

Note: Creosote supports identifying both unused production dependencies and developer dependencies. It all depends on what you would like to achieve.

🤨 FAQ

Which dependency specification tooling/standards are supported?

Tool/standard Supported --deps-file value Example --sections values
PEP-621 :white_check_mark: pyproject.toml project.dependencies,
project.optional-dependencies.<GROUP>
Poetry :white_check_mark: pyproject.toml tool.poetry.dependencies,
tool.poetry.dev-dependencies (legacy),
tool.poetry.group.<GROUP>.dependencies
Pipenv :white_check_mark: pyproject.toml packages,
dev-packages
PEP-508 (requirements.txt, pip-tools) :white_check_mark: *.[txt|in] N/A
Legacy Setuptools (setup.py)

📔 Notes on PEP-508 (requirements.txt)

When using requirements.txt files to specify dependencies, there is no way to tell which part of requirements.txt specifies production vs developer dependencies. Therefore, you have to break your requirements.txt file into e.g. requirements-prod.txt and requirements-dev.txt and use any of them as input. When using pip-tools, you likely want to point Creosote to scan your *.in file(s).

Can I specify multiple toml sections?

Yes, you can specify a list of sections after the --sections argument. It all depends on what your setup looks like and what you set out to achieve.

Can I exclude dependencies from the scan?

Yes, you can use the --exclude-deps argument to specify one or more dependencies you do not wish to get warnings for.

This feature is intended for dependencies you must specify in your dependencies spec file, but which you don't import in your source code. An example of such a dependency are database drivers, which are commonly only defined in connection strings and will signal to the ORM which driver to use.

Can I run Creosote in a GitHub Action workflow?

Yes, please see the action job example in .github/workflows/test.yml.

Can I run Creosote with pre-commit?

Yes, see example in .pre-commit-config.yaml.

Here's another example setup, if already have Creosote installed onto $PATH (via e.g. pipx).
# .pre-commit-config.yaml

repos:
  - repo: local
    hooks:
      - id: system
        name: creosote
        entry: creosote --venv .venv --paths src --deps-file pyproject.toml --sections project.dependencies
        pass_filenames: false
        files: \.(py|toml|txt|in|lock)$
        language: system

What's with the name "creosote"?

This tool has borrowed its name from the Monty Python scene about Mr. Creosote.

👩‍🔬 Development/debugging info

Install in-development builds

You can run in-development versions of Creosote. Examples below:

# Creosote build from main branch
$ pipx install --suffix=@main --force git+https://github.com/fredrikaverpil/creosote.git@main
$ creosote@main --venv .venv ...
$ pipx uninstall creosote@main

# Creosote build from PR #123
$ pipx install --suffix=@123 --force git+https://github.com/fredrikaverpil/creosote.git@refs/pull/123/head
$ creosote@123 --venv .venv ...
$ pipx uninstall creosote@123

Releasing

  1. Bump version in src/creosote/__about__.py and .pre-commit-config.yaml.
  2. GitHub Action will run automatically on creating a release and deploy the release onto PyPi.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

creosote-2.4.0-py3-none-any.whl (12.4 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page