poetry-auto-export 0.2.1

Automatically create requirements.txt on every poetry lock

poetry-auto-export

Automatically export dependencies to requirements.txt on every poetry lock

This plugin enables you to keep a requirements.txt file always up to date with your poetry.lock file. This is helpful if you want to use pip to install your dependencies somewhere, e.g.

• in CI/CD environments
• in docker containers

Usage

Add plugin configuration to your pyproject.toml. The options follow arguments for poetry export.

[tool.poetry-auto-export]
output = "requirements.txt"
without_hashes = true
without = ["dev"]

Then, run poetry commands as usual:

poetry lock

The requirements.txt file will be updated automatically.

The supported commands are:

• lock
• update
• add
• remove

Creating multiple export files

If you need to create multiple requirements files, e.g. dev-requirements.txt and prod-requirements.txt, use the following syntax:

[[tool.poetry-auto-export.exports]]
output = "dev-requirements.txt"
without = ["prod"]

[[tool.poetry-auto-export.exports]]
output = "prod-requirements.txt"
without = ["dev"]

Installation

This is a poetry plugin, so it's meant to be installed inside the global poetry environment, not your project environment like regular pacakges. See poetry's docs.

There are three ways of doing so.

The pipx way

If you are using pipx already, that's easy:

pipx inject poetry poetry-auto-export

The pip way

Otherwise, you can use the pip that comes with poetry. The difficulty is finding that pip! It's not the same one you get with poetry run pip. Here is a best effort guess:

On Linux:

~/.local/share/pypoetry/venv/bin/pip install poetry-auto-export

On Windows:

%APPDATA%\pypoetry\venv\bin\pip install poetry-auto-export

The poetry way

For some reason, this is discouraged by poetry.

poetry self add poetry-auto-export

Roadmap and contributing

The primary goal of the project is to make it more convenient to work with poetry projects in CI/CD and docker. Contributions towards this goal are welcome!

Roadmap:

• more unit tests for the plugin
• integration tests for the plugin
• proper configuration parsing (note: typed dict and dataclasses can't support the with option, since it's a python keyword)
• schema or exhaustive documentation of the supported configuration options