Sign inSign up

fabianemmi/python-poetry

By fabianemmi

•Updated about 1 year ago

Python with poetry for running python packages managed by poetry

Image
1

10K+

fabianemmi/python-poetry repository overview

⁠Python with poetry inside Docker

Docker images with the python dependency management tool poetry⁠.

⁠Usage

Inside the containers, poetry can be used as usual per CLI.

⁠Image size

To reduce image size, it is recommended to use a multi-layer build to avoid having poetry inside the final container. The below example shows how to install a package into a virtual environment in a building layer and then copy it to the executing layer to reach the minimal image size.

# Allow passing of Python and poetry version to the build
ARG PYTHON_VERSION="3.11"
ARG POETRY_VERSION="1.2.2"

# Building layer
FROM fabianemmi/python-poetry:${PYTHON_VERSION}-${POETRY_VERSION}-slim as builder

# Copy package: ideally only `.py` files, `poetry.lock`, and `pyproject.toml`
ADD <package> /opt/<package>
COPY poetry.lock pyproject.toml /opt/
WORKDIR /opt

# Install with poetry
RUN python -m venv /venv && \
  . /venv/bin/activate && \
  pip install --upgrade pip && \
  poetry build && \
  pip install /opt/dist/*.whl

# Delete Python cache files
RUN find . | grep -E "(__pycache__|\.pyc|\.pyo$)" | xargs rm -rf && \
  cd /venv && \
  find . | grep -E "(__pycache__|\.pyc|\.pyo$)" | xargs rm -r

# Execution layer
FROM python:${PYTHON_VERSION}-alpine

# Copy the required files: vritual environment
COPY --from=builder /venv /venv

# Add venv to PATH variable to make installed packages available
ENV PATH=/venv/bin:$PATH

Notes:

  • If you install a package that offers a CLI, you do not have to copy /app/<package> from the builder layer. This is only necessary when executing a source code file of the project.
  • The entrypoint may not need the . /venv/bin/activate but can be made redundant by using 
    ENV PATH=/venv/bin:$PATH
    in the last layer.

⁠Virtual environments

Creation of virtual environments inside Docker containers is not required and, hence, turned off (POETRY_VIRTUALENVS_CREATE=false).

⁠CI caching

To correctly cache dependencies in a CI job, you need to enable creation of virtual environments in a pipeline. Here is an example for GitLab Pipelines:

variables:
  POETRY_CACHE_DIR: ${CI_PROJECT_DIR}/.venv
  PIP_CACHE_DIR: ${CI_PROJECT_DIR}/.cache/pip

.poetry-cache: &poetry-cache
  cache:
    paths:
      - $POETRY_CACHE_DIR
      - $PIP_CACHE_DIR
    key:
      files:
        - poetry.lock
        - .gitlab-ci.yml
      prefix: ${CI_COMMIT_REF_SLUG}-${CI_JOB_NAME}

.python-poetry: &python-poetry
  <<: *poetry-cache
  variables:
    POETRY_CACHE_DIR: ${POETRY_CACHE_DIR}
  image: fabianemmi/python-poetry:latest

...

example job:
  <<: *python-poetry
  before_script:
    - export POETRY_VIRTUALENVS_CREATE=true
    - poetry config virtualenvs.in-project true
    - poetry install ...
...

⁠Contribution

This repository is managed from GitLab⁠. Feel free to open merge requests.

Tag summary

Content type

Image

Digest

sha256:4dcc81191…

Size

399.3 MB

Last updated

about 1 year ago

docker pull fabianemmi/python-poetry:3.12-2.1.4