Skip to content

SUSE/BCI-dockerfile-generator

Repository files navigation

BCI build recipe generator

This repository contains the scripts to generate the build recipes (Dockerfiles or kiwi build descriptions) to build the BCI development project in the Open Build Service.

Find the latest fully rendered documentation here.

Prerequisites

You will need the following tools:

  • Python 3.10 or later
  • poetry
  • osc installed and configured to reach OBS by default.
  • python-dnf installed via your system's package manager (if you want to touch the .Net images)

To get started, clone this repository and run :command:`poetry install` in its root directory.

Overview

This repository contains two major components:

  1. templating logic to autogenerate the :file:`Dockerfile` and kiwi build descriptions for SLE BCI (see :py:mod:`bci_build`) including an updater for our .Net images (see :py:mod:`dotnet`).
  2. Github automation to render the templates into separate branches and synchronize them to the Open Build Service (see :py:mod:`staging`).

Contributing

To contribute a new container or modify an existing one, please check the chapter :ref:`adding-and-modifying-container-images`.

Before submitting your changes, please format your source code with ruff:

poetry run ruff format
# reorder imports:
poetry run ruff check --fix

Additionally, run the unit tests and check whether the documentation builds (additional points if you update it):

# tests
poetry run pytest -vv
# docs
poetry run sphinx-build -M html source build -W

Documentation - READMEs

We are using Jinja2 templates to create Markdown formatted :file:`README.md` files of each container.

To create a new :file:`README.md`, create a new file in the location :file:`src/bci_build/package/$name/README.md.j2` where $name is the container image name (:py:attr:`~bci_build.package.BaseContainerImage.name`). Add a markdown formatted README that explains the specifics of using this container images. Focus on anything container specific, like volumes, the behavior of the entrypoint and differences to the package in SLES/SLFO/Tumbleweed. If there is upstream/downstream documentation, refer to it if applicable.

There are additionally the following helper templates, to enhance the READMEs:

  • licensing_and_eula.j2: Adds a standard licencing & EULA footer. MUST be included at the bottom of every README.
  • badges.j2: Adds a Redistributable & optional supportlevel badge. MUST be included at the top of every README.
  • access_protected_images.j2: Explains how to access an image behind the paywall. It only needs to be added at the bottom of paywalled images.

You can include the helpers via the following directive:

{% include 'licensing_and_eula.j2' %}

The :file:`README.md.j2` template is rendered with the respective :py:class:`~bci_build.package.BaseContainerImage` being passed as the parameter image to the templating engine. This allows you to use all properties of the :py:class:`~bci_build.package.BaseContainerImage` in the README, e.g. the image title (:py:attr:`~bci_build.package.BaseContainerImage.title`) or the reference (:py:attr:`~bci_build.package.BaseContainerImage.pretty_reference`).

Entrypoints

The projects currently provides two entry points. The first is the package build description "dumper" called package. It writes the build description of a single container image into a specified directory:

poetry run package postgres-12-sp4 ~/tmp/postgres/

The first argument is the name of the container image, this is the concatenation of the image name (:py:attr:`~bci_build.package.BaseContainerImage.name`) and the operating system version (:py:attr:`~bci_build.package.BaseContainerImage.os_version`).

The second entry point is the github automation bot, which is not intended for end user usage. You can find some details in the chapter :ref:`staging-bot`.