/bash2gitlab

Compile bash to yaml pipelines to get IDE support for bash and import bash from template repos

Primary LanguagePythonMIT LicenseMIT

bash2gitlab

Compile pure Bash scripts into your .gitlab-ci.yml. Get IDE support for your scripts while keeping them centralized.

Tired of writing Bash inside YAML strings with no syntax highlighting, linting, or testing? bash2gitlab lets you develop your CI logic in .sh files and then compiles them into your GitLab CI configuration, giving you the best of both worlds.

Bash in YAML is Bash without quality gates. Also, includes support for inlining a large number of scripts from other language, from Python to PHP.

tests pre-commit.ci status Downloads Python Version Release


Before

Surely Gitlab has a solution for this? Not as far as I can tell. Here are some of my best workarounds.

Your IDE sees a single YAML string, and your scripts are trapped in one file.

.gitlab-ci.yml:

build-job:
  script:
    - echo "Building project"
    - make build

After

Your IDE provides full support for Bash, and your scripts can be versioned, shared, and tested independently.

scripts/build.sh:

#!/usr/bin/env bash
set -eo pipefail

echo "Building project"
make build

uncompiled.gitlab-ci.yml:

build-job:
  script:
    - ./scripts/build.sh

Run bash2gitlab compile, and the final, valid .gitlab-ci.yml is generated for you.


Who is this for?

This tool is for you if:

  • You manage CI/CD templates in a centralized repository and include: them in many other projects.
  • Your .gitlab-ci.yml files contain thousands of lines of shell scripts.
  • You want to write, test, and debug your CI scripts locally without involving Docker or a full pipeline simulation.
  • Your IDE's lack of Bash support in YAML files is slowing you down.
  • You want to be able to put Python or other non-Bash scripts into your shared templates.

If your CI/CD configuration is simple or contained entirely within a single repository, you might not need this tool.

Installation

bash2gitlab is a standalone command-line tool. Installation with pipx is recommended to avoid dependency conflicts.

Install [all] extras for all commands. On your build server install just bash2gitlab for the core libraries which allows you to run compile, decompile on server. This minimizes supply chain risks.

# Recommended
pipx install bash2gitlab[all]

# Or via pip
pip install bash2gitlab[all]

Getting Started: A Quick Tour

  1. Initialize Your Project Run bash2gitlab init to create a configuration file (.bash2gitlab.toml) and directories to organize your source files.

  2. Decompile an Existing Configuration If you have an existing .gitlab-ci.yml with inline scripts, you can extract them automatically:

bash2gitlab decompile --in .gitlab-ci.yml --out my-project/
  1. Write and Edit Your Scripts Create or edit your .sh files in the scripts directory. Write standard, clean Bash—your IDE will thank you. In your source YAML (uncompiled.yml):
my-job:
  script:
    - ./scripts/my-script.sh
  1. Compile Compile your source YAML and scripts into a final, GitLab-ready configuration:
bash2gitlab compile --in my-project/ --out compiled/

This generates a compiled/.gitlab-ci.yml file, ready to be deployed to your project's root.

Usage and Commands

bash2gitlab provides a few core commands to manage your workflow.

Run with

  • bash2gitlab for CLI
  • bash2gitlab-interactive for CLI question and answer
  • bash2gitlab-tui for Terminal UI
  • bash2gitlab-gui for GUI

Core Compile/Decompile

Command Description
compile Compiles source YAML and .sh files into a final .gitlab-ci.yml.
decompile Extracts inline scripts from a .gitlab-ci.yml into separate .sh files.

Debugging from remote repo

Command Description
copy2local Copies compiled files from a central repo to a local project for testing.
map-deploy Copies compiled files from a central repo to a many local project for testing.
commit-map Copies intential changes in local projects back to the central repo.

Setup

Command Description
init Initializes a new bash2gitlab project and config file.
clean Carefully delete output in target folder.
install-precommit Add git hook to compile before commit
uninstall-precommit Remove precommit hook

Diagnostics

Command Description
lint Call gitlab APIs to lint your yaml
detect-drift Report what unexpected changes were made to the generated files.
show-config Display config after cascade
doctor Look for environment problems
graph Generate graph of inline relationships
detect-uncompiled Detect if you forgot to compile
validate Validate json schema of all yaml in input and output

Simulate Gitlab Pipeline Locally

Command Description
run Best efforts to run bash in a .gitlab-ci.yml file in similar order as a real runner

For detailed options on any command, run bash2gitlab <command> --help.


Advanced Topics

Bash Completion

Enable tab completion in your shell by running the global activation command once:

activate-global-python-argcomplete

Global Variables

To define variables that should be inlined into the global variables: block of your .gitlab-ci.yml, create a global_variables.sh file in your input directory.

Limitations

  • No include: Inlining: This tool only inlines .sh file references. It does not process or merge GitLab's include: statements for other YAML files.
  • Single Statement Invocations: The script invocation must be on its own line. Multi-statement lines like echo "hello" && script.sh are not supported.
  • Comments: Comments in the source YAML may not be preserved in the final compiled output.

How It Compares

  • gitlab-ci-local: This is an excellent tool for running your entire GitLab pipeline in local Docker containers. bash2gitlab is different—it focuses on the "unit testing" of your Bash logic itself, assuming you can and want to execute your scripts on your local machine without the overhead of Docker.
  • GitHub Actions GitHub composite actions do not have this problem. A shared GitHub action can reference a script in the shared action's repo. A GitHub "reusable" workflow is a single yaml file and might suffer from the same problem as Gitlab pipelines.
  • Git Submodules Build runners will need permissions to clone and git is more complicated to use.
  • Base image holds all bash You can only have one base image, so if you are using it for bash and yaml, you can't use other base images.
  • Trigger remote pipeline A remote pipeline has access to the shell files in its own repo.

Project Health & Info

Metric Health Metric Info
Tests Tests License License
Coverage Codecov PyPI PyPI
Lint / Pre-commit pre-commit.ci status Python Versions Python Version
Quality Gate Quality Gate Status Docs Docs
CI Build Build Downloads Downloads
Maintainability Maintainability Rating Last Commit Last Commit
Category Health
Open Issues GitHub issues
Stars GitHub Repo stars