ol-concourse-lib

A Python DSL for programmatically building Concourse CI/CD pipelines.

Instead of writing YAML files by hand, use typed Python classes to construct type-safe pipeline definitions that serialize to valid Concourse pipeline YAML.

Installation

pip install ol-concourse-lib
# or with Pulumi job builder support:
pip install "ol-concourse-lib[pulumi]"

Quick Start

from ol_concourse.lib.models.pipeline import Identifier
from ol_concourse.lib.resources import git_repo, registry_image
from ol_concourse.lib.resource_types import packer_build
from ol_concourse.lib.jobs.infrastructure import packer_jobs
import yaml, json

# Build a pipeline fragment for packer AMI builds
code = git_repo(
    name=Identifier("packer-templates"),
    uri="https://github.com/myorg/packer-templates",
    branch="main",
)

fragment = packer_jobs(
    dependencies=[],
    image_code=code,
    node_types=["web", "worker"],
)

pipeline = fragment.to_pipeline()
print(yaml.dump(json.loads(pipeline.model_dump_json())))

API

Models (ol_concourse.lib.models)

• Pipeline — root pipeline definition
• Job, Resource, ResourceType — core pipeline components
• GetStep, PutStep, TaskStep, InParallelStep — step types
• TaskConfig, Command, Input, Output — task configuration
• PipelineFragment — composable pipeline unit

Builder Functions

Regenerating Pipeline Models

models/pipeline.py is generated from the Concourse pipeline JSON schema:

pip install datamodel-code-generator
datamodel-codegen \
    --url https://raw.githubusercontent.com/nicholasdille/concourse-jsonschema/main/concourse.json \
    --output src/ol_concourse/lib/models/pipeline.py \
    --output-model-type pydantic_v2.BaseModel

License

BSD-3-Clause — Copyright MIT Open Learning