Custom pipeline loading logicΒΆ


For a complete code example click here

When writing a pipeline via a pipeline.yaml file, the most common way to interact is via the command-line interface (e.g., calling ploomber build). However, we may want to add more logic or embed it as part of a script in some cases.

Ploomber represents pipelines using DAG objects. You can convert your pipeline.yaml to a DAG object in Python with the following code:

from ploomber.spec import DAGSpec

def load():
    spec = DAGSpec('pipeline.yaml')
    dag = spec.to_dag()
    return dag

The spec variable is an object of type ploomber.spec.DAGSpec, while dag is of type ploomber.DAG, to learn more about their interfaces, click on any of the links.

Assume the code above exists in a file named; you may load the pipeline from the CLI with:

ploomber build --entry-point pipeline.load

Or the shortcut:

ploomber build -e pipeline.load

The load function may contain extra logic; for example, you may skip tasks based on some custom rules or compute dynamic parameters. For examples with custom logic, click here.