Script local workflow execution

Build the processing images first and run this Python example from the checkout root. It invokes the public cwltool CLI, uses the checked-out typed scatter job, and reads the workflow's JSON output. Adjust the job's acquisition references and AOI before running it.

import json
import subprocess
from pathlib import Path

workspace = Path.cwd()
output = workspace / "runs" / "scripted"
output.mkdir(parents=True, exist_ok=True)
result = subprocess.run(
    [
        "cwltool", "--podman", "--parallel", "--outdir", str(output),
        str(workspace / "cwl-workflow/app-water-bodies-cloud-native.cwl") + "#water-bodies",
        str(workspace / "cwl-workflow/typed-scatter-inputs.yaml"),
    ],
    check=True, capture_output=True, text=True,
)
outputs = json.loads(result.stdout)
print(outputs["stac_catalog"]["location"])

check=True raises an exception on runner failure. Logs are captured in result.stderr on success, or the exception's stderr on failure. The output Directory contains the catalog and referenced masks.

This example uses the CLI from Python so it can share the same inputs as the scatter lab. The historical script scripts/cwltool-scripted-execution.py targets release 1.4.1 and its older input format; it is separate from this checkout-based example.