Validate RunPod Hub files

runpod_doc_worker.testing.hub provides focused assertions that a worker repository can call from its own tests.

from pathlib import Path

from runpod_doc_worker.testing import hub

ROOT = Path(__file__).parents[1]


def test_runpod_metadata() -> None:
    hub.check(ROOT / ".runpod" / "hub.json")
    hub.check_test_inputs(
        ROOT / ".runpod" / "tests.json",
        roots=("/runpod-volume", "/workspace", "/tmp", "/worker"),
    )

Check hub.json

hub.check() verifies that the file exists, parses to a JSON object, includes environment entries, and gives each environment entry a name and non-empty description. Top-level and environment descriptions are limited to 189 characters.

Use hub.problems() when the caller needs the complete list instead of an assertion:

from runpod_doc_worker.testing.hub import load, problems

issues = problems(load(".runpod/hub.json"))

Check volume test fixtures

check_test_inputs() requires a top-level tests key. For each test with input.volume_path, it normalizes the container path and checks that it remains below an allowed POSIX root.

Pass explicit roots when CI has not called configure() or when its configuration differs from the release image. The default comes from config.active().volume_roots; it does not read the runtime <PREFIX>_VOLUME_ROOTS override.

Scope of the validator

These helpers are not a complete RunPod Hub schema validator. They do not check every job field, verify URLs, confirm local file existence, or prove that a listing can be published. Tests without volume_path are skipped.

Malformed JSON can raise the underlying JSON parsing error. Treat that as a failed repository test.

See the repository contributing guide for the full local test command.