Skip the Nextflow resource label check for workflow components

packages/python/openproblems/CHANGELOG.md

@@ -22,6 +22,8 @@
 
 * Improve diagnostic print messages in `check_config` and `run_and_check_output` to be more descriptive.
 
+* `check_config`: Skip the Nextflow resource label check for components whose script is itself a Nextflow workflow. Viash renders those as a workflow rather than a process, so the labels would have no effect.
+
 # openproblems core Python v0.1.1
 
 ## NEW FUNCTIONALITY

packages/python/openproblems/src/openproblems/project/component_tests/check_config.py

@@ -167,27 +167,37 @@ def check_config(config: dict) -> None:
 
     runners = config.get("runners", [])
 
-    print("Check that a Nextflow runner with time, mem, and cpu labels is defined", flush=True)
+    print("Check that a Nextflow runner is defined", flush=True)
     nextflow_runner = next(
         (runner for runner in runners if runner["type"] == "nextflow"),
         None,
     )
 
     assert nextflow_runner, ".runners does not contain a nextflow runner"
-    assert nextflow_runner.get(
-        "directives"
-    ), "directives not a field in nextflow runner"
-    nextflow_labels = nextflow_runner["directives"].get("label")
-    assert nextflow_labels, "label not a field in nextflow runner directives"
-
-    assert [
-        label for label in nextflow_labels if label in TIME_LABELS
-    ], "time label not filled in"
-    assert [
-        label for label in nextflow_labels if label in MEM_LABELS
-    ], "mem label not filled in"
-    assert [
-        label for label in nextflow_labels if label in CPU_LABELS
-    ], "cpu label not filled in"
+
+    # a component whose script is itself a Nextflow workflow is rendered as a
+    # workflow instead of a process, so resource directives have no effect on it
+    is_nextflow_workflow = any(
+        resource.get("type") == "nextflow_script"
+        for resource in config.get("resources") or []
+    )
+
+    if not is_nextflow_workflow:
+        print("Check that the Nextflow runner has time, mem, and cpu labels", flush=True)
+        assert nextflow_runner.get(
+            "directives"
+        ), "directives not a field in nextflow runner"
+        nextflow_labels = nextflow_runner["directives"].get("label")
+        assert nextflow_labels, "label not a field in nextflow runner directives"
+
+        assert [
+            label for label in nextflow_labels if label in TIME_LABELS
+        ], "time label not filled in"
+        assert [
+            label for label in nextflow_labels if label in MEM_LABELS
+        ], "mem label not filled in"
+        assert [
+            label for label in nextflow_labels if label in CPU_LABELS
+        ], "cpu label not filled in"
 
     print("All checks succeeded!", flush=True)

packages/python/openproblems/tests/test_project_check_config.py

@@ -0,0 +1,67 @@
+from typing import Any, Dict
+
+import pytest
+
+from openproblems.project import check_config
+
+
+def _config(**kwargs: Any) -> Dict[str, Any]:
+    # a control method without links or references, so that check_config
+    # does not need network access to validate the metadata
+    config: Dict[str, Any] = {
+        "name": "foo",
+        "label": "Foo",
+        "summary": "A foo control method.",
+        "description": "A foo control method.",
+        "namespace": "control_methods",
+        "info": {"type": "control_method"},
+        "runners": [
+            {
+                "type": "nextflow",
+                "directives": {"label": ["lowtime", "lowmem", "lowcpu"]},
+            }
+        ],
+    }
+    config.update(kwargs)
+    return config
+
+
+def test_check_config_accepts_resource_labels():
+    check_config(_config())
+
+
+def test_check_config_requires_a_nextflow_runner():
+    with pytest.raises(AssertionError, match="does not contain a nextflow runner"):
+        check_config(_config(runners=[{"type": "executable"}]))
+
+
+def test_check_config_requires_resource_labels():
+    with pytest.raises(AssertionError, match="directives not a field"):
+        check_config(_config(runners=[{"type": "nextflow"}]))
+
+    runners = [{"type": "nextflow", "directives": {"tag": "$id"}}]
+    with pytest.raises(AssertionError, match="label not a field"):
+        check_config(_config(runners=runners))
+
+
+def test_check_config_requires_a_label_of_each_kind():
+    for labels, message in [
+        (["lowmem", "lowcpu"], "time label not filled in"),
+        (["lowtime", "lowcpu"], "mem label not filled in"),
+        (["lowtime", "lowmem"], "cpu label not filled in"),
+    ]:
+        runners = [{"type": "nextflow", "directives": {"label": labels}}]
+        with pytest.raises(AssertionError, match=message):
+            check_config(_config(runners=runners))
+
+
+def test_check_config_skips_resource_labels_for_nextflow_workflows():
+    # viash renders a nextflow_script component as a workflow rather than a
+    # process, so it has nothing to attach resource labels to
+    resources = [{"type": "nextflow_script", "path": "main.nf", "entrypoint": "run_wf"}]
+    check_config(
+        _config(
+            resources=resources,
+            runners=[{"type": "nextflow"}],
+        )
+    )