Step name callback (#2109)

Author: tetron

build-cwltool-docker.sh

@@ -8,4 +8,4 @@ ${engine} run -t -v /var/run/docker.sock:/var/run/docker.sock \
 	-v /tmp:/tmp \
 	-v "$PWD":/tmp/cwltool \
 	quay.io/commonwl/cwltool_module /bin/sh -c \
-	"apk add gcc bash git && pip install -r/tmp/cwltool/test-requirements.txt ; pytest -k 'not (test_bioconda or test_double_overwrite or test_env_filtering or test_biocontainers or test_disable_file_overwrite_without_ext or test_disable_file_creation_in_outdir_with_ext or test_write_write_conflict or test_directory_literal_with_real_inputs_inside or test_revsort_workflow or test_stdin_with_id_preset or test_no_compute_chcksum or test_packed_workflow_execution[tests/wf/count-lines1-wf.cwl-tests/wf/wc-job.json-False] or test_sequential_workflow or test_single_process_subwf_subwf_inline_step or test_cache_dockerreq_hint_instead_of_req)' --ignore-glob '*test_udocker.py' -n 0 -v -rs --pyargs cwltool"
+	"apk add gcc bash git && pip install -r/tmp/cwltool/test-requirements.txt ; pytest -k 'not (test_bioconda or test_double_overwrite or test_env_filtering or test_biocontainers or test_disable_file_overwrite_without_ext or test_disable_file_creation_in_outdir_with_ext or test_write_write_conflict or test_directory_literal_with_real_inputs_inside or test_revsort_workflow or test_stdin_with_id_preset or test_no_compute_chcksum or test_packed_workflow_execution[tests/wf/count-lines1-wf.cwl-tests/wf/wc-job.json-False] or test_sequential_workflow or test_single_process_subwf_subwf_inline_step or test_cache_dockerreq_hint_instead_of_req or test_workflow_job_step_name_callback)' --ignore-glob '*test_udocker.py' -n 0 -v -rs --pyargs cwltool"

cwltool/context.py

@@ -29,6 +29,7 @@
     from .process import Process
     from .secrets import SecretStore
     from .software_requirements import DependenciesConfiguration
+    from .workflow_job import WorkflowJobStep
 
 
 class ContextBase:
@@ -191,6 +192,10 @@ def __init__(self, kwargs: Optional[dict[str, Any]] = None) -> None:
         self.default_stderr: Optional[Union[IO[bytes], TextIO]] = None
         self.validate_only: bool = False
         self.validate_stdout: Optional["SupportsWrite[str]"] = None
+        self.workflow_job_step_name_callback: Optional[
+            Callable[[WorkflowJobStep, CWLObjectType], str]
+        ] = None
+
         super().__init__(kwargs)
         if self.tmp_outdir_prefix == "":
             self.tmp_outdir_prefix = self.tmpdir_prefix

cwltool/workflow_job.py

@@ -60,7 +60,12 @@ def job(
     ) -> JobsGeneratorType:
         runtimeContext = runtimeContext.copy()
         runtimeContext.part_of = self.name
-        runtimeContext.name = shortname(self.id)
+
+        if runtimeContext.workflow_job_step_name_callback is not None:
+            vfinputs = {shortname(k): v for k, v in joborder.items()}
+            runtimeContext.name = runtimeContext.workflow_job_step_name_callback(self, vfinputs)
+        else:
+            runtimeContext.name = shortname(self.id)
 
         _logger.info("[%s] start", self.name)
 

tests/test_context.py

@@ -1,10 +1,16 @@
 import subprocess
 import sys
+import logging
+from io import StringIO
+from typing import MutableMapping, cast
 
 from cwltool.context import RuntimeContext
 from cwltool.factory import Factory
+from cwltool.utils import CWLObjectType
+from cwltool.workflow_job import WorkflowJobStep
+from pathlib import Path
 
-from .util import get_data
+from .util import get_data, needs_docker
 
 
 def test_replace_default_stdout_stderr() -> None:
@@ -26,3 +32,59 @@ def test_replace_default_stdout_stderr() -> None:
     assert echo(inp="foo") == {"out": "foo\n"}
     sys.stdout = original_stdout
     sys.stderr = original_stderr
+
+
+@needs_docker
+def test_workflow_job_step_name_callback() -> None:
+    """Test ability to hook custom workflow step naming"""
+
+    stream = StringIO()
+    streamhandler = logging.StreamHandler(stream)
+    _logger = logging.getLogger("cwltool")
+    _logger.addHandler(streamhandler)
+
+    try:
+        runtime_context = RuntimeContext()
+
+        def step_name_hook(step: WorkflowJobStep, job: CWLObjectType) -> str:
+            j1 = cast(MutableMapping[str, CWLObjectType], job)
+            inp = cast(MutableMapping[str, str], j1.get("revtool_input", j1.get("sorted_input")))
+            return "%s on %s" % (
+                step.name,
+                inp.get("basename"),
+            )
+
+        runtime_context.workflow_job_step_name_callback = step_name_hook
+
+        factory = Factory(None, None, runtime_context)
+        revsort = factory.make(get_data("tests/wf/revsort.cwl"))
+
+        result = revsort(
+            workflow_input={
+                "class": "File",
+                "location": Path(get_data("tests/wf/whale.txt")).as_uri(),
+                "format": "https://www.iana.org/assignments/media-types/text/plain",
+            }
+        )
+
+        result = cast(CWLObjectType, result)
+
+        sorted_out = cast(MutableMapping[str, str], result["sorted_output"])
+        loc = sorted_out["location"]
+
+        assert result == {
+            "sorted_output": {
+                "basename": "output.txt",
+                "checksum": "sha1$b9214658cc453331b62c2282b772a5c063dbd284",
+                "class": "File",
+                "http://commonwl.org/cwltool#generation": 0,
+                "nameext": ".txt",
+                "nameroot": "output",
+                "size": 1111,
+                "location": loc,
+            },
+        }
+
+        assert "rev on whale.txt" in stream.getvalue()
+    finally:
+        _logger.removeHandler(streamhandler)