12913: Apply secondaryFiles to inputs to RunInSingleContainer

Fixes bug where mounts mismatch cwl.input.yml due to discovering
secondaryFiles during workflow execution.

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <pamstutz@veritasgenetics.com>

diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py
index 316af0e52847d2caad516ef1b6d126577b948e92..d426c1ad6f32fef90e0a5b15a3691fe811c78760 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvworkflow.py
+++ b/sdk/cwl/arvados_cwl/arvworkflow.py
@@ -17,7 +17,8 @@ from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class
 
 import ruamel.yaml as yaml
 
-from .runner import upload_dependencies, packed_workflow, upload_workflow_collection, trim_anonymous_location, remove_redundant_fields
+from .runner import (upload_dependencies, packed_workflow, upload_workflow_collection,
+                     trim_anonymous_location, remove_redundant_fields, discover_secondary_files)
 from .pathmapper import ArvPathMapper, trim_listing
 from .arvtool import ArvadosCommandTool
 from .perf import Perf
@@ -88,6 +89,8 @@ class ArvadosWorkflow(Workflow):
                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
 
+            discover_secondary_files(self.tool["inputs"], joborder)
+
             with Perf(metrics, "subworkflow upload_deps"):
                 upload_dependencies(self.arvrunner,
                                     os.path.basename(joborder.get("id", "#")),

diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py
index c55e976924f65bd362153a0921152e5a170ca47b..28de7f368a23ccbc52a0cae37fa55d358744d0e1 100644 (file)
--- a/sdk/cwl/arvados_cwl/runner.py
+++ b/sdk/cwl/arvados_cwl/runner.py
@@ -191,12 +191,8 @@ def tag_git_version(packed):
             packed["http://schema.org/version"] = githash
 
 
-def upload_job_order(arvrunner, name, tool, job_order):
-    """Upload local files referenced in the input object and return updated input
-    object with 'location' updated to the proper keep references.
-    """
-
-    for t in tool.tool["inputs"]:
+def discover_secondary_files(inputs, job_order):
+    for t in inputs:
         def setSecondary(fileobj):
             if isinstance(fileobj, dict) and fileobj.get("class") == "File":
                 if "secondaryFiles" not in fileobj:
@@ -209,6 +205,13 @@ def upload_job_order(arvrunner, name, tool, job_order):
         if shortname(t["id"]) in job_order and t.get("secondaryFiles"):
             setSecondary(job_order[shortname(t["id"])])
 
+def upload_job_order(arvrunner, name, tool, job_order):
+    """Upload local files referenced in the input object and return updated input
+    object with 'location' updated to the proper keep references.
+    """
+
+    discover_secondary_files(tool.tool["inputs"], job_order)
+
     jobmapper = upload_dependencies(arvrunner,
                                     name,
                                     tool.doc_loader,