Merge branch '12479-wb-structured-vocabulary'

[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py

diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py
index 22f662e00a1a02579543d1ead9542bf3e74997f7..79bfacd28a10daad1393bf090f2350e7c166244c 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvworkflow.py
+++ b/sdk/cwl/arvados_cwl/arvworkflow.py
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import os
 import json
 import copy
@@ -9,12 +13,13 @@ from cwltool.pack import pack
 from cwltool.load_tool import fetch_document
 from cwltool.process import shortname
 from cwltool.workflow import Workflow, WorkflowException
-from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
+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
-from .pathmapper import trim_listing
+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
 
@@ -79,11 +84,13 @@ class ArvadosWorkflow(Workflow):
         kwargs["work_api"] = self.work_api
         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
         if req:
-            with SourceLine(self.tool, None, WorkflowException):
+            with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
                 if "id" not in self.tool:
                     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", "#")),
@@ -106,15 +113,25 @@ class ArvadosWorkflow(Workflow):
                                         False)
 
             with Perf(metrics, "subworkflow adjust"):
+                joborder_resolved = copy.deepcopy(joborder)
                 joborder_keepmount = copy.deepcopy(joborder)
 
+                reffiles = []
+                visit_class(joborder_keepmount, ("File", "Directory"), lambda x: reffiles.append(x))
+
+                mapper = ArvPathMapper(self.arvrunner, reffiles, kwargs["basedir"],
+                                 "/keep/%s",
+                                 "/keep/%s/%s",
+                                 **kwargs)
+
                 def keepmount(obj):
-                    with SourceLine(obj, None, WorkflowException):
+                    remove_redundant_fields(obj)
+                    with SourceLine(obj, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
                         if "location" not in obj:
                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
-                    with SourceLine(obj, "location", WorkflowException):
+                    with SourceLine(obj, "location", WorkflowException, logger.isEnabledFor(logging.DEBUG)):
                         if obj["location"].startswith("keep:"):
-                            obj["location"] = "/keep/" + obj["location"][5:]
+                            obj["location"] = mapper.mapper(obj["location"]).target
                             if "listing" in obj:
                                 del obj["listing"]
                         elif obj["location"].startswith("_:"):
@@ -122,8 +139,13 @@ class ArvadosWorkflow(Workflow):
                         else:
                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
 
-                adjustFileObjs(joborder_keepmount, keepmount)
-                adjustDirObjs(joborder_keepmount, keepmount)
+                visit_class(joborder_keepmount, ("File", "Directory"), keepmount)
+
+                def resolved(obj):
+                    if obj["location"].startswith("keep:"):
+                        obj["location"] = mapper.mapper(obj["location"]).resolved
+
+                visit_class(joborder_resolved, ("File", "Directory"), resolved)
 
                 if self.wf_pdh is None:
                     adjustFileObjs(packed, keepmount)
@@ -136,7 +158,7 @@ class ArvadosWorkflow(Workflow):
                 "inputs": self.tool["inputs"],
                 "outputs": self.tool["outputs"],
                 "stdout": "cwl.output.json",
-                "requirements": workflowobj["requirements"]+[
+                "requirements": self.requirements+[
                     {
                     "class": "InitialWorkDirRequirement",
                     "listing": [{
@@ -150,11 +172,11 @@ class ArvadosWorkflow(Workflow):
                             "entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
                         }]
                 }],
-                "hints": workflowobj["hints"],
+                "hints": self.hints,
                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
             })
             kwargs["loader"] = self.doc_loader
             kwargs["avsc_names"] = self.doc_schema
-            return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
+            return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder_resolved, output_callback, **kwargs)
         else:
             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)