10129: Ensure that directory literals get preserved when trimming keep directory...
[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py
index 3bf62ad8d46d82feffb4627cff3385bca3a3750d..3bb0a34197594774f81049ff5ccdf606dc2dfcde 100644 (file)
@@ -6,12 +6,12 @@ import logging
 from cwltool.pack import pack
 from cwltool.load_tool import fetch_document
 from cwltool.process import shortname
-from cwltool.workflow import Workflow
-from cwltool.pathmapper import adjustDirObjs
+from cwltool.workflow import Workflow, WorkflowException
+from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
 
 import ruamel.yaml as yaml
 
-from .runner import upload_docker, upload_dependencies
+from .runner import upload_docker, upload_dependencies, del_listing
 from .arvtool import ArvadosCommandTool
 
 logger = logging.getLogger('arvados.cwl-runner')
@@ -23,6 +23,8 @@ def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
 
     packed = pack(document_loader, workflowobj, uri, tool.metadata)
 
+    adjustDirObjs(job_order, del_listing)
+
     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
     for inp in main["inputs"]:
         sn = shortname(inp["id"])
@@ -59,14 +61,41 @@ class ArvadosWorkflow(Workflow):
         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
         if req:
             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
+
             workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
             workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
             packed = pack(document_loader, workflowobj, uri, self.metadata)
 
-            def prune_directories(obj):
-                if obj["location"].startswith("keep:") and "listing" in obj:
-                    del obj["listing"]
-            adjustDirObjs(joborder, prune_directories)
+            upload_dependencies(self.arvrunner,
+                                kwargs.get("name", ""),
+                                document_loader,
+                                packed,
+                                uri,
+                                False)
+
+            upload_dependencies(self.arvrunner,
+                                os.path.basename(joborder.get("id", "#")),
+                                document_loader,
+                                joborder,
+                                joborder.get("id", "#"),
+                                False)
+
+            joborder_keepmount = copy.deepcopy(joborder)
+
+            def keepmount(obj):
+                if obj["location"].startswith("keep:"):
+                    obj["location"] = "/keep/" + obj["location"][5:]
+                    if "listing" in obj:
+                        del obj["listing"]
+                elif obj["location"].startswith("_:"):
+                    del obj["location"]
+                else:
+                    raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
+
+            adjustFileObjs(joborder_keepmount, keepmount)
+            adjustDirObjs(joborder_keepmount, keepmount)
+            adjustFileObjs(packed, keepmount)
+            adjustDirObjs(packed, keepmount)
 
             wf_runner = {
                 "class": "CommandLineTool",
@@ -82,12 +111,12 @@ class ArvadosWorkflow(Workflow):
                             "entryname": "workflow.cwl",
                             "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
                         }, {
-                            "entryname": "cwl.input.json",
-                            "entry": yaml.safe_dump(joborder).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
+                            "entryname": "cwl.input.yml",
+                            "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
                         }]
                 }],
                 "hints": workflowobj["hints"],
-                "arguments": ["--no-container", "--move-outputs", "workflow.cwl", "cwl.input.json"]
+                "arguments": ["--no-container", "--move-outputs", "workflow.cwl#main", "cwl.input.yml"]
             }
             kwargs["loader"] = self.doc_loader
             kwargs["avsc_names"] = self.doc_schema