+cwltool.draft2tool.ACCEPTLIST_RE = re.compile(r".*")
+
+def trim_listing(obj):
+ """Remove 'listing' field from Directory objects that are keep references.
+
+ When Directory objects represent Keep references, it redundant and
+ potentially very expensive to pass fully enumerated Directory objects
+ between instances of cwl-runner (e.g. a submitting a job, or using the
+ RunInSingleContainer feature), so delete the 'listing' field when it is
+ safe to do so.
+ """
+
+ if obj.get("location", "").startswith("keep:") and "listing" in obj:
+ del obj["listing"]
+ if obj.get("location", "").startswith("_:"):
+ del obj["location"]
+
+def upload_dependencies(arvrunner, name, document_loader,
+ workflowobj, uri, loadref_run):
+ """Upload the dependencies of the workflowobj document to Keep.
+
+ Returns a pathmapper object mapping local paths to keep references. Also
+ does an in-place update of references in "workflowobj".
+
+ Use scandeps to find $import, $include, $schemas, run, File and Directory
+ fields that represent external references.
+
+ If workflowobj has an "id" field, this will reload the document to ensure
+ it is scanning the raw document prior to preprocessing.
+ """
+
+ loaded = set()
+ def loadref(b, u):
+ joined = urlparse.urljoin(b, u)
+ defrg, _ = urlparse.urldefrag(joined)
+ if defrg not in loaded:
+ loaded.add(defrg)
+ # Use fetch_text to get raw file (before preprocessing).
+ text = document_loader.fetch_text(defrg)
+ if isinstance(text, bytes):
+ textIO = StringIO(text.decode('utf-8'))
+ else:
+ textIO = StringIO(text)
+ return yaml.safe_load(textIO)
+ else:
+ return {}
+
+ if loadref_run:
+ loadref_fields = set(("$import", "run"))
+ else:
+ loadref_fields = set(("$import",))
+
+ scanobj = workflowobj
+ if "id" in workflowobj:
+ # Need raw file content (before preprocessing) to ensure
+ # that external references in $include and $mixin are captured.
+ scanobj = loadref("", workflowobj["id"])
+
+ sc = scandeps(uri, scanobj,
+ loadref_fields,
+ set(("$include", "$schemas", "location")),
+ loadref)
+
+ normalizeFilesDirs(sc)
+
+ if "id" in workflowobj:
+ sc.append({"class": "File", "location": workflowobj["id"]})
+
+ mapper = ArvPathMapper(arvrunner, sc, "",
+ "keep:%s",
+ "keep:%s/%s",
+ name=name)
+
+ def setloc(p):
+ if "location" in p and (not p["location"].startswith("_:")) and (not p["location"].startswith("keep:")):
+ p["location"] = mapper.mapper(p["location"]).resolved
+ adjustFileObjs(workflowobj, setloc)
+ adjustDirObjs(workflowobj, setloc)
+
+ return mapper
+
+
+def upload_docker(arvrunner, tool):
+ if isinstance(tool, CommandLineTool):
+ (docker_req, docker_is_req) = get_feature(tool, "DockerRequirement")
+ if docker_req:
+ if docker_req.get("dockerOutputDirectory"):
+ # TODO: can be supported by containers API, but not jobs API.
+ raise UnsupportedRequirement("Option 'dockerOutputDirectory' of DockerRequirement not supported.")
+ arv_docker_get_image(arvrunner.api, docker_req, True, arvrunner.project_uuid)
+ elif isinstance(tool, cwltool.workflow.Workflow):
+ for s in tool.steps:
+ upload_docker(arvrunner, s.embedded_tool)
+
+def upload_instance(arvrunner, name, tool, job_order):
+ upload_docker(arvrunner, tool)
+
+ for t in tool.tool["inputs"]:
+ def setSecondary(fileobj):
+ if isinstance(fileobj, dict) and fileobj.get("class") == "File":
+ if "secondaryFiles" not in fileobj:
+ fileobj["secondaryFiles"] = [{"location": substitute(fileobj["location"], sf), "class": "File"} for sf in t["secondaryFiles"]]
+
+ if isinstance(fileobj, list):
+ for e in fileobj:
+ setSecondary(e)
+
+ if shortname(t["id"]) in job_order and t.get("secondaryFiles"):
+ setSecondary(job_order[shortname(t["id"])])
+
+ workflowmapper = upload_dependencies(arvrunner,
+ name,
+ tool.doc_loader,
+ tool.tool,
+ tool.tool["id"],
+ True)
+ jobmapper = upload_dependencies(arvrunner,
+ os.path.basename(job_order.get("id", "#")),
+ tool.doc_loader,
+ job_order,
+ job_order.get("id", "#"),
+ False)
+
+ if "id" in job_order:
+ del job_order["id"]
+
+ return workflowmapper
+
+def arvados_jobs_image(arvrunner):
+ img = "arvados/jobs:"+__version__
+ try:
+ arv_docker_get_image(arvrunner.api, {"dockerPull": img}, True, arvrunner.project_uuid)
+ except Exception as e:
+ raise Exception("Docker image %s is not available\n%s" % (img, e) )
+ return img