def packed_workflow(arvrunner, tool, merged_map):
"""Create a packed workflow.
A "packed" workflow is one where all the components have been combined into a single document."""
rewrites = {}
packed = pack(tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]),
tool.tool["id"], tool.metadata, rewrite_out=rewrites)
rewrite_to_orig = {}
for k,v in rewrites.items():
rewrite_to_orig[v] = k
def visit(v, cur_id):
if isinstance(v, dict):
if v.get("class") in ("CommandLineTool", "Workflow"):
cur_id = rewrite_to_orig.get(v["id"], v["id"])
if "location" in v and not v["location"].startswith("keep:"):
v["location"] = merged_map[cur_id][v["location"]]
for l in v:
visit(v[l], cur_id)
if isinstance(v, list):
for l in v:
visit(l, cur_id)
visit(packed, None)
return packed
def packed_workflow(arvrunner, tool, merged_map):
"""Create a packed workflow.
A "packed" workflow is one where all the components have been combined into a single document."""
rewrites = {}
packed = pack(tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]),
tool.tool["id"], tool.metadata, rewrite_out=rewrites)
rewrite_to_orig = {v: k for k,v in rewrites.items()}
def visit(v, cur_id):
if isinstance(v, dict):
if v.get("class") in ("CommandLineTool", "Workflow"):
if "id" not in v:
raise SourceLine(v, None, Exception).makeError("Embedded process object is missing required 'id' field")
cur_id = rewrite_to_orig.get(v["id"], v["id"])
if "location" in v and not v["location"].startswith("keep:"):
v["location"] = merged_map[cur_id].resolved[v["location"]]
if "location" in v and v["location"] in merged_map[cur_id].secondaryFiles:
v["secondaryFiles"] = merged_map[cur_id].secondaryFiles[v["location"]]
for l in v:
visit(v[l], cur_id)
if isinstance(v, list):
for l in v:
visit(l, cur_id)
visit(packed, None)
return packed
def packed_workflow(arvrunner, tool):
"""Create a packed workflow.
A "packed" workflow is one where all the components have been combined into a single document."""
return pack(tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]),
tool.tool["id"], tool.metadata)
def packed_workflow(arvrunner, tool, merged_map):
"""Create a packed workflow.
A "packed" workflow is one where all the components have been combined into a single document."""
rewrites = {}
packed = pack(tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]),
tool.tool["id"], tool.metadata, rewrite_out=rewrites)
rewrite_to_orig = {v: k for k,v in rewrites.items()}
def visit(v, cur_id):
if isinstance(v, dict):
if v.get("class") in ("CommandLineTool", "Workflow"):
if "id" not in v:
raise SourceLine(v, None, Exception).makeError("Embedded process object is missing required 'id' field")
cur_id = rewrite_to_orig.get(v["id"], v["id"])
if "location" in v and not v["location"].startswith("keep:"):
v["location"] = merged_map[cur_id].resolved[v["location"]]
if "location" in v and v["location"] in merged_map[cur_id].secondaryFiles:
v["secondaryFiles"] = merged_map[cur_id].secondaryFiles[v["location"]]
if v.get("class") == "DockerRequirement":
v["http://arvados.org/cwl#dockerCollectionPDH"] = arvados_cwl.arvdocker.arv_docker_get_image(arvrunner.api, v, True, arvrunner.project_uuid)
for l in v:
visit(v[l], cur_id)
if isinstance(v, list):
for l in v:
visit(l, cur_id)
visit(packed, None)
return packed
def packed_workflow(arvrunner, tool, merged_map):
"""Create a packed workflow.
A "packed" workflow is one where all the components have been combined into a single document."""
rewrites = {}
packed = pack(arvrunner.loadingContext,
tool.tool["id"],
rewrite_out=rewrites,
loader=tool.doc_loader)
rewrite_to_orig = {v: k for k, v in viewitems(rewrites)}
def visit(v, cur_id):
if isinstance(v, dict):
if v.get("class") in ("CommandLineTool", "Workflow",
"ExpressionTool"):
if tool.metadata["cwlVersion"] == "v1.0" and "id" not in v:
raise SourceLine(v, None, Exception).makeError(
"Embedded process object is missing required 'id' field, add an 'id' or use to cwlVersion: v1.1"
)
if "id" in v:
cur_id = rewrite_to_orig.get(v["id"], v["id"])
if "path" in v and "location" not in v:
v["location"] = v["path"]
del v["path"]
if "location" in v and cur_id in merged_map:
if v["location"] in merged_map[cur_id].resolved:
v["location"] = merged_map[cur_id].resolved[v["location"]]
if v["location"] in merged_map[cur_id].secondaryFiles:
v["secondaryFiles"] = merged_map[cur_id].secondaryFiles[
v["location"]]
if v.get("class") == "DockerRequirement":
v["http://arvados.org/cwl#dockerCollectionPDH"] = arvados_cwl.arvdocker.arv_docker_get_image(
arvrunner.api, v, True, arvrunner.project_uuid,
arvrunner.runtimeContext.force_docker_pull,
arvrunner.runtimeContext.tmp_outdir_prefix,
arvrunner.runtimeContext.match_local_docker)
for l in v:
visit(v[l], cur_id)
if isinstance(v, list):
for l in v:
visit(l, cur_id)
visit(packed, None)
return packed
def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
submit_runner_ram=0, name=None):
upload_docker(arvRunner, tool)
document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
packed = pack(document_loader, workflowobj, uri, tool.metadata)
adjustDirObjs(job_order, trim_listing)
main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
for inp in main["inputs"]:
sn = shortname(inp["id"])
if sn in job_order:
inp["default"] = job_order[sn]
if not name:
name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
upload_dependencies(arvRunner, name, document_loader,
packed, uri, False)
# TODO nowhere for submit_runner_ram to go.
body = {
"workflow": {
"name": name,
"description": tool.tool.get("doc", ""),
"definition":yaml.safe_dump(packed)
}}
if project_uuid:
body["workflow"]["owner_uuid"] = project_uuid
if uuid:
call = arvRunner.api.workflows().update(uuid=uuid, body=body)
else:
call = arvRunner.api.workflows().create(body=body)
return call.execute(num_retries=arvRunner.num_retries)["uuid"]
def job(self, joborder, output_callback, runtimeContext):
builder = make_builder(joborder, self.hints, self.requirements, runtimeContext)
runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
if not req:
return super(ArvadosWorkflow, self).job(joborder, output_callback, runtimeContext)
# RunInSingleContainer is true
with SourceLine(self.tool, None, WorkflowException, logger.isEnabledFor(logging.DEBUG)):
if "id" not in self.tool:
raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
discover_secondary_files(self.arvrunner.fs_access, builder,
self.tool["inputs"], joborder)
with Perf(metrics, "subworkflow upload_deps"):
upload_dependencies(self.arvrunner,
os.path.basename(joborder.get("id", "#")),
self.doc_loader,
joborder,
joborder.get("id", "#"),
False)
if self.wf_pdh is None:
packed = pack(self.loadingContext, self.tool["id"], loader=self.doc_loader)
for p in packed["$graph"]:
if p["id"] == "#main":
p["requirements"] = dedup_reqs(self.requirements)
p["hints"] = dedup_reqs(self.hints)
def visit(item):
if "requirements" in item:
item["requirements"] = [i for i in item["requirements"] if i["class"] != "DockerRequirement"]
for t in ("hints", "requirements"):
if t not in item:
continue
for req in item[t]:
if req["class"] == "ResourceRequirement":
dyn = False
for k in max_res_pars + sum_res_pars:
if k in req:
if isinstance(req[k], basestring):
if item["id"] == "#main":
# only the top-level requirements/hints may contain expressions
self.dynamic_resource_req.append(req)
dyn = True
break
else:
with SourceLine(req, k, WorkflowException):
raise WorkflowException("Non-top-level ResourceRequirement in single container cannot have expressions")
if not dyn:
self.static_resource_req.append(req)
visit_class(packed["$graph"], ("Workflow", "CommandLineTool"), visit)
if self.static_resource_req:
self.static_resource_req = [get_overall_res_req(self.static_resource_req)]
upload_dependencies(self.arvrunner,
runtimeContext.name,
self.doc_loader,
packed,
self.tool["id"],
False)
# Discover files/directories referenced by the
# workflow (mainly "default" values)
visit_class(packed, ("File", "Directory"), self.wf_reffiles.append)
if self.dynamic_resource_req:
# Evaluate dynamic resource requirements using current builder
rs = copy.copy(self.static_resource_req)
for dyn_rs in self.dynamic_resource_req:
eval_req = {"class": "ResourceRequirement"}
for a in max_res_pars + sum_res_pars:
if a in dyn_rs:
eval_req[a] = builder.do_eval(dyn_rs[a])
rs.append(eval_req)
job_res_reqs = [get_overall_res_req(rs)]
else:
job_res_reqs = self.static_resource_req
with Perf(metrics, "subworkflow adjust"):
joborder_resolved = copy.deepcopy(joborder)
joborder_keepmount = copy.deepcopy(joborder)
reffiles = []
visit_class(joborder_keepmount, ("File", "Directory"), reffiles.append)
mapper = ArvPathMapper(self.arvrunner, reffiles+self.wf_reffiles, runtimeContext.basedir,
"/keep/%s",
"/keep/%s/%s")
# For containers API, we need to make sure any extra
# referenced files (ie referenced by the workflow but
# not in the inputs) are included in the mounts.
if self.wf_reffiles:
runtimeContext = runtimeContext.copy()
runtimeContext.extra_reffiles = copy.deepcopy(self.wf_reffiles)
def keepmount(obj):
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, logger.isEnabledFor(logging.DEBUG)):
if obj["location"].startswith("keep:"):
obj["location"] = mapper.mapper(obj["location"]).target
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"])
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)
adjustDirObjs(packed, keepmount)
self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
self.loadingContext = self.loadingContext.copy()
self.loadingContext.metadata = self.loadingContext.metadata.copy()
self.loadingContext.metadata["http://commonwl.org/cwltool#original_cwlVersion"] = "v1.0"
if len(job_res_reqs) == 1:
# RAM request needs to be at least 128 MiB or the workflow
# runner itself won't run reliably.
if job_res_reqs[0].get("ramMin", 1024) < 128:
job_res_reqs[0]["ramMin"] = 128
arguments = ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl", "cwl.input.yml"]
if runtimeContext.debug:
arguments.insert(0, '--debug')
wf_runner = cmap({
"class": "CommandLineTool",
"baseCommand": "cwltool",
"inputs": self.tool["inputs"],
"outputs": self.tool["outputs"],
"stdout": "cwl.output.json",
"requirements": self.requirements+job_res_reqs+[
{"class": "InlineJavascriptRequirement"},
{
"class": "InitialWorkDirRequirement",
"listing": [{
"entryname": "workflow.cwl",
"entry": '$({"class": "File", "location": "keep:%s/workflow.cwl"})' % self.wf_pdh
}, {
"entryname": "cwl.input.yml",
"entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}]
}],
"hints": self.hints,
"arguments": arguments,
"id": "#"
})
return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)
def job(self, joborder, output_callback, **kwargs):
kwargs["work_api"] = self.work_api
req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
if req:
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", "#")),
document_loader,
joborder,
joborder.get("id", "#"),
False)
if self.wf_pdh is None:
workflowobj["requirements"] = dedup_reqs(self.requirements)
workflowobj["hints"] = dedup_reqs(self.hints)
packed = pack(document_loader, workflowobj, uri, self.metadata)
upload_dependencies(self.arvrunner,
kwargs.get("name", ""),
document_loader,
packed,
uri,
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):
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, logger.isEnabledFor(logging.DEBUG)):
if obj["location"].startswith("keep:"):
obj["location"] = mapper.mapper(obj["location"]).target
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"])
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)
adjustDirObjs(packed, keepmount)
self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
wf_runner = cmap({
"class": "CommandLineTool",
"baseCommand": "cwltool",
"inputs": self.tool["inputs"],
"outputs": self.tool["outputs"],
"stdout": "cwl.output.json",
"requirements": self.requirements+[
{
"class": "InitialWorkDirRequirement",
"listing": [{
"entryname": "workflow.cwl",
"entry": {
"class": "File",
"location": "keep:%s/workflow.cwl" % self.wf_pdh
}
}, {
"entryname": "cwl.input.yml",
"entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}]
}],
"hints": self.hints,
"arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"],
"id": "#"
})
kwargs["loader"] = self.doc_loader
kwargs["avsc_names"] = self.doc_schema
return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder_resolved, output_callback, **kwargs)
else:
return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)
def job(self, joborder, output_callback, **kwargs):
kwargs["work_api"] = self.work_api
req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
if req:
with SourceLine(self.tool, None, WorkflowException):
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"])
with Perf(metrics, "subworkflow upload_deps"):
upload_dependencies(self.arvrunner,
os.path.basename(joborder.get("id", "#")),
document_loader,
joborder,
joborder.get("id", "#"),
False)
if self.wf_pdh is None:
workflowobj["requirements"] = dedup_reqs(self.requirements)
workflowobj["hints"] = dedup_reqs(self.hints)
packed = pack(document_loader, workflowobj, uri, self.metadata)
upload_dependencies(self.arvrunner,
kwargs.get("name", ""),
document_loader,
packed,
uri,
False)
with Perf(metrics, "subworkflow adjust"):
joborder_keepmount = copy.deepcopy(joborder)
def keepmount(obj):
with SourceLine(obj, None, WorkflowException):
if "location" not in obj:
raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
with SourceLine(obj, "location", WorkflowException):
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)
if self.wf_pdh is None:
adjustFileObjs(packed, keepmount)
adjustDirObjs(packed, keepmount)
self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
wf_runner = cmap({
"class": "CommandLineTool",
"baseCommand": "cwltool",
"inputs": self.tool["inputs"],
"outputs": self.tool["outputs"],
"stdout": "cwl.output.json",
"requirements": workflowobj["requirements"]+[
{
"class": "InitialWorkDirRequirement",
"listing": [{
"entryname": "workflow.cwl",
"entry": {
"class": "File",
"location": "keep:%s/workflow.cwl" % self.wf_pdh
}
}, {
"entryname": "cwl.input.yml",
"entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}]
}],
"hints": workflowobj["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)
else:
return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)
def job(self, joborder, output_callback, runtimeContext):
builder = make_builder(joborder, self.hints, self.requirements, runtimeContext)
runtimeContext = set_cluster_target(self.tool, self.arvrunner, builder, runtimeContext)
req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
if not req:
return super(ArvadosWorkflow, self).job(joborder, output_callback, runtimeContext)
# RunInSingleContainer is true
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", "#")),
document_loader,
joborder,
joborder.get("id", "#"),
False)
if self.wf_pdh is None:
workflowobj["requirements"] = dedup_reqs(self.requirements)
workflowobj["hints"] = dedup_reqs(self.hints)
packed = pack(document_loader, workflowobj, uri, self.metadata)
def visit(item):
for t in ("hints", "requirements"):
if t not in item:
continue
for req in item[t]:
if req["class"] == "ResourceRequirement":
dyn = False
for k in max_res_pars + sum_res_pars:
if k in req:
if isinstance(req[k], basestring):
if item["id"] == "#main":
# only the top-level requirements/hints may contain expressions
self.dynamic_resource_req.append(req)
dyn = True
break
else:
with SourceLine(req, k, WorkflowException):
raise WorkflowException("Non-top-level ResourceRequirement in single container cannot have expressions")
if not dyn:
self.static_resource_req.append(req)
if req["class"] == "DockerRequirement":
if "http://arvados.org/cwl#dockerCollectionPDH" in req:
del req["http://arvados.org/cwl#dockerCollectionPDH"]
visit_class(packed["$graph"], ("Workflow", "CommandLineTool"), visit)
if self.static_resource_req:
self.static_resource_req = [get_overall_res_req(self.static_resource_req)]
upload_dependencies(self.arvrunner,
runtimeContext.name,
document_loader,
packed,
uri,
False)
# Discover files/directories referenced by the
# workflow (mainly "default" values)
visit_class(packed, ("File", "Directory"), self.wf_reffiles.append)
if self.dynamic_resource_req:
# Evaluate dynamic resource requirements using current builder
rs = copy.copy(self.static_resource_req)
for dyn_rs in self.dynamic_resource_req:
eval_req = {"class": "ResourceRequirement"}
for a in max_res_pars + sum_res_pars:
if a in dyn_rs:
eval_req[a] = builder.do_eval(dyn_rs[a])
rs.append(eval_req)
job_res_reqs = [get_overall_res_req(rs)]
else:
job_res_reqs = self.static_resource_req
with Perf(metrics, "subworkflow adjust"):
joborder_resolved = copy.deepcopy(joborder)
joborder_keepmount = copy.deepcopy(joborder)
reffiles = []
visit_class(joborder_keepmount, ("File", "Directory"), reffiles.append)
mapper = ArvPathMapper(self.arvrunner, reffiles+self.wf_reffiles, runtimeContext.basedir,
"/keep/%s",
"/keep/%s/%s")
# For containers API, we need to make sure any extra
# referenced files (ie referenced by the workflow but
# not in the inputs) are included in the mounts.
if self.wf_reffiles:
runtimeContext = runtimeContext.copy()
runtimeContext.extra_reffiles = copy.deepcopy(self.wf_reffiles)
def keepmount(obj):
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, logger.isEnabledFor(logging.DEBUG)):
if obj["location"].startswith("keep:"):
obj["location"] = mapper.mapper(obj["location"]).target
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"])
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)
adjustDirObjs(packed, keepmount)
self.wf_pdh = upload_workflow_collection(self.arvrunner, shortname(self.tool["id"]), packed)
wf_runner = cmap({
"class": "CommandLineTool",
"baseCommand": "cwltool",
"inputs": self.tool["inputs"],
"outputs": self.tool["outputs"],
"stdout": "cwl.output.json",
"requirements": self.requirements+job_res_reqs+[
{"class": "InlineJavascriptRequirement"},
{
"class": "InitialWorkDirRequirement",
"listing": [{
"entryname": "workflow.cwl",
"entry": '$({"class": "File", "location": "keep:%s/workflow.cwl"})' % self.wf_pdh
}, {
"entryname": "cwl.input.yml",
"entry": json.dumps(joborder_keepmount, indent=2, sort_keys=True, separators=(',',': ')).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}]
}],
"hints": self.hints,
"arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"],
"id": "#"
})
return ArvadosCommandTool(self.arvrunner, wf_runner, self.loadingContext).job(joborder_resolved, output_callback, runtimeContext)
def job(self, joborder, output_callback, **kwargs):
kwargs["work_api"] = self.work_api
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"])
with Perf(metrics, "subworkflow upload_deps"):
workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
packed = pack(document_loader, workflowobj, uri, self.metadata)
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)
with Perf(metrics, "subworkflow adjust"):
joborder_keepmount = copy.deepcopy(joborder)
def keepmount(obj):
if "location" not in obj:
raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], 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",
"baseCommand": "cwltool",
"inputs": self.tool["inputs"],
"outputs": self.tool["outputs"],
"stdout": "cwl.output.json",
"requirements": workflowobj["requirements"]+[
{
"class": "InitialWorkDirRequirement",
"listing": [{
"entryname": "workflow.cwl",
"entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}, {
"entryname": "cwl.input.yml",
"entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
}]
}],
"hints": workflowobj["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)
else:
return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)
