X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/84bcaa194dd2dd767aea291cdc17e85cd802325a..b9f7a8693579045f05d142dba8bffd2c5660dfce:/sdk/cwl/arvados_cwl/arvworkflow.py diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py index ab3940998b..8c1db3a817 100644 --- a/sdk/cwl/arvados_cwl/arvworkflow.py +++ b/sdk/cwl/arvados_cwl/arvworkflow.py @@ -3,48 +3,61 @@ import json import copy import logging +from schema_salad.sourceline import SourceLine, cmap + from cwltool.pack import pack from cwltool.load_tool import fetch_document from cwltool.process import shortname -from cwltool.workflow import Workflow +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, trim_listing from .arvtool import ArvadosCommandTool +from .perf import Perf logger = logging.getLogger('arvados.cwl-runner') +metrics = logging.getLogger('arvados.cwl-runner.metrics') -def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid): +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] - name = os.path.basename(tool.tool["id"]) + 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": { - "owner_uuid": project_uuid, - "name": tool.tool.get("label", name), + "name": name, "description": tool.tool.get("doc", ""), - "definition":yaml.safe_dump(packed) + "definition":yaml.round_trip_dump(packed) }} + if project_uuid: + body["workflow"]["owner_uuid"] = project_uuid - if update_uuid: - return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"] + if uuid: + call = arvRunner.api.workflows().update(uuid=uuid, body=body) else: - return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"] + call = arvRunner.api.workflows().create(body=body) + return call.execute(num_retries=arvRunner.num_retries)["uuid"] class ArvadosWorkflow(Workflow): """Wrap cwltool Workflow to override selected methods.""" @@ -60,59 +73,67 @@ class ArvadosWorkflow(Workflow): 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) - - 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:] - else: - raise Exception("Uh oh %s" % obj["location"]) - if "listing" in obj: - del obj["listing"] - adjustFileObjs(joborder_keepmount, keepmount) - adjustDirObjs(joborder_keepmount, keepmount) - adjustFileObjs(packed, keepmount) - adjustDirObjs(packed, keepmount) - - wf_runner = { + 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): + 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) + adjustFileObjs(packed, keepmount) + adjustDirObjs(packed, keepmount) + + wf_runner = cmap({ "class": "CommandLineTool", "baseCommand": "cwltool", "inputs": self.tool["inputs"], "outputs": self.tool["outputs"], "stdout": "cwl.output.json", "requirements": workflowobj["requirements"]+[ - {"class": "InlineJavascriptRequirement"}, { "class": "InitialWorkDirRequirement", "listing": [{ "entryname": "workflow.cwl", - "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${') + "entry": yaml.round_trip_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${') }, { - "entryname": "cwl.input.json", - "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${') + "entryname": "cwl.input.yml", + "entry": yaml.round_trip_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${') }] }], "hints": workflowobj["hints"], - "arguments": ["--no-container", "--move-outputs", "workflow.cwl#main", "cwl.input.json"] - } + "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)