from cwltool.errors import WorkflowException
import cwltool.main
import cwltool.workflow
+import schema_salad
import arvados
import arvados.config
from .arvcontainer import ArvadosContainer, RunnerContainer
from .arvjob import ArvadosJob, RunnerJob, RunnerTemplate
from .arvtool import ArvadosCommandTool
+from .arvworkflow import ArvadosWorkflow
from .fsaccess import CollectionFsAccess
-from .arvworkflow import make_workflow
+from .arvworkflow import upload_workflow
from .perf import Perf
+from cwltool.pack import pack
from cwltool.process import shortname, UnsupportedRequirement
from cwltool.pathmapper import adjustFileObjs
self.work_api = work_api
self.stop_polling = threading.Event()
self.poll_api = None
+ self.pipeline = None
if self.work_api is None:
# todo: autodetect API to use.
if self.work_api not in ("containers", "jobs"):
raise Exception("Unsupported API '%s'" % self.work_api)
- def arvMakeTool(self, toolpath_object, **kwargs):
+ def arv_make_tool(self, toolpath_object, **kwargs):
+ kwargs["work_api"] = self.work_api
if "class" in toolpath_object and toolpath_object["class"] == "CommandLineTool":
- kwargs["work_api"] = self.work_api
return ArvadosCommandTool(self, toolpath_object, **kwargs)
+ elif "class" in toolpath_object and toolpath_object["class"] == "Workflow":
+ return ArvadosWorkflow(self, toolpath_object, **kwargs)
else:
return cwltool.workflow.defaultMakeTool(toolpath_object, **kwargs)
for v in obj:
self.check_writable(v)
- def arvExecutor(self, tool, job_order, **kwargs):
+ def arv_executor(self, tool, job_order, **kwargs):
self.debug = kwargs.get("debug")
tool.visit(self.check_writable)
- if kwargs.get("quiet"):
- logger.setLevel(logging.WARN)
- logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
-
- if self.debug:
- logger.setLevel(logging.DEBUG)
-
useruuid = self.api.users().current().execute()["uuid"]
self.project_uuid = kwargs.get("project_uuid") if kwargs.get("project_uuid") else useruuid
self.pipeline = None
return tmpl.uuid
if kwargs.get("create_workflow") or kwargs.get("update_workflow"):
- return make_workflow(self, tool, job_order, self.project_uuid, kwargs.get("update_workflow"))
+ return upload_workflow(self, tool, job_order, self.project_uuid, kwargs.get("update_workflow"))
self.ignore_docker_for_reuse = kwargs.get("ignore_docker_for_reuse")
return parser
+def add_arv_hints():
+ cache = {}
+ res = pkg_resources.resource_stream(__name__, 'arv-cwl-schema.yml')
+ cache["https://w3id.org/cwl/arv-cwl-schema.yml"] = res.read()
+ res.close()
+ _, cwlnames, _, _ = cwltool.process.get_schema("v1.0")
+ _, extnames, _, _ = schema_salad.schema.load_schema("https://w3id.org/cwl/arv-cwl-schema.yml", cache=cache)
+ for n in extnames.names:
+ if not cwlnames.has_name("http://arvados.org/cwl#"+n, ""):
+ cwlnames.add_name("http://arvados.org/cwl#"+n, "", extnames.get_name(n, ""))
def main(args, stdout, stderr, api_client=None):
parser = arg_parser()
if (arvargs.create_template or arvargs.create_workflow or arvargs.update_workflow) and not arvargs.job_order:
job_order_object = ({}, "")
+ add_arv_hints()
+
try:
if api_client is None:
api_client=arvados.api('v1', model=OrderedJsonModel())
logger.error(e)
return 1
+ if arvargs.debug:
+ logger.setLevel(logging.DEBUG)
+
+ if arvargs.quiet:
+ logger.setLevel(logging.WARN)
+ logging.getLogger('arvados.arv-run').setLevel(logging.WARN)
+
arvargs.conformance_test = None
arvargs.use_container = True
return cwltool.main.main(args=arvargs,
stdout=stdout,
stderr=stderr,
- executor=runner.arvExecutor,
- makeTool=runner.arvMakeTool,
+ executor=runner.arv_executor,
+ makeTool=runner.arv_make_tool,
versionfunc=versionstring,
job_order_object=job_order_object,
make_fs_access=partial(CollectionFsAccess, api_client=api_client))
--- /dev/null
+$base: "http://arvados.org/cwl#"
+$graph:
+- name: RunInSingleContainer
+ type: record
+ doc: |
+ Indicates that a subworkflow should run in a single container
+ and not be scheduled as separate steps.
+ fields:
+ - name: class
+ type: string
+ doc: "Always 'arv:RunInSingleContainer'"
+ jsonldPredicate:
+ _id: "@type"
+ _type: "@vocab"
import logging
import re
import copy
+import json
from cwltool.process import get_feature, shortname
from cwltool.errors import WorkflowException
except WorkflowException as e:
logger.error("Error while collecting job outputs:\n%s", e, exc_info=(e if self.arvrunner.debug else False))
processStatus = "permanentFail"
+ outputs = None
except Exception as e:
logger.exception("Got unknown exception while collecting job outputs:")
processStatus = "permanentFail"
+ outputs = None
self.output_callback(outputs, processStatus)
finally:
import os
import json
import copy
+import logging
from cwltool.pack import pack
from cwltool.load_tool import fetch_document
from cwltool.process import shortname
+from cwltool.workflow import Workflow
import ruamel.yaml as yaml
from .runner import upload_docker, upload_dependencies
+from .arvtool import ArvadosCommandTool
-def make_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
+logger = logging.getLogger('arvados.cwl-runner')
+
+def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
upload_docker(arvRunner, tool)
document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
else:
return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
+
+class ArvadosWorkflow(Workflow):
+ """Wrap cwltool Workflow to override selected methods."""
+
+ def __init__(self, arvrunner, toolpath_object, **kwargs):
+ super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
+ self.arvrunner = arvrunner
+ self.work_api = kwargs["work_api"]
+
+ 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"])
+ workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
+ workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
+ packed = pack(document_loader, workflowobj, uri, self.metadata)
+ wf_runner = {
+ "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.json",
+ "entry": json.dumps(packed, sort_keys=True, indent=4).replace('$(', '\$(').replace('${', '\${')
+ }, {
+ "entryname": "cwl.input.json",
+ "entry": "$(JSON.stringify(inputs))"
+ }]
+ }],
+ "hints": workflowobj["hints"],
+ "arguments": ["--no-container", "--move-outputs", "workflow.json", "cwl.input.json"]
+ }
+ kwargs["loader"] = self.doc_loader
+ kwargs["avsc_names"] = self.doc_schema
+ return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
+ else:
+ return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)
download_url="https://github.com/curoverse/arvados.git",
license='Apache 2.0',
packages=find_packages(),
+ package_data={'arvados_cwl': ['arv-cwl-schema.yml']},
scripts=[
'bin/cwl-runner',
'bin/arvados-cwl-runner'
--- /dev/null
+class: Workflow
+cwlVersion: v1.0
+$namespaces:
+ arv: "http://arvados.org/cwl#"
+inputs:
+ sleeptime:
+ type: int[]
+ default: [44, 29, 14]
+outputs: []
+requirements:
+ SubworkflowFeatureRequirement: {}
+ ScatterFeatureRequirement: {}
+ InlineJavascriptRequirement: {}
+ StepInputExpressionRequirement: {}
+steps:
+ scatterstep:
+ in:
+ sleeptime: sleeptime
+ out: []
+ scatter: sleeptime
+ hints:
+ - class: arv:RunInSingleContainer
+ run:
+ class: Workflow
+ id: mysub
+ inputs:
+ sleeptime: int
+ outputs: []
+ steps:
+ sleep1:
+ in:
+ sleeptime: sleeptime
+ out: [out]
+ run:
+ class: CommandLineTool
+ inputs:
+ sleeptime:
+ type: int
+ inputBinding: {position: 1}
+ outputs:
+ out:
+ type: string
+ outputBinding:
+ outputEval: "out"
+ baseCommand: sleep
+ sleep2:
+ in:
+ sleeptime:
+ source: sleeptime
+ valueFrom: $(self+1)
+ dep: sleep1/out
+ out: []
+ run:
+ class: CommandLineTool
+ inputs:
+ sleeptime:
+ type: int
+ inputBinding: {position: 1}
+ outputs: []
+ baseCommand: sleep
-import arvados_cwl
+
import logging
import mock
import unittest
import os
import functools
-import cwltool.process
-from schema_salad.ref_resolver import Loader
+import json
+import arvados_cwl
+import cwltool.process
from schema_salad.ref_resolver import Loader
if not os.getenv('ARVADOS_DEBUG'):
mock.call().execute(num_retries=0)])
self.assertFalse(api.collections().create.called)
+
+
+class TestWorkflow(unittest.TestCase):
+ # The test passes no builder.resources
+ # Hence the default resources will apply: {'cores': 1, 'ram': 1024, 'outdirSize': 1024, 'tmpdirSize': 1024}
+ def test_run(self):
+ arvados_cwl.add_arv_hints()
+
+ runner = arvados_cwl.ArvCwlRunner(mock.MagicMock())
+ runner.project_uuid = "zzzzz-8i9sb-zzzzzzzzzzzzzzz"
+ runner.ignore_docker_for_reuse = False
+ document_loader, avsc_names, schema_metadata, metaschema_loader = cwltool.process.get_schema("v1.0")
+
+ tool, metadata = document_loader.resolve_ref("tests/wf/scatter2.cwl")
+ metadata["cwlVersion"] = tool["cwlVersion"]
+
+ make_fs_access=functools.partial(arvados_cwl.CollectionFsAccess, api_client=runner.api)
+ arvtool = arvados_cwl.ArvadosWorkflow(runner, tool, work_api="jobs", avsc_names=avsc_names,
+ basedir="", make_fs_access=make_fs_access, loader=document_loader,
+ makeTool=runner.arv_make_tool, metadata=metadata)
+ arvtool.formatgraph = None
+ it = arvtool.job({}, mock.MagicMock(), basedir="", make_fs_access=make_fs_access)
+ it.next().run()
+ it.next().run()
+
+ runner.api.jobs().create.assert_called_with(
+ body={
+ 'minimum_script_version': '9e5b98e8f5f4727856b53447191f9c06e3da2ba6',
+ 'repository': 'arvados',
+ 'script_version': 'master',
+ 'script': 'crunchrunner',
+ 'script_parameters': {
+ 'tasks': [{'task.env': {
+ 'HOME': '$(task.outdir)',
+ 'TMPDIR': '$(task.tmpdir)'},
+ 'task.vwd': {
+ 'workflow.json': '$(task.keep)/f101400a398097d4398cdb3eb5d1a7ca+118/workflow.json',
+ 'cwl.input.json': '$(task.keep)/f101400a398097d4398cdb3eb5d1a7ca+118/cwl.input.json'
+ },
+ 'command': [u'cwltool', u'--no-container', u'--move-outputs', u'workflow.json', u'cwl.input.json'],
+ 'task.stdout': 'cwl.output.json'}]},
+ 'runtime_constraints': {
+ 'min_scratch_mb_per_node': 2048,
+ 'min_cores_per_node': 1,
+ 'docker_image': 'arvados/jobs',
+ 'min_ram_mb_per_node': 1024
+ },
+ 'owner_uuid': 'zzzzz-8i9sb-zzzzzzzzzzzzzzz'},
+ filters=[['repository', '=', 'arvados'],
+ ['script', '=', 'crunchrunner'],
+ ['script_version', 'in git', '9e5b98e8f5f4727856b53447191f9c06e3da2ba6'],
+ ['docker_image_locator', 'in docker', 'arvados/jobs']],
+ find_or_create=True)
projects / arvados.git / commitdiff