[ARVADOS] created: bc6b243539b5480493c7aa758a1272a1a42b58f1
Git user
git at public.curoverse.com
Thu Sep 15 16:13:16 EDT 2016
at bc6b243539b5480493c7aa758a1272a1a42b58f1 (commit)
commit bc6b243539b5480493c7aa758a1272a1a42b58f1
Author: Peter Amstutz <peter.amstutz at curoverse.com>
Date: Thu Sep 15 16:13:08 2016 -0400
10081: Submitting whole workflows as jobs WIP
diff --git a/sdk/cwl/arvados_cwl/__init__.py b/sdk/cwl/arvados_cwl/__init__.py
index 7bfdba8..1256260 100644
--- a/sdk/cwl/arvados_cwl/__init__.py
+++ b/sdk/cwl/arvados_cwl/__init__.py
@@ -22,9 +22,11 @@ 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
@@ -61,10 +63,12 @@ class ArvCwlRunner(object):
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)
@@ -155,18 +159,11 @@ class ArvCwlRunner(object):
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
@@ -180,7 +177,7 @@ class ArvCwlRunner(object):
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")
@@ -384,14 +381,21 @@ def main(args, stdout, stderr, api_client=None):
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))
diff --git a/sdk/cwl/arvados_cwl/arvjob.py b/sdk/cwl/arvados_cwl/arvjob.py
index e6d0cba..7cde273 100644
--- a/sdk/cwl/arvados_cwl/arvjob.py
+++ b/sdk/cwl/arvados_cwl/arvjob.py
@@ -1,6 +1,7 @@
import logging
import re
import copy
+import json
from cwltool.process import get_feature, shortname
from cwltool.errors import WorkflowException
@@ -39,6 +40,7 @@ class ArvadosJob(object):
if self.generatefiles["listing"]:
vwd = arvados.collection.Collection()
script_parameters["task.vwd"] = {}
+ logger.debug("generatefiles is %s", json.dumps(self.generatefiles, indent=4))
generatemapper = InitialWorkDirPathMapper([self.generatefiles], "", "",
separateDirs=False)
for f, p in generatemapper.items():
diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py
index 8e45890..d0d2749 100644
--- a/sdk/cwl/arvados_cwl/arvworkflow.py
+++ b/sdk/cwl/arvados_cwl/arvworkflow.py
@@ -1,16 +1,21 @@
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"])
@@ -39,3 +44,45 @@ def make_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
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.requirements + 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": [
+ {"class": "InlineJavascriptRequirement"},
+ {
+ "class": "InitialWorkDirRequirement",
+ "listing": [{
+ "entryname": "workflow.json",
+ "entry": json.dumps(packed, sort_keys=True, indent=4)
+ }, {
+ "entryname": "cwl.input.json",
+ "entry": "$(JSON.stringify(inputs))"
+ }]
+ }],
+ "arguments": ["--debug", "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)
-----------------------------------------------------------------------
hooks/post-receive
--
More information about the arvados-commits
mailing list