#!/usr/bin/env python
+# Crunch script integration for running arvados-cwl-runner (importing
+# arvados_cwl module) inside a crunch job.
+#
+# This gets the job record, transforms the script parameters into a valid CWL
+# input object, then executes the CWL runner to run the underlying workflow or
+# tool. When the workflow completes, record the output object in an output
+# collection for this runner job.
+
import arvados
import arvados_cwl
import arvados.collection
import logging
import os
import json
+import argparse
from arvados.api import OrderedJsonModel
+from cwltool.process import adjustFiles
+
+# Print package versions
+logging.info(cwltool.main.versionstring())
api = arvados.api("v1")
try:
job_order_object = arvados.current_job()['script_parameters']
- print job_order_object
-
- for k,v in job_order_object.items():
+ def keeppath(v):
if arvados.util.keep_locator_pattern.match(v):
- job_order_object[k] = "file://%s/%s" % (os.environ['TASK_KEEPMOUNT'], v)
+ return "file://%s/%s" % (os.environ['TASK_KEEPMOUNT'], v)
+
+ job_order_object["cwl:tool"] = keeppath(job_order_object["cwl:tool"])
+
+ adjustFiles(job_order_object, keeppath)
runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()))
t = cwltool.main.load_tool(job_order_object, False, True, runner.arvMakeTool, True)
- outputObj = runner.arvExecutor(t, job_order_object, "", None)
+ args = argparse.Namespace()
+ args.project_uuid = arvados.current_job()["owner_uuid"]
+ args.enable_reuse = True
+ args.submit = False
+ args.debug = True
+ args.quiet = False
+ args.ignore_docker_for_reuse = False
+ outputObj = runner.arvExecutor(t, job_order_object, "", args, cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]})
+
+ files = {}
+ def capture(path):
+ sp = path.split("/")
+ col = sp[0][5:]
+ if col not in files:
+ files[col] = set()
+ files[col].add("/".join(sp[1:]))
+ return path
+
+ adjustFiles(outputObj, capture)
+
+ final = arvados.collection.Collection()
+
+ for k,v in files.iteritems():
+ with arvados.collection.Collection(k) as c:
+ for f in c:
+ final.copy(f, f, c, True)
+
+ def makeRelative(path):
+ return "/".join(path.split("/")[1:])
+
+ adjustFiles(outputObj, makeRelative)
- c = arvados.collection.Collection()
- with c.open("cwl.output.json", "w") as f:
+ with final.open("cwl.output.json", "w") as f:
json.dump(outputObj, f, indent=4)
api.job_tasks().update(uuid=arvados.current_task()['uuid'],
body={
- 'output': c.save_new(create_collection_record=False),
+ 'output': final.save_new(create_collection_record=False),
'success': True,
'progress':1.0
}).execute()