sdk/cwl/arvados_cwl/crunch_script.py

   1 # Crunch script integration for running arvados-cwl-runner (importing
   2 # arvados_cwl module) inside a crunch job.
   3 #
   4 # This gets the job record, transforms the script parameters into a valid CWL
   5 # input object, then executes the CWL runner to run the underlying workflow or
   6 # tool.  When the workflow completes, record the output object in an output
   7 # collection for this runner job.
   8
   9 import arvados
  10 import arvados_cwl
  11 import arvados.collection
  12 import arvados.util
  13 import cwltool.main
  14 import logging
  15 import os
  16 import json
  17 import argparse
  18 import re
  19 import functools
  20
  21 from arvados.api import OrderedJsonModel
  22 from cwltool.process import shortname, adjustFileObjs, adjustDirObjs, getListing, normalizeFilesDirs
  23 from cwltool.load_tool import load_tool
  24 from cwltool.errors import WorkflowException
  25
  26 logger = logging.getLogger('arvados.cwl-runner')
  27
  28 def run():
  29     # Print package versions
  30     logger.info(arvados_cwl.versionstring())
  31
  32     api = arvados.api("v1")
  33
  34     arvados_cwl.add_arv_hints()
  35
  36     runner = None
  37     try:
  38         job_order_object = arvados.current_job()['script_parameters']
  39
  40         pdh_path = re.compile(r'^[0-9a-f]{32}\+\d+(/.+)?$')
  41
  42         def keeppath(v):
  43             if pdh_path.match(v):
  44                 return "keep:%s" % v
  45             else:
  46                 return v
  47
  48         def keeppathObj(v):
  49             v["location"] = keeppath(v["location"])
  50
  51         job_order_object["cwl:tool"] = "file://%s/%s" % (os.environ['TASK_KEEPMOUNT'], job_order_object["cwl:tool"])
  52
  53         for k,v in job_order_object.items():
  54             if isinstance(v, basestring) and arvados.util.keep_locator_pattern.match(v):
  55                 job_order_object[k] = {
  56                     "class": "File",
  57                     "location": "keep:%s" % v
  58                 }
  59
  60         adjustFileObjs(job_order_object, keeppathObj)
  61         adjustDirObjs(job_order_object, keeppathObj)
  62         normalizeFilesDirs(job_order_object)
  63         adjustDirObjs(job_order_object, functools.partial(getListing, arvados_cwl.fsaccess.CollectionFsAccess("", api_client=api)))
  64
  65         output_name = None
  66         output_tags = None
  67         enable_reuse = True
  68         if "arv:output_name" in job_order_object:
  69             output_name = job_order_object["arv:output_name"]
  70             del job_order_object["arv:output_name"]
  71
  72         if "arv:output_tags" in job_order_object:
  73             output_tags = job_order_object["arv:output_tags"]
  74             del job_order_object["arv:output_tags"]
  75
  76         if "arv:enable_reuse" in job_order_object:
  77             enable_reuse = job_order_object["arv:enable_reuse"]
  78             del job_order_object["arv:enable_reuse"]
  79
  80         runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()),
  81                                           output_name=output_name, output_tags=output_tags)
  82
  83         t = load_tool(job_order_object, runner.arv_make_tool)
  84
  85         args = argparse.Namespace()
  86         args.project_uuid = arvados.current_job()["owner_uuid"]
  87         args.enable_reuse = enable_reuse
  88         args.submit = False
  89         args.debug = True
  90         args.quiet = False
  91         args.ignore_docker_for_reuse = False
  92         args.basedir = os.getcwd()
  93         args.name = None
  94         args.cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]}
  95         outputObj = runner.arv_executor(t, job_order_object, **vars(args))
  96     except Exception as e:
  97         if isinstance(e, WorkflowException):
  98             logging.info("Workflow error %s", e)
  99         else:
 100             logging.exception("Unhandled exception")
 101         if runner and runner.final_output_collection:
 102             outputCollection = runner.final_output_collection.portable_data_hash()
 103         else:
 104             outputCollection = None
 105         api.job_tasks().update(uuid=arvados.current_task()['uuid'],
 106                                              body={
 107                                                  'output': outputCollection,
 108                                                  'success': False,
 109                                                  'progress':1.0
 110                                              }).execute()