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, normalizeFilesDirs
  23 from cwltool.load_tool import load_tool
  24 from cwltool.errors import WorkflowException
  25
  26 from .fsaccess import CollectionFetcher
  27
  28 logger = logging.getLogger('arvados.cwl-runner')
  29
  30 def run():
  31     # Timestamps are added by crunch-job, so don't print redundant timestamps.
  32     arvados.log_handler.setFormatter(logging.Formatter('%(name)s %(levelname)s: %(message)s'))
  33
  34     # Print package versions
  35     logger.info(arvados_cwl.versionstring())
  36
  37     api = arvados.api("v1")
  38
  39     arvados_cwl.add_arv_hints()
  40
  41     runner = None
  42     try:
  43         job_order_object = arvados.current_job()['script_parameters']
  44         toolpath = "file://%s/%s" % (os.environ['TASK_KEEPMOUNT'], job_order_object.pop("cwl:tool"))
  45
  46         pdh_path = re.compile(r'^[0-9a-f]{32}\+\d+(/.+)?$')
  47
  48         def keeppath(v):
  49             if pdh_path.match(v):
  50                 return "keep:%s" % v
  51             else:
  52                 return v
  53
  54         def keeppathObj(v):
  55             v["location"] = keeppath(v["location"])
  56
  57         for k,v in job_order_object.items():
  58             if isinstance(v, basestring) and arvados.util.keep_locator_pattern.match(v):
  59                 job_order_object[k] = {
  60                     "class": "File",
  61                     "location": "keep:%s" % v
  62                 }
  63
  64         adjustFileObjs(job_order_object, keeppathObj)
  65         adjustDirObjs(job_order_object, keeppathObj)
  66         normalizeFilesDirs(job_order_object)
  67
  68         output_name = None
  69         output_tags = None
  70         enable_reuse = True
  71         on_error = "continue"
  72         if "arv:output_name" in job_order_object:
  73             output_name = job_order_object["arv:output_name"]
  74             del job_order_object["arv:output_name"]
  75
  76         if "arv:output_tags" in job_order_object:
  77             output_tags = job_order_object["arv:output_tags"]
  78             del job_order_object["arv:output_tags"]
  79
  80         if "arv:enable_reuse" in job_order_object:
  81             enable_reuse = job_order_object["arv:enable_reuse"]
  82             del job_order_object["arv:enable_reuse"]
  83
  84         if "arv:on_error" in job_order_object:
  85             on_error = job_order_object["arv:on_error"]
  86             del job_order_object["arv:on_error"]
  87
  88         runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()),
  89                                           output_name=output_name, output_tags=output_tags)
  90
  91         t = load_tool(toolpath, runner.arv_make_tool,
  92                       fetcher_constructor=functools.partial(CollectionFetcher,
  93                                                             api_client=api,
  94                                                             keep_client=arvados.keep.KeepClient(api_client=api, num_retries=4)))
  95
  96         args = argparse.Namespace()
  97         args.project_uuid = arvados.current_job()["owner_uuid"]
  98         args.enable_reuse = enable_reuse
  99         args.on_error = on_error
 100         args.submit = False
 101         args.debug = False
 102         args.quiet = False
 103         args.ignore_docker_for_reuse = False
 104         args.basedir = os.getcwd()
 105         args.name = None
 106         args.cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]}
 107         runner.arv_executor(t, job_order_object, **vars(args))
 108     except Exception as e:
 109         if isinstance(e, WorkflowException):
 110             logging.info("Workflow error %s", e)
 111         else:
 112             logging.exception("Unhandled exception")
 113         if runner and runner.final_output_collection:
 114             outputCollection = runner.final_output_collection.portable_data_hash()
 115         else:
 116             outputCollection = None
 117         api.job_tasks().update(uuid=arvados.current_task()['uuid'],
 118                                              body={
 119                                                  'output': outputCollection,
 120                                                  'success': False,
 121                                                  'progress':1.0
 122                                              }).execute()