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, CollectionFsAccess
  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             if "location" in v:
  56                 v["location"] = keeppath(v["location"])
  57
  58         for k,v in job_order_object.items():
  59             if isinstance(v, basestring) and arvados.util.keep_locator_pattern.match(v):
  60                 job_order_object[k] = {
  61                     "class": "File",
  62                     "location": "keep:%s" % v
  63                 }
  64
  65         adjustFileObjs(job_order_object, keeppathObj)
  66         adjustDirObjs(job_order_object, keeppathObj)
  67         normalizeFilesDirs(job_order_object)
  68
  69         output_name = None
  70         output_tags = None
  71         enable_reuse = True
  72         on_error = "continue"
  73         if "arv:output_name" in job_order_object:
  74             output_name = job_order_object["arv:output_name"]
  75             del job_order_object["arv:output_name"]
  76
  77         if "arv:output_tags" in job_order_object:
  78             output_tags = job_order_object["arv:output_tags"]
  79             del job_order_object["arv:output_tags"]
  80
  81         if "arv:enable_reuse" in job_order_object:
  82             enable_reuse = job_order_object["arv:enable_reuse"]
  83             del job_order_object["arv:enable_reuse"]
  84
  85         if "arv:on_error" in job_order_object:
  86             on_error = job_order_object["arv:on_error"]
  87             del job_order_object["arv:on_error"]
  88
  89         runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()),
  90                                           output_name=output_name, output_tags=output_tags)
  91
  92         make_fs_access = functools.partial(CollectionFsAccess,
  93                                  collection_cache=runner.collection_cache)
  94
  95         t = load_tool(toolpath, runner.arv_make_tool,
  96                       fetcher_constructor=functools.partial(CollectionFetcher,
  97                                                   api_client=runner.api,
  98                                                   fs_access=make_fs_access(""),
  99                                                   num_retries=runner.num_retries))
 100
 101         args = argparse.Namespace()
 102         args.project_uuid = arvados.current_job()["owner_uuid"]
 103         args.enable_reuse = enable_reuse
 104         args.on_error = on_error
 105         args.submit = False
 106         args.debug = False
 107         args.quiet = False
 108         args.ignore_docker_for_reuse = False
 109         args.basedir = os.getcwd()
 110         args.name = None
 111         args.cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]}
 112         args.make_fs_access = make_fs_access
 113
 114         runner.arv_executor(t, job_order_object, **vars(args))
 115     except Exception as e:
 116         if isinstance(e, WorkflowException):
 117             logging.info("Workflow error %s", e)
 118         else:
 119             logging.exception("Unhandled exception")
 120         if runner and runner.final_output_collection:
 121             outputCollection = runner.final_output_collection.portable_data_hash()
 122         else:
 123             outputCollection = None
 124         api.job_tasks().update(uuid=arvados.current_task()['uuid'],
 125                                              body={
 126                                                  'output': outputCollection,
 127                                                  'success': False,
 128                                                  'progress':1.0
 129                                              }).execute()