Merge branch '12195-nodemanager-quota-error'
[arvados.git] / sdk / cwl / arvados_cwl / arvjob.py
index 877f4238a33e8f41952f37793c4d1c168da89ca0..794c99446c7bdf5ce442d714816d93f150a623ce 100644 (file)
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import logging
 import re
 import copy
@@ -9,7 +13,7 @@ from cwltool.errors import WorkflowException
 from cwltool.draft2tool import revmap_file, CommandLineTool
 from cwltool.load_tool import fetch_document
 from cwltool.builder import Builder
-from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
+from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, visit_class
 
 from schema_salad.sourceline import SourceLine
 
@@ -19,7 +23,7 @@ import arvados.collection
 from arvados.errors import ApiError
 
 from .arvdocker import arv_docker_get_image
-from .runner import Runner, arvados_jobs_image, packed_workflow, upload_workflow_collection, trim_anonymous_location
+from .runner import Runner, arvados_jobs_image, packed_workflow, upload_workflow_collection, trim_anonymous_location, remove_redundant_fields
 from .pathmapper import VwdPathMapper, trim_listing
 from .perf import Perf
 from . import done
@@ -28,7 +32,7 @@ from ._version import __version__
 logger = logging.getLogger('arvados.cwl-runner')
 metrics = logging.getLogger('arvados.cwl-runner.metrics')
 
-crunchrunner_re = re.compile(r"^\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.(tmpdir|outdir|keep)\)=(.*)")
+crunchrunner_re = re.compile(r"^.*crunchrunner: \$\(task\.(tmpdir|outdir|keep)\)=(.*)$")
 
 crunchrunner_git_commit = 'a3f2cb186e437bfce0031b024b2157b73ed2717d'
 
@@ -218,12 +222,13 @@ class ArvadosJob(object):
                                                                    keep_client=self.arvrunner.keep_client,
                                                                    num_retries=self.arvrunner.num_retries)
                         log = logc.open(logc.keys()[0])
-                        dirs = {}
-                        tmpdir = None
-                        outdir = None
-                        keepdir = None
+                        dirs = {
+                            "tmpdir": "/tmpdir",
+                            "outdir": "/outdir",
+                            "keep": "/keep"
+                        }
                         for l in log:
-                            # Determine the tmpdir, outdir and keepdir paths from
+                            # Determine the tmpdir, outdir and keep paths from
                             # the job run.  Unfortunately, we can't take the first
                             # values we find (which are expected to be near the
                             # top) and stop scanning because if the node fails and
@@ -278,8 +283,8 @@ class RunnerJob(Runner):
             self.job_order["cwl:tool"] = "%s/workflow.cwl#main" % wf_pdh
 
         adjustDirObjs(self.job_order, trim_listing)
-        adjustFileObjs(self.job_order, trim_anonymous_location)
-        adjustDirObjs(self.job_order, trim_anonymous_location)
+        visit_class(self.job_order, ("File", "Directory"), trim_anonymous_location)
+        visit_class(self.job_order, ("File", "Directory"), remove_redundant_fields)
 
         if self.output_name:
             self.job_order["arv:output_name"] = self.output_name