Merge branch '8784-dir-listings'

[arvados.git] / sdk / cwl / arvados_cwl / crunch_script.py

diff --git a/sdk/cwl/arvados_cwl/crunch_script.py b/sdk/cwl/arvados_cwl/crunch_script.py
index 71c65bca88c17e1196b7f9280646be7b83bea500..7fbbd29d50d72f385b5e1ea95f2b093405518baf 100644 (file)
--- a/sdk/cwl/arvados_cwl/crunch_script.py
+++ b/sdk/cwl/arvados_cwl/crunch_script.py
@@ -1,3 +1,7 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 # Crunch script integration for running arvados-cwl-runner (importing
 # arvados_cwl module) inside a crunch job.
 #
@@ -19,10 +23,13 @@ import re
 import functools
 
 from arvados.api import OrderedJsonModel
-from cwltool.process import shortname, adjustFileObjs, adjustDirObjs, getListing, normalizeFilesDirs
+from cwltool.process import shortname
+from cwltool.pathmapper import adjustFileObjs, adjustDirObjs, normalizeFilesDirs
 from cwltool.load_tool import load_tool
 from cwltool.errors import WorkflowException
 
+from .fsaccess import CollectionFetcher, CollectionFsAccess
+
 logger = logging.getLogger('arvados.cwl-runner')
 
 def run():
@@ -50,7 +57,8 @@ def run():
                 return v
 
         def keeppathObj(v):
-            v["location"] = keeppath(v["location"])
+            if "location" in v:
+                v["location"] = keeppath(v["location"])
 
         for k,v in job_order_object.items():
             if isinstance(v, basestring) and arvados.util.keep_locator_pattern.match(v):
@@ -62,11 +70,11 @@ def run():
         adjustFileObjs(job_order_object, keeppathObj)
         adjustDirObjs(job_order_object, keeppathObj)
         normalizeFilesDirs(job_order_object)
-        adjustDirObjs(job_order_object, functools.partial(getListing, arvados_cwl.fsaccess.CollectionFsAccess("", api_client=api)))
 
         output_name = None
         output_tags = None
         enable_reuse = True
+        on_error = "continue"
         if "arv:output_name" in job_order_object:
             output_name = job_order_object["arv:output_name"]
             del job_order_object["arv:output_name"]
@@ -86,7 +94,14 @@ def run():
         runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()),
                                           output_name=output_name, output_tags=output_tags)
 
-        t = load_tool(toolpath, runner.arv_make_tool)
+        make_fs_access = functools.partial(CollectionFsAccess,
+                                 collection_cache=runner.collection_cache)
+
+        t = load_tool(toolpath, runner.arv_make_tool,
+                      fetcher_constructor=functools.partial(CollectionFetcher,
+                                                  api_client=runner.api,
+                                                  fs_access=make_fs_access(""),
+                                                  num_retries=runner.num_retries))
 
         args = argparse.Namespace()
         args.project_uuid = arvados.current_job()["owner_uuid"]
@@ -99,7 +114,11 @@ def run():
         args.basedir = os.getcwd()
         args.name = None
         args.cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]}
-        outputObj = runner.arv_executor(t, job_order_object, **vars(args))
+        args.make_fs_access = make_fs_access
+        args.trash_intermediate = False
+        args.intermediate_output_ttl = 0
+
+        runner.arv_executor(t, job_order_object, **vars(args))
     except Exception as e:
         if isinstance(e, WorkflowException):
             logging.info("Workflow error %s", e)