14198: Bringing unit tests up to date WIP
[arvados.git] / sdk / cwl / arvados_cwl / crunch_script.py
index 500fbdd0d7aeb87b6db1cdc1f123b4e438fe3345..7512d5bef27f28014f650d897d24e3d59cb7b3c4 100644 (file)
@@ -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.
 #
@@ -19,9 +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 arvados_cwl.context import ArvRuntimeContext
+
+from .fsaccess import CollectionFetcher, CollectionFsAccess
 
 logger = logging.getLogger('arvados.cwl-runner')
 
@@ -50,7 +58,8 @@ 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):
@@ -62,12 +71,13 @@ 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"
+        debug = False
+
         if "arv:output_name" in job_order_object:
             output_name = job_order_object["arv:output_name"]
             del job_order_object["arv:output_name"]
@@ -84,23 +94,50 @@ def run():
             on_error = job_order_object["arv:on_error"]
             del job_order_object["arv:on_error"]
 
-        runner = arvados_cwl.ArvCwlRunner(api_client=arvados.api('v1', model=OrderedJsonModel()),
-                                          output_name=output_name, output_tags=output_tags)
+        if "arv:debug" in job_order_object:
+            debug = job_order_object["arv:debug"]
+            del job_order_object["arv:debug"]
+
+        arvargs = argparse.Namespace()
+        arvargs.work_api = "jobs"
+        arvargs.output_name = output_name
+        arvargs.output_tags = output_tags
+        arvargs.thread_count = 1
+
+        runner = arvados_cwl.ArvCwlExecutor(api_client=arvados.safeapi.ThreadSafeApiCache(
+            api_params={"model": OrderedJsonModel()}, keep_params={"num_retries": 4}),
+                                          arvargs=arvargs)
+
+        make_fs_access = functools.partial(CollectionFsAccess,
+                                 collection_cache=runner.collection_cache)
+
+        t = load_tool(toolpath, runner.loadingContext)
 
-        t = load_tool(toolpath, runner.arv_make_tool)
+        if debug:
+            logger.setLevel(logging.DEBUG)
+            logging.getLogger('arvados').setLevel(logging.DEBUG)
+            logging.getLogger("cwltool").setLevel(logging.DEBUG)
 
-        args = argparse.Namespace()
+        args = ArvRuntimeContext(vars(arvargs))
         args.project_uuid = arvados.current_job()["owner_uuid"]
         args.enable_reuse = enable_reuse
         args.on_error = on_error
         args.submit = False
-        args.debug = False
+        args.debug = debug
         args.quiet = False
         args.ignore_docker_for_reuse = False
         args.basedir = os.getcwd()
         args.name = None
         args.cwl_runner_job={"uuid": arvados.current_job()["uuid"], "state": arvados.current_job()["state"]}
-        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
+        args.priority = arvados_cwl.DEFAULT_PRIORITY
+        args.do_validate = True
+        args.disable_js_validation = False
+        args.tmp_outdir_prefix = "tmp"
+
+        runner.arv_executor(t, job_order_object, args, logger=logger)
     except Exception as e:
         if isinstance(e, WorkflowException):
             logging.info("Workflow error %s", e)