10460: Add check for 'dockerOutputDirectory' and raise UnsupportedError for now.

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

diff --git a/sdk/cwl/arvados_cwl/arvjob.py b/sdk/cwl/arvados_cwl/arvjob.py
index 759bf0bf34abc0ae8ddc1a5c7b14baccd8f15d49..8a62204f8fb9ec22298abec15529411ace70ed9e 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvjob.py
+++ b/sdk/cwl/arvados_cwl/arvjob.py
@@ -22,9 +22,7 @@ from ._version import __version__
 logger = logging.getLogger('arvados.cwl-runner')
 metrics = logging.getLogger('arvados.cwl-runner.metrics')
 
-tmpdirre = re.compile(r"^\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.tmpdir\)=(.*)")
-outdirre = re.compile(r"^\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.outdir\)=(.*)")
-keepre = re.compile(r"^\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.keep\)=(.*)")
+crunchrunner_re = re.compile(r"^\S+ \S+ \d+ \d+ stderr \S+ \S+ crunchrunner: \$\(task\.(tmpdir|outdir|keep)\)=(.*)")
 
 class ArvadosJob(object):
     """Submit and manage a Crunch job for executing a CWL CommandLineTool."""
@@ -87,6 +85,8 @@ class ArvadosJob(object):
         with Perf(metrics, "arv_docker_get_image %s" % self.name):
             (docker_req, docker_is_req) = get_feature(self, "DockerRequirement")
             if docker_req and kwargs.get("use_container") is not False:
+                if docker_req.get("dockerOutputDirectory"):
+                    raise UnsupportedRequirement("Option 'dockerOutputDirectory' of DockerRequirement not supported.")
                 runtime_constraints["docker_image"] = arv_docker_get_image(self.arvrunner.api, docker_req, pull_image, self.arvrunner.project_uuid)
             else:
                 runtime_constraints["docker_image"] = arvados_jobs_image(self.arvrunner)
@@ -184,6 +184,7 @@ 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
@@ -195,19 +196,13 @@ class ArvadosJob(object):
                             # the job restarts on a different node these values
                             # will different runs, and we need to know about the
                             # final run that actually produced output.
-
-                            g = tmpdirre.match(l)
-                            if g:
-                                tmpdir = g.group(1)
-                            g = outdirre.match(l)
-                            if g:
-                                outdir = g.group(1)
-                            g = keepre.match(l)
+                            g = crunchrunner_re.match(l)
                             if g:
-                                keepdir = g.group(1)
+                                dirs[g.group(1)] = g.group(2)
 
                     with Perf(metrics, "output collection %s" % self.name):
-                        outputs = done.done(self, record, tmpdir, outdir, keepdir)
+                        outputs = done.done(self, record, dirs["tmpdir"],
+                                            dirs["outdir"], dirs["keep"])
             except WorkflowException as e:
                 logger.error("Error while collecting job outputs:\n%s", e, exc_info=(e if self.arvrunner.debug else False))
                 processStatus = "permanentFail"