X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/6c0bf267d795a3ca49c3258c9490714c9e18d333..cc3857d9fdcd7588af329a9b2452e0ef09eb68a4:/sdk/cwl/arvados_cwl/runner.py

diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py
index a4132ca3f7..49d37ebd5a 100644
--- a/sdk/cwl/arvados_cwl/runner.py
+++ b/sdk/cwl/arvados_cwl/runner.py
@@ -18,6 +18,7 @@ import ruamel.yaml as yaml
 
 from .arvdocker import arv_docker_get_image
 from .pathmapper import ArvPathMapper
+from ._version import __version__
 
 logger = logging.getLogger('arvados.cwl-runner')
 
@@ -112,46 +113,56 @@ def upload_docker(arvrunner, tool):
         for s in tool.steps:
             upload_docker(arvrunner, s.embedded_tool)
 
+def upload_instance(arvrunner, name, tool, job_order):
+        upload_docker(arvrunner, tool)
+
+        workflowmapper = upload_dependencies(arvrunner,
+                                             name,
+                                             tool.doc_loader,
+                                             tool.tool,
+                                             tool.tool["id"],
+                                             True)
+
+        jobmapper = upload_dependencies(arvrunner,
+                                        os.path.basename(job_order.get("id", "#")),
+                                        tool.doc_loader,
+                                        job_order,
+                                        job_order.get("id", "#"),
+                                        False)
+
+        if "id" in job_order:
+            del job_order["id"]
+
+        return workflowmapper
+
+def arvados_jobs_image(arvrunner):
+    img = "arvados/jobs:"+__version__
+    try:
+        arv_docker_get_image(arvrunner.api, {"dockerPull": img}, True, arvrunner.project_uuid)
+    except Exception as e:
+        raise Exception("Docker image %s is not available\n%s" % (img, e) )
+    return img
 
 class Runner(object):
-    def __init__(self, runner, tool, job_order, enable_reuse):
+    def __init__(self, runner, tool, job_order, enable_reuse, output_name):
         self.arvrunner = runner
         self.tool = tool
         self.job_order = job_order
         self.running = False
         self.enable_reuse = enable_reuse
         self.uuid = None
+        self.final_output = None
+        self.output_name = output_name
 
     def update_pipeline_component(self, record):
         pass
 
     def arvados_job_spec(self, *args, **kwargs):
-        upload_docker(self.arvrunner, self.tool)
-
         self.name = os.path.basename(self.tool.tool["id"])
-
-        workflowmapper = upload_dependencies(self.arvrunner,
-                                             self.name,
-                                             self.tool.doc_loader,
-                                             self.tool.tool,
-                                             self.tool.tool["id"],
-                                             True)
-
-        jobmapper = upload_dependencies(self.arvrunner,
-                                        os.path.basename(self.job_order.get("id", "#")),
-                                        self.tool.doc_loader,
-                                        self.job_order,
-                                        self.job_order.get("id", "#"),
-                                        False)
-
+        workflowmapper = upload_instance(self.arvrunner, self.name, self.tool, self.job_order)
         adjustDirObjs(self.job_order, trim_listing)
-
-        if "id" in self.job_order:
-            del self.job_order["id"]
-
         return workflowmapper
 
-
     def done(self, record):
         if record["state"] == "Complete":
             if record.get("exit_code") is not None:
@@ -169,7 +180,11 @@ class Runner(object):
         outputs = None
         try:
             try:
-                outc = arvados.collection.Collection(record["output"])
+                self.final_output = record["output"]
+                outc = arvados.collection.CollectionReader(self.final_output,
+                                                           api_client=self.arvrunner.api,
+                                                           keep_client=self.arvrunner.keep_client,
+                                                           num_retries=self.arvrunner.num_retries)
                 with outc.open("cwl.output.json") as f:
                     outputs = json.load(f)
                 def keepify(fileobj):