X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/c9b5db991553e3a6eb6ec1f268e94f4a7b2a64d0..0561bd0c3c07257fd58ded6c7cfa5feeae97af57:/sdk/cwl/arvados_cwl/arvdocker.py

diff --git a/sdk/cwl/arvados_cwl/arvdocker.py b/sdk/cwl/arvados_cwl/arvdocker.py
index 3cc1a42ae2..f42e6d8c9c 100644
--- a/sdk/cwl/arvados_cwl/arvdocker.py
+++ b/sdk/cwl/arvados_cwl/arvdocker.py
@@ -1,37 +1,79 @@
+# Copyright (C) The Arvados Authors. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
 import logging
+import sys
+import threading
+import copy
+
+from schema_salad.sourceline import SourceLine
+
 import cwltool.docker
+from cwltool.errors import WorkflowException
 import arvados.commands.keepdocker
-import sys
 
 logger = logging.getLogger('arvados.cwl-runner')
 
+cached_lookups = {}
+cached_lookups_lock = threading.Lock()
+
 def arv_docker_get_image(api_client, dockerRequirement, pull_image, project_uuid):
     """Check if a Docker image is available in Keep, if not, upload it using arv-keepdocker."""
 
     if "dockerImageId" not in dockerRequirement and "dockerPull" in dockerRequirement:
+        dockerRequirement = copy.deepcopy(dockerRequirement)
         dockerRequirement["dockerImageId"] = dockerRequirement["dockerPull"]
+        if hasattr(dockerRequirement, 'lc'):
+            dockerRequirement.lc.data["dockerImageId"] = dockerRequirement.lc.data["dockerPull"]
+
+    global cached_lookups
+    global cached_lookups_lock
+    with cached_lookups_lock:
+        if dockerRequirement["dockerImageId"] in cached_lookups:
+            return dockerRequirement["dockerImageId"]
+
+    with SourceLine(dockerRequirement, "dockerImageId", WorkflowException):
+        sp = dockerRequirement["dockerImageId"].split(":")
+        image_name = sp[0]
+        image_tag = sp[1] if len(sp) > 1 else None
+
+        images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
+                                                                image_name=image_name,
+                                                                image_tag=image_tag)
+
+        if not images:
+            # Fetch Docker image if necessary.
+            cwltool.docker.get_image(dockerRequirement, pull_image)
 
-    sp = dockerRequirement["dockerImageId"].split(":")
-    image_name = sp[0]
-    image_tag = sp[1] if len(sp) > 1 else None
+            # Upload image to Arvados
+            args = []
+            if project_uuid:
+                args.append("--project-uuid="+project_uuid)
+            args.append(image_name)
+            if image_tag:
+                args.append(image_tag)
+            logger.info("Uploading Docker image %s", ":".join(args[1:]))
+            try:
+                arvados.commands.keepdocker.main(args, stdout=sys.stderr)
+            except SystemExit as e:
+                if e.code:
+                    raise WorkflowException("keepdocker exited with code %s" % e.code)
 
-    images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
-                                                            image_name=image_name,
-                                                            image_tag=image_tag)
+            images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
+                                                                    image_name=image_name,
+                                                                    image_tag=image_tag)
 
-    if not images:
-        imageId = cwltool.docker.get_image(dockerRequirement, pull_image)
-        args = ["--project-uuid="+project_uuid, image_name]
-        if image_tag:
-            args.append(image_tag)
-        logger.info("Uploading Docker image %s", ":".join(args[1:]))
-        arvados.commands.keepdocker.main(args, stdout=sys.stderr)
+        if not images:
+            raise WorkflowException("Could not find Docker image %s:%s" % (image_name, image_tag))
 
-    images = arvados.commands.keepdocker.list_images_in_arv(api_client, 3,
-                                                            image_name=image_name,
-                                                            image_tag=image_tag)
+        with cached_lookups_lock:
+            cached_lookups[dockerRequirement["dockerImageId"]] = True
 
-    #return dockerRequirement["dockerImageId"]
+    return dockerRequirement["dockerImageId"]
 
-    pdh = api_client.collections().get(uuid=images[0][0]).execute()["portable_data_hash"]
-    return pdh
+def arv_docker_clear_cache():
+    global cached_lookups
+    global cached_lookups_lock
+    with cached_lookups_lock:
+        cached_lookups = {}