13976: Add code comment about only_real in upload_dependencies

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

diff --git a/sdk/cwl/arvados_cwl/runner.py b/sdk/cwl/arvados_cwl/runner.py
index 9c422a42f76975d300952d004f5e2849d283c2a3..29c0535d93c8a0ff164af01949ae152f056591d7 100644 (file)
--- a/sdk/cwl/arvados_cwl/runner.py
+++ b/sdk/cwl/arvados_cwl/runner.py
@@ -7,7 +7,7 @@ import urlparse
 from functools import partial
 import logging
 import json
-import subprocess
+import subprocess32 as subprocess
 from collections import namedtuple
 
 from StringIO import StringIO
@@ -129,6 +129,8 @@ def upload_dependencies(arvrunner, name, document_loader,
 
     sc = []
     def only_real(obj):
+        # Only interested in local files than need to be uploaded,
+        # don't include file literals, keep references, etc.
         if obj.get("location", "").startswith("file:"):
             sc.append(obj)
 
@@ -168,8 +170,13 @@ def upload_dependencies(arvrunner, name, document_loader,
 
     visit_class(workflowobj, ("CommandLineTool", "Workflow"), discover_default_secondary_files)
 
-    for d in discovered:
-        sc.extend(discovered[d])
+    for d in list(discovered.keys()):
+        # Only interested in discovered secondaryFiles which are local
+        # files that need to be uploaded.
+        if d.startswith("file:"):
+            sc.extend(discovered[d])
+        else:
+            del discovered[d]
 
     mapper = ArvPathMapper(arvrunner, sc, "",
                            "keep:%s",