9552: import compute_checksums from cwltool.

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

diff --git a/sdk/cwl/arvados_cwl/__init__.py b/sdk/cwl/arvados_cwl/__init__.py
index fd9e74f8575167f7aa003ef32d9c449596810408..fe3249cc026dc51892b0a3e5e247332d679700e5 100644 (file)
--- a/sdk/cwl/arvados_cwl/__init__.py
+++ b/sdk/cwl/arvados_cwl/__init__.py
@@ -27,6 +27,7 @@ from .fsaccess import CollectionFsAccess
 
 from cwltool.process import shortname, UnsupportedRequirement
 from cwltool.pathmapper import adjustFileObjs
+from cwltool.draft2tool import compute_checksums
 from arvados.api import OrderedJsonModel
 
 logger = logging.getLogger('arvados.cwl-runner')
@@ -114,6 +115,8 @@ class ArvCwlRunner(object):
         useruuid = self.api.users().current().execute()["uuid"]
         self.project_uuid = kwargs.get("project_uuid") if kwargs.get("project_uuid") else useruuid
         self.pipeline = None
+        make_fs_access = kwargs.get("make_fs_access") or partial(CollectionFsAccess, api_client=self.api)
+        self.fs_access = make_fs_access(kwargs["basedir"])
 
         if kwargs.get("create_template"):
             tmpl = RunnerTemplate(self, tool, job_order, kwargs.get("enable_reuse"))
@@ -124,8 +127,6 @@ class ArvCwlRunner(object):
         self.debug = kwargs.get("debug")
         self.ignore_docker_for_reuse = kwargs.get("ignore_docker_for_reuse")
 
-        make_fs_access = kwargs.get("make_fs_access") or partial(CollectionFsAccess, api_client=self.api)
-        self.fs_access = make_fs_access(kwargs["basedir"])
         kwargs["make_fs_access"] = make_fs_access
         kwargs["enable_reuse"] = kwargs.get("enable_reuse")
         kwargs["use_container"] = True
@@ -230,17 +231,7 @@ class ArvCwlRunner(object):
             raise WorkflowException("Workflow did not return a result.")
 
         if kwargs.get("compute_checksum"):
-            def compute_checksums(fileobj):
-                if "checksum" not in fileobj:
-                    checksum = hashlib.sha1()
-                    with self.fs_access.open(fileobj["location"], "rb") as f:
-                        contents = f.read(1024*1024)
-                        while contents != "":
-                            checksum.update(contents)
-                            contents = f.read(1024*1024)
-                    fileobj["checksum"] = "sha1$%s" % checksum.hexdigest()
-
-            adjustFileObjs(self.final_output, compute_checksums)
+            adjustFileObjs(self.final_output, partial(compute_checksums, self.fs_access))
 
         return self.final_output