cwltool: do not read all file contents when compute_checksum is not set.

Author: radhika

cwltool/draft2tool.py

@@ -107,10 +107,13 @@ def __init__(self, job, output_callback, cachebuilder, jobcache):
 
     def run(self, **kwargs):
         # type: (**Any) -> None
+        compute_checksum = kwargs.get("compute_checksum")
+        if compute_checksum == None:
+            compute_checksum = True
         self.output_callback(self.job.collect_output_ports(self.job.tool["outputs"],
                                                            self.cachebuilder,
                                                            self.outdir,
-                                                           kwargs.get("compute_checksum")),
+                                                           compute_checksum),
                                             "success")
 
 # map files to assigned path inside a container. We need to also explicitly
@@ -334,9 +337,12 @@ def rm_pending_output_callback(output_callback, jobcachepending,
         else:
             j.command_line = flatten(map(builder.generate_arg, builder.bindings))
 
+        compute_checksum = kwargs.get("compute_checksum")
+        if compute_checksum == None:
+            compute_checksum = True
         j.pathmapper = builder.pathmapper
         j.collect_outputs = partial(
-                self.collect_output_ports, self.tool["outputs"], builder, compute_checksum=kwargs.get("compute_checksum"))
+                self.collect_output_ports, self.tool["outputs"], builder, compute_checksum=compute_checksum)
         j.output_callback = output_callback
 
         yield j
@@ -409,19 +415,21 @@ def collect_output(self, schema, builder, outdir, compute_checksum=True):
                     if files["class"] == "Directory" and "listing" not in files:
                         getListing(builder.fs_access, files)
                     else:
-                        checksum = hashlib.sha1()
                         with builder.fs_access.open(files["location"], "rb") as f:
+                            filesize = 0
                             contents = f.read(CONTENT_LIMIT)
                             if binding.get("loadContents"):
                                 files["contents"] = contents
-                            filesize = 0
-                            while contents != "":
-                                if compute_checksum:
+                            if compute_checksum:
+                                checksum = hashlib.sha1()
+                                while contents != "":
                                     checksum.update(contents)
-                                filesize += len(contents)
-                                contents = f.read(1024*1024)
-                        if compute_checksum:
-                            files["checksum"] = "sha1$%s" % checksum.hexdigest()
+                                    filesize += len(contents)
+                                    contents = f.read(1024*1024)
+                                files["checksum"] = "sha1$%s" % checksum.hexdigest()
+                            else:
+                                f.seek(0, 2)
+                                filesize = f.tell()
                         files["size"] = filesize
                         if "format" in schema:
                             files["format"] = builder.do_eval(schema["format"], context=files)