Merge pull request #825 from wtsi-hgi/fix-823

Fix #823 (file contents interpreted as bytes)

Author: mr-c

cwltool/builder.py

@@ -282,7 +282,7 @@ def bind_input(self,
                 self.files.append(datum)
                 if (binding and binding.get("loadContents")) or schema.get("loadContents"):
                     with self.fs_access.open(datum["location"], "rb") as f:
-                        datum["contents"] = f.read(CONTENT_LIMIT)
+                        datum["contents"] = f.read(CONTENT_LIMIT).decode("utf-8")
 
                 if "secondaryFiles" in schema:
                     if "secondaryFiles" not in datum:

cwltool/command_line_tool.py

@@ -685,7 +685,7 @@ def collect_output(self,
                             if binding.get("loadContents") or compute_checksum:
                                 contents = f.read(CONTENT_LIMIT)
                             if binding.get("loadContents"):
-                                files["contents"] = contents
+                                files["contents"] = contents.decode("utf-8") 
                             if compute_checksum:
                                 checksum = hashlib.sha1()
                                 while contents != b"":

cwltool/expression.py

@@ -19,10 +19,6 @@ def jshead(engine_config, rootvars):
 
     # make sure all the byte strings are converted
     # to str in `rootvars` dict.
-    # TODO: need to make sure the `rootvars dict`
-    # contains no bytes type in the first place.
-    if six.PY3:
-        rootvars = bytes2str_in_dicts(rootvars)  # type: ignore
 
     return u"\n".join(
         engine_config + [u"var {} = {};".format(k, json_dumps(v, indent=4))
@@ -271,6 +267,11 @@ def do_eval(ex,                       # type: Union[Text, Dict]
         u"self": context,
         u"runtime": runtime}
 
+    # TODO: need to make sure the `rootvars dict`
+    # contains no bytes type in the first place.
+    if six.PY3:
+        rootvars = bytes2str_in_dicts(rootvars)  # type: ignore
+
     if needs_parsing(ex):
         assert isinstance(ex, string_types)
         fullJS = False

cwltool/workflow.py

@@ -336,7 +336,7 @@ def postScatterEval(io):
                 for k, v in io.items():
                     if k in loadContents and v.get("contents") is None:
                         with fs_access.open(v["location"], "rb") as f:
-                            v["contents"] = f.read(CONTENT_LIMIT)
+                            v["contents"] = f.read(CONTENT_LIMIT).decode("utf-8")
 
                 def valueFromFunc(k, v):  # type: (Any, Any) -> Any
                     if k in valueFrom: