converting hash_value to a standalone function; edits in is_existing_file

Author: djarecka

pydra/engine/helpers.py

@@ -10,6 +10,7 @@
 import subprocess as sp
 
 from .specs import Runtime, File, attr_fields
+from .helpers_file import is_existing_file, hash_file
 
 
 def ensure_list(obj, tuple2list=False):
@@ -396,6 +397,27 @@ def hash_function(obj):
     return sha256(str(obj).encode()).hexdigest()
 
 
+def hash_value(value, tp=None, metadata=None):
+    """calculating hash or returning values recursively"""
+    if isinstance(value, (tuple, list)):
+        return [hash_value(el, tp, metadata) for el in value]
+    elif isinstance(value, dict):
+        dict_hash = {k: hash_value(v, tp, metadata) for (k, v) in value.items()}
+        # returning a sorted object
+        return sorted(dict_hash.items(), key=lambda x: x[0])
+    else:  # not a container
+        if (
+            (tp is File or "pydra.engine.specs.File" in str(tp))
+            and is_existing_file(value)
+            and "container_path" not in metadata
+        ):
+            return hash_file(value)
+        elif isinstance(value, tuple):
+            return list(value)
+        else:
+            return value
+
+
 def output_names_from_inputfields(inputs):
     """
     Collect outputs from input fields with output_file_template.

pydra/engine/helpers_file.py

@@ -514,12 +514,9 @@ def is_local_file(f):
     return f.type is File and "container_path" not in f.metadata
 
 
-def is_existed_file(f):
-    """ checking if object is an existed file: using multiple path classes"""
-    from py import path
-
-    # for windows might need more path options
-    if isinstance(f, (str, Path, path.local)) and Path(f).exists():
-        return True
-    else:
+def is_existing_file(f):
+    """ checking if an object is an existing file"""
+    try:
+        return Path(f).exists()
+    except TypeError:
         return False

pydra/engine/specs.py

@@ -2,7 +2,6 @@
 import attr
 from pathlib import Path
 import typing as ty
-from .helpers_file import hash_file, is_existed_file
 
 
 def attr_fields(x):
@@ -40,15 +39,15 @@ def collect_additional_outputs(self, input_spec, inputs, output_dir):
     @property
     def hash(self):
         """Compute a basic hash for any given set of fields."""
-        from .helpers import hash_function
+        from .helpers import hash_value, hash_function
 
         inp_dict = {}
         for field in attr_fields(self):
             if field.name in ["_graph_checksums", "bindings"] or field.metadata.get(
                 "output_file_template"
             ):
                 continue
-            inp_dict[field.name] = self.hash_value(
+            inp_dict[field.name] = hash_value(
                 value=getattr(self, field.name), tp=field.type, metadata=field.metadata
             )
         inp_hash = hash_function(inp_dict)
@@ -57,28 +56,6 @@ def hash(self):
         else:
             return inp_hash
 
-    def hash_value(self, value, tp=None, metadata=None):
-        """calculating hash or returning values recursively"""
-        if isinstance(value, (tuple, list)):
-            return [self.hash_value(el, tp, metadata) for el in value]
-        elif isinstance(value, dict):
-            dict_hash = {
-                k: self.hash_value(v, tp, metadata) for (k, v) in value.items()
-            }
-            # returning a sorted object
-            return sorted(dict_hash.items(), key=lambda x: x[0])
-        else:  # not a container
-            if (
-                (tp is File or "pydra.engine.specs.File" in str(tp))
-                and is_existed_file(value)
-                and "container_path" not in metadata
-            ):
-                return hash_file(value)
-            elif isinstance(value, tuple):
-                return list(value)
-            else:
-                return value
-
     def retrieve_values(self, wf, state_index=None):
         """Get values contained by this spec."""
         temp_values = {}