bugfix. not returning group or dataset from attrs

Author: matthiasprobst

h5rdmtoolbox/convention/toolbox_validators.py

@@ -1,12 +1,12 @@
 """general validation functions of the toolbox usable by convention. If users wish to user
 their own validators, they need to define them separately. The respective python script then
 must be provided during initialization of a Convention"""
+import re
+from typing import Union, Dict
 
 import pint
-import re
 import typing_extensions
 from pydantic.functional_validators import WrapValidator
-from typing import Union, Dict
 from typing_extensions import Annotated
 
 from h5rdmtoolbox import get_ureg, errors
@@ -70,7 +70,7 @@ def __validate_standard_name(value, handler, info) -> "StandardName":
         data_scale = parent.attrs.get('DATA_SCALE', None)
         if data_scale is not None:
             # scale_ds = parent.rootparent[data_scale]
-            ds_scale_units = data_scale.attrs.raw.get('units', '')
+            ds_scale_units = parent.rootparent[data_scale].attrs.get('units', '')
             ureg = get_ureg()
             units = str(ureg.Unit(ds_scale_units) * units)
 

h5rdmtoolbox/wrapper/h5attr.py

@@ -86,17 +86,6 @@ def __str__(self) -> str:
         return self.__repr__()
 
 
-# def Attribute(value, definition: Optional[str] = None,
-#               rdf_predicate=None, rdf_object=None) -> AttributeValue:
-#     """attribute interface function"""
-#     return AttributeValue(
-#         value,
-#         definition,
-#         rdf_predicate,
-#         rdf_object
-#     )
-
-
 def pop_hdf_attributes(attrs: Dict) -> Dict:
     """Remove HDF attributes like NAME, CLASS, .... from the input dictionary
 
@@ -156,29 +145,29 @@ def _parse_return_value(_id, ret):
                     if isinstance(v, str):
                         if not v:
                             dictionary[k] = ''
-                        else:
-                            if v[0] == '/':
-                                if isinstance(_id, h5py.h5g.GroupID):
-                                    rootgrp = get_rootparent(h5py.Group(_id))
-                                    dictionary[k] = rootgrp.get(v)
-                                elif isinstance(_id, h5py.h5d.DatasetID):
-                                    rootgrp = get_rootparent(h5py.Dataset(_id).parent)
-                                    dictionary[k] = rootgrp.get(v)
+                        # else:
+                        #     if v[0] == '/':
+                        #         if isinstance(_id, h5py.h5g.GroupID):
+                        #             rootgrp = get_rootparent(h5py.Group(_id))
+                        #             dictionary[k] = rootgrp.get(v)
+                        #         elif isinstance(_id, h5py.h5d.DatasetID):
+                        #             rootgrp = get_rootparent(h5py.Dataset(_id).parent)
+                        #             dictionary[k] = rootgrp.get(v)
                 return dictionary
-            if ret[0] == '/':
-                # it may be group or dataset path or actually just a filepath stored by the user
-                if isinstance(_id, h5py.h5g.GroupID):
-                    # call like this, otherwise recursive call!
-                    from .core import Group
-                    rootgrp = get_rootparent(Group(_id))
-                    if rootgrp.get(ret) is None:
-                        # not a dataset or group, maybe just a filename that has been stored
-                        return ret
-                    return rootgrp.get(ret)
-                else:
-                    from .core import Dataset
-                    rootgrp = get_rootparent(Dataset(_id).parent)
-                    return rootgrp.get(ret)
+            # if ret[0] == '/':
+            #     # it may be group or dataset path or actually just a filepath stored by the user
+            #     if isinstance(_id, h5py.h5g.GroupID):
+            #         # call like this, otherwise recursive call!
+            #         from .core import Group
+            #         rootgrp = get_rootparent(Group(_id))
+            #         if rootgrp.get(ret) is None:
+            #             # not a dataset or group, maybe just a filename that has been stored
+            #             return ret
+            #         return rootgrp.get(ret)
+            #     else:
+            #         from .core import Dataset
+            #         rootgrp = get_rootparent(Dataset(_id).parent)
+            #         return rootgrp.get(ret)
             if ret[0] == '(':
                 if ret[-1] == ')':
                     # might be a tuple object

tests/wrapper/test_core.py

@@ -147,6 +147,16 @@ def test_setattr(self):
             h5.attrs['none_attr'] = None
             self.assertEqual(h5.attrs['none_attr'], 'None')
 
+    def test_filename_attr(self):
+        with h5tbx.File() as h5:
+
+            h5.create_dataset('ds', data=np.arange(10))
+            h5['ds'].attrs['filename'] = '/path/to/file.csv'
+            self.assertEqual(h5['ds'].attrs['filename'], '/path/to/file.csv')
+
+            h5.attrs['filename'] = '/path/to/file.csv'
+            self.assertEqual(h5.attrs['filename'], '/path/to/file.csv')
+
     def test_rootparent(self):
         with h5tbx.File() as h5:
             g = h5.create_group('g')
@@ -541,15 +551,15 @@ def test_special_attribute_types(self):
 
                 # testing links:
                 obj.attrs['link_to_group'] = h5['/']
-                self.assertEqual(obj.attrs['link_to_group'], h5['/'])
-                self.assertIsInstance(obj.attrs['link_to_group'], h5py.Group)
+                self.assertEqual(obj.attrs['link_to_group'], '/')
+                self.assertIsInstance(obj.attrs['link_to_group'], str)
                 obj.attrs['link_to_ds'] = ds
                 self.assertEqual(obj.attrs['link_to_ds'], ds)
-                self.assertIsInstance(obj.attrs['link_to_ds'], h5py.Dataset)
+                self.assertIsInstance(obj.attrs['link_to_ds'], str)
                 obj.attrs['attribute_of_links_to_ds'] = {'ds': ds, 'grp': grp, 'astr': 'test', 'afloat': 3.1}
                 self.assertIsInstance(obj.attrs['attribute_of_links_to_ds'], dict)
-                self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['ds'], h5py.Dataset)
-                self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['grp'], h5py.Group)
+                self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['ds'], str)
+                self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['grp'], str)
                 self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['astr'], str)
                 self.assertIsInstance(obj.attrs['attribute_of_links_to_ds']['afloat'], float)
 
@@ -920,7 +930,7 @@ def test_attr_group_link_and_xarray(self):
 
             ds.attrs['link to grp'] = grp
 
-            self.assertIsInstance(ds.attrs['link to grp'], h5py.Group)
+            self.assertIsInstance(ds.attrs['link to grp'], str)
 
             da = ds[()]
             self.assertIsInstance(da, xr.DataArray)

tests/wrapper/test_rdf.py

@@ -252,7 +252,7 @@ def test_set_single_PSO(self):
             method_grp.rdf.subject = M4I.Method
             method_grp.attrs['has_participants', OBO.RO_0000057] = h5['contact_person']  # has participants
 
-            self.assertEqual(method_grp.attrs['has_participants'], h5['contact_person'])
+            self.assertEqual(method_grp.attrs['has_participants'], h5['contact_person'].name)
 
             grp.attrs['arbitrary'] = 'arbitrary'
             self.assertEqual(grp.rdf.predicate.get('arbitrary', 'invalid'), 'invalid')