move to bald__latest

Author: marqh

lib/bald/__init__.py

@@ -135,7 +135,7 @@ def validate_hdf5(afilepath):
             sattrs = dict(fhandle.attrs).copy()
             sattrs.update(dataset.attrs)
             dset = Subject(sattrs)
-            dset_val = bv.DatasetValidation(name, dataset, fhandle=fhandle,
+            dset_val = bv.ArrayValidation(name, dataset, fhandle=fhandle,
                                             subject=dset)
             sval.stored_exceptions += dset_val.exceptions()
 

lib/bald/tests/integration/test_netcdf.py

@@ -8,11 +8,21 @@
 from bald.tests import BaldTestCase
 
 def _fattrs(f):
-    f.bald__ = 'http://binary-array-ld.net/experimental/'
+    f.bald__ = 'http://binary-array-ld.net/latest/'
     f.rdf__ = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
     f.rdf__type =  'bald__Container'
     return f
 
+# def _create_parent_child(f, pshape, cshape):
+#     dsetp = f.create_dataset("parent_dataset", pshape, dtype='i')
+#     dsetc = f.create_dataset("child_dataset", cshape, dtype='i')
+#     dsetp.attrs['rdf__type'] = 'bald__Dataset'
+#     dsetp.attrs['bald__references'] = dsetc.ref
+#     dsetc.attrs['rdf__type'] = 'bald__Dataset'
+#     dsetc.attrs['rdf__type'] = 'bald__Reference'
+#     dsetc.attrs['bald__dataset'] = dsetc.ref
+#     return f
+
 
 class Test(BaldTestCase):
 
@@ -35,6 +45,26 @@ def test_invalid_uri(self):
             validation = bald.validate_netcdf(tfile)
             self.assertFalse(validation.is_valid())
 
+# class TestArrayReference(BaldTestCase):
+#     def test_match(self):
+#         with self.temp_filename('.nc') as tfile:
+#             f = netCDF4.Dataset(tfile, "w", format="NETCDF4")
+#             f = _fattrs(f)
+#             f = _create_parent_child(f, (11, 17), (11, 17))
+#             f.close()
+#             validation = bald.validate_netcdf(tfile)
+#             self.assertTrue(validation.is_valid())
+
+#     def test_mismatch_zeroth(self):
+#         with self.temp_filename('.nc') as tfile:
+#             f = netCDF4.Dataset(tfile, "w", format="NETCDF4")
+#             f = _fattrs(f)
+#             f = _create_parent_child(f, (11, 17), (11, 13))
+#             f.close()
+#             validation = bald.validate_netcdf(tfile)
+#             self.assertFalse(validation.is_valid())
+
+
 
 if __name__ == '__main__':
     unittest.main()

lib/bald/tests/integration/test_validation.py

@@ -7,19 +7,19 @@
 from bald.tests import BaldTestCase
 
 def _fattrs(f):
-    f.attrs['bald__'] = 'http://binary-array-ld.net/experimental/'
+    f.attrs['bald__'] = 'http://binary-array-ld.net/latest/'
     f.attrs['rdf__'] = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'
     f.attrs['rdf__type'] = 'bald__Container'
     return f
 
 def _create_parent_child(f, pshape, cshape):
     dsetp = f.create_dataset("parent_dataset", pshape, dtype='i')
     dsetc = f.create_dataset("child_dataset", cshape, dtype='i')
-    dsetp.attrs['rdf__type'] = 'bald__Dataset'
+    dsetp.attrs['rdf__type'] = 'bald__Array'
     dsetp.attrs['bald__references'] = dsetc.ref
-    dsetc.attrs['rdf__type'] = 'bald__Dataset'
+    dsetc.attrs['rdf__type'] = 'bald__Array'
     dsetc.attrs['rdf__type'] = 'bald__Reference'
-    dsetc.attrs['bald__dataset'] = dsetc.ref
+    dsetc.attrs['bald__array'] = dsetc.ref
     return f
 
 
@@ -31,7 +31,8 @@ def test_valid_uri(self):
             f = _fattrs(f)
             f = _create_parent_child(f, (11, 17), (11, 17))
             f.close()
-            self.assertTrue(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertTrue(validation.is_valid())
 
     def test_invalid_uri(self):
         with self.temp_filename('.hdf') as tfile:
@@ -40,7 +41,8 @@ def test_invalid_uri(self):
             f = _create_parent_child(f, (11, 17), (11, 17))
             f.attrs['bald__turtle'] = 'bald__walnut'
             f.close()
-            self.assertFalse(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
 
 class TestArrayReference(BaldTestCase):
     def test_match(self):
@@ -49,23 +51,26 @@ def test_match(self):
             f = _fattrs(f)
             f = _create_parent_child(f, (11, 17), (11, 17))
             f.close()
-            self.assertTrue(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertTrue(validation.is_valid())
 
     def test_mismatch_zeroth(self):
         with self.temp_filename('.hdf') as tfile:
             f = h5py.File(tfile, "w")
             f = _fattrs(f)
             f = _create_parent_child(f, (11, 17), (11, 13))
             f.close()
-            self.assertFalse(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
 
     def test_mismatch_oneth(self):
         with self.temp_filename('.hdf') as tfile:
             f = h5py.File(tfile, "w")
             f = _fattrs(f)
             f = _create_parent_child(f, (11, 17), (13, 17))
             f.close()
-            self.assertFalse(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
 
     def test_match_plead_dim(self):
         with self.temp_filename('.hdf') as tfile:
@@ -74,7 +79,8 @@ def test_match_plead_dim(self):
             # parent has leading dimension wrt child
             f = _create_parent_child(f, (4, 13, 17), (13, 17))
             f.close()
-            self.assertTrue(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertTrue(validation.is_valid())
 
     def test_match_clead_dim(self):
         with self.temp_filename('.hdf') as tfile:
@@ -83,7 +89,8 @@ def test_match_clead_dim(self):
             # child has leading dimension wrt parent
             f = _create_parent_child(f, (13, 17), (7, 13, 17))
             f.close()
-            self.assertTrue(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertTrue(validation.is_valid())
 
     def test_mismatch_pdisjc_lead_dim(self):
         with self.temp_filename('.hdf') as tfile:
@@ -93,7 +100,8 @@ def test_mismatch_pdisjc_lead_dim(self):
             f = _create_parent_child(f, (4, 13, 17), (7, 13, 17))
 
             f.close()
-            self.assertFalse(bald.validate_hdf5(tfile).is_valid())
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
 
     def test_mismatch_pdisjc_trail_dim(self):
         with self.temp_filename('.hdf') as tfile:
@@ -102,8 +110,8 @@ def test_mismatch_pdisjc_trail_dim(self):
             # child and parent have disjoint trailing dimensions
             f = _create_parent_child(f, (13, 17, 2), (13, 17, 9))
             f.close()
-            self.assertFalse(bald.validate_hdf5(tfile).is_valid())
-
+            validation = bald.validate_hdf5(tfile)
+            self.assertFalse(validation.is_valid())
 
 
     # def test_match_(self):

lib/bald/validation.py

@@ -59,7 +59,6 @@ def exceptions(self):
 
 
 class SubjectValidation(Validation):
-
     def __init__(self, subject, fhandle, httpcache=None):
         self.subject = subject
         self.fhandle = fhandle
@@ -145,39 +144,39 @@ def __init__(self, **kwargs):
         super(ContainerValidation, self).__init__(**kwargs)
 
 
-class DatasetValidation(SubjectValidation):
+class ArrayValidation(SubjectValidation):
 
-    def __init__(self, name, dataset, **kwargs):
-        self.dataset = dataset
+    def __init__(self, name, array, **kwargs):
+        self.array = array
         self.name = name
-        super(DatasetValidation, self).__init__(**kwargs)
+        super(ArrayValidation, self).__init__(**kwargs)
 
     def _extra_exceptions(self, exceptions):
         exceptions = self.check_array_references(exceptions)
         return exceptions
 
     def check_array_references(self, exceptions):
-        def _check_ref(pdataset, parray, cdataset, carray):
+        def _check_ref(parraysubj, parray, carraysubj, carray):
             if not valid_array_reference(parray, carray):
                 msg = ('{} declares a child of {} but the arrays'
                        'do not conform to the bald array reference'
                        'rules')
-                msg = msg.format(pdataset, cdataset)
+                msg = msg.format(parraysubj, carraysubj)
                 exceptions.append(ValueError(msg))
             return exceptions
 
-        # if this Dataset has a bald__references
+        # if this Array has a bald__references
         # not implemented yet: and it's a singleton
         if self.subject.attrs.get('bald__references', ''):
-            # then it must have a bald_dataset
+            # then it must have a bald_array
             ref_dset = self.fhandle[self.subject.attrs.get('bald__references')]
-            if not ref_dset.attrs.get('bald__dataset', ''):
+            if not ref_dset.attrs.get('bald__array', ''):
                 exceptions.append(ValueError('A bald__Reference must link to '
-                                             'one and only one bald__Dataset'))
+                                             'one and only one bald__Array'))
             else:
                 # and we impose bald broadcasting rules on it
-                child_dset = self.fhandle[ref_dset.attrs.get('bald__dataset')]
-                parray = np.zeros(self.dataset.shape)
+                child_dset = self.fhandle[ref_dset.attrs.get('bald__array')]
+                parray = np.zeros(self.array.shape)
                 carray = np.zeros(child_dset.shape)
 
                 exceptions = _check_ref('p', parray, 'c', carray)