use bald objects

Author: marqh

lib/bald/__init__.py

@@ -37,7 +37,8 @@ def check_uri(self, uri):
 
 
 class Subject(object):
-    def __init__(self, attrs=None, prefixes=None, aliases=None):
+    def __init__(self, attrs=None, prefixes=None, aliases=None,
+                 rdftype='bald__Subject'):
         """
         A subject of metadata statements.
 
@@ -50,13 +51,35 @@ def __init__(self, attrs=None, prefixes=None, aliases=None):
         if aliases is None:
             aliases = {}
         self.attrs = attrs
+        if 'rdf__type' not in attrs:
+            attrs['rdf__type'] = rdftype
+        elif rdftype not in attrs['rdf__type']:
+            if isinstance(attrs['rdf__type'], list):
+                attrs['rdf__type'].append(rdftype)
+            elif isinstance(attrs['rdf__type'], set):
+                attrs['rdf__type'].add(rdftype)
+            else:
+                attrs['rdf__type'] = set((attrs['rdf__type'], rdftype))
+
         self.aliases = aliases
         self._prefixes = prefixes
         self._prefix_suffix = re.compile('(^(?:(?!__).)*)__((?!.*__).*$)')
         _http_p = 'http[s]?://.*'
         self._http_uri = re.compile('{}'.format(_http_p))
         self._http_uri_prefix = re.compile('{}/|#'.format(_http_p))
 
+    def __str__(self):
+        return '{}: {}'.format(type(self), self.attrs)
+
+    def __repr__(self):
+        return str(self)
+
+    def __getattr__(self, attr):
+        if attr not in self.attrs:
+            msg = '{} object has no attribute {}'.format(type(self), attr)
+            raise AttributeError(msg)
+        return self.attrs[attr]
+
     def prefixes(self):
         prefixes = {}
         for key, value in self._prefixes.iteritems():
@@ -80,7 +103,30 @@ def unpack_uri(self, astring):
             result = self.aliases[astring]
         return result
 
+    def graph(self):
+        graph = rdflib.Graph()
+        
+        return graph
+
+class Array(Subject):
+    def __init__(self, attrs=None, prefixes=None, aliases=None,
+                 shape=None):
+        self.shape = shape
+        rdftype = 'bald__Array'
+        super(Array, self).__init__(attrs, prefixes, aliases, rdftype)
 
+    @property
+    def array_references(self):
+        return self.attrs.get('bald__references', [])
+        #return []
+
+class Container(Subject):
+    def __init__(self, attrs=None, prefixes=None, aliases=None,
+                 shape=None):
+        rdftype = 'bald__Container'
+        super(Container, self).__init__(attrs, prefixes, aliases, rdftype)
+
+    
 @contextlib.contextmanager
 def load(afilepath):
     if afilepath.endswith('.hdf'):
@@ -108,78 +154,136 @@ def load(afilepath):
     finally:
         f.close()
 
-
-def validate_netcdf(afilepath):
+def load_netcdf(afilepath):
     """
-    Validate a file with respect to binarry-array-linked-data.
+    Validate a file with respect to binary-array-linked-data.
     Returns a :class:`bald.validation.Validation`
     """
 
     with load(afilepath) as fhandle:
-        sval = bv.StoredValidation()
-        prefix_group = fhandle[fhandle.bald__isPrefixedBy] if hasattr(fhandle, 'bald__isPrefixedBy') else {}
+        prefix_group = (fhandle[fhandle.bald__isPrefixedBy] if
+                        hasattr(fhandle, 'bald__isPrefixedBy') else {})
         prefixes = {}
         if prefix_group:
-            prefixes = dict([(prefix, getattr(prefix_group, prefix)) for prefix in prefix_group.ncattrs()])
+            prefixes = (dict([(prefix, getattr(prefix_group, prefix)) for
+                              prefix in prefix_group.ncattrs()]))
         else:
             for k in fhandle.ncattrs():
                 if k.endswith('__'):
                     prefixes[k] = getattr(fhandle, k)
-        alias_group = fhandle[fhandle.bald__isAliasedBy] if hasattr(fhandle, 'bald__isAliasedBy') else {}
+        alias_group = (fhandle[fhandle.bald__isAliasedBy]
+                       if hasattr(fhandle, 'bald__isAliasedBy') else {})
         aliases = {}
         if alias_group:
-            aliases = dict([(alias, getattr(alias_group, alias)) for alias in alias_group.ncattrs()])
+            aliases = (dict([(alias, getattr(alias_group, alias))
+                             for alias in alias_group.ncattrs()]))
 
         attrs = {}
         for k in fhandle.ncattrs():
             attrs[k] = getattr(fhandle, k)
-        root_container = Subject(attrs, prefixes=prefixes, aliases=aliases)
-        root_val = bv.ContainerValidation(subject=root_container,
-                                          fhandle=fhandle)
-        sval.stored_exceptions += root_val.exceptions()
+        # It would be nice to use the URI of the file if it is known. 
+        attrs['@id'] = 'root'
+        root_container = Container(attrs, prefixes=prefixes,
+                                   aliases=aliases)
+        root_container.attrs['bald__contains'] = []
+
         for name in fhandle.variables:
-            sattrs = fhandle.__dict__.copy()
-            sattrs.update(fhandle.variables[name].__dict__.copy())
-            var = Subject(sattrs, prefixes=prefixes, aliases=aliases)
-            var_val = bv.ArrayValidation(name, fhandle.variables[name], fhandle=fhandle,
-                                         subject=var)
-            sval.stored_exceptions += var_val.exceptions()
+
+            sattrs = fhandle.variables[name].__dict__.copy()
+            sattrs['@id'] = '/{}'.format(name)
+
+            if 'bald__references' in fhandle.variables[name].ncattrs():
+                child_dset = fhandle.variables[fhandle.variables[name].bald__references]
+                cattrs = child_dset.__dict__.copy()
+                cattrs['@id'] = '/{}'.format(child_dset.name)
+                carray = Array(cattrs, prefixes, aliases, child_dset.shape)
+                sattrs['bald__references'] = [carray]
+            
+            var = Array(sattrs, prefixes=prefixes, aliases=aliases,
+                        shape=fhandle.variables[name].shape)
+            root_container.attrs['bald__contains'].append(var)
 
-        return sval
+
+    return root_container
+
+
+def validate_netcdf(afilepath):
+    """
+    Validate a file with respect to binary-array-linked-data.
+    Returns a :class:`bald.validation.Validation`
+
+    """
+    root_container = load_netcdf(afilepath)
+    return validate(root_container)
 
 
 def validate_hdf5(afilepath):
     """
-    Validate a file with respect ot binarry-array-linked-data.
+    Validate a file with respect to binary-array-linked-data.
+    Returns a :class:`bald.validation.Validation`
+
+    """
+    root_container = load_hdf5(afilepath)
+    return validate(root_container)
+
+def validate(root_container):
+    """
+    Validate a Container with respect to binary-array-linked-data.
     Returns a :class:`bald.validation.Validation`
+
     """
+    sval = bv.StoredValidation()
+
+    root_val = bv.ContainerValidation(subject=root_container)
+    sval.stored_exceptions += root_val.exceptions()
+    for subject in root_container.attrs.get('bald__contains', []):
+        
+        # a dataset's attribute collection inherits from and
+        # specialises it's container's attrbiute collection
+        # this only helps with prefixes, afaik, hence:
+        # #
+        array_val = bv.ArrayValidation(subject)
+        sval.stored_exceptions += array_val.exceptions()
+
+    return sval
 
+def load_hdf5(afilepath):
     with load(afilepath) as fhandle:
-        sval = bv.StoredValidation()
+        # unused?
         cache = {}
         prefix_group = fhandle.attrs.get('bald__isPrefixedBy')
         prefixes = {}
         if prefix_group:
-            prefixes = fhandle[prefix_group].attrs
+            prefixes = dict(fhandle[prefix_group].attrs)
         alias_group = fhandle.attrs.get('bald__isAliasedBy')
         aliases = {}
         if alias_group:
-            aliases = dict(fhandle[alias_group].attrs.iteritems())
-        root_container = Subject(fhandle.attrs, prefixes=prefixes, aliases=aliases)
-        root_val = bv.ContainerValidation(subject=root_container,
-                                          fhandle=fhandle)
-        sval.stored_exceptions += root_val.exceptions()
+            aliases = dict(fhandle[alias_group].attrs)
+        attrs = dict(fhandle.attrs)
+        attrs['@id'] = 'root'
+        root_container = Container(attrs, prefixes=prefixes, aliases=aliases)
+
+        root_container.attrs['bald__contains'] = []
+
         # iterate through the datasets
         for name, dataset in fhandle.items():
-            # a dataset's attribute collection inherits from and
-            # specialises it's container's attrbiute collection
-            # this only helps with prefixes, afaik, hence:
-            # #
-            sattrs = dict(fhandle.attrs).copy()
-            sattrs.update(dataset.attrs)
-            dset = Subject(sattrs, prefixes, aliases)
-            dset_val = bv.ArrayValidation(name, dataset, fhandle=fhandle,
-                                            subject=dset)
-            sval.stored_exceptions += dset_val.exceptions()
-
-        return sval
+            if hasattr(dataset, 'shape'):
+                sattrs = dict(dataset.attrs)
+                sattrs['@id'] = dataset.name
+                ref = sattrs.get('bald__references', '')
+                if ref:
+                    ref_dset = fhandle[ref]
+                    child_dset = fhandle[ref_dset.attrs.get('bald__array',
+                                                            None)]
+                    if child_dset:
+                        cattrs = dict(child_dset.attrs)
+                        cattrs['@id'] = child_dset.name
+                        carray = Array(cattrs, prefixes, aliases,
+                                       child_dset.shape)
+                        sattrs['bald__references'] = [carray]
+                dset = Array(sattrs, prefixes, aliases, dataset.shape)
+                root_container.attrs['bald__contains'].append(dset)
+            
+    return root_container
+
+    

lib/bald/tests/__init__.pyc

@@ -3,7 +3,7 @@ dimensions:
 	i = 600 ;
 	j = 180 ;
 	k = 47 ;
-	time = 50521 ;
+	time = 5 ;
 variables:
 	double zc(k) ;
 		zc:axis = "Z" ;

lib/bald/tests/integration/CDL/ereefs-gbr4_ncld.cdl

@@ -55,12 +55,6 @@ def test_invalid_uri(self):
             validation = bald.validate_hdf5(tfile)
             exns = validation.exceptions()
             expected = ['http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
                         'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).']
             self.assertTrue((not validation.is_valid()) and exns == expected,
                              msg='{}  != {}'.format(exns, expected))

lib/bald/tests/integration/test_aliases.py

@@ -35,14 +35,18 @@ def test_process_chain(self):
             subprocess.check_call(['ncgen', '-o', tfile, cdl_file])
             validation = bald.validate_netcdf(tfile)
             exns = validation.exceptions()
-            self.assertTrue(validation.is_valid(), msg='{}  != []'.format(exns))
+            self.assertTrue(validation.is_valid(), msg='{} != []'.format(exns))
 
     def test_ereef(self):
         with self.temp_filename('.nc') as tfile:
             cdl_file = os.path.join(self.cdl_path, 'ereefs-gbr4_ncld.cdl')
             subprocess.check_call(['ncgen', '-o', tfile, cdl_file])
             validation = bald.validate_netcdf(tfile)
             exns = validation.exceptions()
-            #self.assertTrue(validation.is_valid(), msg='{} != []'.format(exns))
-            self.assertFalse(validation.is_valid(), msg='AssertionError: [\'http://qudt.org/vocab/unit#Meter is not resolving as a resource (404).\', \'http://qudt.org/vocab/unit#MeterPerSecond is not resolving as a resource (404).\', \'http://qudt.org/vocab/unit#MeterPerSecond is not resolving as a resource (404).\', \'http://qudt.org/vocab/unit#DegreeCelsius is not resolving as a resource (404).\']. Also {} != []'.format(exns))
-            
+            xs = '{} is not resolving as a resource (404).'
+            expected = [xs.format('http://qudt.org/vocab/unit#Meter'),
+                        xs.format('http://qudt.org/vocab/unit#MeterPerSecond'),
+                        xs.format('http://qudt.org/vocab/unit#MeterPerSecond'),
+                        xs.format('http://qudt.org/vocab/unit#DegreeCelsius')]
+            self.assertTrue(not validation.is_valid() and exns == expected,
+                            msg='{} \n!= \n{}'.format(exns, expected))

lib/bald/tests/integration/test_cdl.py

@@ -0,0 +1,27 @@
+import os
+import subprocess
+import unittest
+
+import netCDF4
+import numpy as np
+
+import bald
+from bald.tests import BaldTestCase
+
+class Test(BaldTestCase):
+    def setUp(self):
+        self.cdl_path = os.path.join(os.path.dirname(__file__), 'CDL')
+        
+    def test_array_reference(self):
+        with self.temp_filename('.nc') as tfile:
+            cdl_file = os.path.join(self.cdl_path, 'array_reference.cdl')
+            subprocess.check_call(['ncgen', '-o', tfile, cdl_file])
+            root_container = bald.load_netcdf(tfile)
+
+            print root_container
+            
+
+    # def test_alias(self):
+    #     with self.temp_filename('.nc') as tfile:
+    #         cdl_file = os.path.join(self.cdl_path, 'array_alias.cdl')
+    #         subprocess.check_call(['ncgen', '-o', tfile, cdl_file])

lib/bald/tests/integration/test_cdl_graph.py

@@ -47,12 +47,6 @@ def test_invalid_uri(self):
             validation = bald.validate_hdf5(tfile)
             exns = validation.exceptions()
             expected = ['http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).',
-                        'http://binary-array-ld.net/latest/turtle is not resolving as a resource (404).',
                         'http://binary-array-ld.net/latest/walnut is not resolving as a resource (404).']
             self.assertTrue((not validation.is_valid()) and exns == expected,
                              msg='{}  != {}'.format(exns, expected))