aliases

Author: marqh

lib/bald/__init__.py

@@ -15,9 +15,12 @@ class HttpCache(object):
     def __init__(self):
         self.cache = {}
 
+    def is_http_uri(self, item):
+        return item.startswith('http://') or item.startswith('https://')
+
     def __getitem__(self, item):
 
-        if not item.startswith('http://') or item.startswith('https://'):
+        if not self.is_http_uri(item):
             raise ValueError('{} is not a HTTP URI.'.format(item))
         if item not in self.cache:
             headers = {'Accept': 'text/turtle'}
@@ -33,7 +36,7 @@ def check_uri(self, uri):
 
 
 class Subject(object):
-    def __init__(self, attrs=None, prefixes=None):
+    def __init__(self, attrs=None, prefixes=None, aliases=None):
         """
         A subject of metadata statements.
 
@@ -43,7 +46,10 @@ def __init__(self, attrs=None, prefixes=None):
             attrs = {}
         if prefixes is None:
             prefixes = {}
+        if aliases is None:
+            aliases = {}
         self.attrs = attrs
+        self.aliases = aliases
         self._prefixes = prefixes
         self._prefix_suffix = re.compile('(^(?:(?!__).)*)__((?!.*__).*$)')
         _http_p = 'http[s]?://.*'
@@ -69,6 +75,8 @@ def unpack_uri(self, astring):
                 if self._http_uri.match(self.prefixes()[prefix]):
                     result = astring.replace('{}__'.format(prefix),
                                              self.prefixes()[prefix])
+        elif astring in self.aliases:
+            result = self.aliases[astring]
         return result
 
 
@@ -147,7 +155,11 @@ def validate_hdf5(afilepath):
         prefixes = {}
         if prefix_group:
             prefixes = fhandle[prefix_group].attrs
-        root_container = Subject(fhandle.attrs, prefixes=prefixes)
+        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()
@@ -159,7 +171,7 @@ def validate_hdf5(afilepath):
             # #
             sattrs = dict(fhandle.attrs).copy()
             sattrs.update(dataset.attrs)
-            dset = Subject(sattrs, prefixes)
+            dset = Subject(sattrs, prefixes, aliases)
             dset_val = bv.ArrayValidation(name, dataset, fhandle=fhandle,
                                             subject=dset)
             sval.stored_exceptions += dset_val.exceptions()

lib/bald/tests/integration/test_aliases.py

@@ -32,8 +32,8 @@ def test_valid_uri(self):
             f = h5py.File(tfile, "w")
             f = _fattrs(f)
             group_alias = f.create_group('bald__alias_list')
-            f.attrs['bald__aliases'] = group_alias.ref
-            group_alias['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#prefLabel'
+            f.attrs['bald__isAliasedBy'] = group_alias.ref
+            group_alias.attrs['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#prefLabel'
             dsetp = f.create_dataset("parent_dataset", (11, 17), dtype='i')
             dsetp.attrs['skosPrefLabel'] = 'alabel'
             f.close()
@@ -46,8 +46,8 @@ def test_invalid_uri(self):
             f = _fattrs(f)
             f.attrs['bald__turtle'] = 'bald__walnut'
             group_alias = f.create_group('bald__alias_list')
-            f.attrs['bald__aliases'] = group_alias.ref            
-            group_alias['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#notThisPrefLabel'
+            f.attrs['bald__isAliasedBy'] = group_alias.ref            
+            group_alias.attrs['skosPrefLabel'] = 'http://www.w3.org/2004/02/skos/core#notThisPrefLabel'
             dsetp = f.create_dataset("parent_dataset", (11, 17), dtype='i')
             dsetp.attrs['skosPrefLabel'] = 'alabel'
             f.close()

lib/bald/validation.py

@@ -91,12 +91,16 @@ def _check_uri(uri, exceptions):
         for pref, uri in self.subject.prefixes().iteritems():
             exceptions = _check_uri(self.subject.unpack_uri(uri),
                                     exceptions)
+        for alias, uri in self.subject.aliases.iteritems():
+            exceptions = _check_uri(self.subject.unpack_uri(uri),
+                                    exceptions)
         for attr, value in self.subject.attrs.iteritems():
             exceptions = _check_uri(self.subject.unpack_uri(attr),
                                     exceptions)
             if isinstance(value, str):
-                exceptions = _check_uri(self.subject.unpack_uri(value),
-                                        exceptions)
+                val = self.subject.unpack_uri(value)
+                if self.cache.is_http_uri(val):
+                    exceptions = _check_uri(val, exceptions)
         return exceptions
 
     def check_attr_domain_range(self, exceptions):
@@ -106,7 +110,11 @@ def check_attr_domain_range(self, exceptions):
                 # thus we have a payload
                 # go rdf
                 g = rdflib.Graph()
-                g.parse(data=self.cache[uri].text, format="n3")
+                data=self.cache[uri].text
+                try:
+                    g.parse(data=self.cache[uri].text, format="n3")
+                except Exception, e:
+                    g.parse(data=self.cache[uri].text, format="xml")
                 query = ('SELECT ?s \n'
                          '(GROUP_CONCAT(?domain; SEPARATOR=" | ") AS ?domains)'
                          ' \n'