Merge pull request #496 from RDFLib/fix_canonicalization

fixed #494 canonicalization sometimes collapses BNodes

2 files changed, 111 insertions, 7 deletions

diff --git a/rdflib/compare.py b/rdflib/compare.py
index 0f93639c..06ba9750 100644
--- a/rdflib/compare.py
+++ b/rdflib/compare.py
@@ -204,7 +204,9 @@ class Color:
                 return unicode(x)
         if isinstance(color, Node):
             return stringify(color)
-        value = sum(map(self.hashfunc, ' '.join([stringify(x) for x in color])))
+        value = 0
+        for triple in color:
+            value += self.hashfunc(' '.join([stringify(x) for x in triple]))
         val = u"%x" % value
         self._hash_cache[color] = val
         return val
@@ -290,7 +292,7 @@ class _TripleCanonicalizer(object):
 
     def _individuate(self, color, individual):
         new_color = list(color.color)
-        new_color.append((len(color.nodes)))
+        new_color.append((len(color.nodes),))
 
         color.nodes.remove(individual)
         c = Color([individual], self.hashfunc, tuple(new_color),
@@ -320,6 +322,7 @@ class _TripleCanonicalizer(object):
                         sequence = sequence[:si] + colors + sequence[si+1:]
                     except ValueError:
                         sequence = colors[1:] + sequence
+
         return coloring
 
     @_runtime("to_hash_runtime")
@@ -407,7 +410,6 @@ class _TripleCanonicalizer(object):
                     stats['prunings'] += 1
         discrete = [x for x in best if self._discrete(x)]
         if len(discrete) == 0:
-            very_best = None
             best_score = None
             best_depth = None
             for coloring in best:
@@ -434,6 +436,7 @@ class _TripleCanonicalizer(object):
         if stats is not None:
             stats['initial_coloring_runtime'] = _total_seconds(datetime.now() - start_coloring)
             stats['initial_color_count'] = len(coloring)
+
         if not self._discrete(coloring):
             depth = [0]
             coloring = self._traces(coloring, stats=stats, depth=depth)
diff --git a/test/test_canonicalization.py b/test/test_canonicalization.py
index 2745f490..87b5eeaa 100644
--- a/test/test_canonicalization.py
+++ b/test/test_canonicalization.py
@@ -37,10 +37,10 @@ def negative_graph_match_test():
     True
     ],
     [ unicode('''@prefix : <http://example.org/ns#> .
-     :linear_two_step_symmetry_start :related [ :related [ :related :linear_two_step_symmatry_end]], 
+     :linear_two_step_symmetry_start :related [ :related [ :related :linear_two_step_symmatry_end]],
                                               [ :related [ :related :linear_two_step_symmatry_end]].'''),
     unicode('''@prefix : <http://example.org/ns#> .
-     :linear_two_step_symmetry_start :related [ :related [ :related :linear_two_step_symmatry_end]], 
+     :linear_two_step_symmetry_start :related [ :related [ :related :linear_two_step_symmatry_end]],
                                               [ :related [ :related :linear_two_step_symmatry_end]].'''),
     True
     ],
@@ -68,7 +68,7 @@ def negative_graph_match_test():
           ].'''),
     False
     ],
-    # This test fails because the algorithm purposefully breaks the symmetry of symetric 
+    # This test fails because the algorithm purposefully breaks the symmetry of symetric
     [ unicode('''@prefix : <http://example.org/ns#> .
      _:a :rel [
          :rel [
@@ -144,8 +144,109 @@ def negative_graph_match_test():
     def fn(rdf1, rdf2, identical):
         digest1 = get_digest_value(rdf1,"text/turtle")
         digest2 = get_digest_value(rdf2,"text/turtle")
+        print rdf1
         print digest1
+        print rdf2
         print digest2
         assert (digest1 == digest2) == identical
     for inputs in testInputs:
-        yield fn, inputs[0], inputs[1], inputs[2]
\ No newline at end of file
+        yield fn, inputs[0], inputs[1], inputs[2]
+
+def test_issue494_collapsing_bnodes():
+    """Test for https://github.com/RDFLib/rdflib/issues/494 collapsing BNodes"""
+    g = Graph()
+    g += [
+        (BNode('Na1a8fbcf755f41c1b5728f326be50994'),
+         RDF['object'],
+         URIRef(u'source')),
+        (BNode('Na1a8fbcf755f41c1b5728f326be50994'),
+         RDF['predicate'],
+         BNode('vcb3')),
+        (BNode('Na1a8fbcf755f41c1b5728f326be50994'),
+         RDF['subject'],
+         BNode('vcb2')),
+        (BNode('Na1a8fbcf755f41c1b5728f326be50994'),
+         RDF['type'],
+         RDF['Statement']),
+        (BNode('Na713b02f320d409c806ff0190db324f4'),
+         RDF['object'],
+         URIRef(u'target')),
+        (BNode('Na713b02f320d409c806ff0190db324f4'),
+         RDF['predicate'],
+         BNode('vcb0')),
+        (BNode('Na713b02f320d409c806ff0190db324f4'),
+         RDF['subject'],
+         URIRef(u'source')),
+        (BNode('Na713b02f320d409c806ff0190db324f4'),
+         RDF['type'],
+         RDF['Statement']),
+        (BNode('Ndb804ba690a64b3dbb9063c68d5e3550'),
+         RDF['object'],
+         BNode('vr0KcS4')),
+        (BNode('Ndb804ba690a64b3dbb9063c68d5e3550'),
+         RDF['predicate'],
+         BNode('vrby3JV')),
+        (BNode('Ndb804ba690a64b3dbb9063c68d5e3550'),
+         RDF['subject'],
+         URIRef(u'source')),
+        (BNode('Ndb804ba690a64b3dbb9063c68d5e3550'),
+         RDF['type'],
+         RDF['Statement']),
+        (BNode('Ndfc47fb1cd2d4382bcb8d5eb7835a636'),
+         RDF['object'],
+         URIRef(u'source')),
+        (BNode('Ndfc47fb1cd2d4382bcb8d5eb7835a636'),
+         RDF['predicate'],
+         BNode('vcb5')),
+        (BNode('Ndfc47fb1cd2d4382bcb8d5eb7835a636'),
+         RDF['subject'],
+         URIRef(u'target')),
+        (BNode('Ndfc47fb1cd2d4382bcb8d5eb7835a636'),
+         RDF['type'],
+         RDF['Statement']),
+        (BNode('Nec6864ef180843838aa9805bac835c98'),
+         RDF['object'],
+         URIRef(u'source')),
+        (BNode('Nec6864ef180843838aa9805bac835c98'),
+         RDF['predicate'],
+         BNode('vcb4')),
+        (BNode('Nec6864ef180843838aa9805bac835c98'),
+         RDF['subject'],
+         URIRef(u'source')),
+        (BNode('Nec6864ef180843838aa9805bac835c98'),
+         RDF['type'],
+         RDF['Statement']),
+    ]
+
+    print 'graph length: %d, nodes: %d' % (len(g), len(g.all_nodes()))
+    print 'triple_bnode degrees:'
+    for triple_bnode in g.subjects(RDF['type'], RDF['Statement']):
+        print len(list(g.triples([triple_bnode, None, None])))
+    print 'all node degrees:'
+    g_node_degs = sorted([
+        len(list(g.triples([node, None, None])))
+        for node in g.all_nodes()
+    ], reverse=True)
+    print g_node_degs
+
+    cg = to_canonical_graph(g)
+    print 'graph length: %d, nodes: %d' % (len(cg), len(cg.all_nodes()))
+    print 'triple_bnode degrees:'
+    for triple_bnode in cg.subjects(RDF['type'], RDF['Statement']):
+        print len(list(cg.triples([triple_bnode, None, None])))
+    print 'all node degrees:'
+    cg_node_degs = sorted([
+        len(list(cg.triples([node, None, None])))
+        for node in cg.all_nodes()
+    ], reverse=True)
+    print cg_node_degs
+
+    assert len(g) == len(cg), \
+        'canonicalization changed number of triples in graph'
+    assert len(g.all_nodes()) == len(cg.all_nodes()), \
+        'canonicalization changed number of nodes in graph'
+    assert len(list(g.subjects(RDF['type'], RDF['Statement']))) == \
+        len(list(cg.subjects(RDF['type'], RDF['Statement']))), \
+        'canonicalization changed number of statements'
+    assert g_node_degs == cg_node_degs, \
+        'canonicalization changed node degrees'