Graph_diff fails on presence of unrelated blank nodes

[aucampia]

This tracks #1294 which was a bug but incorrectly converted to a discussion #1610.

The xfail test for this is here:

rdflib/test/test_graph/test_canonicalization.py

Lines 526 to 568 in 1cba9d8

	class TestConsistency(unittest.TestCase):
	@expectedFailure
	def test_consistent_ids(self) -> None:
	"""
	This test verifies that `to_canonical_graph` creates consistent
	identifiers for blank nodes even when the graph changes.
	It does this by creating two triple sets `g0_ts` and `g1_ts`
	and then first creating a canonical graph with only the first
	triple set (cg0), and then a canonical graph with both triple
	sets (cg1), and then confirming the triples in cg0 is a subset
	of cg1.
	This will fail if the `to_canonical_graph` does not generate
	consistent identifiers for blank nodes when the graph changes.
	This property is essential for `to_canonical_graph` to
	be useful for diffing graphs.
	"""
	bnode = BNode()
	g0_ts: _TripleSet = {
	(bnode, FOAF.name, Literal("Golan Trevize")),
	(bnode, RDF.type, FOAF.Person),
	}
	bnode = BNode()
	g1_ts: _TripleSet = {
	(bnode, FOAF.name, Literal("Janov Pelorat")),
	(bnode, RDF.type, FOAF.Person),
	}
	g0 = Graph()
	g0 += g0_ts
	cg0 = to_canonical_graph(g0)
	cg0_ts = GraphHelper.triple_set(cg0)
	g1 = Graph()
	g1 += g1_ts
	cg1 = to_canonical_graph(g1)
	cg1_ts = GraphHelper.triple_set(cg1)
	assert cg0_ts.issubset(
	cg1_ts
	), "canonical triple set cg0_ts should be a subset of canonical triple set cg1_ts"

I likely will never have time to rewrite this properly, but I will refer to this issue to keep track of it.

Discussed in #1610

^{Originally posted by rob-metalinkage April 23, 2021}
If both graph has multiple blank nodes graph diff fails to detect matches. It succeeds for a single blank node case.

code:

from pprint import pprint

from rdflib import Graph
from rdflib.compare import graph_diff, to_canonical_graph

g1t = """@prefix : <http://test.org#> .
@prefix owl:   <http://www.w3.org/2002/07/owl#> .
@prefix xsd:   <http://www.w3.org/2001/XMLSchema#> .
@prefix rdfs:  <http://www.w3.org/2000/01/rdf-schema#> .
@prefix rdf:   <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .

:C1  a          owl:Class ;
        rdfs:subClassOf  [ a                   owl:Restriction ;
                           owl:minCardinality  "1"^^xsd:int ;
                           owl:onProperty      rdfs:label
                         ] ;
        rdfs:label   "C1"@en ."""

g2t =  g1t + """ :C2  a          owl:Class ;
        rdfs:subClassOf  [ a                   owl:Restriction ;
                           owl:minCardinality  "1"^^xsd:int ;
                           owl:onProperty      rdfs:comment
                         ] ;
        rdfs:label   "C2"@en ."""
g1 = Graph()
g2 = Graph()

g1.parse(data=g1t, format='turtle')
# g1.parse(source="profile.ttl", format='turtle')
#g2.parse(source="source.ttl", format='turtle')
g2.parse(data=g2t, format='turtle')


in_both, in_first, in_second = graph_diff(g1,g2)
print("should be empty as all in g1 is in g2")
pprint(in_first.serialize(format='turtle'))

in_both, in_first, in_second = graph_diff(g1,g1)

print("should be empty as all in g1 is in g1")
pprint(in_first.serialize(format='turtle'))


output: (which should be things only in first (in g1) - g2 is g1 + stuff so this should always be empty..

should be empty as all in g1 is in g2

(b'@prefix ns1: <http://www.w3.org/2002/07/owl#> .\n@prefix rdfs: <http://ww'
 b'w.w3.org/2000/01/rdf-schema#> .\n@prefix xsd: <http://www.w3.org/2001/XML'
 b'Schema#> .\n\n<http://test.org#C1> rdfs:subClassOf [ a ns1:Restriction ;\n '
 b'           ns1:minCardinality "1"^^xsd:int ;\n            ns1:onProperty '
 b'rdfs:label ] .\n\n')

should be empty as all in g1 is in g1

b'\n'