class TestGraphSet(unittest.TestCase):
resp_agent = 'http://resp_agent.test/'
def setUp(self):
self.graph_set = GraphSet("http://test/", "./info_dir/", "", False)
def test_get_entity(self):
ar = self.graph_set.add_ar(self.resp_agent)
ref = ar.res
result = self.graph_set.get_entity(ref)
self.assertIsNotNone(result)
self.assertIs(result, ar)
def test_add_an(self):
an = self.graph_set.add_an(self.resp_agent)
self.assertIsNotNone(an)
self.assertIsInstance(an, ReferenceAnnotation)
self.assertEqual(str(an.g.identifier), self.graph_set.g_an)
def test_add_ar(self):
ar = self.graph_set.add_ar(self.resp_agent)
self.assertIsNotNone(ar)
self.assertIsInstance(ar, AgentRole)
self.assertEqual(str(ar.g.identifier), self.graph_set.g_ar)
def test_add_be(self):
be = self.graph_set.add_be(self.resp_agent)
self.assertIsNotNone(be)
self.assertIsInstance(be, BibliographicReference)
self.assertEqual(str(be.g.identifier), self.graph_set.g_be)
def test_add_br(self):
br = self.graph_set.add_br(self.resp_agent)
self.assertIsNotNone(br)
self.assertIsInstance(br, BibliographicResource)
self.assertEqual(str(br.g.identifier), self.graph_set.g_br)
def test_add_ci(self):
ci = self.graph_set.add_ci(self.resp_agent)
self.assertIsNotNone(ci)
self.assertIsInstance(ci, Citation)
self.assertEqual(str(ci.g.identifier), self.graph_set.g_ci)
def test_add_de(self):
de = self.graph_set.add_de(self.resp_agent)
self.assertIsNotNone(de)
self.assertIsInstance(de, DiscourseElement)
self.assertEqual(str(de.g.identifier), self.graph_set.g_de)
def test_add_id(self):
identifier = self.graph_set.add_id(self.resp_agent)
self.assertIsNotNone(identifier)
self.assertIsInstance(identifier, Identifier)
self.assertEqual(str(identifier.g.identifier), self.graph_set.g_id)
def test_add_pl(self):
pl = self.graph_set.add_pl(self.resp_agent)
self.assertIsNotNone(pl)
self.assertIsInstance(pl, PointerList)
self.assertEqual(str(pl.g.identifier), self.graph_set.g_pl)
def test_add_rp(self):
rp = self.graph_set.add_rp(self.resp_agent)
self.assertIsNotNone(rp)
self.assertIsInstance(rp, ReferencePointer)
self.assertEqual(str(rp.g.identifier), self.graph_set.g_rp)
def test_add_ra(self):
ra = self.graph_set.add_ra(self.resp_agent)
self.assertIsNotNone(ra)
self.assertIsInstance(ra, ResponsibleAgent)
self.assertEqual(str(ra.g.identifier), self.graph_set.g_ra)
def test_add_re(self):
re = self.graph_set.add_re(self.resp_agent)
self.assertIsNotNone(re)
self.assertIsInstance(re, ResourceEmbodiment)
self.assertEqual(str(re.g.identifier), self.graph_set.g_re)
def test_graphs(self):
count = 10
for i in range(count):
self.graph_set.add_ar(self.resp_agent)
result = self.graph_set.graphs()
self.assertIsNotNone(result)
self.assertEqual(len(result), count)
for graph in result:
self.assertIsInstance(graph, Graph)
def test_get_graph_iri(self):
ar = self.graph_set.add_ar(self.resp_agent)
iri = str(ar.g.identifier)
result = GraphSet.get_graph_iri(ar.g)
self.assertIsNotNone(result)
self.assertEqual(iri, result)
def test_get_orphans(self):
br = self.graph_set.add_br(self.resp_agent)
ar = self.graph_set.add_ar(self.resp_agent)
ra = self.graph_set.add_ra(self.resp_agent)
with self.subTest("subtest 1"):
orphans = self.graph_set.get_orphans()
self.assertIsNotNone(orphans)
orphans_set = {o.res for o in orphans}
self.assertSetEqual({br.res, ar.res, ra.res}, orphans_set)
with self.subTest("subtest 2"):
# Here we link br and ar:
br.has_contributor(ar)
orphans = self.graph_set.get_orphans()
self.assertIsNotNone(orphans)
orphans_set = {o.res for o in orphans}
self.assertSetEqual({br.res, ra.res}, orphans_set)
with self.subTest("subtest 3"):
# Here we link ar and ra:
ar.is_held_by(ra)
orphans = self.graph_set.get_orphans()
self.assertIsNotNone(orphans)
orphans_set = {o.res for o in orphans}
self.assertSetEqual({br.res}, orphans_set)
class TestProvSet(unittest.TestCase):
resp_agent = 'http://resp_agent.test/'
def setUp(self):
self.graph_set = GraphSet("http://test/", "./info_dir/", "", False)
self.prov_set = ProvSet(self.graph_set, "http://test/", "./info_dir/",
False)
def test_add_se(self):
prov_subj = self.graph_set.add_br(self.resp_agent)
se = self.prov_set.add_se(prov_subj)
self.assertIsNotNone(se)
self.assertIsInstance(se, SnapshotEntity)
self.assertEqual(str(se.g.identifier), str(prov_subj.res) + "/prov/")
def test_generate_provenance(self):
cur_time = 1607375859.846196
cur_time_str = '2020-12-07T22:17:39'
with self.subTest('Creation [Merged entity]'):
a = self.graph_set.add_br(self.resp_agent)
b = self.graph_set.add_br(self.resp_agent)
a.merge(b)
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
se_a = self.prov_set.get_entity(URIRef(a.res + '/prov/se/1'))
self.assertIsNotNone(se_a)
self.assertIsInstance(se_a, SnapshotEntity)
self.assertEqual(a.res, se_a.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a.get_resp_agent()))
self.assertEqual(f"The entity '{a.res}' has been created.",
se_a.get_description())
with self.subTest('No snapshot [Merged entity]'):
a = self.graph_set.add_br(self.resp_agent)
b = self.graph_set.add_br(self.resp_agent)
a.merge(b)
se_a_1 = self.prov_set.add_se(a)
# This avoids that the presence of the mandatory rdf:type gets interpreted
# as a modification with respect to an empty preexisting_graph:
a.remove_every_triple()
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNone(se_a_2)
with self.subTest('Modification [Merged entity]'):
title = "TEST TITLE"
a = self.graph_set.add_br(self.resp_agent)
b = self.graph_set.add_br(self.resp_agent)
b.has_title(title)
a.merge(b)
se_a_1 = self.prov_set.add_se(a)
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
self.assertEqual(cur_time_str, se_a_1.get_invalidation_time())
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNotNone(se_a_2)
self.assertIsInstance(se_a_2, SnapshotEntity)
self.assertEqual(a.res, se_a_2.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a_2.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a_2.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a_2.get_resp_agent()))
self.assertSetEqual({se_a_1}, set(se_a_2.get_derives_from()))
self.assertIsNotNone(se_a_2.get_update_action())
self.assertEqual(f"The entity '{a.res}' has been modified.",
se_a_2.get_description())
with self.subTest('Merge [Merged entity]'):
a = self.graph_set.add_br(self.resp_agent)
b = self.graph_set.add_br(self.resp_agent)
c = self.graph_set.add_br(self.resp_agent)
a.merge(b)
a.merge(c)
se_a_1 = self.prov_set.add_se(a)
se_b_1 = self.prov_set.add_se(b)
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
self.assertEqual(cur_time_str, se_a_1.get_invalidation_time())
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNotNone(se_a_2)
self.assertIsInstance(se_a_2, SnapshotEntity)
self.assertEqual(a.res, se_a_2.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a_2.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a_2.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a_2.get_resp_agent()))
self.assertSetEqual({se_a_1, se_b_1},
set(se_a_2.get_derives_from()))
self.assertEqual(
f"The entity '{a.res}' has been merged with '{b.res}'.",
se_a_2.get_description())
with self.subTest('Creation [Non-Merged entity]'):
a = self.graph_set.add_br(self.resp_agent)
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
se_a = self.prov_set.get_entity(URIRef(a.res + '/prov/se/1'))
self.assertIsNotNone(se_a)
self.assertIsInstance(se_a, SnapshotEntity)
self.assertEqual(a.res, se_a.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a.get_resp_agent()))
self.assertEqual(f"The entity '{a.res}' has been created.",
se_a.get_description())
with self.subTest('No snapshot [Non-Merged entity] (Scenario 1)'):
a = self.graph_set.add_br(self.resp_agent)
a.mark_as_to_be_deleted()
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
se_a = self.prov_set.get_entity(URIRef(a.res + '/prov/se/1'))
self.assertIsNone(se_a)
with self.subTest('No snapshot [Merged entity] (Scenario 2)'):
a = self.graph_set.add_br(self.resp_agent)
se_a_1 = self.prov_set.add_se(a)
# This avoids that the presence of the mandatory rdf:type gets interpreted
# as a modification with respect to an empty preexisting_graph:
a.remove_every_triple()
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNone(se_a_2)
with self.subTest('Deletion [Non-Merged entity]'):
a = self.graph_set.add_br(self.resp_agent)
se_a_1 = self.prov_set.add_se(a)
a.has_title('ciao')
a.mark_as_to_be_deleted()
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
self.assertEqual(cur_time_str, se_a_1.get_invalidation_time())
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNotNone(se_a_2)
self.assertIsInstance(se_a_2, SnapshotEntity)
self.assertEqual(a.res, se_a_2.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a_2.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a_2.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a_2.get_resp_agent()))
self.assertSetEqual({se_a_1}, set(se_a_2.get_derives_from()))
self.assertEqual(f"The entity '{a.res}' has been deleted.",
se_a_2.get_description())
with self.subTest('Modification [Non-Merged entity]'):
title = "TEST TITLE"
a = self.graph_set.add_br(self.resp_agent)
se_a_1 = self.prov_set.add_se(a)
a.has_title(title)
result = self.prov_set.generate_provenance(cur_time)
self.assertIsNone(result)
self.assertEqual(cur_time_str, se_a_1.get_invalidation_time())
se_a_2 = self.prov_set.get_entity(URIRef(a.res + '/prov/se/2'))
self.assertIsNotNone(se_a_2)
self.assertIsInstance(se_a_2, SnapshotEntity)
self.assertEqual(a.res, se_a_2.get_is_snapshot_of())
self.assertEqual(cur_time_str, se_a_2.get_generation_time())
if a.source is not None:
self.assertEqual(a.source, str(se_a_2.get_primary_source()))
if a.resp_agent is not None:
self.assertEqual(a.resp_agent, str(se_a_2.get_resp_agent()))
self.assertSetEqual({se_a_1}, set(se_a_2.get_derives_from()))
self.assertIsNotNone(se_a_2.get_update_action())
self.assertEqual(f"The entity '{a.res}' has been modified.",
se_a_2.get_description())
def test_retrieve_last_snapshot(self):
br = self.graph_set.add_br(self.resp_agent)
br_res = br.res
result = self.prov_set._retrieve_last_snapshot(br_res)
self.assertIsNone(result)
se = self.prov_set.add_se(br)
result = self.prov_set._retrieve_last_snapshot(br_res)
self.assertIsNotNone(result)
self.assertEqual(str(result), str(se))
prov_subject = URIRef('https://w3id.org/oc/corpus/br/0')
self.assertRaises(ValueError, self.prov_set._retrieve_last_snapshot,
prov_subject)
prov_subject = URIRef('https://w3id.org/oc/corpus/br/-1')
self.assertRaises(ValueError, self.prov_set._retrieve_last_snapshot,
prov_subject)
prov_subject = URIRef('https://w3id.org/oc/corpus/br/abc')
self.assertRaises(ValueError, self.prov_set._retrieve_last_snapshot,
prov_subject)
def import_entities_from_graph(g_set: GraphSet,
graph: Graph,
resp_agent: str,
enable_validation: bool = False,
closed: bool = False) -> List[GraphEntity]:
if enable_validation:
graph = Reader.graph_validation(graph, closed)
imported_entities: List[GraphEntity] = []
for subject in Reader._extract_subjects(graph):
types: List[term] = []
for o in graph.objects(subject, RDF.type):
types.append(o)
# ReferenceAnnotation
if GraphEntity.iri_note in types:
imported_entities.append(
g_set.add_an(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# AgentRole
elif GraphEntity.iri_role_in_time in types:
imported_entities.append(
g_set.add_ar(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# BibliographicReference
elif GraphEntity.iri_bibliographic_reference in types:
imported_entities.append(
g_set.add_be(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# BibliographicResource
elif GraphEntity.iri_expression in types:
imported_entities.append(
g_set.add_br(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# Citation
elif GraphEntity.iri_citation in types:
imported_entities.append(
g_set.add_ci(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# DiscourseElement
elif GraphEntity.iri_discourse_element in types:
imported_entities.append(
g_set.add_de(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# Identifier
elif GraphEntity.iri_identifier in types:
imported_entities.append(
g_set.add_id(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# PointerList
elif GraphEntity.iri_singleloc_pointer_list in types:
imported_entities.append(
g_set.add_pl(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# ResponsibleAgent
elif GraphEntity.iri_agent in types:
imported_entities.append(
g_set.add_ra(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# ResourceEmbodiment
elif GraphEntity.iri_manifestation in types:
imported_entities.append(
g_set.add_re(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
# ReferencePointer
elif GraphEntity.iri_intextref_pointer in types:
imported_entities.append(
g_set.add_rp(
resp_agent=resp_agent,
res=subject,
preexisting_graph=Reader.get_graph_from_subject(
graph, subject)))
return imported_entities
class TestSupport(unittest.TestCase):
resp_agent = 'http://resp_agent.test/'
pro = Namespace("http://purl.org/spar/pro/")
def setUp(self):
self.graph_set = GraphSet("http://test/", "./info_dir/", "", False)
self.prov_set = ProvSet(self.graph_set, "http://test/", "./info_dir/",
False)
self.br = self.graph_set.add_br(self.resp_agent)
def _prepare_ordered_authors_list(self, list_len):
# First of all, we must cleanup the GraphSet:
self.br.remove_contributor()
for ar in self.graph_set.get_ar():
del self.graph_set.res_to_entity[ar.res]
for ra in self.graph_set.get_ra():
del self.graph_set.res_to_entity[ra.res]
# Then, we initialize a new well-formed linked list of authors:
ar_ordered_list = []
for i in range(list_len):
ra = self.graph_set.add_ra(self.resp_agent)
ar = self.graph_set.add_ar(self.resp_agent)
ar.create_author()
ar.is_held_by(ra)
self.br.has_contributor(ar)
ar_ordered_list.append(ar)
# Here each node of the list gets linked to the next one:
for i in range(list_len - 1):
ar_ordered_list[i].has_next(ar_ordered_list[i + 1])
return ar_ordered_list
@staticmethod
def _extract_ra_list(ar_list):
# Here the RA list is built and returned:
ra_list = []
for i in range(len(ar_list)):
ra = ar_list[i].get_is_held_by()
if ra is not None:
ra_list.append(ra)
return ra_list
def test_get_ordered_contributors_from_br(self):
list_len = 100
with self.subTest("Empty linked list"):
result = get_ordered_contributors_from_br(self.br, self.pro.author)
self.assertIsNotNone(result)
self.assertListEqual([], result)
with self.subTest("Well-formed linked list"):
correct_list = self._prepare_ordered_authors_list(list_len)
result = get_ordered_contributors_from_br(self.br, self.pro.author)
self.assertIsNotNone(result)
ar_list = self._extract_ra_list(correct_list)
self.assertListEqual(ar_list, result)
with self.subTest("Linked list with a loop"):
correct_list = self._prepare_ordered_authors_list(list_len)
# Here we corrupt the well-formed linked list introducing a loop:
correct_list[80].has_next(correct_list[50])
self.assertRaises(ValueError, get_ordered_contributors_from_br,
self.br, self.pro.author)
with self.subTest("Linked list split in two sub-lists"):
correct_list = self._prepare_ordered_authors_list(list_len)
# Here we corrupt the well-formed linked list introducing a loop:
correct_list[64].remove_next()
self.assertRaises(ValueError, get_ordered_contributors_from_br,
self.br, self.pro.author)