def create_objects_from_relations(relationships: List[Tuple],
metadata: List[Tuple[dict]] = None):
"""Given a list of relationships, create all corresponding DB objects.
Optional 'metadata' list can be passed, which contains corresponding
metadata items for each of the relationships, i.e., a triplet of Source,
Relation and Target metadata.
E.g.:
relationships = [
('A', Relation.Cites, 'B'),
('C', Relation.Cites, 'D'),
]
metadata = [
({<source-1>}, {<relation-1>}, {<target-1>}),
({<source-2>}, {<relation-2>}, {<target-2>}),
]
Will create Identifier, Relationship, Group and all M2M objects.
"""
if not metadata:
metadata = [({}, {}, {}) for _ in range(len(relationships))]
assert len(relationships) == len(metadata)
identifiers = sorted(set(sum([[a, b] for a, _, b in relationships], [])))
groups = [] # Contains pairs of (Identifier2Group, Group2Group)
for i in identifiers:
id_ = Identifier(value=i, scheme='doi')
db.session.add(id_)
groups.append(get_or_create_groups(id_))
rel_obj = []
for (src, rel, tar), (src_m, rel_m, tar_m) in zip(relationships, metadata):
src_, tar_ = Identifier.get(src, 'doi'), \
Identifier.get(tar, 'doi')
r = Relationship(source=src_, target=tar_, relation=rel)
db.session.add(r)
rel_obj.append(r)
s_id_gr, s_ver_gr = groups[identifiers.index(src)]
t_id_gr, t_ver_gr = groups[identifiers.index(tar)]
id_gr_rel = GroupRelationship(
source=s_id_gr, target=t_id_gr, relation=rel,
type=GroupType.Identity, id=uuid.uuid4())
s_id_gr.data.update(src_m, validate=False)
t_id_gr.data.update(tar_m, validate=False)
grm = GroupRelationshipMetadata(group_relationship_id=id_gr_rel.id)
db.session.add(grm)
grm.update(rel_m, validate=False)
db.session.add(Relationship2GroupRelationship(
relationship=r, group_relationship=id_gr_rel))
db.session.add(id_gr_rel)
ver_gr_rel = GroupRelationship(
source=s_ver_gr, target=t_ver_gr, relation=rel,
type=GroupType.Version)
db.session.add(GroupRelationshipM2M(
relationship=ver_gr_rel, subrelationship=id_gr_rel))
db.session.add(ver_gr_rel)
db.session.commit()
def test_get_or_create_groups(db):
"""Test creating groups (Identity and Version) for an identifier."""
id1 = Identifier(value='A', scheme='doi')
db.session.add(id1)
# id2 = Identifier(value='B', scheme='doi')
# rel = Relationship(source=id1, target=id2,
# relation=Relation.IsIdenticalTo)
assert not Group.query.count()
assert not GroupM2M.query.count()
assert not Identifier2Group.query.count()
id_g, ver_g = get_or_create_groups(id1)
db.session.commit()
def _check_groups(identifier, id_g, ver_g):
assert Group.query.count() == 2
assert GroupM2M.query.count() == 1
assert Identifier2Group.query.count() == 1
assert Group.query.filter_by(type=GroupType.Identity).one() == id_g
assert Group.query.filter_by(type=GroupType.Version).one() == ver_g
id2g = Identifier2Group.query.one()
g2g = GroupM2M.query.one()
assert id2g.identifier == identifier
assert id2g.group == id_g
assert g2g.group == ver_g
assert g2g.subgroup == id_g
_check_groups(id1, id_g, ver_g)
# Fetch the ID again and try to create groups again
id2 = Identifier.get('A', 'doi')
assert id2
id_g, ver_g = get_or_create_groups(id1)
db.session.commit()
# Make sure nothing changed
_check_groups(id2, id_g, ver_g)
# Add a new, separate identifier
id3 = Identifier(value='B', scheme='doi')
db.session.add(id3)
id_g, ver_g = get_or_create_groups(id3)
assert Group.query.count() == 4
assert GroupM2M.query.count() == 2
assert Identifier2Group.query.count() == 2
def id_obj(identifier, scheme):
"""Identifier DB object generator."""
return Identifier(value=identifier, scheme=scheme)
def assert_grouping(grouping):
"""Determine if database state corresponds to 'grouping' definition.
See tests in test_grouping.py for example input.
"""
groups, relationships, relationship_groups = grouping
group_types = [
(GroupType.Identity if isinstance(g[0], str) else GroupType.Version)
for g in groups]
# Mapping 'relationship_types' is a mapping between relationship index to:
# * None if its a regular Relation between Identifiers
# * GroupType.Identity if it's a relation between 'Identity'-type Groups
# * GroupType.Version if it's a relation between 'Version'-type Groups
relationship_types = [None if isinstance(r[0], str) else group_types[r[0]]
for r in relationships]
id_groups = [g for g, t in zip(groups, group_types)
if t == GroupType.Identity]
uniqe_ids = set(sum(id_groups, []))
# id_map is a mapping of str -> Identifier
# E.g.: 'A' -> Instance('A', 'doi')
id_map = dict(map(lambda x: (x, Identifier.get(x, 'doi')), uniqe_ids))
group_map = []
for g in groups:
if isinstance(g[0], str): # Identity group
group_map.append(
Identifier2Group.query.filter_by(
identifier=id_map[g[0]]).one().group)
elif isinstance(g[0], int): # GroupM2M
group_map.append(
GroupM2M.query.filter_by(
subgroup=group_map[g[0]]).one().group)
rel_map = []
for r in relationships:
obj_a, relation, obj_b = r
if isinstance(obj_a, str) and isinstance(obj_b, str):
# Identifiers relation
rel_map.append(
Relationship.query.filter_by(
source=id_map[obj_a], target=id_map[obj_b],
relation=relation).one()
)
elif isinstance(obj_a, int) and isinstance(obj_b, int):
# Groups relation
rel_map.append(
GroupRelationship.query.filter_by(
source=group_map[obj_a], target=group_map[obj_b],
relation=relation).one()
)
# Make sure all loaded identifiers are unique
assert len(set(map(lambda x: x[1].id, id_map.items()))) == len(id_map)
assert Identifier.query.count() == len(id_map)
# Make sure there's correct number of Identitfier2Group records
# and 'Identity'-type groups
assert Identifier2Group.query.count() == len(id_map)
assert Group.query.filter_by(
type=GroupType.Identity).count() == len(id_groups)
assert GroupMetadata.query.count() == len(id_groups)
# Make sure that all loaded groups are unique
assert len(set(map(lambda x: x.id, group_map))) == len(group_map)
assert Group.query.count() == len(group_map)
# Make sure there's correct number of GroupM2M records
# and 'Version'-type groups
m2m_groups = [g for g in groups if isinstance(g[0], int)]
assert Group.query.filter_by(
type=GroupType.Version).count() == len(m2m_groups)
# There are as many M2M groups as there are Identity groups
assert GroupM2M.query.count() == len(id_groups)
# Make sure that all loaded relationships are unique
id_rels = [r for r, t in zip(rel_map, relationship_types)
if t is None]
assert len(set(map(lambda x: x.id, id_rels))) == len(id_rels)
assert Relationship.query.count() == len(id_rels)
grp_rels = [r for r, t in zip(rel_map, relationship_types)
if t is not None]
# Make sure that all loaded groups relationships are unique
assert len(set(map(lambda x: x.id, grp_rels))) == len(grp_rels)
assert GroupRelationship.query.count() == len(grp_rels)
# Make sure that GroupRelationshipM2M are correct
id_grp_rels = [r for r, t in zip(rel_map, relationship_types)
if t == GroupType.Identity]
# There are as many GroupRelationshipM2M objects as Identity Groups
assert GroupRelationshipM2M.query.count() == len(id_grp_rels)
# Same number of GroupRelationshipMetadata as GRelationships of type ID
assert GroupRelationshipMetadata.query.count() == len(id_grp_rels)
# There are as many Relationship to GR items as Relationships
n_rel2grrels = sum([len(x[1]) for x in relationship_groups
if isinstance(rel_map[x[1][0]], Relationship)])
assert Relationship2GroupRelationship.query.count() == n_rel2grrels
# Make sure that all GroupRelationshipM2M are matching
for group_rel, group_subrels in relationship_groups:
for group_subrel in group_subrels:
if isinstance(rel_map[group_subrel], Relationship):
assert Relationship2GroupRelationship.query.filter_by(
relationship=rel_map[group_subrel],
group_relationship=rel_map[group_rel]).one()
else: # isinstance(rel_map[group_rel], GroupRelationship):
assert GroupRelationshipM2M.query.filter_by(
relationship=rel_map[group_rel],
subrelationship=rel_map[group_subrel]).one()