def add(self, subject, canonical):
subject, _ = Namespace.parse(get_entity_id(subject))
canonical, _ = Namespace.parse(get_entity_id(canonical))
# Don't do no-ops.
if subject == canonical:
return
if subject is None or canonical is None:
return
cluster = Cluster(canonical, subject)
cluster = self.clusters.get(canonical, cluster)
if subject in self.clusters:
previous = self.clusters.get(subject)
cluster.update(previous.entities)
for entity in cluster.entities:
self.clusters[entity] = cluster
def delete_aggregator_entity(collection, entity_id):
aggregator = get_aggregator(collection)
try:
entity_id = collection.ns.sign(entity_id)
aggregator.delete(entity_id=entity_id)
base_id, _ = Namespace.parse(entity_id)
aggregator.delete(entity_id=base_id)
finally:
aggregator.close()
def by_id(cls, document_id, collection=None):
try:
document_id, _ = Namespace.parse(document_id)
document_id = int(document_id)
except Exception:
return
q = cls.all()
q = q.filter(cls.id == document_id)
if collection is not None:
q = q.filter(cls.collection_id == collection.id)
return q.first()
def save(cls,
session,
subject,
candidate,
score=None,
judgement=None,
priority=None):
obj = cls.by_id(session, subject, candidate)
if obj is None:
obj = cls()
obj.id = cls.make_id(subject, candidate)
obj.subject, _ = Namespace.parse(get_entity_id(subject))
obj.candidate, _ = Namespace.parse(get_entity_id(candidate))
priority = priority or DEFAULT_PRIORITY
if score is not None:
obj.score = score
obj.priority = score * priority
if judgement is not None:
obj.judgement = judgement
obj.updated_at = now()
session.add(obj)
return obj
def by_id(cls, entity_id, collection_id=None):
entity_id, _ = Namespace.parse(entity_id)
q = cls.all()
q = q.filter(cls.id == entity_id)
return q.first()
def check_entity_id(value):
value, _ = Namespace.parse(value)
if not registry.entity.validate(value):
msg = gettext('Invalid entity ID: %s')
raise ValueError(msg % value)
return True
def make_id(cls, subject, candidate):
subject, _ = Namespace.parse(get_entity_id(subject))
candidate, _ = Namespace.parse(get_entity_id(candidate))
return '.'.join((subject, candidate))
def __init__(self, subject, canonical, judgement):
self.subject, _ = Namespace.parse(get_entity_id(subject))
self.canonical, _ = Namespace.parse(get_entity_id(canonical))
self.judgement = judgement or self.UNSURE
