def merge(self, other):
if self.id == other.id:
raise ValueError("Cannot merge an entity with itself.")
if self.collection_id != other.collection_id:
raise ValueError(
"Cannot merge entities from different collections.") # noqa
self.schema = model.precise_schema(self.schema, other.schema)
self.foreign_ids = string_set(self.foreign_ids, self.foreign_ids)
self.created_at = min((self.created_at, other.created_at))
self.updated_at = datetime.utcnow()
data = merge_data(self.data, other.data)
if self.name != other.name:
data = merge_data(data, {'alias': [other.name]})
self.data = data
# update alerts
from aleph.model.alert import Alert
q = db.session.query(Alert).filter(Alert.entity_id == other.id)
q.update({Alert.entity_id: self.id})
# delete source entities
other.delete()
db.session.add(self)
db.session.commit()
db.session.refresh(other)
def test_model_precise_schema(self):
assert model.precise_schema('Thing', 'Thing') == 'Thing'
assert model.precise_schema('Thing', 'Person') == 'Person'
assert model.precise_schema('Person', 'Thing') == 'Person'
assert model.precise_schema('Person', 'Company') == 'LegalEntity'
with assert_raises(InvalidData):
model.precise_schema('Person', 'Directorship')
def compare(left, right):
"""Compare two entities and return number between 0 and 1.
Returned number indicates probability that two entities are the same.
"""
left_schema = model.get(left.get('schema'))
right_schema = model.get(right.get('schema'))
if right_schema not in list(left_schema.matchable_schemata):
return 0
schema = model.precise_schema(left_schema, right_schema)
score = compare_fingerprints(left, right) * FP_WEIGHT
left_properties = left.get('properties', {})
right_properties = right.get('properties', {})
for name, prop in schema.properties.items():
weight = MATCH_WEIGHTS.get(prop.type, 0)
if weight == 0:
continue
left_values = left_properties.get(name)
right_values = right_properties.get(name)
prop_score = prop.type.compare_sets(left_values, right_values)
score = score + prop_score * weight
return max(0.0, min(1.0, score)) * 0.9
def test_model_precise_schema(self):
assert model.precise_schema('Thing', 'Thing') == 'Thing'
assert model.precise_schema('Thing', 'Person') == 'Person'
assert model.precise_schema('Person', 'Thing') == 'Person'
assert model.precise_schema('Person', 'Company') == 'LegalEntity'