def to_dict(self):
from nomenklatura.model.entity import Entity
num_aliases = Entity.all(self).filter(Entity.canonical_id!=None).count()
num_review = Entity.all(self).filter_by(reviewed=False).count()
num_entities = Entity.all(self).count()
num_invalid = Entity.all(self).filter_by(invalid=True).count()
return {
'id': self.id,
'name': self.name,
'label': self.label,
'owner': self.owner.to_dict(),
'stats': {
'num_aliases': num_aliases,
'num_entities': num_entities,
'num_review': num_review,
'num_invalid': num_invalid
},
'ignore_case': self.ignore_case,
'match_aliases': self.match_aliases,
'public_edit': self.public_edit,
'normalize_text': self.normalize_text,
'enable_invalid': self.enable_invalid,
'created_at': self.created_at,
'updated_at': self.updated_at
}
def to_dict(self):
from nomenklatura.model.entity import Entity
num_aliases = Entity.all(self).filter(
Entity.canonical_id != None).count()
num_review = Entity.all(self).filter_by(reviewed=False).count()
num_entities = Entity.all(self).count()
num_invalid = Entity.all(self).filter_by(invalid=True).count()
return {
'id': self.id,
'name': self.name,
'label': self.label,
'owner': self.owner.to_dict(),
'stats': {
'num_aliases': num_aliases,
'num_entities': num_entities,
'num_review': num_review,
'num_invalid': num_invalid
},
'ignore_case': self.ignore_case,
'match_aliases': self.match_aliases,
'public_edit': self.public_edit,
'normalize_text': self.normalize_text,
'enable_invalid': self.enable_invalid,
'created_at': self.created_at,
'updated_at': self.updated_at
}
def lookup(cls, dataset, data, account, match_entity=True, readonly=False):
data = AliasLookupSchema().to_python(data)
if match_entity:
entity = Entity.by_name(dataset, data['name'])
if entity is not None:
return entity
else:
entity = None
alias = cls.by_name(dataset, data['name'])
if alias is not None:
return alias
choices = match_op(data['name'], dataset)
choices = filter(lambda (c, v, s): s > 99.9, choices)
if len(choices) == 1:
c, entity_id, s = choices.pop()
entity = Entity.by_id(dataset, entity_id)
if readonly:
return entity
alias = cls()
alias.creator = account
alias.dataset = dataset
alias.entity = entity
alias.is_matched = entity is not None
alias.name = data['name']
alias.data = data['data']
db.session.add(alias)
db.session.flush()
if entity is not None:
add_candidate_to_cache(dataset, alias.name, entity.id)
return alias
def lookup(cls, dataset, data, account, match_entity=True,
readonly=False):
data = AliasLookupSchema().to_python(data)
if match_entity:
entity = Entity.by_name(dataset, data['name'])
if entity is not None:
return entity
else:
entity = None
alias = cls.by_name(dataset, data['name'])
if alias is not None:
return alias
choices = match_op(data['name'], dataset)
choices = filter(lambda (c,v,s): s > 99.9, choices)
if len(choices)==1:
c, entity_id, s = choices.pop()
entity = Entity.by_id(dataset, entity_id)
if readonly:
return entity
alias = cls()
alias.creator = account
alias.dataset = dataset
alias.entity = entity
alias.is_matched = entity is not None
alias.name = data['name']
alias.data = data['data']
db.session.add(alias)
db.session.flush()
if entity is not None:
add_candidate_to_cache(dataset, alias.name, entity.id)
return alias
def __iter__(self):
rp = db.engine.execute(self.lq)
rows = rp.fetchall()
ids = [r[0] for r in rows]
entities = Entity.id_map(ids)
for (id, score) in rows:
yield {'score': int(score), 'entity': entities.get(id)}
def __iter__(self):
rp = db.engine.execute(self.lq)
rows = rp.fetchall()
ids = [r[0] for r in rows]
entities = Entity.id_map(ids)
for (id, score) in rows:
yield {'score': int(score), 'entity': entities.get(id)}
def _to_python(self, value, state):
if value == 'NEW':
return value
elif value == 'INVALID' and state.dataset.enable_invalid:
return value
entity = Entity.by_id(state.dataset, value)
if entity is not None:
return entity
raise Invalid('No such entity.', value, None)
def _to_python(self, value, state):
if value == 'NEW':
return value
elif value == 'INVALID' and state.dataset.enable_invalid:
return value
entity = Entity.by_id(state.dataset, value)
if entity is not None:
return entity
raise Invalid('No such entity.', value, None)
def match(self, dataset, data, account):
state = AliasMatchState(dataset)
data = AliasMatchSchema().to_python(data, state)
self.is_matched = True
self.matcher = account
if data['choice'] == 'INVALID':
self.entity = None
self.is_invalid = True
elif data['choice'] == 'NEW':
self.entity = Entity.create(dataset, data, account)
self.is_invalid = False
else:
self.entity = data['choice']
self.is_invalid = False
db.session.add(self)
db.session.flush()
def match(self, dataset, data, account):
state = AliasMatchState(dataset)
data = AliasMatchSchema().to_python(data, state)
self.is_matched = True
self.matcher = account
if data['choice'] == 'INVALID':
self.entity = None
self.is_invalid = True
elif data['choice'] == 'NEW':
self.entity = Entity.create(dataset, data, account)
self.is_invalid = False
else:
self.entity = data['choice']
self.is_invalid = False
db.session.add(self)
db.session.flush()
def delete(self):
from nomenklatura.model.entity import Entity
Entity.all(self).delete()
db.session.delete(self)
db.session.flush()
