class ProblemLabel(db.Model):
__tablename__ = 'problem_label'
id = db.Column(
UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True
)
problem_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Problem.id),
nullable=False,
)
problem = db.relationship(Problem, backref='labels')
label = db.Column(
db.Unicode(255),
nullable=False,
)
order_index = db.Column(
db.Integer,
nullable=False,
)
def __repr__(self):
return '<Problemlabel label=%r>' % (self.label,)
class UserProblem(db.Model):
__tablename__ = 'user_problem'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
problem_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Problem.id),
nullable=False,
)
problem = db.relationship(Problem)
user_id = db.Column(UUIDType(binary=False),
db.ForeignKey(User.id),
nullable=False,
index=True)
user = db.relationship(User, backref='problems')
def __repr__(self):
return '<UserProblem problem=%r user=%r>' % (self.problem.label,
self.user.username)
class LabelEvent(db.Model):
__tablename__ = 'label_event'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
data_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Dataset.id, ondelete='CASCADE'),
nullable=False,
index=True,
)
data = db.relationship(Dataset, backref='label_events')
label_id = db.Column(
UUIDType(binary=False),
db.ForeignKey('problem_label.id', ondelete='CASCADE'),
nullable=False,
index=True,
)
label = db.relationship('ProblemLabel')
label_matches = db.Column(db.Boolean(), nullable=True)
created_at = db.Column(db.DateTime,
server_default=db.func.now(),
nullable=False)
def __repr__(self):
return '<LabelEvent label=%r>' % self.label
class User(db.Model, UserMixin):
__tablename__ = 'user'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
username = db.Column(db.Unicode(255),
nullable=False,
unique=True,
index=True)
password = db.Column(PasswordType(schemes=['pbkdf2_sha512']),
nullable=False)
created_at = db.Column(db.DateTime,
server_default=db.func.now(),
nullable=False)
is_superuser = db.Column(db.Boolean(),
nullable=False,
default=False,
server_default='FALSE')
@hybrid_method
def can_access_problem(self, problem):
from annotator.models import UserProblem
return self.is_superuser or db.session.query(UserProblem.id).filter(
UserProblem.user_id == self.id, UserProblem.problem_id
== problem.id).count() > 0
@can_access_problem.expression
def can_access_problem(cls, problem):
from annotator.models import Problem, UserProblem
return cls.is_superuser | Problem.id.in_(
db.session.query(
UserProblem.problem_id).filter(UserProblem.user_id == cls.id))
def __repr__(self):
return '<User username=%r>' % self.username
class Problem(db.Model):
__tablename__ = 'problem'
query_class = ProblemQuery
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
name = db.Column(
db.Unicode(255),
nullable=False,
)
created_at = db.Column(db.DateTime,
server_default=db.func.now(),
nullable=False)
classification_type = db.Column(db.Enum('binary',
'multi-label',
'multi-class',
name='classification_type_enum'),
default='binary',
server_default='binary')
@hybrid_property
def dataset_count(self):
from annotator.models import Dataset
return object_session(self).query(Dataset).filter(
Dataset.problem == self).count()
@dataset_count.expression
def dataset_count(cls):
from annotator.models import Dataset
return select([
func.count(Dataset.id)
]).where(Dataset.problem_id == cls.id).label('dataset_count')
def __repr__(self):
return '<Problem name=%r>' % self.name
class DatasetLabelProbability(db.Model):
__tablename__ = 'dataset_label_probability'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
data_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Dataset.id, ondelete='CASCADE'),
nullable=False,
index=True,
)
data = db.relationship(Dataset, backref='probabilities')
label_id = db.Column(
UUIDType(binary=False),
db.ForeignKey('problem_label.id', ondelete='CASCADE'),
nullable=False,
index=True,
)
label = db.relationship('ProblemLabel')
probability = db.Column(db.Float(), nullable=True)
__table_args__ = (
db.CheckConstraint((0 <= probability) & (probability <= 1),
name='chk_dataset_probability'),
db.UniqueConstraint(
'data_id',
'label_id',
name='uq_dataset_label_probability_data_id_label_id'),
)
def __repr__(self):
return '<Problemlabel label=%r>' % (self.label, )
class TrainingJob(db.Model):
__tablename__ = 'training_job'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
accuracy = db.Column(db.Float(), nullable=False)
problem_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Problem.id, ondelete='CASCADE'),
nullable=False,
index=True,
)
problem = db.relationship(Problem, backref='training_jobs')
created_at = db.Column(db.DateTime,
server_default=db.func.now(),
nullable=False)
def __repr__(self):
return '<TrainingJob problem=%r>' % self.problem.id
class Dataset(db.Model):
__tablename__ = 'dataset'
id = db.Column(UUIDType(binary=False),
server_default=db.func.uuid_generate_v4(),
primary_key=True)
entity_id = db.Column(db.Unicode(255), index=True)
problem_id = db.Column(
UUIDType(binary=False),
db.ForeignKey(Problem.id, ondelete='CASCADE'),
nullable=False,
index=True,
)
problem = db.relationship(Problem, backref='dataset')
table_name = db.Column(
db.Unicode(255),
nullable=False,
)
free_text = db.Column(db.Text(), nullable=False)
meta = db.Column(JSONB)
sort_value = db.Column(db.Float(), nullable=True, index=True)
__table_args__ = (db.UniqueConstraint(
'problem_id',
'table_name',
'entity_id',
name='uq_dataset_problem_idtable_name_entity_id'), )
def matching_label_events(self, matches=True):
return [x for x in self.label_events if x.label_matches == matches]
def __repr__(self):
return '<Dataset table_name=%r>' % self.table_name