def exact_aggregates(self, query):
child_scope = Scope()
if self.options.row_privacy:
keycount_expr = AggFunction("COUNT", None, AllColumns())
else:
key_col = self.key_col(query)
keycount_expr = AggFunction("COUNT", "DISTINCT", Column(key_col))
child_scope.push_name(keycount_expr.expression)
for ne in query.select.namedExpressions:
child_scope.push_name(ne.expression)
keycount = NamedExpression("keycount", keycount_expr)
select = Seq([keycount] + [ne for ne in query.select.namedExpressions])
select = Select(None, select)
subquery = Query(child_scope.select(), query.source, query.where, query.agg, None, None, None)
if self.options.reservoir_sample and not self.options.row_privacy:
subquery = self.per_key_random(subquery)
subquery = [AliasedRelation(subquery, "per_key_random")]
filtered = Where(BooleanCompare(Column("per_key_random.row_num"), "<=", Literal(str(self.options.max_contrib), self.options.max_contrib)))
return Query(select, From(subquery), filtered, query.agg, None, None, None)
else:
subquery = self.per_key_clamped(subquery)
subquery = [AliasedRelation(subquery, "per_key_all")]
return Query(select, From(subquery), None, query.agg, None, None, None)
def calculate_variance(self, exp, scope):
"""
Calculate the variance from avg of squares and square of averages
"""
expr = exp.expression
quant = exp.quantifier
avg_of_square = self.calculate_avg(AggFunction("AVG", quant, ArithmeticExpression(expr, '*', expr)), scope)
avg = self.calculate_avg(AggFunction("AVG", quant, expr), scope)
avg_squared = ArithmeticExpression(avg, '*', avg)
new_exp = ArithmeticExpression(avg_of_square, "-", avg_squared)
return new_exp
def calculate_avg(self, exp, scope):
"""
Takes an expression for a noisy mean and rewrites
to a noisy sum and a noisy count
"""
expr = exp.expression
quant = exp.quantifier
sum_expr = self.push_sum_or_count(AggFunction("SUM", quant, expr), scope)
count_expr = self.push_sum_or_count(AggFunction("COUNT", quant, expr), scope)
new_exp = NestedExpression(ArithmeticExpression(sum_expr, "/", count_expr))
return new_exp
def calculate_stddev(self, exp, scope):
"""
Calculate the standard deviation from the variance
"""
expr = AggFunction('STD', exp.quantifier, exp.expression)
var_expr = self.calculate_variance(expr, scope)
new_exp = MathFunction("SQRT", var_expr)
return new_exp
def push_sum_or_count(self, exp, scope):
"""
Push a sum or count expression to child scope
and convert to a sum
"""
new_name = scope.push_name(AggFunction(exp.name, exp.quantifier, exp.expression))
new_exp = Column(new_name)
return new_exp