def test_format_simple():
assert format_expr(BinaryOp('+', 1, 2)) == '1 + 2'
assert format_expr(BinaryOp('-', 1, 2)) == '1 - 2'
assert format_expr(BinaryOp('*', 1, 2)) == '1 * 2'
assert format_expr(BinaryOp('/', 1, 2)) == '1 / 2'
assert format_expr(UnaryOp('+', 1)) == '+1'
assert format_expr(UnaryOp('-', 1)) == '-1'
def test_parser(self):
e1 = "1"
self.assertEqual(parse(e1), NumberLit(1))
e2 = "(1)"
self.assertEqual(parse(e2), NumberLit(1))
e3 = "((1))"
self.assertEqual(parse(e3), NumberLit(1))
e4 = "(1 + 2)"
r4 = BinaryOp(NumberLit(1), Op.Plus, NumberLit(2))
self.assertEqual(parse(e4), r4)
e5 = "1 + 2"
self.assertEqual(parse(e5), r4)
e6 = "(1 * 2)"
r6 = BinaryOp(NumberLit(1), Op.Mult, NumberLit(2))
self.assertEqual(parse(e6), r6)
e7 = "1 * 2"
self.assertEqual(parse(e7), r6)
e8 = "1 + 2 * 3"
r8 = BinaryOp(NumberLit(1), Op.Plus,
BinaryOp(NumberLit(2), Op.Mult, NumberLit(3)))
self.assertEqual(parse(e8), r8)
def compute(self, context, expr, filter=None, skip_na=True):
# FIXME: either take "contextual filter" into account here (by using
# self._getfilter), or don't do it in sum & gini
if filter is not None:
tmpvar = self.add_tmp_var(context, filter)
if getdtype(expr, context) is bool:
# convert expr to int because mul_bbb is not implemented in
# numexpr
# expr *= 1
expr = BinaryOp('*', expr, 1)
# expr *= filter_values
expr = BinaryOp('*', expr, tmpvar)
else:
filter = True
values = expr_eval(expr, context)
values = np.asarray(values)
if skip_na:
# we should *not* use an inplace operation because filter can be a
# simple variable
filter = filter & ispresent(values)
if filter is True:
numrows = len(values)
else:
numrows = np.sum(filter)
if numrows:
if skip_na:
return na_sum(values) / float(numrows)
else:
return np.sum(values) / float(numrows)
else:
return float('nan')
def build_regression_expr(self, expr, mult=0.0, error_var=None):
if error_var is not None:
# expr += error_var
expr = BinaryOp('+', expr, Variable(None, error_var))
if mult:
# expr += normal(0, 1) * mult
expr = BinaryOp('+', expr, BinaryOp('*', Normal(0, 1), mult))
return expr
def test_expr_str(self):
n = NumberLit(11)
m = NumberLit(22)
self.assertEqual(str(n), "11")
self.assertEqual(str(m), "22")
e1 = BinaryOp(n, Op.Plus, m)
self.assertEqual(str(e1), "(11 + 22)")
e2 = BinaryOp(n, Op.Mult, m)
self.assertEqual(str(e2), "(11 * 22)")
e3 = BinaryOp(e1, Op.Mult, e1)
self.assertEqual(str(e3), "((11 + 22) * (11 + 22))")
def test_evaluator(self):
n = NumberLit(2)
m = NumberLit(3)
self.assertEqual(evaluate(n), 2)
self.assertEqual(evaluate(m), 3)
e1 = BinaryOp(n, Op.Plus, m)
self.assertEqual(evaluate(e1), 5)
e2 = BinaryOp(n, Op.Mult, m)
self.assertEqual(evaluate(e2), 6)
e3 = BinaryOp(e1, Op.Mult, e1)
self.assertEqual(evaluate(e3), 25)
def compute(self, context, expr, filter=None, skip_na=True):
filter_expr = self._getfilter(context, filter)
if filter_expr is not None:
expr = BinaryOp('*', expr, filter_expr)
values = expr_eval(expr, context)
values = np.asarray(values)
return na_sum(values) if skip_na else np.sum(values)
def test_expr_str(self):
n = NumberLit(11)
m = NumberLit(22)
self.assertEqual(str(n), "11")
self.assertEqual(str(m), "22")
x = Variable("x")
self.assertEqual(str(x), "x")
e1 = BinaryOp(n, Op.Plus, m)
self.assertEqual(str(e1), "(11 + 22)")
e2 = BinaryOp(n, Op.Mult, m)
self.assertEqual(str(e2), "(11 * 22)")
e3 = BinaryOp(e1, Op.Mult, e1)
self.assertEqual(str(e3), "((11 + 22) * (11 + 22))")
y = Variable("y")
self.assertEqual(str(y), "y")
e4 = BinaryOp(x, Op.Plus, y)
self.assertEqual(str(e4), "(x + y)")
e5 = LetIn("x", n, x)
self.assertEqual(str(e5), "(let x := 11 in x)")
e6 = LetIn("x", n, BinaryOp(x, Op.Plus, x))
self.assertEqual(str(e6), "(let x := 11 in (x + x))")
def build_expr(self, context, expr):
if isinstance(expr, basestring):
# assume it is a filename
expr = ExtExpr(expr)
u = Uniform()
# expr in (0, 0.0, False, '')
if not isinstance(expr, Expr) and not expr:
expr = u
else:
epsilon = Logit(u)
# expr = logistic(expr - epsilon)
expr = Logistic(BinaryOp('-', expr, epsilon))
return expr
def build_expr(self, context, expr):
# log(x / (1 - x))
return Log(DivisionOp('/', expr, BinaryOp('-', 1.0, expr)))
def test_parser(self):
v1 = "x"
self.assertEqual(parse(v1), Variable("x"))
v2 = "(x)"
self.assertEqual(parse(v2), Variable("x"))
v3 = "thisisalongvariable"
self.assertEqual(parse(v3), Variable(v3))
e1 = "1"
self.assertEqual(parse(e1), NumberLit(1))
e2 = "(1)"
self.assertEqual(parse(e2), NumberLit(1))
e3 = "((1))"
self.assertEqual(parse(e3), NumberLit(1))
e4 = "(1 + 2)"
r4 = BinaryOp(
NumberLit(1),
Op.Plus,
NumberLit(2))
self.assertEqual(parse(e4), r4)
e5 = "1 + 2"
self.assertEqual(parse(e5), r4)
e6 = "(1 * 2)"
r6 = BinaryOp(
NumberLit(1),
Op.Mult,
NumberLit(2))
self.assertEqual(parse(e6), r6)
e7 = "1 * 2"
self.assertEqual(parse(e7), r6)
e8 = "1 + 2 * 3"
r8 = BinaryOp(
NumberLit(1),
Op.Plus,
BinaryOp(
NumberLit(2),
Op.Mult,
NumberLit(3)
))
self.assertEqual(parse(e8), r8)
e9 = "let x := 1 in x end"
r9 = LetIn(
"x",
NumberLit(1),
Variable("x")
)
self.assertEqual(parse(e9), r9)
e10 = "2 + let x := 1 in x end"
r10 = BinaryOp(
NumberLit(2),
Op.Plus,
LetIn(
"x",
NumberLit(1),
Variable("x")
))
self.assertEqual(parse(e10), r10)
e11 = "let x := 1 in x + 2 end"
r11 = LetIn(
"x",
NumberLit(1),
BinaryOp(
Variable("x"),
Op.Plus,
NumberLit(2)
))
self.assertEqual(parse(e11), r11)
e12 = "let x := 1 in x end + 2"
r12 = BinaryOp(
LetIn(
"x",
NumberLit(1),
Variable("x")),
Op.Plus,
NumberLit(2)
)
self.assertEqual(parse(e12), r12)
def _mul(a, b):
return BinaryOp('*', a, b)
def _plus(a, b):
return BinaryOp('+', a, b)
def build_expr(self, context, expr):
# 1 / (1 + exp(-x))
return DivisionOp('/', 1.0, BinaryOp('+', 1.0, Exp(UnaryOp('-',
expr))))
def test_simplify():
# Constants
assert format_expr(simplify_expr(BinaryOp('+', 1, 3))) == '4'
assert format_expr(simplify_expr(UnaryOp('-', 1))) == '-1'
# Variables
assert format_expr(simplify_expr(BinaryOp('+', 2, VarExpr('x')))) == '2 + x'
assert format_expr(simplify_expr(BinaryOp('-', 2, VarExpr('x')))) == '2 - x'
assert format_expr(simplify_expr(BinaryOp('*', 2, VarExpr('x')))) == '2 * x'
assert format_expr(simplify_expr(BinaryOp('/', 2, VarExpr('x')))) == '2 / x'
assert format_expr(simplify_expr(BinaryOp('+', VarExpr('x'), 2))) == 'x + 2'
assert format_expr(simplify_expr(BinaryOp('-', VarExpr('x'), 2))) == 'x - 2'
assert format_expr(simplify_expr(BinaryOp('*', VarExpr('x'), 2))) == 'x * 2'
assert format_expr(simplify_expr(BinaryOp('/', VarExpr('x'), 2))) == 'x / 2'
# Add zero
assert format_expr(simplify_expr(BinaryOp('+', 0, VarExpr('x')))) == 'x'
assert format_expr(simplify_expr(BinaryOp('+', VarExpr('x'), 0))) == 'x'
# Subtract zero
assert format_expr(simplify_expr(BinaryOp('-', 0, VarExpr('x')))) == '-x'
assert format_expr(simplify_expr(BinaryOp('-', VarExpr('x'), 0))) == 'x'
# Multiply zero
assert format_expr(simplify_expr(BinaryOp('*', 0, VarExpr('x')))) == '0'
assert format_expr(simplify_expr(BinaryOp('*', VarExpr('x'), 0))) == '0'
# Multiply one
assert format_expr(simplify_expr(BinaryOp('*', 1, VarExpr('x')))) == 'x'
assert format_expr(simplify_expr(BinaryOp('*', VarExpr('x'), 1))) == 'x'
# Divide one
assert format_expr(simplify_expr(BinaryOp('/', VarExpr('x'), 1))) == 'x'
# Nested
assert format_expr(simplify_expr(
BinaryOp('+',
BinaryOp('*', 1, VarExpr('x')),
BinaryOp('+', 0, VarExpr('y'))
))) == 'x + y'
# Unary plus
assert format_expr(simplify_expr(UnaryOp('+', VarExpr('x')))) == 'x'
# Double minus
assert format_expr(simplify_expr(UnaryOp('-', UnaryOp('-', VarExpr('x'))))) == 'x'
def test_format_complex():
# Same operator
assert format_expr(BinaryOp('+', BinaryOp('+', 1, 2), 3)) == '1 + 2 + 3'
assert format_expr(BinaryOp('+', 1, BinaryOp('+', 2, 3))) == '1 + (2 + 3)'
assert format_expr(BinaryOp('-', BinaryOp('-', 1, 2), 3)) == '1 - 2 - 3'
assert format_expr(BinaryOp('-', 1, BinaryOp('-', 2, 3))) == '1 - (2 - 3)'
assert format_expr(BinaryOp('*', BinaryOp('*', 1, 2), 3)) == '1 * 2 * 3'
assert format_expr(BinaryOp('*', 1, BinaryOp('*', 2, 3))) == '1 * (2 * 3)'
assert format_expr(BinaryOp('/', BinaryOp('/', 1, 2), 3)) == '1 / 2 / 3'
assert format_expr(BinaryOp('/', 1, BinaryOp('/', 2, 3))) == '1 / (2 / 3)'
# Equal precedence
assert format_expr(BinaryOp('+', BinaryOp('-', 1, 2), 3)) == '1 - 2 + 3'
assert format_expr(BinaryOp('-', BinaryOp('+', 1, 2), 3)) == '1 + 2 - 3'
assert format_expr(BinaryOp('+', 1, BinaryOp('-', 2, 3))) == '1 + (2 - 3)'
assert format_expr(BinaryOp('-', 1, BinaryOp('+', 2, 3))) == '1 - (2 + 3)'
assert format_expr(BinaryOp('*', BinaryOp('/', 1, 2), 3)) == '1 / 2 * 3'
assert format_expr(BinaryOp('/', BinaryOp('*', 1, 2), 3)) == '1 * 2 / 3'
assert format_expr(BinaryOp('*', 1, BinaryOp('/', 2, 3))) == '1 * (2 / 3)'
assert format_expr(BinaryOp('/', 1, BinaryOp('*', 2, 3))) == '1 / (2 * 3)'
# Different precedence
assert format_expr(BinaryOp('+', BinaryOp('*', 1, 2), 3)) == '1 * 2 + 3'
assert format_expr(BinaryOp('*', BinaryOp('+', 1, 2), 3)) == '(1 + 2) * 3'
assert format_expr(BinaryOp('+', 1, BinaryOp('*', 2, 3))) == '1 + 2 * 3'
assert format_expr(BinaryOp('*', 1, BinaryOp('+', 2, 3))) == '1 * (2 + 3)'
def test_eval_binary():
assert eval_expr(BinaryOp('+', 1, 2)) == 3
assert eval_expr(BinaryOp('-', 1, 2)) == -1
assert eval_expr(BinaryOp('*', 1, 2)) == 2
assert eval_expr(BinaryOp('/', 1, 2)) == 0.5
