def test_simulate_models():
assert parse_bql_string(
'simulate dependence probability of a with b from models of t;') == [
ast.SimulateModels([
ast.SimCol(ast.ExpBQLDepProb('a', 'b'), None),
], 't', None)
]
assert parse_bql_string(
'simulate dependence probability of a with b AS q, '
'mutual information of c with d given (e, r=2.7) '
'using 100 samples as g '
'from models of p modeled by z') == [
ast.SimulateModels([
ast.SimCol(ast.ExpBQLDepProb('a', 'b'), 'q'),
ast.SimCol(
ast.ExpBQLMutInf('c', 'd',
[('e', ast.ExpLit(ast.LitNull(0))),
('r', ast.ExpLit(ast.LitFloat(2.7)))],
ast.ExpLit(ast.LitInt(100))), 'g'),
], 'p', 'z')
]
assert parse_bql_string(
'simulate probability of (a=2, c=1.1) given (b=0.5) '
'from models of p') == [
ast.SimulateModels([
ast.SimCol(
ast.ExpBQLProb([('a', ast.ExpLit(ast.LitInt(2))),
('c', ast.ExpLit(ast.LitFloat(1.1)))], [
('b', ast.ExpLit(ast.LitFloat(0.5)))
]), None),
], 'p', None)
]
for temp, ifnotexists in itertools.product(('temp', ''),
('if not exists', '')):
assert parse_bql_string(
'create %s table %s f as '
'simulate dependence probability of a with b AS q, '
'mutual information of c with d '
'given (e, r=2.7) using 100 samples as g '
'from models of p modeled by z' % (temp, ifnotexists)) == [
ast.CreateTabSimModels(
bool(temp), bool(ifnotexists), 'f',
ast.SimulateModels([
ast.SimCol(ast.ExpBQLDepProb('a', 'b'), 'q'),
ast.SimCol(
ast.ExpBQLMutInf('c', 'd', [
('e', ast.ExpLit(ast.LitNull(0))),
('r', ast.ExpLit(ast.LitFloat(2.7)))
], ast.ExpLit(ast.LitInt(100))), 'g'),
], 'p', 'z'))
]
def test_is_bql():
assert ast.is_bql(ast.ExpLit(ast.LitInt(0))) == False
assert ast.is_bql(ast.ExpNumpar(0)) == False
assert ast.is_bql(ast.ExpNampar(0, 'x')) == False
assert ast.is_bql(ast.ExpCol('t', 'c')) == False
# ...
assert ast.is_bql(ast.ExpBQLPredProb('c'))
assert ast.is_bql(ast.ExpBQLProb([('c', ast.ExpLit(ast.LitInt(0)))], []))
assert ast.is_bql(ast.ExpBQLProbFn(ast.ExpLit(ast.LitInt(0)), []))
assert ast.is_bql(ast.ExpBQLSim(ast.ExpLit(ast.LitInt(0)), []))
assert ast.is_bql(ast.ExpBQLDepProb('c0', 'c1'))
assert ast.is_bql(ast.ExpBQLMutInf('c0', 'c1', None, 100))
assert ast.is_bql(ast.ExpBQLCorrel('c0', 'c1'))
assert ast.is_bql(ast.ExpBQLPredict('c', ast.ExpLit(ast.LitInt(.5)), None))
assert ast.is_bql(ast.ExpBQLPredictConf('c', None))
def p_bqlfn_depprob(self, cols):
return ast.ExpBQLDepProb(*cols)
def test_select_bql():
assert parse_bql_string('select predictive probability of c from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLPredProb('c'), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select predictive probability of c, * from t;') \
== \
[ast.Select(ast.SELQUANT_ALL,
[
ast.SelColExp(ast.ExpBQLPredProb('c'), None),
ast.SelColAll(None),
],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select c, predictive probability of d from t;') \
== \
[ast.Select(ast.SELQUANT_ALL,
[
ast.SelColExp(ast.ExpCol(None, 'c'), None),
ast.SelColExp(ast.ExpBQLPredProb('d'), None),
],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select predictive probability of c, d from t;') \
== \
[ast.Select(ast.SELQUANT_ALL,
[
ast.SelColExp(ast.ExpBQLPredProb('c'), None),
ast.SelColExp(ast.ExpCol(None, 'd'), None),
],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select probability of c = 42 from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLProb([('c', ast.ExpLit(ast.LitInt(42)))],
[]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select similarity from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLSim(None, [ast.ColListAll()]), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select similarity to (rowid=8) from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(8))
)),
[ast.ColListAll()]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select similarity with respect to c from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLSim(None, [ast.ColListLit(['c'])]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string(
'select similarity to (rowid=8) with respect to c from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(8)),
)),
[ast.ColListLit(['c'])]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string(
'select similarity to (rowid=5) with respect to age from t1;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(5)),
)),
[ast.ColListLit(['age'])]),
None)],
[ast.SelTab('t1', None)], None, None, None, None)]
assert parse_bql_string(
'select similarity to (rowid=8) with respect to c, d from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[
ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(8)),
)),
[ast.ColListLit(['c'])]),
None),
ast.SelColExp(ast.ExpCol(None, 'd'), None),
],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select similarity to (rowid=8)'
' with respect to (c, d) from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(8)),
)),
[ast.ColListLit(['c']), ast.ColListLit(['d'])]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select similarity to (rowid=8) with respect to' +
' (estimate * from columns of t order by ' +
' probability of value 4 limit 1)' +
' from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLSim(
ast.ExpOp(ast.OP_EQ, (
ast.ExpCol(None, 'rowid'),
ast.ExpLit(ast.LitInt(8)),
)),
[ast.ColListSub(
ast.EstCols([ast.SelColAll(None)], 't', None, None,
[ast.Ord(ast.ExpBQLProbFn(
ast.ExpLit(ast.LitInt(4)),
[]),
ast.ORD_ASC)],
ast.Lim(ast.ExpLit(ast.LitInt(1)), None))
)]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select dependence probability with c from t;') ==\
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLDepProb('c', None), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string(
'select dependence probability of c with d from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLDepProb('c', 'd'), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select mutual information with c from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLMutInf('c', None, None, None), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string(
'select mutual information of c with d from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLMutInf('c', 'd', None, None), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select mutual information of c with d' +
' using (1+2) samples from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLMutInf(
'c', 'd', None,
ast.op(
ast.OP_ADD, ast.ExpLit(ast.LitInt(1)),
ast.ExpLit(ast.LitInt(2)))),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('''
select mutual information with c given (d, a=1) using
10 samples from t;
''') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLMutInf(
'c', None,
[('d', ast.ExpLit(ast.LitNull(0))),
('a',ast.ExpLit(ast.LitInt(1)))],
ast.ExpLit(ast.LitInt(10))
),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('''
select mutual information of b with c
given (d, a=1, e, r=2) from t;''') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(
ast.ExpBQLMutInf(
'b', 'c',
[
('d', ast.ExpLit(ast.LitNull(0))),
('a',ast.ExpLit(ast.LitInt(1))),
('e', ast.ExpLit(ast.LitNull(0))),
('r', ast.ExpLit(ast.LitInt(2))),
],
None
),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select correlation with c from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLCorrel('c', None), None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select correlation of c with d from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLCorrel('c', 'd'), None)],
[ast.SelTab('t', None)], None, None, None, None)]
# XXX This got broken a while ago: parenthesization in PROBABILITY
# OF X = E is too permissive. I didn't notice because before I
# introduced BQLParseError, this simply caught Exception -- which
# covered the AssertionError that this turned into.
#
# with pytest.raises(parse.BQLParseError):
# parse_bql_string('select probability of x = 1 -' +
# ' probability of y = 0 from t;')
# # XXX Should really be this test, but getting the grammar to
# # admit this unambiguously is too much of a pain at the
# # moment.
# assert parse_bql_string('select probability of x = 1 -' +
# ' probability of y = 0 from t;') == \
# [ast.Select(ast.SELQUANT_ALL,
# [ast.SelColExp(ast.ExpBQLProb([('x',
# ast.ExpOp(ast.OP_SUB, (
# ast.ExpLit(ast.LitInt(1)),
# ast.ExpBQLProb([('y',
# ast.ExpLit(ast.LitInt(0)))],
# []),
# )))],
# []),
# None)],
# [ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select probability of c1 = f(c2) from t;') == \
[ast.Select(ast.SELQUANT_ALL,
[ast.SelColExp(ast.ExpBQLProb([('c1',
ast.ExpApp(False, 'f', [ast.ExpCol(None, 'c2')]))],
[]),
None)],
[ast.SelTab('t', None)], None, None, None, None)]
assert parse_bql_string('select key, t.(estimate * from columns of t'
' order by dependence probability with c desc limit 4)'
' from t order by key asc') == \
[ast.Select(ast.SELQUANT_ALL, [
ast.SelColExp(ast.ExpCol(None, 'key'), None),
ast.SelColSub('t',
ast.EstCols([ast.SelColAll(None)], 't', None, None,
[ast.Ord(ast.ExpBQLDepProb('c', None), ast.ORD_DESC)],
ast.Lim(ast.ExpLit(ast.LitInt(4)), None)))
],
[ast.SelTab('t', None)],
None, None,
[ast.Ord(ast.ExpCol(None, 'key'), ast.ORD_ASC)],
None)]
def p_bqlfn_depprob(self, cols): return ast.ExpBQLDepProb(*cols)
def p_bqlfn_mutinf(self, cols, constraints, nsamp):