def make_array_type(nullable, value_type=value_type):
return dt.Array(value_type, nullable=nullable)
def test_array_with_string_value_type():
assert dt.Array('int32') == dt.Array(dt.int32)
assert dt.Array(dt.Array('array<map<string, double>>')) == (dt.Array(
dt.Array(dt.Array(dt.Map(dt.string, dt.double)))))
class MyOp(ops.ValueOp):
value = rlz.value(dt.Array(dt.double))
output_type = rlz.typeof('value')
def test_complex_datatype_builtins(benchmark, func):
datatype = dt.Array(
dt.Struct.from_dict(
dict(a=dt.Array(dt.string),
b=dt.Map(dt.string, dt.Array(dt.int64)))))
benchmark(func, datatype)
def times_two(x, scope=None):
return x * 2.0
@udf.analytic(input_type=['double'], output_type='double')
def zscore(series):
return (series - series.mean()) / series.std()
@udf.elementwise([], dt.int64)
def a_single_number(**kwargs):
return 1
@udf.reduction(
input_type=[dt.double, dt.Array(dt.double)],
output_type=dt.Array(dt.double),
)
def quantiles(series, quantiles):
return list(series.quantile(quantiles))
def test_udf(t, df):
expr = my_string_length(t.a)
assert isinstance(expr, ir.ColumnExpr)
result = expr.execute()
expected = df.a.str.len().mul(2)
tm.assert_series_equal(result, expected)
@pytest.fixture(scope='module')
def df3():
return pd.DataFrame(
{
'key': list('ac'),
'other_value': [4.0, 6.0],
'key2': list('ae'),
'key3': list('fe'),
}
)
t_schema = {
'decimal': dt.Decimal(4, 3),
'array_of_float64': dt.Array(dt.double),
'array_of_int64': dt.Array(dt.int64),
'array_of_strings': dt.Array(dt.string),
'map_of_strings_integers': dt.Map(dt.string, dt.int64),
'map_of_integers_strings': dt.Map(dt.int64, dt.string),
'map_of_complex_values': dt.Map(dt.string, dt.Array(dt.int64)),
}
@pytest.fixture(scope='module')
def t(client):
return client.table('df', schema=t_schema)
@pytest.fixture(scope='module')
def lahman(batting_df, awards_players_df):
}
return my_len(s);
""";
SELECT (my_len_0('abcd') + my_len_0('abcd')) + my_len_1('abcd') AS `tmp`'''
assert sql == expected
@pytest.mark.parametrize(
('argument_type', 'return_type'),
[
param(dt.int64, dt.float64, marks=pytest.mark.xfail(raises=TypeError)),
param(dt.float64, dt.int64, marks=pytest.mark.xfail(raises=TypeError)),
# complex argument type, valid return type
param(
dt.Array(dt.int64),
dt.float64,
marks=pytest.mark.xfail(raises=TypeError),
),
# valid argument type, complex invalid return type
param(
dt.float64,
dt.Array(dt.int64),
marks=pytest.mark.xfail(raises=TypeError),
),
# both invalid
param(
dt.Array(dt.Array(dt.int64)),
dt.int64,
marks=pytest.mark.xfail(raises=TypeError),
),
def test_array():
assert dt.dtype('ARRAY<DOUBLE>') == dt.Array(dt.double)
def test_nested_array():
assert dt.dtype('array<array<string>>') == dt.Array(dt.Array(dt.string))
def test_nested_map():
assert dt.validate_type('map<int64, array<map<string, int8>>>') == dt.Map(
dt.int64, dt.Array(dt.Map(dt.string, dt.int8)))
# numpy types
(np.int8(5), dt.int8),
(np.int16(-1), dt.int16),
(np.int32(2), dt.int32),
(np.int64(-5), dt.int64),
(np.uint8(5), dt.uint8),
(np.uint16(50), dt.uint16),
(np.uint32(500), dt.uint32),
(np.uint64(5000), dt.uint64),
(np.float32(5.5), dt.float32),
(np.float32(5.5), dt.float),
(np.float64(5.55), dt.float64),
(np.float64(5.55), dt.double),
(np.bool_(True), dt.boolean),
(np.bool_(False), dt.boolean),
(np.arange(5, dtype='int32'), dt.Array(dt.int32)),
# pandas types
(pd.Timestamp('2015-01-01 12:00:00',
tz='US/Eastern'), dt.Timestamp('US/Eastern')),
# parametric types
(list('abc'), dt.Array(dt.string)),
([1, 2, 3], dt.Array(dt.int8)),
([1, 128], dt.Array(dt.int16)),
([1, 128, 32768], dt.Array(dt.int32)),
([1, 128, 32768, 2147483648], dt.Array(dt.int64)),
({
'a': 1,
'b': 2,
'c': 3
def test_whole_schema():
customers = ibis.table(
[('cid', 'int64'), ('mktsegment', 'string'),
('address', ('struct<city: string, street: string, '
'street_number: int32, zip: int16>')),
('phone_numbers', 'array<string>'),
('orders', """array<struct<
oid: int64,
status: string,
totalprice: decimal(12, 2),
order_date: string,
items: array<struct<
iid: int64,
name: string,
price: decimal(12, 2),
discount_perc: decimal(12, 2),
shipdate: string
>>
>>"""),
('web_visits', ('map<string, struct<user_agent: string, '
'client_ip: string, visit_date: string, '
'duration_ms: int32>>')),
('support_calls', ('array<struct<agent_id: int64, '
'call_date: string, duration_ms: int64, '
'issue_resolved: boolean, '
'agent_comment: string>>'))],
name='customers',
)
expected = ibis.Schema.from_tuples([
('cid', dt.int64),
('mktsegment', dt.string),
(
'address',
dt.Struct.from_tuples([('city', dt.string), ('street', dt.string),
('street_number', dt.int32),
('zip', dt.int16)]),
),
('phone_numbers', dt.Array(dt.string)),
('orders',
dt.Array(
dt.Struct.from_tuples([('oid', dt.int64), ('status', dt.string),
('totalprice', dt.Decimal(12, 2)),
('order_date', dt.string),
('items',
dt.Array(
dt.Struct.from_tuples([
('iid', dt.int64),
('name', dt.string),
('price', dt.Decimal(12, 2)),
('discount_perc', dt.Decimal(
12, 2)),
('shipdate', dt.string),
])))]))),
('web_visits',
dt.Map(
dt.string,
dt.Struct.from_tuples([
('user_agent', dt.string),
('client_ip', dt.string),
('visit_date', dt.string),
('duration_ms', dt.int32),
]))),
('support_calls',
dt.Array(
dt.Struct.from_tuples([('agent_id', dt.int64),
('call_date', dt.string),
('duration_ms', dt.int64),
('issue_resolved', dt.boolean),
('agent_comment', dt.string)]))),
], )
assert customers.schema() == expected
class StringSplit(Value):
arg = rlz.string
delimiter = rlz.string
output_shape = rlz.shape_like("arg")
output_dtype = dt.Array(dt.string)
@pytest.mark.parametrize(
('datatype', 'expected'),
[
(dt.float32, 'FLOAT64'),
(dt.float64, 'FLOAT64'),
(dt.uint8, 'INT64'),
(dt.uint16, 'INT64'),
(dt.uint32, 'INT64'),
(dt.int8, 'INT64'),
(dt.int16, 'INT64'),
(dt.int32, 'INT64'),
(dt.int64, 'INT64'),
(dt.string, 'STRING'),
(dt.Array(dt.int64), 'ARRAY<INT64>'),
(dt.Array(dt.string), 'ARRAY<STRING>'),
(
dt.Struct.from_tuples([('a', dt.int64), ('b', dt.string),
('c', dt.Array(dt.string))]),
'STRUCT<a INT64, b STRING, c ARRAY<STRING>>',
),
(dt.date, 'DATE'),
(dt.timestamp, 'TIMESTAMP'),
param(
dt.Timestamp(timezone='US/Eastern'),
'TIMESTAMP',
marks=pytest.mark.xfail(raises=TypeError,
reason='Not supported in BigQuery'),
),
('array<struct<a: string>>', 'ARRAY<STRUCT<a STRING>>'),
def infer_array(value):
# TODO(kszucs): infer series
return dt.Array(dt.dtype(value.dtype.name))
def test_nested_map():
expected = dt.Map(dt.int64, dt.Array(dt.Map(dt.string, dt.int8)))
assert dt.dtype('map<int64, array<map<string, int8>>>') == expected
return x + 1.0
@udf.elementwise([dt.double], dt.double)
def times_two(x):
return x * 2.0
@udf.analytic(input_type=['double'], output_type='double')
def zscore(series):
return (series - series.mean()) / series.std()
@udf.reduction(
input_type=[dt.double],
output_type=dt.Array(dt.double),
)
def quantiles(series, *, quantiles):
return np.array(series.quantile(quantiles))
def test_udf(t, df):
expr = my_string_length(t.a)
assert isinstance(expr, ir.ColumnExpr)
result = expr.execute()
expected = df.a.str.len().mul(2)
tm.assert_series_equal(result, expected)
kwargs = {kind: (begin, end)}
with pytest.raises(com.IbisInputError):
ibis.window(**kwargs)
@pytest.mark.parametrize(
('left', 'right', 'expected'),
[
(ibis.literal(1), ibis.literal(1.0), dt.float64),
(ibis.literal('a'), ibis.literal('b'), dt.string),
(ibis.literal(1.0), ibis.literal(1), dt.float64),
(ibis.literal(1), ibis.literal(1), dt.int8),
(ibis.literal(1), ibis.literal(1000), dt.int16),
(ibis.literal(2**16), ibis.literal(2**17), dt.int32),
(ibis.literal(2**50), ibis.literal(1000), dt.int64),
(ibis.literal([1, 2]), ibis.literal([1, 2]), dt.Array(dt.int8)),
(ibis.literal(['a']), ibis.literal([]), dt.Array(dt.string)),
(ibis.literal([]), ibis.literal(['a']), dt.Array(dt.string)),
(ibis.literal([]), ibis.literal([]), dt.Array(dt.null)),
],
)
def test_nullif_type(left, right, expected):
assert left.nullif(right).type() == expected
@pytest.mark.parametrize(('left', 'right'),
[(ibis.literal(1), ibis.literal('a'))])
def test_nullif_fail(left, right):
with pytest.raises(com.IbisTypeError):
left.nullif(right)
with pytest.raises(com.IbisTypeError):
def test_array_schema(array_types):
assert array_types.x.type() == dt.Array(dt.int64)
assert array_types.y.type() == dt.Array(dt.string)
assert array_types.z.type() == dt.Array(dt.double)
class FooNode(ops.ValueOp):
value = Arg(rlz.value(dt.Array(dt.int64)))
def output_type(self):
return Foo
@pytest.mark.parametrize(
"dtypes",
[
pytest.param(
[
obj for _, obj in inspect.getmembers(
dt,
lambda obj: isinstance(obj, dt.DataType),
)
],
id="singletons",
),
pytest.param(
dt.Array(
dt.Struct.from_dict(
dict(
a=dt.Array(dt.string),
b=dt.Map(dt.string, dt.Array(dt.int64)),
))),
id="complex",
),
],
)
def test_eq_datatypes(benchmark, dtypes):
def eq(a, b):
assert a == b
benchmark(eq, dtypes, copy.deepcopy(dtypes))
def multiple_joins(table, num_joins):
for _ in range(num_joins):
def test_map_keys_output_type():
mapping = ibis.literal({'a': 1, 'b': 2})
assert mapping.keys().type() == dt.Array(dt.string)
def test_array_schema(array_types, column, value_type):
assert array_types[column].type() == dt.Array(value_type)
def test_map_values_output_type():
mapping = ibis.literal({'a': 1, 'b': 2})
assert mapping.values().type() == dt.Array(dt.int8)
def test_columns_types_with_additional_argument(con):
sql_types = ["toFixedString('foo', 8) AS fixedstring_col"]
if parse_version(con.version).base_version >= '1.1.54337':
sql_types.append(
"toDateTime('2018-07-02 00:00:00', 'UTC') AS datetime_col")
sql = 'SELECT {}'.format(', '.join(sql_types))
df = con.sql(sql).execute()
assert df.fixedstring_col.dtype.name == 'object'
if parse_version(con.version).base_version >= '1.1.54337':
assert df.datetime_col.dtype.name == 'datetime64[ns]'
@pytest.mark.parametrize(
('ch_type', 'ibis_type'),
[
('Array(Int8)', dt.Array(dt.Int8(nullable=False))),
('Array(Int16)', dt.Array(dt.Int16(nullable=False))),
('Array(Int32)', dt.Array(dt.Int32(nullable=False))),
('Array(Int64)', dt.Array(dt.Int64(nullable=False))),
('Array(UInt8)', dt.Array(dt.UInt8(nullable=False))),
('Array(UInt16)', dt.Array(dt.UInt16(nullable=False))),
('Array(UInt32)', dt.Array(dt.UInt32(nullable=False))),
('Array(UInt64)', dt.Array(dt.UInt64(nullable=False))),
('Array(Float32)', dt.Array(dt.Float32(nullable=False))),
('Array(Float64)', dt.Array(dt.Float64(nullable=False))),
('Array(String)', dt.Array(dt.String(nullable=False))),
('Array(FixedString(32))', dt.Array(dt.String(nullable=False))),
('Array(Date)', dt.Array(dt.Date(nullable=False))),
('Array(DateTime)', dt.Array(dt.Timestamp(nullable=False))),
('Array(DateTime64)', dt.Array(dt.Timestamp(nullable=False))),
('Array(Nothing)', dt.Array(dt.Null(nullable=False))),
def test_literal_array():
what = []
expr = api.literal(what)
assert isinstance(expr, ir.ArrayValue)
assert expr.type().equals(dt.Array(dt.null))
def test_map_with_string_value_type():
assert dt.Map('int32', 'double') == dt.Map(dt.int32, dt.double)
assert dt.Map('int32', 'array<double>') == \
dt.Map(dt.int32, dt.Array(dt.double))
def __init__(self, value_type, *args, **kwargs):
super(ArrayValueTyped, self).__init__(dt.Array(value_type), *args,
**kwargs)
import pytest
import pytz
from multipledispatch.conflict import ambiguities
import ibis
import ibis.expr.datatypes as dt
def test_validate_type():
assert dt.validate_type is dt.dtype
@pytest.mark.parametrize(
('spec', 'expected'),
[
('ARRAY<DOUBLE>', dt.Array(dt.double)),
('array<array<string>>', dt.Array(dt.Array(dt.string))),
('map<string, double>', dt.Map(dt.string, dt.double)),
(
'map<int64, array<map<string, int8>>>',
dt.Map(dt.int64, dt.Array(dt.Map(dt.string, dt.int8))),
),
('set<uint8>', dt.Set(dt.uint8)),
([dt.uint8], dt.Array(dt.uint8)),
([dt.float32, dt.float64], dt.Array(dt.float64)),
({dt.string}, dt.Set(dt.string)),
('point', dt.point),
('point;4326', dt.point),
('point;4326:geometry', dt.point),
('point;4326:geography', dt.point),
('linestring', dt.linestring),
@pytest.mark.parametrize(
("datatype", "expected"),
[
(dt.float32, "FLOAT64"),
(dt.float64, "FLOAT64"),
(dt.uint8, "INT64"),
(dt.uint16, "INT64"),
(dt.uint32, "INT64"),
(dt.int8, "INT64"),
(dt.int16, "INT64"),
(dt.int32, "INT64"),
(dt.int64, "INT64"),
(dt.string, "STRING"),
(dt.Array(dt.int64), "ARRAY<INT64>"),
(dt.Array(dt.string), "ARRAY<STRING>"),
(dt.date, "DATE"),
(dt.timestamp, "TIMESTAMP"),
param(
dt.Timestamp(timezone="US/Eastern"),
"TIMESTAMP",
),
],
)
def test_simple(datatype, expected):
context = TypeTranslationContext()
assert ibis_type_to_cloud_spanner_type(datatype, context) == expected
@pytest.mark.parametrize("datatype", [dt.uint64, dt.Decimal(8, 3)])