singlestoredb 0.4.0__py3-none-any.whl → 1.0.4__py3-none-any.whl

singlestoredb/tests/ext_funcs/__init__.py

@@ -0,0 +1,385 @@
+#!/usr/bin/env python3
+# type: ignore
+from typing import Optional
+from typing import Tuple
+
+from singlestoredb.functions.decorator import udf
+from singlestoredb.functions.dtypes import BIGINT
+from singlestoredb.functions.dtypes import FLOAT
+from singlestoredb.functions.dtypes import MEDIUMINT
+from singlestoredb.functions.dtypes import SMALLINT
+from singlestoredb.functions.dtypes import TINYINT
+from singlestoredb.functions.dtypes import VARCHAR
+
+
+@udf
+def double_mult(x: float, y: float) -> float:
+    return x * y
+
+
+@udf.pandas
+def pandas_double_mult(x: float, y: float) -> float:
+    return x * y
+
+
+@udf.numpy
+def numpy_double_mult(x: float, y: float) -> float:
+    return x * y
+
+
+@udf.arrow
+def arrow_double_mult(x: float, y: float) -> float:
+    import pyarrow.compute as pc
+    return pc.multiply(x, y)
+
+
+@udf.polars
+def polars_double_mult(x: float, y: float) -> float:
+    return x * y
+
+
+@udf
+def nullable_double_mult(x: Optional[float], y: Optional[float]) -> Optional[float]:
+    if x is None or y is None:
+        return None
+    return x * y
+
+
+@udf(args=[FLOAT(nullable=False), FLOAT(nullable=False)], returns=FLOAT(nullable=False))
+def float_mult(x: float, y: float) -> float:
+    return x * y
+
+
+@udf(args=[FLOAT(nullable=True), FLOAT(nullable=True)], returns=FLOAT(nullable=True))
+def nullable_float_mult(x: Optional[float], y: Optional[float]) -> Optional[float]:
+    if x is None or y is None:
+        return None
+    return x * y
+
+
+def _int_mult(x: int, y: int) -> int:
+    if x is None or y is None:
+        return None
+    return x * y
+
+
+def _arrow_int_mult(x: int, y: int) -> int:
+    import pyarrow.compute as pc
+    return pc.multiply(x, y)
+
+
+def _int_mult_with_masks(x: Tuple[int, bool], y: Tuple[int, bool]) -> Tuple[int, bool]:
+    x_data, x_nulls = x
+    y_data, y_nulls = y
+    return (x_data * y_data, x_nulls | y_nulls)
+
+
+def _arrow_int_mult_with_masks(
+    x: Tuple[int, bool],
+    y: Tuple[int, bool],
+) -> Tuple[int, bool]:
+    import pyarrow.compute as pc
+    x_data, x_nulls = x
+    y_data, y_nulls = y
+    return (pc.multiply(x_data, y_data), pc.or_(x_nulls, y_nulls))
+
+
+int_mult = udf(_int_mult, name='int_mult')
+
+tinyint_mult = udf(
+    _int_mult,
+    name='tinyint_mult',
+    args=[TINYINT(nullable=False), TINYINT(nullable=False)],
+    returns=TINYINT(nullable=False),
+)
+
+pandas_tinyint_mult = udf.pandas(
+    _int_mult,
+    name='pandas_tinyint_mult',
+    args=[TINYINT(nullable=False), TINYINT(nullable=False)],
+    returns=TINYINT(nullable=False),
+)
+
+polars_tinyint_mult = udf.polars(
+    _int_mult,
+    name='polars_tinyint_mult',
+    args=[TINYINT(nullable=False), TINYINT(nullable=False)],
+    returns=TINYINT(nullable=False),
+)
+
+numpy_tinyint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_tinyint_mult',
+    args=[TINYINT(nullable=False), TINYINT(nullable=False)],
+    returns=TINYINT(nullable=False),
+)
+
+arrow_tinyint_mult = udf.arrow(
+    _arrow_int_mult,
+    name='arrow_tinyint_mult',
+    args=[TINYINT(nullable=False), TINYINT(nullable=False)],
+    returns=TINYINT(nullable=False),
+)
+
+smallint_mult = udf(
+    _int_mult,
+    name='smallint_mult',
+    args=[SMALLINT(nullable=False), SMALLINT(nullable=False)],
+    returns=SMALLINT(nullable=False),
+)
+
+pandas_smallint_mult = udf.pandas(
+    _int_mult,
+    name='pandas_smallint_mult',
+    args=[SMALLINT(nullable=False), SMALLINT(nullable=False)],
+    returns=SMALLINT(nullable=False),
+)
+
+polars_smallint_mult = udf.polars(
+    _int_mult,
+    name='polars_smallint_mult',
+    args=[SMALLINT(nullable=False), SMALLINT(nullable=False)],
+    returns=SMALLINT(nullable=False),
+)
+
+numpy_smallint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_smallint_mult',
+    args=[SMALLINT(nullable=False), SMALLINT(nullable=False)],
+    returns=SMALLINT(nullable=False),
+)
+
+arrow_smallint_mult = udf.arrow(
+    _arrow_int_mult,
+    name='arrow_smallint_mult',
+    args=[SMALLINT(nullable=False), SMALLINT(nullable=False)],
+    returns=SMALLINT(nullable=False),
+)
+
+mediumint_mult = udf(
+    _int_mult,
+    name='mediumint_mult',
+    args=[MEDIUMINT(nullable=False), MEDIUMINT(nullable=False)],
+    returns=MEDIUMINT(nullable=False),
+)
+
+pandas_mediumint_mult = udf.pandas(
+    _int_mult,
+    name='pandas_mediumint_mult',
+    args=[MEDIUMINT(nullable=False), MEDIUMINT(nullable=False)],
+    returns=MEDIUMINT(nullable=False),
+)
+
+polars_mediumint_mult = udf.polars(
+    _int_mult,
+    name='polars_mediumint_mult',
+    args=[MEDIUMINT(nullable=False), MEDIUMINT(nullable=False)],
+    returns=MEDIUMINT(nullable=False),
+)
+
+numpy_mediumint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_mediumint_mult',
+    args=[MEDIUMINT(nullable=False), MEDIUMINT(nullable=False)],
+    returns=MEDIUMINT(nullable=False),
+)
+
+arrow_mediumint_mult = udf.arrow(
+    _arrow_int_mult,
+    name='arrow_mediumint_mult',
+    args=[MEDIUMINT(nullable=False), MEDIUMINT(nullable=False)],
+    returns=MEDIUMINT(nullable=False),
+)
+
+bigint_mult = udf(
+    _int_mult,
+    name='bigint_mult',
+    args=[BIGINT(nullable=False), BIGINT(nullable=False)],
+    returns=BIGINT(nullable=False),
+)
+
+pandas_bigint_mult = udf.pandas(
+    _int_mult,
+    name='pandas_bigint_mult',
+    args=[BIGINT(nullable=False), BIGINT(nullable=False)],
+    returns=BIGINT(nullable=False),
+)
+
+polars_bigint_mult = udf.polars(
+    _int_mult,
+    name='polars_bigint_mult',
+    args=[BIGINT(nullable=False), BIGINT(nullable=False)],
+    returns=BIGINT(nullable=False),
+)
+
+numpy_bigint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_bigint_mult',
+    args=[BIGINT(nullable=False), BIGINT(nullable=False)],
+    returns=BIGINT(nullable=False),
+)
+
+arrow_bigint_mult = udf.arrow(
+    _arrow_int_mult,
+    name='arrow_bigint_mult',
+    args=[BIGINT(nullable=False), BIGINT(nullable=False)],
+    returns=BIGINT(nullable=False),
+)
+
+nullable_tinyint_mult = udf(
+    _int_mult,
+    name='nullable_tinyint_mult',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+)
+
+pandas_nullable_tinyint_mult = udf.pandas(
+    _int_mult,
+    name='pandas_nullable_tinyint_mult',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+)
+
+pandas_nullable_tinyint_mult_with_masks = udf.pandas(
+    _int_mult_with_masks,
+    name='pandas_nullable_tinyint_mult_with_masks',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+    include_masks=True,
+)
+
+polars_nullable_tinyint_mult = udf.polars(
+    _int_mult,
+    name='polars_nullable_tinyint_mult',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+)
+
+polars_nullable_tinyint_mult_with_masks = udf.polars(
+    _int_mult_with_masks,
+    name='polars_nullable_tinyint_mult_with_masks',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+    include_masks=True,
+)
+
+numpy_nullable_tinyint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_nullable_tinyint_mult',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+)
+
+numpy_nullable_tinyint_mult_with_masks = udf.numpy(
+    _int_mult_with_masks,
+    name='numpy_nullable_tinyint_mult_with_masks',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+    include_masks=True,
+)
+
+arrow_nullable_tinyint_mult = udf.arrow(
+    _arrow_int_mult,
+    name='arrow_nullable_tinyint_mult',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+)
+
+arrow_nullable_tinyint_mult_with_masks = udf.arrow(
+    _arrow_int_mult_with_masks,
+    name='arrow_nullable_tinyint_mult_with_masks',
+    args=[TINYINT, TINYINT],
+    returns=TINYINT,
+    include_masks=True,
+)
+
+nullable_smallint_mult = udf(
+    _int_mult,
+    name='nullable_smallint_mult',
+    args=[SMALLINT, SMALLINT],
+    returns=SMALLINT,
+)
+
+nullable_mediumint_mult = udf(
+    _int_mult,
+    name='nullable_mediumint_mult',
+    args=[MEDIUMINT, MEDIUMINT],
+    returns=MEDIUMINT,
+)
+
+nullable_bigint_mult = udf(
+    _int_mult,
+    name='nullable_bigint_mult',
+    args=[BIGINT, BIGINT],
+    returns=BIGINT,
+)
+
+numpy_nullable_bigint_mult = udf.numpy(
+    _int_mult,
+    name='numpy_nullable_bigint_mult',
+    args=[BIGINT, BIGINT],
+    returns=BIGINT,
+)
+
+numpy_nullable_bigint_mult_with_masks = udf.numpy(
+    _int_mult_with_masks,
+    name='numpy_nullable_bigint_mult',
+    args=[BIGINT, BIGINT],
+    returns=BIGINT,
+    include_masks=True,
+)
+
+
+@udf
+def nullable_int_mult(x: Optional[int], y: Optional[int]) -> Optional[int]:
+    if x is None or y is None:
+        return None
+    return x * y
+
+
+@udf
+def string_mult(x: str, times: int) -> str:
+    return x * times
+
+
+@udf.pandas
+def pandas_string_mult(x: str, times: int) -> str:
+    return x * times
+
+
+@udf.numpy
+def numpy_string_mult(x: str, times: int) -> str:
+    return x * times
+
+
+# @udf.polars
+# def polars_string_mult(x: str, times: int) -> str:
+#     print(type(x), x, type(times), times)
+#     return x * times
+
+
+# @udf.arrow
+# def arrow_string_mult(x: str, times: int) -> str:
+#     print(type(x), x, type(times), times)
+#     import pyarrow.compute as pc
+#     return pc.multiply(x, times)
+#     return x * times
+
+
+@udf
+def nullable_string_mult(x: Optional[str], times: Optional[int]) -> Optional[str]:
+    if x is None or times is None:
+        return None
+    return x * times
+
+
+@udf(args=dict(x=VARCHAR(20, nullable=False)))
+def varchar_mult(x: str, times: int) -> str:
+    return x * times
+
+
+@udf(args=dict(x=VARCHAR(20, nullable=True)))
+def nullable_varchar_mult(x: Optional[str], times: Optional[int]) -> Optional[str]:
+    if x is None or times is None:
+        return None
+    return x * times

singlestoredb/tests/test.sql

@@ -1,3 +1,4 @@
+
 CREATE ROWSTORE TABLE IF NOT EXISTS data (
     id VARCHAR(255) NOT NULL,
     name VARCHAR(255) NOT NULL,
@@ -13,6 +14,21 @@ INSERT INTO data SET id='e', name='elephants', value=0;
 
 COMMIT;
 
+CREATE ROWSTORE TABLE IF NOT EXISTS data_with_nulls (
+    id VARCHAR(255) NOT NULL,
+    name VARCHAR(255),
+    value BIGINT,
+    PRIMARY KEY (id) USING HASH
+) DEFAULT CHARSET = utf8 COLLATE = utf8_unicode_ci;
+
+INSERT INTO data_with_nulls SET id='a', name='antelopes', value=2;
+INSERT INTO data_with_nulls SET id='b', name=NULL, value=2;
+INSERT INTO data_with_nulls SET id='c', name=NULL, value=5;
+INSERT INTO data_with_nulls SET id='d', name='dogs', value=NULL;
+INSERT INTO data_with_nulls SET id='e', name='elephants', value=0;
+
+COMMIT;
+
 CREATE OR REPLACE PROCEDURE get_animal(nm VARCHAR(255) NOT NULL COLLATE utf8_unicode_ci) AS
 BEGIN
     ECHO SELECT value FROM data WHERE name = nm; --
@@ -186,4 +202,198 @@ INSERT INTO alltypes SET
     `bit`=NULL
 ;
 
+-- Minimum values
+INSERT INTO alltypes SET
+    `id`=2,
+    `tinyint`=-128,
+    `unsigned_tinyint`=0,
+    `bool`=-128,
+    `boolean`=-128,
+    `smallint`=-32768,
+    `unsigned_smallint`=0,
+    `mediumint`=-8388608,
+    `unsigned_mediumint`=0,
+    `int24`=-8388608,
+    `unsigned_int24`=0,
+    `int`=-2147483648,
+    `unsigned_int`=0,
+    `integer`=-2147483648,
+    `unsigned_integer`=0,
+    `bigint`=-9223372036854775808,
+    `unsigned_bigint`=0,
+    `float`=0,
+    `double`=-1.7976931348623158e308,
+    `real`=-1.7976931348623158e308,
+    `decimal`=-99999999999999.999999,
+    `dec`=-99999999999999.999999,
+    `fixed`=-99999999999999.999999,
+    `numeric`=-99999999999999.999999,
+    `date`='1000-01-01',
+    `time`='-838:59:59',
+    `time_6`='-838:59:59.000000',
+    `datetime`='1000-01-01 00:00:00',
+    `datetime_6`='1000-01-01 00:00:00.000000',
+    `timestamp`='1970-01-01 00:00:01',
+    `timestamp_6`='1970-01-01 00:00:01.000000',
+    `year`=1901,
+    `char_100`='',
+    `binary_100`=x'',
+    `varchar_200`='',
+    `varbinary_200`=x'',
+    `longtext`='',
+    `mediumtext`='',
+    `text`='',
+    `tinytext`='',
+    `longblob`=x'',
+    `mediumblob`=x'',
+    `blob`=x'',
+    `tinyblob`=x'',
+    `json`='{}',
+    `enum`='one',
+    `set`='two',
+    `bit`=0
+;
+
+-- Maximum values
+INSERT INTO alltypes SET
+    `id`=3,
+    `tinyint`=127,
+    `unsigned_tinyint`=255,
+    `bool`=127,
+    `boolean`=127,
+    `smallint`=32767,
+    `unsigned_smallint`=65535,
+    `mediumint`=8388607,
+    `unsigned_mediumint`=16777215,
+    `int24`=8388607,
+    `unsigned_int24`=16777215,
+    `int`=2147483647,
+    `unsigned_int`=4294967295,
+    `integer`=2147483647,
+    `unsigned_integer`=4294967295,
+    `bigint`=9223372036854775807,
+    `unsigned_bigint`=18446744073709551615,
+    `float`=0,
+    `double`=1.7976931348623158e308,
+    `real`=1.7976931348623158e308,
+    `decimal`=99999999999999.999999,
+    `dec`=99999999999999.999999,
+    `fixed`=99999999999999.999999,
+    `numeric`=99999999999999.999999,
+    `date`='9999-12-31',
+    `time`='838:59:59',
+    `time_6`='838:59:59.999999',
+    `datetime`='9999-12-31 23:59:59',
+    `datetime_6`='9999-12-31 23:59:59.999999',
+    `timestamp`='2038-01-18 21:14:07',
+    `timestamp_6`='2038-01-18 21:14:07.999999',
+    `year`=2155,
+    `char_100`='',
+    `binary_100`=x'',
+    `varchar_200`='',
+    `varbinary_200`=x'',
+    `longtext`='',
+    `mediumtext`='',
+    `text`='',
+    `tinytext`='',
+    `longblob`=x'',
+    `mediumblob`=x'',
+    `blob`=x'',
+    `tinyblob`=x'',
+    `json`='{}',
+    `enum`='one',
+    `set`='two',
+    `bit`=18446744073709551615
+;
+
+-- Zero values
+--
+-- Note that v8 of SingleStoreDB does not allow zero date/times by
+-- default, so they are set to NULL here.
+--
+INSERT INTO alltypes SET
+    `id`=4,
+    `tinyint`=0,
+    `unsigned_tinyint`=0,
+    `bool`=0,
+    `boolean`=0,
+    `smallint`=0,
+    `unsigned_smallint`=0,
+    `mediumint`=0,
+    `unsigned_mediumint`=0,
+    `int24`=0,
+    `unsigned_int24`=0,
+    `int`=0,
+    `unsigned_int`=0,
+    `integer`=0,
+    `unsigned_integer`=0,
+    `bigint`=0,
+    `unsigned_bigint`=0,
+    `float`=0,
+    `double`=0.0,
+    `real`=0.0,
+    `decimal`=0.0,
+    `dec`=0.0,
+    `fixed`=0.0,
+    `numeric`=0.0,
+    `date`=NULL,
+    `time`='00:00:00',
+    `time_6`='00:00:00.000000',
+    `datetime`=NULL,
+    `datetime_6`=NULL,
+    `timestamp`=NULL,
+    `timestamp_6`=NULL,
+    `year`=NULL,
+    `char_100`='',
+    `binary_100`=x'',
+    `varchar_200`='',
+    `varbinary_200`=x'',
+    `longtext`='',
+    `mediumtext`='',
+    `text`='',
+    `tinytext`='',
+    `longblob`=x'',
+    `mediumblob`=x'',
+    `blob`=x'',
+    `tinyblob`=x'',
+    `json`='{}',
+    `enum`='one',
+    `set`='two',
+    `bit`=0
+;
+
+
+--
+-- Table of extended data types
+--
+CREATE ROWSTORE TABLE IF NOT EXISTS `extended_types` (
+    `id` INT(11),
+    `geography` GEOGRAPHY,
+    `geographypoint` GEOGRAPHYPOINT,
+    `vectors` BLOB,
+    `dt` DATETIME,
+    `d` DATE,
+    `t` TIME,
+    `td` TIME,
+    `testkey` LONGTEXT
+)
+COLLATE='utf8_unicode_ci';
+
+
+--
+-- Invalid utf8 table
+--
+-- These sequences were breaking during fetch on a customer's machine
+-- however, they seem to work fine in our tests.
+--
+CREATE TABLE IF NOT EXISTS `badutf8` (
+    `text` TEXT CHARACTER SET utf8mb4 COLLATE utf8mb4_general_ci
+)
+COLLATE='utf8_unicode_ci';
+
+
+INSERT INTO `badutf8` VALUES ('🥷🧙👻.eth');
+INSERT INTO `badutf8` VALUES ('🥒rick.eth');
+
+
 COMMIT;

singlestoredb/tests/test2.sql

@@ -0,0 +1 @@
+-- Nearly empty file for Stages test