def testMap(self):
expr = self.expr.id.map(lambda a: float(a + 1), rtype=types.float64)
self.assertIsInstance(expr, MappedExpr)
self.assertIs(expr._data_type, types.float64)
if not LESS_PY35:
l = locals().copy()
six.exec_(
textwrap.dedent("""
from typing import Optional
def fun(v) -> float:
return float(v + 1)
expr = self.expr.id.map(fun)
"""), globals(), l)
expr = l['expr']
self.assertIsInstance(expr, MappedExpr)
self.assertIsInstance(expr._data_type, types.Float)
l = locals().copy()
six.exec_(
textwrap.dedent("""
from typing import Optional
def fun(v) -> Optional[float]:
return float(v + 1)
expr = self.expr.id.map(fun)
"""), globals(), l)
expr = l['expr']
self.assertIsInstance(expr, MappedExpr)
self.assertIsInstance(expr._data_type, types.Float)
def testApplyFunction(self):
def my_func(row):
return row.name, row.id
self.engine.compile(self.expr.apply(my_func, axis=1, names=['name', 'id'], types=['string', 'int']))
udtf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udtf, globals(), locals())
udtf = locals()[UDF_CLASS_NAME]
self.assertEqual([('name1', 1), ('name2', 2)],
runners.simple_run(udtf, [('name1', 1, None), ('name2', 2, None)]))
def testApplyToSequenceFuntion(self):
def my_func(row):
return row.name + str(row.id)
self.engine.compile(self.expr.apply(my_func, axis=1, reduce=True).rename('test'))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertEqual(['name1', 'name2'],
runners.simple_run(udf, [('name', 1, None), ('name', 2, None)]))
def testSimpleLambda(self):
self.engine.compile(self.expr.id.map(lambda x: x + 1))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf)
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([
4,
], runners.simple_run(udf, [
(3, ),
]))
def testApplyFunction(self):
def my_func(row):
return row.name, row.id
self.engine.compile(self.expr.apply(my_func, axis=1, names=['name', 'id'], types=['string', 'int']))
udtf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udtf, globals(), locals())
udtf = locals()[UDF_CLASS_NAME]
self.assertEqual([('name1', 1), ('name2', 2)],
runners.simple_run(udtf, [('name1', 1, None), ('name2', 2, None)]))
def testApplyToSequenceFuntion(self):
def my_func(row):
return row.name + str(row.id)
self.engine.compile(self.expr.apply(my_func, axis=1, reduce=True).rename('test'))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertEqual(['name1', 'name2'],
runners.simple_run(udf, [('name', 1, None), ('name', 2, None)]))
def testApplyGeneratorFunction(self):
def my_func(row):
for n in row.name.split(','):
yield n
self.engine.compile(self.expr.apply(my_func, axis=1, names='name'))
udtf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udtf, globals(), locals())
udtf = locals()[UDF_CLASS_NAME]
self.assertEqual(['name1', 'name2', 'name3', 'name4'],
runners.simple_run(udtf, [('name1,name2', 1, None), ('name3,name4', 2, None)]))
def testApplyGeneratorFunction(self):
def my_func(row):
for n in row.name.split(','):
yield n
self.engine.compile(self.expr.apply(my_func, axis=1, names='name'))
udtf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udtf, globals(), locals())
udtf = locals()[UDF_CLASS_NAME]
self.assertEqual(['name1', 'name2', 'name3', 'name4'],
runners.simple_run(udtf, [('name1,name2', 1, None), ('name3,name4', 2, None)]))
def testSimpleFunction(self):
def my_func(x):
if x < 0:
return -1
elif x == 0:
return 0
else:
return 1
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf)
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-1, 0, 1], runners.simple_run(udf, [(-3, ), (0, ), (5, )]))
def testSimpleFunction(self):
def my_func(x):
if x < 0:
return -1
elif x == 0:
return 0
else:
return 1
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf)
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-1, 0, 1], runners.simple_run(udf, [(-3, ), (0, ), (5, )]))
def testGlobalVarFunction(self):
global_val = 10
def my_func(x):
if x < global_val:
return -1
elif x == global_val:
return 0
else:
return 1
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-1, 0, 1], runners.simple_run(udf, [(-9, ), (10, ), (15, )]))
def testGlobalVarFunction(self):
global_val = 10
def my_func(x):
if x < global_val:
return -1
elif x == global_val:
return 0
else:
return 1
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-1, 0, 1], runners.simple_run(udf, [(-9, ), (10, ), (15, )]))
def testNestFunction(self):
def my_func(x):
def inner(y):
if y < 0:
return -2
elif y == 0:
return 0
else:
return 2
return inner(x)
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-2, 0, 2], runners.simple_run(udf, [(-3, ), (0, ), (5, )]))
def testReduceApply(self):
expr = self.expr[self.expr.id, self.expr['name', 'id'].apply(
lambda row: row.name + row.id, axis=1, reduce=True).rename('nameid')]
self.assertIsInstance(expr._fields[1], MappedExpr)
if not LESS_PY35:
l = locals().copy()
six.exec_(textwrap.dedent("""
def fun(r) -> float:
return r.id + r.fid
expr = self.expr[self.expr.id, self.expr['id', 'fid'].apply(fun, axis=1, reduce=True).rename('idfid')]
"""), globals(), l)
expr = l['expr']
self.assertIsInstance(expr._fields[1], MappedExpr)
self.assertIsInstance(expr._fields[1]._data_type, types.Float)
def testNestFunction(self):
def my_func(x):
def inner(y):
if y < 0:
return -2
elif y == 0:
return 0
else:
return 2
return inner(x)
self.engine.compile(self.expr.id.map(my_func))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf, globals(), locals())
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([-2, 0, 2], runners.simple_run(udf, [(-3, ), (0, ), (5, )]))
def testBizarreField(self):
def my_func(row):
return getattr(row, '012') * 2.0
datatypes = lambda *types: [validate_data_type(t) for t in types]
schema = Schema.from_lists(['name', 'id', 'fid', '012'],
datatypes('string', 'int64', 'float64', 'float64'))
table = MockTable(name='pyodps_test_expr_table', schema=schema)
expr = CollectionExpr(_source_data=table, _schema=schema)
self.engine.compile(expr.apply(my_func, axis=1, names=['out_col'], types=['float64']))
udtf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udtf, globals(), locals())
udtf = locals()[UDF_CLASS_NAME]
self.assertEqual([20, 40],
runners.simple_run(udtf, [('name1', 1, None, 10), ('name2', 2, None, 20)]))
def testAgg(self):
class Agg(object):
def buffer(self):
return [0]
def __call__(self, buffer, val):
buffer[0] += val
def merge(self, buffer, pbuffer):
buffer[0] += pbuffer[0]
def getvalue(self, buffer):
return buffer[0]
expr = self.expr.int64.agg(Agg)
self.assertIsInstance(expr, Aggregation)
self.assertEqual(expr.dtype, types.int64)
if not LESS_PY35:
l = locals().copy()
six.exec_(
textwrap.dedent("""
class Agg(object):
def buffer(self):
return [0]
def __call__(self, buffer, val):
buffer[0] += val
def merge(self, buffer, pbuffer):
buffer[0] += pbuffer[0]
def getvalue(self, buffer) -> float:
return buffer[0]
expr = self.expr.int64.agg(Agg)
"""), globals(), l)
expr = l['expr']
self.assertIsInstance(expr, Aggregation)
self.assertIsInstance(expr.dtype, types.Float)
def get_function(source, fun_name):
d = dict()
six.exec_(source, d, d)
return d[fun_name]
def _gen_class_builder_func():
out_closure = 10
def _gen_nested_class_obj():
class BuildCls(BuildBase, metaclass=BuildMeta):
a = 10
def b(self, add_val):
print(self.a)
return self.a + add_val + out_closure
return BuildCls
return _gen_nested_class_obj
""")
my_locs = locals().copy()
six.exec_(py3_code, globals(), my_locs)
_gen_class_builder_func = my_locs.get('_gen_class_builder_func')
if sys.version_info[:2] < (3, 6):
def _gen_format_string_func():
out_closure = 4.0
def _format_fun(arg):
return 'Formatted stuff {0}: {1:>5}'.format(arg, out_closure)
return _format_fun
else:
py36_code = textwrap.dedent("""
def _gen_format_string_func():
out_closure = 4.0
def _gen_class_builder_func():
out_closure = 10
def _gen_nested_class_obj():
class BuildCls(BuildBase, metaclass=BuildMeta):
a = 10
def b(self, add_val):
print(self.a)
return self.a + add_val + out_closure
return BuildCls
return _gen_nested_class_obj
""")
my_locs = locals().copy()
six.exec_(py3_code, globals(), my_locs)
_gen_class_builder_func = my_locs.get('_gen_class_builder_func')
def _gen_nested_fun():
out_closure = 10
def _gen_nested_obj():
# class NestedClass(object):
def nested_method(add_val):
return out_closure + add_val
return nested_method
return lambda v: _gen_nested_obj()(v)
# under the License.
from odps.tests.core import TestBase
from odps.compat import unittest, six
from odps.models import Schema
from odps.udf.tools import runners
from odps.df.types import validate_data_type
from odps.df.backends.odpssql.engine import ODPSEngine, UDF_CLASS_NAME
from odps.df.expr.expressions import CollectionExpr
from odps.df.expr.tests.core import MockTable
# required by cloudpickle tests
six.exec_(
"""
import base64
from collections import namedtuple
import inspect
from odps.lib.cloudpickle import *
from odps.lib.importer import *
""", globals(), locals())
class Test(TestBase):
def setup(self):
datatypes = lambda *types: [validate_data_type(t) for t in types]
schema = Schema.from_lists(['name', 'id', 'fid'],
datatypes('string', 'int64', 'float64'))
table = MockTable(name='pyodps_test_expr_table', schema=schema)
self.expr = CollectionExpr(_source_data=table, _schema=schema)
def testSimpleLambda(self):
self.engine.compile(self.expr.id.map(lambda x: x + 1))
udf = list(self.engine._ctx._func_to_udfs.values())[0]
six.exec_(udf)
udf = locals()[UDF_CLASS_NAME]
self.assertSequenceEqual([4, ], runners.simple_run(udf, [(3, ), ]))
# under the License.
from odps.tests.core import TestBase
from odps.compat import unittest, six
from odps.models import Schema
from odps.udf.tools import runners
from odps.df.types import validate_data_type
from odps.df.backends.odpssql.engine import ODPSEngine, UDF_CLASS_NAME
from odps.df.expr.expressions import CollectionExpr
from odps.df.expr.tests.core import MockTable
# required by cloudpickle tests
six.exec_("""
import base64
from collections import namedtuple
import inspect
import functools
from odps.lib.cloudpickle import *
from odps.lib.importer import *
""", globals(), locals())
class Test(TestBase):
def setup(self):
datatypes = lambda *types: [validate_data_type(t) for t in types]
schema = Schema.from_lists(['name', 'id', 'fid'],
datatypes('string', 'int64', 'float64'))
table = MockTable(name='pyodps_test_expr_table', schema=schema)
self.expr = CollectionExpr(_source_data=table, _schema=schema)
from odps.udf.tools import runners
from odps.df.types import validate_data_type
from odps.df.backends.odpssql.engine import ODPSSQLEngine, UDF_CLASS_NAME
from odps.df.expr.expressions import CollectionExpr
from odps.df.expr.tests.core import MockTable
# required by cloudpickle tests
six.exec_("""
import sys
import base64
from collections import namedtuple
import inspect
import functools
from odps.compat import OrderedDict
from odps.lib.cloudpickle import *
from odps.lib.importer import *
PY2 = sys.version_info[0] == 2
if PY2:
string_type = unicode
else:
string_type = str
""", globals(), locals())
from odps.df.backends.odpssql.codegen import X_NAMED_TUPLE
six.exec_(X_NAMED_TUPLE, globals(), locals())
class ODPSEngine(ODPSSQLEngine):
