cpython/Lib/test/test_copy.py

984 lines
29 KiB
Python
Raw Normal View History

"""Unit tests for the copy module."""
import copy
import copyreg
import weakref
import abc
from operator import le, lt, ge, gt, eq, ne, attrgetter
import unittest
from test import support
order_comparisons = le, lt, ge, gt
equality_comparisons = eq, ne
comparisons = order_comparisons + equality_comparisons
class TestCopy(unittest.TestCase):
# Attempt full line coverage of copy.py from top to bottom
def test_exceptions(self):
self.assertIs(copy.Error, copy.error)
self.assertTrue(issubclass(copy.Error, Exception))
# The copy() method
def test_copy_basic(self):
x = 42
y = copy.copy(x)
self.assertEqual(x, y)
def test_copy_copy(self):
class C(object):
def __init__(self, foo):
self.foo = foo
def __copy__(self):
return C(self.foo)
x = C(42)
y = copy.copy(x)
self.assertEqual(y.__class__, x.__class__)
self.assertEqual(y.foo, x.foo)
def test_copy_registry(self):
class C(object):
def __new__(cls, foo):
obj = object.__new__(cls)
obj.foo = foo
return obj
def pickle_C(obj):
return (C, (obj.foo,))
x = C(42)
self.assertRaises(TypeError, copy.copy, x)
copyreg.pickle(C, pickle_C, C)
y = copy.copy(x)
self.assertIsNot(x, y)
self.assertEqual(type(y), C)
self.assertEqual(y.foo, x.foo)
def test_copy_reduce_ex(self):
class C(object):
def __reduce_ex__(self, proto):
c.append(1)
return ""
def __reduce__(self):
self.fail("shouldn't call this")
c = []
x = C()
y = copy.copy(x)
self.assertIs(y, x)
self.assertEqual(c, [1])
def test_copy_reduce(self):
class C(object):
def __reduce__(self):
c.append(1)
return ""
c = []
x = C()
y = copy.copy(x)
self.assertIs(y, x)
self.assertEqual(c, [1])
def test_copy_cant(self):
class C(object):
def __getattribute__(self, name):
if name.startswith("__reduce"):
raise AttributeError(name)
return object.__getattribute__(self, name)
x = C()
self.assertRaises(copy.Error, copy.copy, x)
# Type-specific _copy_xxx() methods
def test_copy_atomic(self):
class NewStyle:
pass
def f():
pass
class WithMetaclass(metaclass=abc.ABCMeta):
pass
tests = [None, ..., NotImplemented,
42, 2**100, 3.14, True, False, 1j,
"hello", "hello\u1234", f.__code__,
b"world", bytes(range(256)), range(10), slice(1, 10, 2),
NewStyle, max, WithMetaclass, property()]
for x in tests:
self.assertIs(copy.copy(x), x)
def test_copy_list(self):
x = [1, 2, 3]
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
x = []
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
def test_copy_tuple(self):
x = (1, 2, 3)
self.assertIs(copy.copy(x), x)
x = ()
self.assertIs(copy.copy(x), x)
x = (1, 2, 3, [])
self.assertIs(copy.copy(x), x)
def test_copy_dict(self):
x = {"foo": 1, "bar": 2}
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
x = {}
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
def test_copy_set(self):
x = {1, 2, 3}
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
x = set()
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
def test_copy_frozenset(self):
x = frozenset({1, 2, 3})
self.assertIs(copy.copy(x), x)
x = frozenset()
self.assertIs(copy.copy(x), x)
def test_copy_bytearray(self):
x = bytearray(b'abc')
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
x = bytearray()
y = copy.copy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
def test_copy_inst_vanilla(self):
class C:
def __init__(self, foo):
self.foo = foo
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
def test_copy_inst_copy(self):
class C:
def __init__(self, foo):
self.foo = foo
def __copy__(self):
return C(self.foo)
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
def test_copy_inst_getinitargs(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getinitargs__(self):
return (self.foo,)
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
def test_copy_inst_getnewargs(self):
class C(int):
def __new__(cls, foo):
self = int.__new__(cls)
self.foo = foo
return self
def __getnewargs__(self):
return self.foo,
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
y = copy.copy(x)
self.assertIsInstance(y, C)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertEqual(y.foo, x.foo)
def test_copy_inst_getnewargs_ex(self):
class C(int):
def __new__(cls, *, foo):
self = int.__new__(cls)
self.foo = foo
return self
def __getnewargs_ex__(self):
return (), {'foo': self.foo}
def __eq__(self, other):
return self.foo == other.foo
x = C(foo=42)
y = copy.copy(x)
self.assertIsInstance(y, C)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertEqual(y.foo, x.foo)
def test_copy_inst_getstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getstate__(self):
return {"foo": self.foo}
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
def test_copy_inst_setstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __setstate__(self, state):
self.foo = state["foo"]
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
def test_copy_inst_getstate_setstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getstate__(self):
return self.foo
def __setstate__(self, state):
self.foo = state
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C(42)
self.assertEqual(copy.copy(x), x)
# State with boolean value is false (issue #25718)
x = C(0.0)
self.assertEqual(copy.copy(x), x)
# The deepcopy() method
def test_deepcopy_basic(self):
x = 42
y = copy.deepcopy(x)
self.assertEqual(y, x)
def test_deepcopy_memo(self):
# Tests of reflexive objects are under type-specific sections below.
# This tests only repetitions of objects.
x = []
x = [x, x]
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y[0], x[0])
self.assertIs(y[0], y[1])
def test_deepcopy_issubclass(self):
# XXX Note: there's no way to test the TypeError coming out of
# issubclass() -- this can only happen when an extension
# module defines a "type" that doesn't formally inherit from
# type.
class Meta(type):
pass
class C(metaclass=Meta):
pass
self.assertEqual(copy.deepcopy(C), C)
def test_deepcopy_deepcopy(self):
class C(object):
def __init__(self, foo):
self.foo = foo
def __deepcopy__(self, memo=None):
return C(self.foo)
x = C(42)
y = copy.deepcopy(x)
self.assertEqual(y.__class__, x.__class__)
self.assertEqual(y.foo, x.foo)
def test_deepcopy_registry(self):
class C(object):
def __new__(cls, foo):
obj = object.__new__(cls)
obj.foo = foo
return obj
def pickle_C(obj):
return (C, (obj.foo,))
x = C(42)
self.assertRaises(TypeError, copy.deepcopy, x)
copyreg.pickle(C, pickle_C, C)
y = copy.deepcopy(x)
self.assertIsNot(x, y)
self.assertEqual(type(y), C)
self.assertEqual(y.foo, x.foo)
def test_deepcopy_reduce_ex(self):
class C(object):
def __reduce_ex__(self, proto):
c.append(1)
return ""
def __reduce__(self):
self.fail("shouldn't call this")
c = []
x = C()
y = copy.deepcopy(x)
self.assertIs(y, x)
self.assertEqual(c, [1])
def test_deepcopy_reduce(self):
class C(object):
def __reduce__(self):
c.append(1)
return ""
c = []
x = C()
y = copy.deepcopy(x)
self.assertIs(y, x)
self.assertEqual(c, [1])
def test_deepcopy_cant(self):
class C(object):
def __getattribute__(self, name):
if name.startswith("__reduce"):
raise AttributeError(name)
return object.__getattribute__(self, name)
x = C()
self.assertRaises(copy.Error, copy.deepcopy, x)
# Type-specific _deepcopy_xxx() methods
def test_deepcopy_atomic(self):
class NewStyle:
pass
def f():
pass
tests = [None, ..., NotImplemented, 42, 2**100, 3.14, True, False, 1j,
b"bytes", "hello", "hello\u1234", f.__code__,
NewStyle, range(10), max, property()]
for x in tests:
self.assertIs(copy.deepcopy(x), x)
def test_deepcopy_list(self):
x = [[1, 2], 3]
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(x, y)
self.assertIsNot(x[0], y[0])
def test_deepcopy_reflexive_list(self):
x = []
x.append(x)
y = copy.deepcopy(x)
for op in comparisons:
self.assertRaises(RecursionError, op, y, x)
self.assertIsNot(y, x)
self.assertIs(y[0], y)
self.assertEqual(len(y), 1)
def test_deepcopy_empty_tuple(self):
x = ()
y = copy.deepcopy(x)
self.assertIs(x, y)
def test_deepcopy_tuple(self):
x = ([1, 2], 3)
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(x, y)
self.assertIsNot(x[0], y[0])
def test_deepcopy_tuple_of_immutables(self):
x = ((1, 2), 3)
y = copy.deepcopy(x)
self.assertIs(x, y)
def test_deepcopy_reflexive_tuple(self):
x = ([],)
x[0].append(x)
y = copy.deepcopy(x)
for op in comparisons:
self.assertRaises(RecursionError, op, y, x)
self.assertIsNot(y, x)
self.assertIsNot(y[0], x[0])
self.assertIs(y[0][0], y)
def test_deepcopy_dict(self):
x = {"foo": [1, 2], "bar": 3}
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(x, y)
self.assertIsNot(x["foo"], y["foo"])
def test_deepcopy_reflexive_dict(self):
x = {}
x['foo'] = x
y = copy.deepcopy(x)
for op in order_comparisons:
self.assertRaises(TypeError, op, y, x)
for op in equality_comparisons:
self.assertRaises(RecursionError, op, y, x)
self.assertIsNot(y, x)
self.assertIs(y['foo'], y)
self.assertEqual(len(y), 1)
def test_deepcopy_keepalive(self):
memo = {}
x = []
y = copy.deepcopy(x, memo)
self.assertIs(memo[id(memo)][0], x)
def test_deepcopy_dont_memo_immutable(self):
memo = {}
x = [1, 2, 3, 4]
y = copy.deepcopy(x, memo)
self.assertEqual(y, x)
# There's the entry for the new list, and the keep alive.
self.assertEqual(len(memo), 2)
memo = {}
x = [(1, 2)]
y = copy.deepcopy(x, memo)
self.assertEqual(y, x)
# Tuples with immutable contents are immutable for deepcopy.
self.assertEqual(len(memo), 2)
def test_deepcopy_inst_vanilla(self):
class C:
def __init__(self, foo):
self.foo = foo
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_deepcopy(self):
class C:
def __init__(self, foo):
self.foo = foo
def __deepcopy__(self, memo):
return C(copy.deepcopy(self.foo, memo))
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_getinitargs(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getinitargs__(self):
return (self.foo,)
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_getnewargs(self):
class C(int):
def __new__(cls, foo):
self = int.__new__(cls)
self.foo = foo
return self
def __getnewargs__(self):
return self.foo,
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertIsInstance(y, C)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertEqual(y.foo, x.foo)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_getnewargs_ex(self):
class C(int):
def __new__(cls, *, foo):
self = int.__new__(cls)
self.foo = foo
return self
def __getnewargs_ex__(self):
return (), {'foo': self.foo}
def __eq__(self, other):
return self.foo == other.foo
x = C(foo=[42])
y = copy.deepcopy(x)
self.assertIsInstance(y, C)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertEqual(y.foo, x.foo)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_getstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getstate__(self):
return {"foo": self.foo}
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_setstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __setstate__(self, state):
self.foo = state["foo"]
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_inst_getstate_setstate(self):
class C:
def __init__(self, foo):
self.foo = foo
def __getstate__(self):
return self.foo
def __setstate__(self, state):
self.foo = state
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.foo == other.foo
x = C([42])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
# State with boolean value is false (issue #25718)
x = C([])
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y, x)
self.assertIsNot(y.foo, x.foo)
def test_deepcopy_reflexive_inst(self):
class C:
pass
x = C()
x.foo = x
y = copy.deepcopy(x)
self.assertIsNot(y, x)
self.assertIs(y.foo, y)
# _reconstruct()
def test_reconstruct_string(self):
class C(object):
def __reduce__(self):
return ""
x = C()
y = copy.copy(x)
self.assertIs(y, x)
y = copy.deepcopy(x)
self.assertIs(y, x)
def test_reconstruct_nostate(self):
class C(object):
def __reduce__(self):
return (C, ())
x = C()
x.foo = 42
y = copy.copy(x)
self.assertIs(y.__class__, x.__class__)
y = copy.deepcopy(x)
self.assertIs(y.__class__, x.__class__)
def test_reconstruct_state(self):
class C(object):
def __reduce__(self):
return (C, (), self.__dict__)
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.__dict__ == other.__dict__
x = C()
x.foo = [42]
y = copy.copy(x)
self.assertEqual(y, x)
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y.foo, x.foo)
def test_reconstruct_state_setstate(self):
class C(object):
def __reduce__(self):
return (C, (), self.__dict__)
def __setstate__(self, state):
self.__dict__.update(state)
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return self.__dict__ == other.__dict__
x = C()
x.foo = [42]
y = copy.copy(x)
self.assertEqual(y, x)
y = copy.deepcopy(x)
self.assertEqual(y, x)
self.assertIsNot(y.foo, x.foo)
def test_reconstruct_reflexive(self):
class C(object):
pass
x = C()
x.foo = x
y = copy.deepcopy(x)
self.assertIsNot(y, x)
self.assertIs(y.foo, y)
# Additions for Python 2.3 and pickle protocol 2
def test_reduce_4tuple(self):
class C(list):
def __reduce__(self):
return (C, (), self.__dict__, iter(self))
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return (list(self) == list(other) and
self.__dict__ == other.__dict__)
x = C([[1, 2], 3])
y = copy.copy(x)
self.assertEqual(x, y)
self.assertIsNot(x, y)
self.assertIs(x[0], y[0])
y = copy.deepcopy(x)
self.assertEqual(x, y)
self.assertIsNot(x, y)
self.assertIsNot(x[0], y[0])
def test_reduce_5tuple(self):
class C(dict):
def __reduce__(self):
return (C, (), self.__dict__, None, self.items())
Restructure comparison dramatically. There is no longer a default *ordering* between objects; there is only a default equality test (defined by an object being equal to itself only). Read the comment in object.c. The current implementation never uses a three-way comparison to compute a rich comparison, but it does use a rich comparison to compute a three-way comparison. I'm not quite done ripping out all the calls to PyObject_Compare/Cmp, or replacing tp_compare implementations with tp_richcompare implementations; but much of that has happened (to make most unit tests pass). The following tests still fail, because I need help deciding or understanding: test_codeop -- depends on comparing code objects test_datetime -- need Tim Peters' opinion test_marshal -- depends on comparing code objects test_mutants -- need help understanding it The problem with test_codeop and test_marshal is this: these tests compare two different code objects and expect them to be equal. Is that still a feature we'd like to support? I've temporarily removed the comparison and hash code from code objects, so they use the default (equality by pointer only) comparison. For the other two tests, run them to see for yourself. (There may be more failing test with "-u all".) A general problem with getting lots of these tests to pass is the reality that for object types that have a natural total ordering, implementing __cmp__ is much more convenient than implementing __eq__, __ne__, __lt__, and so on. Should we go back to allowing __cmp__ to provide a total ordering? Should we provide some other way to implement rich comparison with a single method override? Alex proposed a __key__() method; I've considered a __richcmp__() method. Or perhaps __cmp__() just shouldn't be killed off...
2006-08-23 21:41:19 -03:00
def __eq__(self, other):
return (dict(self) == dict(other) and
self.__dict__ == other.__dict__)
x = C([("foo", [1, 2]), ("bar", 3)])
y = copy.copy(x)
self.assertEqual(x, y)
self.assertIsNot(x, y)
self.assertIs(x["foo"], y["foo"])
y = copy.deepcopy(x)
self.assertEqual(x, y)
self.assertIsNot(x, y)
self.assertIsNot(x["foo"], y["foo"])
def test_reduce_6tuple(self):
def state_setter(*args, **kwargs):
self.fail("shouldn't call this")
class C:
def __reduce__(self):
return C, (), self.__dict__, None, None, state_setter
x = C()
with self.assertRaises(TypeError):
copy.copy(x)
with self.assertRaises(TypeError):
copy.deepcopy(x)
def test_reduce_6tuple_none(self):
class C:
def __reduce__(self):
return C, (), self.__dict__, None, None, None
x = C()
with self.assertRaises(TypeError):
copy.copy(x)
with self.assertRaises(TypeError):
copy.deepcopy(x)
def test_copy_slots(self):
class C(object):
__slots__ = ["foo"]
x = C()
x.foo = [42]
y = copy.copy(x)
self.assertIs(x.foo, y.foo)
def test_deepcopy_slots(self):
class C(object):
__slots__ = ["foo"]
x = C()
x.foo = [42]
y = copy.deepcopy(x)
self.assertEqual(x.foo, y.foo)
self.assertIsNot(x.foo, y.foo)
def test_deepcopy_dict_subclass(self):
class C(dict):
def __init__(self, d=None):
if not d:
d = {}
self._keys = list(d.keys())
super().__init__(d)
def __setitem__(self, key, item):
super().__setitem__(key, item)
if key not in self._keys:
self._keys.append(key)
x = C(d={'foo':0})
y = copy.deepcopy(x)
self.assertEqual(x, y)
self.assertEqual(x._keys, y._keys)
self.assertIsNot(x, y)
x['bar'] = 1
self.assertNotEqual(x, y)
self.assertNotEqual(x._keys, y._keys)
def test_copy_list_subclass(self):
class C(list):
pass
x = C([[1, 2], 3])
x.foo = [4, 5]
y = copy.copy(x)
self.assertEqual(list(x), list(y))
self.assertEqual(x.foo, y.foo)
self.assertIs(x[0], y[0])
self.assertIs(x.foo, y.foo)
def test_deepcopy_list_subclass(self):
class C(list):
pass
x = C([[1, 2], 3])
x.foo = [4, 5]
y = copy.deepcopy(x)
self.assertEqual(list(x), list(y))
self.assertEqual(x.foo, y.foo)
self.assertIsNot(x[0], y[0])
self.assertIsNot(x.foo, y.foo)
def test_copy_tuple_subclass(self):
class C(tuple):
pass
x = C([1, 2, 3])
self.assertEqual(tuple(x), (1, 2, 3))
y = copy.copy(x)
self.assertEqual(tuple(y), (1, 2, 3))
def test_deepcopy_tuple_subclass(self):
class C(tuple):
pass
x = C([[1, 2], 3])
self.assertEqual(tuple(x), ([1, 2], 3))
y = copy.deepcopy(x)
self.assertEqual(tuple(y), ([1, 2], 3))
self.assertIsNot(x, y)
self.assertIsNot(x[0], y[0])
def test_getstate_exc(self):
class EvilState(object):
def __getstate__(self):
raise ValueError("ain't got no stickin' state")
self.assertRaises(ValueError, copy.copy, EvilState())
def test_copy_function(self):
self.assertEqual(copy.copy(global_foo), global_foo)
def foo(x, y): return x+y
self.assertEqual(copy.copy(foo), foo)
bar = lambda: None
self.assertEqual(copy.copy(bar), bar)
def test_deepcopy_function(self):
self.assertEqual(copy.deepcopy(global_foo), global_foo)
def foo(x, y): return x+y
self.assertEqual(copy.deepcopy(foo), foo)
bar = lambda: None
self.assertEqual(copy.deepcopy(bar), bar)
def _check_weakref(self, _copy):
class C(object):
pass
obj = C()
x = weakref.ref(obj)
y = _copy(x)
self.assertIs(y, x)
del obj
y = _copy(x)
self.assertIs(y, x)
def test_copy_weakref(self):
self._check_weakref(copy.copy)
def test_deepcopy_weakref(self):
self._check_weakref(copy.deepcopy)
def _check_copy_weakdict(self, _dicttype):
class C(object):
pass
a, b, c, d = [C() for i in range(4)]
u = _dicttype()
u[a] = b
u[c] = d
v = copy.copy(u)
self.assertIsNot(v, u)
self.assertEqual(v, u)
self.assertEqual(v[a], b)
self.assertEqual(v[c], d)
self.assertEqual(len(v), 2)
del c, d
support.gc_collect() # For PyPy or other GCs.
self.assertEqual(len(v), 1)
x, y = C(), C()
# The underlying containers are decoupled
v[x] = y
2010-01-23 11:40:09 -04:00
self.assertNotIn(x, u)
def test_copy_weakkeydict(self):
self._check_copy_weakdict(weakref.WeakKeyDictionary)
def test_copy_weakvaluedict(self):
self._check_copy_weakdict(weakref.WeakValueDictionary)
def test_deepcopy_weakkeydict(self):
class C(object):
def __init__(self, i):
self.i = i
a, b, c, d = [C(i) for i in range(4)]
u = weakref.WeakKeyDictionary()
u[a] = b
u[c] = d
# Keys aren't copied, values are
v = copy.deepcopy(u)
self.assertNotEqual(v, u)
self.assertEqual(len(v), 2)
self.assertIsNot(v[a], b)
self.assertIsNot(v[c], d)
self.assertEqual(v[a].i, b.i)
self.assertEqual(v[c].i, d.i)
del c
support.gc_collect() # For PyPy or other GCs.
self.assertEqual(len(v), 1)
def test_deepcopy_weakvaluedict(self):
class C(object):
def __init__(self, i):
self.i = i
a, b, c, d = [C(i) for i in range(4)]
u = weakref.WeakValueDictionary()
u[a] = b
u[c] = d
# Keys are copied, values aren't
v = copy.deepcopy(u)
self.assertNotEqual(v, u)
self.assertEqual(len(v), 2)
(x, y), (z, t) = sorted(v.items(), key=lambda pair: pair[0].i)
self.assertIsNot(x, a)
self.assertEqual(x.i, a.i)
self.assertIs(y, b)
self.assertIsNot(z, c)
self.assertEqual(z.i, c.i)
self.assertIs(t, d)
del x, y, z, t
del d
support.gc_collect() # For PyPy or other GCs.
self.assertEqual(len(v), 1)
def test_deepcopy_bound_method(self):
class Foo(object):
def m(self):
pass
f = Foo()
f.b = f.m
g = copy.deepcopy(f)
self.assertEqual(g.m, g.b)
self.assertIs(g.b.__self__, g)
g.b()
class TestReplace(unittest.TestCase):
def test_unsupported(self):
self.assertRaises(TypeError, copy.replace, 1)
self.assertRaises(TypeError, copy.replace, [])
self.assertRaises(TypeError, copy.replace, {})
def f(): pass
self.assertRaises(TypeError, copy.replace, f)
class A: pass
self.assertRaises(TypeError, copy.replace, A)
self.assertRaises(TypeError, copy.replace, A())
def test_replace_method(self):
class A:
def __new__(cls, x, y=0):
self = object.__new__(cls)
self.x = x
self.y = y
return self
def __init__(self, *args, **kwargs):
self.z = self.x + self.y
def __replace__(self, **changes):
x = changes.get('x', self.x)
y = changes.get('y', self.y)
return type(self)(x, y)
attrs = attrgetter('x', 'y', 'z')
a = A(11, 22)
self.assertEqual(attrs(copy.replace(a)), (11, 22, 33))
self.assertEqual(attrs(copy.replace(a, x=1)), (1, 22, 23))
self.assertEqual(attrs(copy.replace(a, y=2)), (11, 2, 13))
self.assertEqual(attrs(copy.replace(a, x=1, y=2)), (1, 2, 3))
def test_namedtuple(self):
from collections import namedtuple
from typing import NamedTuple
PointFromCall = namedtuple('Point', 'x y', defaults=(0,))
class PointFromInheritance(PointFromCall):
pass
class PointFromClass(NamedTuple):
x: int
y: int = 0
for Point in (PointFromCall, PointFromInheritance, PointFromClass):
with self.subTest(Point=Point):
p = Point(11, 22)
self.assertIsInstance(p, Point)
self.assertEqual(copy.replace(p), (11, 22))
self.assertIsInstance(copy.replace(p), Point)
self.assertEqual(copy.replace(p, x=1), (1, 22))
self.assertEqual(copy.replace(p, y=2), (11, 2))
self.assertEqual(copy.replace(p, x=1, y=2), (1, 2))
with self.assertRaisesRegex(TypeError, 'unexpected field name'):
copy.replace(p, x=1, error=2)
def test_dataclass(self):
from dataclasses import dataclass
@dataclass
class C:
x: int
y: int = 0
attrs = attrgetter('x', 'y')
c = C(11, 22)
self.assertEqual(attrs(copy.replace(c)), (11, 22))
self.assertEqual(attrs(copy.replace(c, x=1)), (1, 22))
self.assertEqual(attrs(copy.replace(c, y=2)), (11, 2))
self.assertEqual(attrs(copy.replace(c, x=1, y=2)), (1, 2))
with self.assertRaisesRegex(TypeError, 'unexpected keyword argument'):
copy.replace(c, x=1, error=2)
class MiscTestCase(unittest.TestCase):
def test__all__(self):
support.check__all__(self, copy, not_exported={"dispatch_table", "error"})
def global_foo(x, y): return x+y
if __name__ == "__main__":
unittest.main()