diff --git a/Lib/copy.py b/Lib/copy.py index c64fc076179..6149301ad13 100644 --- a/Lib/copy.py +++ b/Lib/copy.py @@ -100,14 +100,14 @@ def copy(x): return _reconstruct(x, None, *rv) -_copy_atomic_types = {types.NoneType, int, float, bool, complex, str, tuple, - bytes, frozenset, type, range, slice, property, +_copy_atomic_types = frozenset({types.NoneType, int, float, bool, complex, str, tuple, + bytes, frozendict, frozenset, type, range, slice, property, types.BuiltinFunctionType, types.EllipsisType, types.NotImplementedType, types.FunctionType, types.CodeType, - weakref.ref, super} -_copy_builtin_containers = {list, dict, set, bytearray} + weakref.ref, super}) +_copy_builtin_containers = frozenset({list, dict, set, bytearray}) -def deepcopy(x, memo=None, _nil=[]): +def deepcopy(x, memo=None): """Deep copy operation on arbitrary Python objects. See the module's __doc__ string for more info. @@ -122,8 +122,8 @@ def deepcopy(x, memo=None, _nil=[]): if memo is None: memo = {} else: - y = memo.get(d, _nil) - if y is not _nil: + y = memo.get(d, None) + if y is not None: return y copier = _deepcopy_dispatch.get(cls) @@ -162,9 +162,9 @@ def deepcopy(x, memo=None, _nil=[]): _keep_alive(x, memo) # Make sure x lives at least as long as d return y -_atomic_types = {types.NoneType, types.EllipsisType, types.NotImplementedType, +_atomic_types = frozenset({types.NoneType, types.EllipsisType, types.NotImplementedType, int, float, bool, complex, bytes, str, types.CodeType, type, range, - types.BuiltinFunctionType, types.FunctionType, weakref.ref, property} + types.BuiltinFunctionType, types.FunctionType, weakref.ref, property}) _deepcopy_dispatch = d = {} @@ -203,6 +203,21 @@ def _deepcopy_dict(x, memo, deepcopy=deepcopy): return y d[dict] = _deepcopy_dict +def _deepcopy_frozendict(x, memo, deepcopy=deepcopy): + y = {} + for key, value in x.items(): + y[deepcopy(key, memo)] = deepcopy(value, memo) + + # We're not going to put the frozendict in the memo, but it's still + # important we check for it, in case the frozendict contains recursive + # mutable structures. + try: + return memo[id(x)] + except KeyError: + pass + return frozendict(y) +d[frozendict] = _deepcopy_frozendict + def _deepcopy_method(x, memo): # Copy instance methods return type(x)(x.__func__, deepcopy(x.__self__, memo)) d[types.MethodType] = _deepcopy_method @@ -230,7 +245,7 @@ def _reconstruct(x, memo, func, args, *, deepcopy=deepcopy): deep = memo is not None if deep and args: - args = (deepcopy(arg, memo) for arg in args) + args = [deepcopy(arg, memo) for arg in args] y = func(*args) if deep: memo[id(x)] = y diff --git a/Lib/test/mapping_tests.py b/Lib/test/mapping_tests.py index 1358200add5..ae2fb3f5f44 100644 --- a/Lib/test/mapping_tests.py +++ b/Lib/test/mapping_tests.py @@ -4,7 +4,7 @@ from test import support -class BasicTestMappingProtocol(unittest.TestCase): +class BasicTestImmutableMappingProtocol(unittest.TestCase): # This base class can be used to check that an object conforms to the # mapping protocol @@ -20,12 +20,9 @@ def _empty_mapping(self): """Return an empty mapping object""" return self.type2test() def _full_mapping(self, data): - """Return a mapping object with the value contained in data + """Return a mapping object with the values contained in data dictionary""" - x = self._empty_mapping() - for key, value in data.items(): - x[key] = value - return x + return self.type2test(data) def __init__(self, *args, **kw): unittest.TestCase.__init__(self, *args, **kw) @@ -88,6 +85,72 @@ def check_iterandlist(iter, lst, ref): self.assertEqual(d.get(knownkey, knownvalue), knownvalue) self.assertNotIn(knownkey, d) + def test_constructor(self): + self.assertEqual(self._empty_mapping(), self._empty_mapping()) + + def test_bool(self): + self.assertTrue(not self._empty_mapping()) + self.assertTrue(self.reference) + self.assertFalse(bool(self._empty_mapping())) + self.assertTrue(bool(self.reference)) + + def test_keys(self): + d = self._empty_mapping() + self.assertEqual(list(d.keys()), []) + d = self.reference + self.assertIn(list(self.inmapping.keys())[0], d.keys()) + self.assertNotIn(list(self.other.keys())[0], d.keys()) + self.assertRaises(TypeError, d.keys, None) + + def test_values(self): + d = self._empty_mapping() + self.assertEqual(list(d.values()), []) + + self.assertRaises(TypeError, d.values, None) + + def test_items(self): + d = self._empty_mapping() + self.assertEqual(list(d.items()), []) + + self.assertRaises(TypeError, d.items, None) + + def test_len(self): + d = self._empty_mapping() + self.assertEqual(len(d), 0) + + def test_getitem(self): + d = self.reference + self.assertEqual(d[list(self.inmapping.keys())[0]], + list(self.inmapping.values())[0]) + + self.assertRaises(TypeError, d.__getitem__) + + # no test_fromkeys or test_copy as both os.environ and selves don't support it + + def test_get(self): + d = self._empty_mapping() + self.assertIsNone(d.get(list(self.other.keys())[0])) + self.assertEqual(d.get(list(self.other.keys())[0], 3), 3) + d = self.reference + self.assertIsNone(d.get(list(self.other.keys())[0])) + self.assertEqual(d.get(list(self.other.keys())[0], 3), 3) + self.assertEqual(d.get(list(self.inmapping.keys())[0]), + list(self.inmapping.values())[0]) + self.assertEqual(d.get(list(self.inmapping.keys())[0], 3), + list(self.inmapping.values())[0]) + self.assertRaises(TypeError, d.get) + self.assertRaises(TypeError, d.get, None, None, None) + + +class BasicTestMappingProtocol(BasicTestImmutableMappingProtocol): + def _full_mapping(self, data): + """Return a mapping object with the values contained in data + dictionary""" + x = self._empty_mapping() + for key, value in data.items(): + x[key] = value + return x + def test_write(self): # Test for write operations on mapping p = self._empty_mapping() @@ -130,46 +193,6 @@ def test_write(self): p=self._empty_mapping() self.assertRaises(KeyError, p.popitem) - def test_constructor(self): - self.assertEqual(self._empty_mapping(), self._empty_mapping()) - - def test_bool(self): - self.assertTrue(not self._empty_mapping()) - self.assertTrue(self.reference) - self.assertTrue(bool(self._empty_mapping()) is False) - self.assertTrue(bool(self.reference) is True) - - def test_keys(self): - d = self._empty_mapping() - self.assertEqual(list(d.keys()), []) - d = self.reference - self.assertIn(list(self.inmapping.keys())[0], d.keys()) - self.assertNotIn(list(self.other.keys())[0], d.keys()) - self.assertRaises(TypeError, d.keys, None) - - def test_values(self): - d = self._empty_mapping() - self.assertEqual(list(d.values()), []) - - self.assertRaises(TypeError, d.values, None) - - def test_items(self): - d = self._empty_mapping() - self.assertEqual(list(d.items()), []) - - self.assertRaises(TypeError, d.items, None) - - def test_len(self): - d = self._empty_mapping() - self.assertEqual(len(d), 0) - - def test_getitem(self): - d = self.reference - self.assertEqual(d[list(self.inmapping.keys())[0]], - list(self.inmapping.values())[0]) - - self.assertRaises(TypeError, d.__getitem__) - def test_update(self): # mapping argument d = self._empty_mapping() @@ -265,22 +288,6 @@ def __next__(self): self.assertRaises(ValueError, d.update, [(1, 2, 3)]) - # no test_fromkeys or test_copy as both os.environ and selves don't support it - - def test_get(self): - d = self._empty_mapping() - self.assertTrue(d.get(list(self.other.keys())[0]) is None) - self.assertEqual(d.get(list(self.other.keys())[0], 3), 3) - d = self.reference - self.assertTrue(d.get(list(self.other.keys())[0]) is None) - self.assertEqual(d.get(list(self.other.keys())[0], 3), 3) - self.assertEqual(d.get(list(self.inmapping.keys())[0]), - list(self.inmapping.values())[0]) - self.assertEqual(d.get(list(self.inmapping.keys())[0], 3), - list(self.inmapping.values())[0]) - self.assertRaises(TypeError, d.get) - self.assertRaises(TypeError, d.get, None, None, None) - def test_setdefault(self): d = self._empty_mapping() self.assertRaises(TypeError, d.setdefault) @@ -622,7 +629,7 @@ def __repr__(self): d = self._full_mapping({1: BadRepr()}) self.assertRaises(Exc, repr, d) - @support.run_with_limited_c_stack() + @support.skip_if_huge_c_stack() @support.skip_wasi_stack_overflow() @support.skip_emscripten_stack_overflow() @support.skip_if_sanitizer("requires deep stack", ub=True) diff --git a/Lib/test/pickletester.py b/Lib/test/pickletester.py index 41e4087bacb..75e2afa7ae6 100644 --- a/Lib/test/pickletester.py +++ b/Lib/test/pickletester.py @@ -4394,8 +4394,6 @@ def test_fast_save_enter_frozenset(self): def test_fast_save_enter_set(self): self.fast_save_enter(lambda i: set([i])) - # TODO: RUSTPYTHON; frozendict support: https://github.com/RustPython/RustPython/pull/8951 - @unittest.expectedFailure def test_fast_save_enter_frozendict(self): if self.py_version < (3, 15): self.skipTest('need frozendict') @@ -4443,8 +4441,6 @@ def test_deep_nested_struct_set(self): depth=FAST_NESTING_LIMIT+1, compare_equal=False) - # TODO: RUSTPYTHON; frozendict support: https://github.com/RustPython/RustPython/pull/8951 - @unittest.expectedFailure def test_deep_nested_struct_frozendict(self): if self.py_version < (3, 15): self.skipTest('need frozendict') diff --git a/Lib/test/test_builtin.py b/Lib/test/test_builtin.py index 8fdee358acf..2a1a876660e 100644 --- a/Lib/test/test_builtin.py +++ b/Lib/test/test_builtin.py @@ -4,6 +4,7 @@ import builtins import collections import contextlib +import copy import decimal import fractions import gc @@ -21,6 +22,7 @@ import typing import unittest import warnings +import weakref from contextlib import ExitStack from functools import partial from inspect import CO_COROUTINE @@ -31,6 +33,7 @@ from test import support from test.support import cpython_only, swap_attr from test.support import async_yield, run_yielding_async_fn +from test.support import warnings_helper from test.support.import_helper import import_module from test.support.os_helper import (EnvironmentVarGuard, TESTFN, unlink) from test.support.script_helper import assert_python_ok @@ -51,6 +54,10 @@ # used as proof of globals being used A_GLOBAL_VALUE = 123 +A_SENTINEL = sentinel("A_SENTINEL") + +class SentinelContainer: + CLASS_SENTINEL = sentinel("SentinelContainer.CLASS_SENTINEL") class Squares: @@ -245,7 +252,7 @@ def test_any(self): S = [10, 20, 30] self.assertEqual(any(x > 42 for x in S), False) - def test_all_any_tuple_optimization(self): + def test_all_any_tuple_list_set_optimization(self): def f_all(): return all(x-2 for x in [1,2,3]) @@ -255,7 +262,13 @@ def f_any(): def f_tuple(): return tuple(2*x for x in [1,2,3]) - funcs = [f_all, f_any, f_tuple] + def f_list(): + return list(2*x for x in [1,2,3]) + + def f_set(): + return set(2*x for x in [1,2,3]) + + funcs = [f_all, f_any, f_tuple, f_list, f_set] for f in funcs: # check that generator code object is not duplicated @@ -265,33 +278,35 @@ def f_tuple(): # check the overriding the builtins works - global all, any, tuple - saved = all, any, tuple + global all, any, tuple, list, set + saved = all, any, tuple, list, set try: all = lambda x : "all" any = lambda x : "any" tuple = lambda x : "tuple" + list = lambda x : "list" + set = lambda x : "set" overridden_outputs = [f() for f in funcs] finally: - all, any, tuple = saved - - self.assertEqual(overridden_outputs, ['all', 'any', 'tuple']) + all, any, tuple, list, set = saved + self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list', 'set']) # Now repeat, overriding the builtins module as well - saved = all, any, tuple + saved = all, any, tuple, list, set try: builtins.all = all = lambda x : "all" builtins.any = any = lambda x : "any" builtins.tuple = tuple = lambda x : "tuple" + builtins.list = list = lambda x : "list" + builtins.set = set = lambda x : "set" overridden_outputs = [f() for f in funcs] finally: - all, any, tuple = saved - builtins.all, builtins.any, builtins.tuple = saved - - self.assertEqual(overridden_outputs, ['all', 'any', 'tuple']) + all, any, tuple, list, set = saved + builtins.all, builtins.any, builtins.tuple, builtins.list, builtins.set = saved + self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list', 'set']) def test_builtin_call_async_genexpr_no_crash(self): async def f_all(): @@ -527,6 +542,10 @@ async def sleep(delay, result=None): '''a = [x async for x in (x async for x in arange(5))][1]''', '''a, = [1 for x in {x async for x in arange(1)}]''', '''a = [await sleep(0, x) async for x in arange(2)][1]''', + '''a = [await sleep(0, 1) for _ in [0]][0]''', + '''a = {await sleep(0, 1) for _ in [0]}.pop()''', + '''a = {0: await sleep(0, 1) for _ in [0]}[0]''', + '''a = (lambda x=[await sleep(0, 1) for _ in [0]]: x)()[0]''', # gh-121637: Make sure we correctly handle the case where the # async code is optimized away '''assert not await sleep(0); a = 1''', @@ -595,7 +614,26 @@ async def __aexit__(self, *exc_info): '''def f(): async with Lock() as l: a = 1 - ''' + ''', + '''class C: + [await x for x in y] + ''', + '''class C: + [x async for x in arange(10)] + ''', + '''async def f(): + class C: + [await x for x in y] + ''', + '''lambda: [await x for x in y]''', + '''class C: + def f(self, x=[await y for y in z]): + pass + ''', + '''type T = [await x for x in y]''', + '''async def f[T=[await x for x in y]](): + pass + ''', ] for mode, code_sample in product(modes, code_samples): source = dedent(code_sample) @@ -796,6 +834,16 @@ def __getitem__(self, key): raise ValueError self.assertRaises(ValueError, eval, "foo", {}, X()) + def test_eval_frozendict(self): + ns = frozendict(x=1, data=[], __builtins__=__builtins__) + eval("data.append(x)", ns, ns) + self.assertEqual(ns['data'], [1]) + + ns = frozendict() + errmsg = "cannot assign __builtins__ to frozendict globals" + with self.assertRaisesRegex(TypeError, errmsg): + eval("", ns, ns) + def test_eval_kwargs(self): data = {"A_GLOBAL_VALUE": 456} self.assertEqual(eval("globals()['A_GLOBAL_VALUE']", globals=data), 456) @@ -894,6 +942,21 @@ def test_exec(self): del l['__builtins__'] self.assertEqual((g, l), ({'a': 1}, {'b': 2})) + def test_exec_frozendict(self): + ns = frozendict(x=1, data=[], __builtins__=__builtins__) + exec("data.append(x)", ns, ns) + self.assertEqual(ns['data'], [1]) + + ns = frozendict(__builtins__=__builtins__) + errmsg = "'frozendict' object does not support item assignment" + with self.assertRaisesRegex(TypeError, errmsg): + exec("x = 1", ns, ns) + + ns = frozendict() + errmsg = "cannot assign __builtins__ to frozendict globals" + with self.assertRaisesRegex(TypeError, errmsg): + exec("", ns, ns) + def test_exec_kwargs(self): g = {} exec('global z\nz = 1', globals=g) @@ -1108,6 +1171,30 @@ def four_freevars(): three_freevars.__globals__, closure=my_closure) + @unittest.expectedFailure # TODO: RUSTPYTHON; SyntaxError: invalid decimal literal + def test_exec_filter_syntax_warnings_by_module(self): + filename = support.findfile('test_import/data/syntax_warnings.py') + with open(filename, 'rb') as f: + source = f.read() + with warnings.catch_warnings(record=True) as wlog: + warnings.simplefilter('error') + warnings.filterwarnings('always', module=r'\z') + exec(source, {}) + self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21]) + for wm in wlog: + self.assertEqual(wm.filename, '') + self.assertIs(wm.category, SyntaxWarning) + + with warnings.catch_warnings(record=True) as wlog: + warnings.simplefilter('error') + warnings.filterwarnings('always', module=r'package.module\z') + warnings.filterwarnings('error', module=r'') + exec(source, {'__name__': 'package.module', '__file__': filename}) + self.assertEqual(sorted(wm.lineno for wm in wlog), [4, 7, 10, 13, 14, 21]) + for wm in wlog: + self.assertEqual(wm.filename, '') + self.assertIs(wm.category, SyntaxWarning) + def test_filter(self): self.assertEqual(list(filter(lambda c: 'a' <= c <= 'z', 'Hello World')), list('elloorld')) @@ -1868,6 +1955,115 @@ class C: __repr__ = None self.assertRaises(TypeError, repr, C()) + def test_sentinel(self): + missing = sentinel("MISSING") + other = sentinel("MISSING") + + self.assertIsInstance(missing, sentinel) + self.assertIs(type(missing), sentinel) + self.assertEqual(missing.__name__, "MISSING") + self.assertEqual(missing.__module__, __name__) + self.assertIsNot(missing, other) + self.assertEqual(repr(missing), "MISSING") + self.assertTrue(missing) + self.assertIs(copy.copy(missing), missing) + self.assertIs(copy.deepcopy(missing), missing) + self.assertEqual(missing, missing) + self.assertNotEqual(missing, other) + self.assertRaises(TypeError, sentinel) + self.assertRaises(TypeError, sentinel, "MISSING", "EXTRA") + self.assertRaises(TypeError, sentinel, name="MISSING") + with self.assertRaisesRegex(TypeError, "must be str"): + sentinel(1) + self.assertTrue(sentinel.__flags__ & support._TPFLAGS_IMMUTABLETYPE) + self.assertTrue(sentinel.__flags__ & support._TPFLAGS_HAVE_GC) + self.assertFalse(sentinel.__flags__ & support._TPFLAGS_BASETYPE) + with self.assertRaises(TypeError): + class SubSentinel(sentinel): + pass + + def test_sentinel_attributes(self): + missing = sentinel("MISSING") + with self.assertRaises(TypeError): + sentinel.attribute = "value" + with self.assertRaises(AttributeError): + missing.attribute = "value" + with self.assertRaises(AttributeError): + missing.__name__ = "CHANGED" + missing.__module__ = "changed" + self.assertEqual(missing.__module__, "changed") + with self.assertRaises(AttributeError): + del missing.__name__ + del missing.__module__ + with self.assertRaises(AttributeError): + missing.__module__ + + def test_sentinel_repr(self): + with_repr = sentinel("WITH_REPR", repr="custom") + without_repr = sentinel("WITHOUT_REPR", repr=None) + self.assertEqual(repr(with_repr), "custom") + self.assertEqual(repr(without_repr), "WITHOUT_REPR") + self.assertEqual(str(with_repr), "custom") + self.assertEqual(str(without_repr), "WITHOUT_REPR") + + with self.assertRaisesRegex(TypeError, "repr.*str or None"): + sentinel("BAD_REPR", repr=42) + + def test_sentinel_pickle(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(protocol=proto): + self.assertIs( + pickle.loads(pickle.dumps(A_SENTINEL, protocol=proto)), + A_SENTINEL) + self.assertIs( + pickle.loads(pickle.dumps( + SentinelContainer.CLASS_SENTINEL, protocol=proto)), + SentinelContainer.CLASS_SENTINEL) + + missing = sentinel("MISSING") + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(protocol=proto): + with self.assertRaises(pickle.PicklingError): + pickle.dumps(missing, protocol=proto) + + def test_sentinel_str_subclass_name_cycle(self): + class Name(str): + pass + + name = Name("MISSING") + missing = sentinel(name) + self.assertIs(missing.__name__, name) + self.assertTrue(gc.is_tracked(missing)) + + name.missing = missing + ref = weakref.ref(name) + del name, missing + support.gc_collect() + self.assertIsNone(ref()) + + def test_sentinel_union(self): + missing = sentinel("MISSING") + + self.assertIsInstance(missing | int, typing.Union) + self.assertEqual((missing | int).__args__, (missing, int)) + self.assertIsInstance(int | missing, typing.Union) + self.assertEqual((int | missing).__args__, (int, missing)) + self.assertIs(missing | missing, missing) + self.assertEqual(repr(int | missing), "int | MISSING") + self.assertIsInstance(missing | None, typing.Union) + self.assertEqual((missing | None).__args__, (missing, type(None))) + self.assertIsInstance(None | missing, typing.Union) + self.assertEqual((None | missing).__args__, (type(None), missing)) + self.assertIsInstance(missing | list[int], typing.Union) + self.assertEqual((missing | list[int]).__args__, (missing, list[int])) + self.assertIsInstance(missing | (int | str), typing.Union) + self.assertEqual((missing | (int | str)).__args__, (missing, int, str)) + + with self.assertRaises(TypeError): + missing | 1 + with self.assertRaises(TypeError): + 1 | missing + def test_round(self): self.assertEqual(round(0.0), 0.0) self.assertEqual(type(round(0.0)), int) @@ -2594,6 +2790,7 @@ def run_child(self, child, terminal_input): finally: signal.signal(signal.SIGHUP, old_sighup) + @warnings_helper.ignore_fork_in_thread_deprecation_warnings() def _run_child(self, child, terminal_input): r, w = os.pipe() # Pipe test results from child back to parent try: @@ -2983,6 +3180,12 @@ def test_type_doc(self): A.__doc__ = doc self.assertEqual(A.__doc__, doc) + def test_type_frozendict(self): + A = type('A', (), frozendict({'x': 4, 'y': 2})) + self.assertEqual(A.x, 4) + self.assertEqual(A.y, 2) + self.assertEqual(A.__name__, 'A') + def test_bad_args(self): with self.assertRaises(TypeError): type() diff --git a/Lib/test/test_copy.py b/Lib/test/test_copy.py index b452ddbef5c..13b7ce29565 100644 --- a/Lib/test/test_copy.py +++ b/Lib/test/test_copy.py @@ -60,7 +60,7 @@ class C(object): def __reduce_ex__(self, proto): c.append(1) return "" - def __reduce__(self): + def __reduce__(*args): self.fail("shouldn't call this") c = [] x = C() @@ -70,9 +70,15 @@ def __reduce__(self): def test_copy_reduce(self): class C(object): + def __reduce_ex__(*args): + self.fail("shouldn't call this") def __reduce__(self): c.append(1) return "" + def __getattribute__(self, name): + if name == "__reduce_ex__": + raise AttributeError(name) + return object.__getattribute__(self, name) c = [] x = C() y = copy.copy(x) @@ -133,6 +139,12 @@ def test_copy_dict(self): self.assertEqual(y, x) self.assertIsNot(y, x) + def test_copy_frozendict(self): + x = frozendict(x=1, y=2) + self.assertIs(copy.copy(x), x) + x = frozendict() + self.assertIs(copy.copy(x), x) + def test_copy_set(self): x = {1, 2, 3} y = copy.copy(x) @@ -323,7 +335,7 @@ class C(object): def __reduce_ex__(self, proto): c.append(1) return "" - def __reduce__(self): + def __reduce__(*args): self.fail("shouldn't call this") c = [] x = C() @@ -333,9 +345,15 @@ def __reduce__(self): def test_deepcopy_reduce(self): class C(object): + def __reduce_ex__(*args): + self.fail("shouldn't call this") def __reduce__(self): c.append(1) return "" + def __getattribute__(self, name): + if name == "__reduce_ex__": + raise AttributeError(name) + return object.__getattribute__(self, name) c = [] x = C() y = copy.deepcopy(x) @@ -421,6 +439,30 @@ def test_deepcopy_dict(self): self.assertIsNot(x, y) self.assertIsNot(x["foo"], y["foo"]) + def test_deepcopy_frozendict(self): + x = frozendict({"foo": [1, 2], "bar": 3}) + y = copy.deepcopy(x) + self.assertEqual(y, x) + self.assertIsNot(x, y) + self.assertIsNot(x["foo"], y["foo"]) + + # recursive frozendict + x = frozendict(foo=[]) + x['foo'].append(x) + y = copy.deepcopy(x) + self.assertEqual(y.keys(), x.keys()) + self.assertIsNot(x, y) + self.assertIsNot(x["foo"], y["foo"]) + self.assertIs(y['foo'][0], y) + + x = frozendict(foo=[]) + x['foo'].append(x) + x = x['foo'] + y = copy.deepcopy(x) + self.assertIsNot(x, y) + self.assertIsNot(x[0], y[0]) + self.assertIs(y[0]['foo'], y) + @support.skip_if_huge_c_stack() @support.skip_emscripten_stack_overflow() @support.skip_wasi_stack_overflow() diff --git a/Lib/test/test_dict.py b/Lib/test/test_dict.py index 195466e6658..e0d2dbb92b3 100644 --- a/Lib/test/test_dict.py +++ b/Lib/test/test_dict.py @@ -678,7 +678,7 @@ def __repr__(self): d = {1: BadRepr()} self.assertRaises(Exc, repr, d) - @support.run_with_limited_c_stack() + @support.skip_if_huge_c_stack() @support.skip_wasi_stack_overflow() @support.skip_emscripten_stack_overflow() def test_repr_deep(self): @@ -951,7 +951,6 @@ def test_empty_presized_dict_in_freelist(self): 'f': None, 'g': None, 'h': None} d = {} - @unittest.expectedFailure # TODO: RUSTPYTHON def test_container_iterator(self): # Bug #3680: tp_traverse was not implemented for dictiter and # dictview objects. @@ -1264,7 +1263,6 @@ def test_free_after_iterating(self): support.check_free_after_iterating(self, lambda d: iter(d.values()), dict) support.check_free_after_iterating(self, lambda d: iter(d.items()), dict) - @unittest.expectedFailure # TODO: RUSTPYTHON def test_equal_operator_modifying_operand(self): # test fix for seg fault reported in bpo-27945 part 3. class X(): @@ -1426,6 +1424,32 @@ def test_reversed_dict_after_clear_and_restore(self): for it in iterators: self.assertEqual(list(it), []) + @unittest.expectedFailure # TODO: RUSTPYTHON; AssertionError: RuntimeError not raised + def test_reversed_dict_keys_changed_during_iteration(self): + d = dict.fromkeys(range(10)) + for i in range(7): + del d[i] + + iterators = ( + reversed(d), + reversed(d.keys()), + reversed(d.values()), + reversed(d.items()), + ) + for it in iterators: + next(it) + + # Same size as before, but with different keys below + # the iterators' current position. + d.clear() + d.update(dict.fromkeys(range(10))) + for i in range(3, 10): + del d[i] + + for it in iterators: + with self.assertRaisesRegex(RuntimeError, 'keys changed'): + list(it) + def test_dict_copy_order(self): # bpo-34320 od = collections.OrderedDict([('a', 1), ('b', 2)]) @@ -1634,7 +1658,7 @@ def check_unhashable_key(): with check_unhashable_key(): d.get(key) - # Only TypeError exception is overriden, + # Only TypeError exception is overridden, # other exceptions are left unchanged. class HashError: def __hash__(self): @@ -1830,6 +1854,250 @@ class Dict(dict): class SubclassMappingTests(mapping_tests.BasicTestMappingProtocol): type2test = Dict +class FrozenDictMappingTests(mapping_tests.BasicTestImmutableMappingProtocol): + type2test = frozendict + + +class FrozenDict(frozendict): + pass + + +class FrozenDictSlots(frozendict): + __slots__ = ('slot_attr',) + + +class FrozenDictTests(unittest.TestCase): + def test_constructor(self): + # frozendict.__init__() has no effect + d = frozendict(a=1, b=2, c=3) + d.__init__(x=1) + self.assertEqual(d, frozendict(a=1, b=2, c=3)) + + # dict constructor cannot be used on frozendict + with self.assertRaises(TypeError): + dict.__init__(d, x=1) + + # Avoid copy if it's frozendict type + d2 = frozendict(d) + self.assertIs(d2, d) + d2 = FrozenDict(d) + self.assertIsNot(d2, d) + self.assertEqual(d2, d) + + def test_copy(self): + d = frozendict(x=1, y=2) + d2 = d.copy() + self.assertIs(d2, d) + + d = FrozenDict(x=1, y=2) + d2 = d.copy() + self.assertIsNot(d2, d) + self.assertEqual(d2, frozendict(x=1, y=2)) + self.assertEqual(type(d2), frozendict) + + def test_merge(self): + # test "a | b" operator + self.assertEqual(frozendict(x=1) | frozendict(y=2), + frozendict({'x': 1, 'y': 2})) + self.assertEqual(frozendict(x=1) | dict(y=2), + frozendict({'x': 1, 'y': 2})) + self.assertEqual(frozendict(x=1, y=2) | frozendict(y=5), + frozendict({'x': 1, 'y': 5})) + self.assertEqual(FrozenDict(x=1, y=2) | FrozenDict(y=5), + frozendict({'x': 1, 'y': 5})) + + fd = frozendict(x=1, y=2) + self.assertIs(fd | frozendict(), fd) + self.assertIs(fd | {}, fd) + self.assertIs(frozendict() | fd, fd) + + fd = FrozenDict(x=1, y=2) + self.assertEqual(fd | frozendict(), fd) + self.assertEqual(fd | {}, fd) + self.assertEqual(frozendict() | fd, fd) + + # gh-149676: Test hash(frozendict | frozendict) + a = frozendict({"a": 1}) + b = frozendict({"b": 2}) + self.assertEqual(hash(a | b), hash(frozendict({"a": 1, "b": 2}))) + + def test_update(self): + # test "a |= b" operator + d = frozendict(x=1) + copy = d + self.assertIs(copy, d) + d |= frozendict(y=2) + self.assertIsNot(copy, d) + self.assertEqual(d, frozendict({'x': 1, 'y': 2})) + self.assertEqual(copy, frozendict({'x': 1})) + + def test_items_xor(self): + # test "a ^ b" operator on items views + res = frozendict(a=1, b=2).items() ^ frozendict(b=2, c=3).items() + self.assertEqual(res, {('a', 1), ('c', 3)}) + + def test_repr(self): + d = frozendict() + self.assertEqual(repr(d), "frozendict()") + + d = frozendict(x=1, y=2) + self.assertEqual(repr(d), "frozendict({'x': 1, 'y': 2})") + + d = FrozenDict(x=1, y=2) + self.assertEqual(repr(d), "FrozenDict({'x': 1, 'y': 2})") + + def test_hash(self): + # hash() doesn't rely on the items order + self.assertEqual(hash(frozendict(x=1, y=2)), + hash(frozendict(y=2, x=1))) + + # Check that hash() computes the hash of (key, value) pairs + cases = [ + frozendict(a=False, b=True, c=True), + frozendict(a=True, b=False, c=True), + frozendict(a=True, b=True, c=False), + frozendict({False: "a", "b": True, "c": True}), + frozendict({"a": "b", False: True, True: "c"}), + ] + hashes = {hash(fd) for fd in cases} + self.assertEqual(len(hashes), len(cases)) + + fd = frozendict(x=[1], y=[2]) + with self.assertRaisesRegex(TypeError, "unhashable type: 'list'"): + hash(fd) + + @support.cpython_only + def test_hash_cpython(self): + # Check that hash(frozendict) implementation is: + # hash(frozenset(fd.items())) + for fd in ( + frozendict(), + frozendict(x=1, y=2), + frozendict(y=2, x=1), + frozendict(a=False, b=True, c=True), + frozendict.fromkeys('abc'), + ): + with self.subTest(fd=fd): + self.assertEqual(hash(fd), hash(frozenset(fd.items()))) + + def test_fromkeys(self): + self.assertEqual(frozendict.fromkeys('abc'), + frozendict(a=None, b=None, c=None)) + + # Subclass which overrides the constructor + created = frozendict(x=1) + class FrozenDictSubclass(frozendict): + def __new__(cls, *args, **kwargs): + if args or kwargs: + return super().__new__(cls, *args, **kwargs) + else: + return created + + fd = FrozenDictSubclass.fromkeys("abc") + self.assertEqual(fd, frozendict(x=1, a=None, b=None, c=None)) + self.assertEqual(type(fd), FrozenDictSubclass) + self.assertEqual(created, frozendict(x=1)) + + fd = FrozenDictSubclass.fromkeys(frozendict(y=2)) + self.assertEqual(fd, frozendict(x=1, y=None)) + self.assertEqual(type(fd), FrozenDictSubclass) + self.assertEqual(created, frozendict(x=1)) + + # Dict subclass with a constructor which returns a frozendict + # by default + class DictSubclass(dict): + def __new__(cls, *args, **kwargs): + if args or kwargs: + return super().__new__(cls, *args, **kwargs) + else: + return created + + fd = DictSubclass.fromkeys("abc") + self.assertEqual(fd, frozendict(x=1, a=None, b=None, c=None)) + self.assertEqual(type(fd), DictSubclass) + self.assertEqual(created, frozendict(x=1)) + + # Subclass which doesn't override the constructor + class FrozenDictSubclass2(frozendict): + pass + + fd = FrozenDictSubclass2.fromkeys("abc") + self.assertEqual(fd, frozendict(a=None, b=None, c=None)) + self.assertEqual(type(fd), FrozenDictSubclass2) + + def test_pickle(self): + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + for fd in ( + frozendict(), + frozendict(x=1, y=2), + FrozenDict(x=1, y=2), + FrozenDictSlots(x=1, y=2), + ): + if type(fd) == FrozenDict: + fd.attr = 123 + if type(fd) == FrozenDictSlots: + fd.slot_attr = 456 + with self.subTest(fd=fd, proto=proto): + if proto >= 2: + p = pickle.dumps(fd, proto) + fd2 = pickle.loads(p) + self.assertEqual(fd2, fd) + self.assertEqual(type(fd2), type(fd)) + if type(fd) == FrozenDict: + self.assertEqual(fd2.attr, 123) + if type(fd) == FrozenDictSlots: + self.assertEqual(fd2.slot_attr, 456) + else: + # protocol 0 and 1 don't support frozendict + with self.assertRaises(TypeError): + pickle.dumps(fd, proto) + + def test_pickle_iter(self): + fd = frozendict(c=1, b=2, a=3, d=4, e=5, f=6) + for method_name in (None, 'keys', 'values', 'items'): + if method_name is not None: + meth = getattr(fd, method_name) + else: + meth = lambda: fd + expected = list(meth())[1:] + for proto in range(pickle.HIGHEST_PROTOCOL + 1): + with self.subTest(method_name=method_name, protocol=proto): + it = iter(meth()) + next(it) + p = pickle.dumps(it, proto) + unpickled = pickle.loads(p) + self.assertEqual(list(unpickled), expected) + self.assertEqual(list(it), expected) + + def test_unhashable_key(self): + d = frozendict(a=1) + key = [1, 2, 3] + + def check_unhashable_key(): + msg = "cannot use 'list' as a frozendict key (unhashable type: 'list')" + return self.assertRaisesRegex(TypeError, re.escape(msg)) + + with check_unhashable_key(): + key in d + with check_unhashable_key(): + d[key] + with check_unhashable_key(): + d.get(key) + + # Only TypeError exception is overridden, + # other exceptions are left unchanged. + class HashError: + def __hash__(self): + raise KeyError('error') + + key2 = HashError() + with self.assertRaises(KeyError): + key2 in d + with self.assertRaises(KeyError): + d[key2] + with self.assertRaises(KeyError): + d.get(key2) + if __name__ == "__main__": unittest.main() diff --git a/Lib/test/test_doctest/test_doctest.py b/Lib/test/test_doctest/test_doctest.py index 1945cd3f0dd..b9f67f07ebc 100644 --- a/Lib/test/test_doctest/test_doctest.py +++ b/Lib/test/test_doctest/test_doctest.py @@ -742,7 +742,7 @@ def non_Python_modules(): r""" >>> import builtins >>> tests = doctest.DocTestFinder().find(builtins) - >>> 750 < len(tests) < 800 # approximate number of objects with docstrings + >>> 750 < len(tests) < 850 # approximate number of objects with docstrings True >>> real_tests = [t for t in tests if len(t.examples) > 0] >>> len(real_tests) # objects that actually have doctests diff --git a/Lib/test/test_dynamic.py b/Lib/test/test_dynamic.py index 948a2029f1e..1df010732c1 100644 --- a/Lib/test/test_dynamic.py +++ b/Lib/test/test_dynamic.py @@ -135,7 +135,7 @@ def test_eval_gives_lambda_custom_globals(self): self.assertEqual(foo(), 7) - @unittest.expectedFailure # TODO: RUSTPYTHON + @unittest.expectedFailureIf(sys.getrecursionlimit() == 256, "TODO: RUSTPYTHON; default debug recursion limit is 256") @skip_wasi_stack_overflow() def test_load_global_specialization_failure_keeps_oparg(self): # https://github.com/python/cpython/issues/91625 diff --git a/Lib/test/test_import/data/syntax_warnings.py b/Lib/test/test_import/data/syntax_warnings.py new file mode 100644 index 00000000000..103f07b6187 --- /dev/null +++ b/Lib/test/test_import/data/syntax_warnings.py @@ -0,0 +1,21 @@ +# Syntax warnings emitted in different parts of the Python compiler. + +# Parser/lexer/lexer.c +x = 1or 0 # line 4 + +# Parser/tokenizer/helpers.c +'\z' # line 7 + +# Parser/string_parser.c +'\400' # line 10 + +# _PyCompile_Warn() in Python/codegen.c +assert(x, 'message') # line 13 +x is 1 # line 14 + +# _PyErr_EmitSyntaxWarning() in Python/ast_preprocess.c +def f(): + try: + pass + finally: + return 42 # line 21 diff --git a/Lib/test/test_pickle.py b/Lib/test/test_pickle.py index 71cd4baa942..1b85eceac17 100644 --- a/Lib/test/test_pickle.py +++ b/Lib/test/test_pickle.py @@ -107,6 +107,10 @@ def loads(self, buf, **kwds): def test_c_methods(self): return super().test_c_methods() + @unittest.expectedFailureIf(sys.getrecursionlimit() == 256, "TODO: RUSTPYTHON; default debug recursion limit is 256") + def test_deep_nested_struct_frozendict(self): + return super().test_deep_nested_struct_frozendict() + @@ -408,6 +412,10 @@ class CPicklerTests(PyPicklerTests): pickler = _pickle.Pickler unpickler = _pickle.Unpickler + @unittest.expectedSuccess # TODO: RUSTPYTHON; native pickler passes at the default debug recursion limit + def test_deep_nested_struct_frozendict(self): + return super().test_deep_nested_struct_frozendict() + class CPersPicklerTests(PyPersPicklerTests): pickler = _pickle.Pickler unpickler = _pickle.Unpickler @@ -421,6 +429,10 @@ class CDumpPickle_LoadPickle(PyPicklerTests): pickler = _pickle.Pickler unpickler = pickle._Unpickler + @unittest.expectedSuccess # TODO: RUSTPYTHON; native pickler passes at the default debug recursion limit + def test_deep_nested_struct_frozendict(self): + return super().test_deep_nested_struct_frozendict() + class DumpPickle_CLoadPickle(PyPicklerTests): pickler = pickle._Pickler unpickler = _pickle.Unpickler diff --git a/Lib/test/test_unittest/testmock/testhelpers.py b/Lib/test/test_unittest/testmock/testhelpers.py index bcae14c5b3c..0e82c723ec3 100644 --- a/Lib/test/test_unittest/testmock/testhelpers.py +++ b/Lib/test/test_unittest/testmock/testhelpers.py @@ -45,7 +45,6 @@ def test_any_and_datetime(self): mock.assert_called_with(ANY, foo=ANY) - @unittest.expectedFailure # TODO: RUSTPYTHON def test_any_mock_calls_comparison_order(self): mock = Mock() class Foo(object): diff --git a/crates/capi/src/ceval.rs b/crates/capi/src/ceval.rs index ace7896163c..036c259454b 100644 --- a/crates/capi/src/ceval.rs +++ b/crates/capi/src/ceval.rs @@ -3,10 +3,10 @@ use crate::pystate::with_vm; use crate::unicodeobject::decode_fsdefault_and_size; use crate::util::{CStrExt, FfiPtrExt}; use core::ffi::{CStr, c_char, c_int}; -use rustpython_vm::builtins::{PyCode, PyDict}; +use rustpython_vm::builtins::{PyAnyDict, PyCode}; use rustpython_vm::function::ArgMapping; use rustpython_vm::scope::Scope; -use rustpython_vm::{AsObject, PyObject, TryFromObject}; +use rustpython_vm::{AsObject, PyObject, PyRef, TryFromObject}; use rustpython_vm::{PyObjectRef, version}; #[unsafe(no_mangle)] @@ -32,12 +32,17 @@ pub unsafe extern "C" fn PyEval_EvalCode( ) -> *mut PyObject { with_vm(|vm| { let code = unsafe { co.assume_borrowed_and_cast::(vm) }?; - let globals = unsafe { globals.assume_borrowed_and_cast::(vm) }?; + let globals = PyRef::::try_from_object( + vm, + unsafe { globals.assume_borrowed() }.to_owned(), + )?; let locals = unsafe { locals.assume_borrowed_or_opt() } .map(|obj| ArgMapping::try_from_object(vm, obj.to_owned())) .transpose()?; - let scope = Scope::with_builtins(locals, globals.to_owned(), vm); + let locals = locals.or_else(|| Some(ArgMapping::from_anydict_exact(globals.clone()))); + // Unlike Python's eval/exec, this API does not insert __builtins__. + let scope = Scope::new(locals, globals); vm.run_code_obj(code.to_owned(), scope) }) diff --git a/crates/capi/src/dictobject.rs b/crates/capi/src/dictobject.rs index 2c3eedff3ba..1e6fab1b131 100644 --- a/crates/capi/src/dictobject.rs +++ b/crates/capi/src/dictobject.rs @@ -4,9 +4,8 @@ use crate::pystate::with_vm; use crate::util::{CStrExt, FfiPtrExt}; use core::ffi::{c_char, c_int}; use core::ptr::NonNull; -use rustpython_vm::AsObject; -use rustpython_vm::PyPayload; -use rustpython_vm::builtins::PyDict; +use rustpython_vm::builtins::{PyAnyDict, PyDict, PyFrozenDict}; +use rustpython_vm::{AsObject, Py, PyPayload, PyResult, VirtualMachine}; define_py_check!(fn PyDict_Check, types.dict_type); define_py_check!(exact fn PyDict_CheckExact, types.dict_type); @@ -14,6 +13,27 @@ define_py_check!(fn PyDictKeys_Check, types.dict_keys_type); define_py_check!(fn PyDictValues_Check, types.dict_values_type); define_py_check!(fn PyDictItems_Check, types.dict_items_type); +fn any_dict<'a>(dict: &'a PyObject, vm: &VirtualMachine) -> PyResult<&'a PyAnyDict> { + PyAnyDict::from_object(dict) + .ok_or_else(|| vm.new_system_error("bad argument to internal function")) +} + +fn writable_dict<'a>( + dict: &'a PyObject, + operation: &str, + vm: &VirtualMachine, +) -> PyResult<&'a Py> { + dict.downcast_ref::().ok_or_else(|| { + if dict.downcast_ref::().is_some() { + vm.new_type_error(format!( + "frozendict object does not support item {operation}" + )) + } else { + vm.new_system_error("bad argument to internal function") + } + }) +} + #[unsafe(no_mangle)] pub extern "C" fn PyDict_New() -> *mut PyObject { with_vm(|vm| vm.ctx.new_dict()) @@ -21,10 +41,10 @@ pub extern "C" fn PyDict_New() -> *mut PyObject { #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Clear(dict: *mut PyObject) { - with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; - dict.clear(); - Ok(()) + with_vm(|_vm| { + if let Some(dict) = unsafe { dict.assume_borrowed() }.downcast_ref::() { + dict.clear(); + } }) } @@ -35,7 +55,7 @@ pub unsafe extern "C" fn PyDict_SetItem( val: *mut PyObject, ) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let key = unsafe { key.assume_borrowed() }; let value = unsafe { val.assume_borrowed() }.to_owned(); dict.inner_setitem(key, value, vm) @@ -49,8 +69,8 @@ pub unsafe extern "C" fn PyDict_SetItemString( val: *mut PyObject, ) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; let key = unsafe { key.try_as_str(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let value = unsafe { val.assume_borrowed() }.to_owned(); dict.inner_setitem(key, value, vm) }) @@ -59,12 +79,14 @@ pub unsafe extern "C" fn PyDict_SetItemString( #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_GetItem(dict: *mut PyObject, key: *mut PyObject) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let Some(dict) = PyAnyDict::from_object(unsafe { dict.assume_borrowed() }) else { + return core::ptr::null_mut(); + }; let key = unsafe { key.assume_borrowed() }; match dict.inner_getitem_opt(key, vm) { - Ok(Some(value)) => Ok(value.as_object().as_raw().cast_mut()), - Ok(None) | Err(_) => Ok(core::ptr::null_mut()), + Ok(Some(value)) => value.as_object().as_raw().cast_mut(), + Ok(None) | Err(_) => core::ptr::null_mut(), } }) } @@ -75,7 +97,7 @@ pub unsafe extern "C" fn PyDict_GetItemWithError( key: *mut PyObject, ) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = any_dict(unsafe { dict.assume_borrowed() }, vm)?; let key = unsafe { key.assume_borrowed() }; if let Some(value) = dict.inner_getitem_opt(key, vm)? { @@ -92,12 +114,16 @@ pub unsafe extern "C" fn PyDict_GetItemString( key: *const c_char, ) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; - let key = unsafe { key.try_as_str(vm) }?; + let Ok(key) = (unsafe { key.try_as_str(vm) }) else { + return core::ptr::null_mut(); + }; + let Some(dict) = PyAnyDict::from_object(unsafe { dict.assume_borrowed() }) else { + return core::ptr::null_mut(); + }; - match dict.inner_getitem_opt(key, vm)? { - Some(value) => Ok(value.as_object().as_raw().cast_mut()), - None => Ok(core::ptr::null_mut()), + match dict.inner_getitem_opt(key, vm) { + Ok(Some(value)) => value.as_object().as_raw().cast_mut(), + Ok(None) | Err(_) => core::ptr::null_mut(), } }) } @@ -112,8 +138,8 @@ pub unsafe extern "C" fn PyDict_GetItemStringRef( unsafe { *result = core::ptr::null_mut(); } - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; let key = unsafe { key.try_as_str(vm) }?; + let dict = any_dict(unsafe { dict.assume_borrowed() }, vm)?; if let Some(value) = dict.inner_getitem_opt(key, vm)? { unsafe { @@ -134,7 +160,7 @@ pub unsafe extern "C" fn PyDict_GetItemRef( ) -> c_int { with_vm(|vm| { unsafe { *result = core::ptr::null_mut() }; - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = any_dict(unsafe { dict.assume_borrowed() }, vm)?; let key = unsafe { key.assume_borrowed() }; if let Some(value) = dict.inner_getitem_opt(key, vm)? { @@ -165,7 +191,7 @@ pub unsafe extern "C" fn PyDict_SetDefaultRef( result.write(core::ptr::null_mut()); } } - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let key = unsafe { key.assume_borrowed() }; if let Some(value) = dict.inner_getitem_opt(key, vm)? { @@ -191,7 +217,9 @@ pub unsafe extern "C" fn PyDict_SetDefaultRef( #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Size(dict: *mut PyObject) -> isize { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed_or_opt() } + .ok_or_else(|| vm.new_system_error("bad argument to internal function"))?; + let dict = any_dict(dict, vm)?; Ok(dict.__len__()) }) } @@ -199,7 +227,7 @@ pub unsafe extern "C" fn PyDict_Size(dict: *mut PyObject) -> isize { #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Contains(dict: *mut PyObject, key: *mut PyObject) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = any_dict(unsafe { dict.assume_borrowed() }, vm)?; let key = unsafe { key.assume_borrowed() }; Ok(dict.inner_getitem_opt(key, vm)?.is_some()) }) @@ -208,7 +236,9 @@ pub unsafe extern "C" fn PyDict_Contains(dict: *mut PyObject, key: *mut PyObject #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Copy(dict: *mut PyObject) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed_or_opt() } + .and_then(|dict| dict.downcast_ref::()) + .ok_or_else(|| vm.new_system_error("bad argument to internal function"))?; Ok(dict.copy().into_ref(&vm.ctx)) }) } @@ -216,8 +246,12 @@ pub unsafe extern "C" fn PyDict_Copy(dict: *mut PyObject) -> *mut PyObject { #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_DelItem(dict: *mut PyObject, key: *mut PyObject) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed() }; let key = unsafe { key.assume_borrowed() }; + if dict.downcast_ref::().is_some() { + PyFrozenDict::key_hash(key, vm)?; + } + let dict = writable_dict(dict, "deletion", vm)?; dict.del_item(key, vm) }) } @@ -225,8 +259,8 @@ pub unsafe extern "C" fn PyDict_DelItem(dict: *mut PyObject, key: *mut PyObject) #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_DelItemString(dict: *mut PyObject, key: *const c_char) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; let key = unsafe { key.try_as_str(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "deletion", vm)?; dict.del_item(key, vm) }) } @@ -234,7 +268,9 @@ pub unsafe extern "C" fn PyDict_DelItemString(dict: *mut PyObject, key: *const c #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Items(dict: *mut PyObject) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed_or_opt() } + .ok_or_else(|| vm.new_system_error("bad argument to internal function"))?; + let dict = any_dict(dict, vm)?; let items = dict .items_vec() .into_iter() @@ -247,7 +283,9 @@ pub unsafe extern "C" fn PyDict_Items(dict: *mut PyObject) -> *mut PyObject { #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Keys(dict: *mut PyObject) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed_or_opt() } + .ok_or_else(|| vm.new_system_error("bad argument to internal function"))?; + let dict = any_dict(dict, vm)?; Ok(vm.ctx.new_list(dict.keys_vec())) }) } @@ -255,7 +293,9 @@ pub unsafe extern "C" fn PyDict_Keys(dict: *mut PyObject) -> *mut PyObject { #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Values(dict: *mut PyObject) -> *mut PyObject { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = unsafe { dict.assume_borrowed_or_opt() } + .ok_or_else(|| vm.new_system_error("bad argument to internal function"))?; + let dict = any_dict(dict, vm)?; Ok(vm.ctx.new_list(dict.values_vec())) }) } @@ -267,7 +307,7 @@ pub unsafe extern "C" fn PyDict_Merge( override_: c_int, ) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let other = unsafe { other.assume_borrowed() }.to_owned(); if override_ != 0 { dict.merge_object(other, vm) @@ -280,7 +320,7 @@ pub unsafe extern "C" fn PyDict_Merge( #[unsafe(no_mangle)] pub unsafe extern "C" fn PyDict_Update(dict: *mut PyObject, other: *mut PyObject) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let other = unsafe { other.assume_borrowed() }.to_owned(); dict.merge_object(other, vm) }) @@ -293,7 +333,7 @@ pub unsafe extern "C" fn PyDict_MergeFromSeq2( override_: c_int, ) -> c_int { with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + let dict = writable_dict(unsafe { dict.assume_borrowed() }, "assignment", vm)?; let seq2 = unsafe { seq2.assume_borrowed() }; dict.merge_from_seq2(seq2, override_ != 0, vm) }) @@ -306,8 +346,10 @@ pub unsafe extern "C" fn PyDict_Next( key: *mut *mut PyObject, value: *mut *mut PyObject, ) -> c_int { - with_vm(|vm| { - let dict = unsafe { dict.assume_borrowed_and_cast::(vm) }?; + with_vm(|_vm| { + let Some(dict) = PyAnyDict::from_object(unsafe { dict.assume_borrowed() }) else { + return false; + }; let index = unsafe { *pos } as usize; if let Some((next_pos, k, v)) = dict.next_entry(index) { @@ -320,9 +362,9 @@ pub unsafe extern "C" fn PyDict_Next( value.write(v.as_object().as_raw().cast_mut()); } } - Ok(true) + true } else { - Ok(false) + false } }) } diff --git a/crates/capi/src/pyframe.rs b/crates/capi/src/pyframe.rs index 2214ee55820..c3e81c235b3 100644 --- a/crates/capi/src/pyframe.rs +++ b/crates/capi/src/pyframe.rs @@ -2,9 +2,9 @@ use crate::PyObject; use crate::pystate::with_vm; use crate::util::FfiPtrExt; use core::ffi::c_int; -use rustpython_vm::Py; use rustpython_vm::builtins::PyCode; use rustpython_vm::frame::FrameObject; +use rustpython_vm::{Py, PyObjectRef}; pub type PyFrameObject = Py; pub type PyCodeObject = Py; @@ -26,7 +26,7 @@ pub unsafe extern "C" fn PyFrame_GetLocals(frame: *mut PyFrameObject) -> *mut Py #[unsafe(no_mangle)] pub unsafe extern "C" fn PyFrame_GetGlobals(frame: *mut PyFrameObject) -> *mut PyObject { - with_vm(|_vm| Ok(unsafe { frame.assume_borrowed() }.f_globals())) + with_vm(|_vm| PyObjectRef::from(unsafe { frame.assume_borrowed() }.f_globals())) } #[unsafe(no_mangle)] diff --git a/crates/vm/src/builtins/dict.rs b/crates/vm/src/builtins/dict.rs index a98ba4a4356..9d876f2f906 100644 --- a/crates/vm/src/builtins/dict.rs +++ b/crates/vm/src/builtins/dict.rs @@ -7,22 +7,26 @@ use crate::common::lock::LazyLock; use crate::object::{Traverse, TraverseFn}; use crate::stdlib::PyOrderedDictItems; use crate::{ - AsObject, Context, Py, PyExact, PyObject, PyObjectRef, PyPayload, PyRef, PyRefExact, PyResult, + AsObject, Context, Py, PyExact, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, TryFromObject, atomic_func, builtins::{PyList, PyTuple, iter::builtins_iter, type_::PyAttributes}, class::{PyClassDef, PyClassImpl}, - common::{ascii, hash::PyHash}, + common::{ + ascii, + hash::{self, PyHash}, + }, dict_inner::{self, DictKey}, function::{ArgIterable, FuncArgs, KwArgs, OptionalArg, PyArithmeticValue, PyComparisonValue}, protocol::{PyIter, PyIterReturn, PyMappingMethods, PyNumberMethods, PySequenceMethods}, recursion::ReprGuard, types::{ AsMapping, AsNumber, AsSequence, Callable, Comparable, Constructor, DefaultConstructor, - Initializer, IterNext, Iterable, PyComparisonOp, Representable, SelfIter, + Hashable, Initializer, IterNext, Iterable, PyComparisonOp, Representable, SelfIter, }, vm::VirtualMachine, }; use alloc::fmt; +use core::borrow::Borrow; use core::cell::Cell; use core::ptr::NonNull; use rustpython_common::atomic::{Ordering, PyAtomic, Radium}; @@ -31,15 +35,264 @@ use rustpython_common::wtf8::Wtf8Buf; pub(crate) type DictContentType = dict_inner::Dict; +pub(crate) fn key_hash( + key: &K, + mapping_name: &str, + vm: &VirtualMachine, +) -> PyResult { + match key.key_hash(vm) { + // Exact TypeError only: exception subclasses retain their identity. + Err(exc) if exc.class().is(vm.ctx.exceptions.type_error) => { + let name = key.to_pyobject(vm).class().fully_qualified_name(vm)?; + // Match BaseException_str without the diagnostic formatter's + // fallback: argument __str__/__repr__ failures must propagate. + let args = exc.args(); + let message = match args.as_slice() { + [] => vm.ctx.empty_str.to_owned(), + [arg] => arg.str(vm)?, + _ => args.as_object().str(vm)?, + }; + Err(vm.new_type_error(format!( + "cannot use '{name}' as a {mapping_name} key ({message})" + ))) + } + result => result, + } +} + #[pyclass(module = false, name = "dict", unhashable = true, traverse = "manual")] #[derive(Default)] // OrderedDict contains eight-byte-aligned fields on 32-bit targets too. #[repr(align(8))] pub struct PyDict { - entries: DictContentType, + pub(crate) entries: DictContentType, } pub type PyDictRef = PyRef; +/// A borrowed view of a dict or frozendict, including subclasses. +/// +/// This is an erased header view, not a shared payload layout. Every payload +/// access downcasts to the actual alternative, preserving its mutability. +#[repr(transparent)] +pub struct PyAnyDict { + object: PyObject, +} + +// SAFETY: this owner contributes exactly one strong reference to the object. +unsafe impl Traverse for PyRef { + fn traverse(&self, tracer: &mut TraverseFn<'_>) { + tracer(self.as_object()); + } +} + +impl crate::convert::ToPyObject for PyRef { + fn to_pyobject(self, _vm: &VirtualMachine) -> PyObjectRef { + self.into_object() + } +} + +impl Borrow for PyAnyDict { + fn borrow(&self) -> &PyObject { + &self.object + } +} + +impl Borrow for PyRef { + fn borrow(&self) -> &PyAnyDict { + self + } +} + +impl ToOwned for PyAnyDict { + type Owned = PyRef; + + fn to_owned(&self) -> PyRef { + // SAFETY: this borrowed view already guarantees a dictionary payload. + unsafe { PyRef::::from_object_unchecked(self.object.to_owned()) } + } +} + +impl From for PyRef { + fn from(dict: PyDictRef) -> Self { + // SAFETY: the typed owner guarantees the dict alternative. + unsafe { Self::from_object_unchecked(dict.into()) } + } +} + +impl From> for PyRef { + fn from(dict: PyRef) -> Self { + // SAFETY: the typed owner guarantees the frozendict alternative. + unsafe { Self::from_object_unchecked(dict.into()) } + } +} + +impl From> for PyObjectRef { + fn from(dict: PyRef) -> Self { + dict.into_object() + } +} + +// SAFETY: the transparent PyRef owner has the same layout as PyObjectRef. +// A successful check proves the actual dict or frozendict payload invariant. +unsafe impl crate::convert::TransmuteFromObject for PyRef { + fn check(vm: &VirtualMachine, obj: &PyObject) -> PyResult<()> { + if PyAnyDict::from_object(obj).is_some() { + Ok(()) + } else { + Err(vm.new_type_error(format!( + "expected dict or frozendict, got {}", + obj.class().name() + ))) + } + } +} + +impl TryFromObject for PyRef { + fn try_from_object(vm: &VirtualMachine, obj: PyObjectRef) -> PyResult { + ::check(vm, &obj)?; + // SAFETY: the shared checker proved the union view's payload invariant. + Ok(unsafe { Self::from_object_unchecked(obj) }) + } +} + +impl PyAnyDict { + pub fn from_object(obj: &PyObject) -> Option<&Self> { + if obj.downcast_ref::().is_some() || obj.downcast_ref::().is_some() { + // SAFETY: PyAnyDict is transparent over PyObject and the payload was checked. + Some(unsafe { &*(obj as *const PyObject).cast::() }) + } else { + None + } + } + + pub(crate) fn as_mutable(&self) -> Option<&Py> { + self.object.downcast_ref() + } + + pub(crate) fn is_frozen(&self) -> bool { + self.object.downcastable::() + } + + // The raw payload is private: callers outside the VM cannot use it to mutate a frozendict. + pub(crate) fn as_dict(&self) -> &PyDict { + if let Some(dict) = self.as_mutable() { + dict + } else { + &self.object.downcast_ref::().unwrap().dict + } + } + + pub(crate) fn uses_builtin_iter(&self, vm: &VirtualMachine) -> bool { + let base = if self.is_frozen() { + vm.ctx.types.frozendict_type + } else { + vm.ctx.types.dict_type + }; + self.as_object() + .class() + .slots + .iter + .load() + .map(crate::types::fn_addr) + == base.slots.iter.load().map(crate::types::fn_addr) + } + + #[must_use] + pub fn __len__(&self) -> usize { + self.as_dict().__len__() + } + #[must_use] + pub fn is_empty(&self) -> bool { + self.as_dict().is_empty() + } + #[must_use] + pub fn keys_vec(&self) -> Vec { + self.as_dict().keys_vec() + } + #[must_use] + pub fn values_vec(&self) -> Vec { + self.as_dict().values_vec() + } + #[must_use] + pub fn items_vec(&self) -> Vec<(PyObjectRef, PyObjectRef)> { + self.as_dict().items_vec() + } + #[must_use] + pub fn next_entry(&self, position: usize) -> Option<(usize, PyObjectRef, PyObjectRef)> { + self.as_dict().next_entry(position) + } + pub(crate) fn size(&self) -> crate::dict_inner::DictSize { + self.as_dict().size() + } + pub fn inner_getitem_opt( + &self, + key: &K, + vm: &VirtualMachine, + ) -> PyResult> { + if let Some(dict) = self.as_mutable() { + dict.inner_getitem_opt(key, vm) + } else { + self.object + .downcast_ref::() + .unwrap() + .inner_getitem_opt(key, vm) + } + } + pub fn get_item_opt( + &self, + key: &K, + vm: &VirtualMachine, + ) -> PyResult> { + if let Some(dict) = self.as_mutable() { + return dict.get_item_opt(key, vm); + } + if self.object.class().is(vm.ctx.types.frozendict_type) { + return self.inner_getitem_opt(key, vm); + } + match self.object.get_item(key, vm) { + Ok(value) => Ok(Some(value)), + Err(exc) if exc.fast_isinstance(vm.ctx.exceptions.key_error) => Ok(None), + Err(exc) => Err(exc), + } + } + pub fn get_item(&self, key: &K, vm: &VirtualMachine) -> PyResult { + if let Some(dict) = self.as_mutable() { + dict.get_item(key, vm) + } else { + self.object.get_item(key, vm) + } + } + pub fn contains_key(&self, key: &K, vm: &VirtualMachine) -> bool { + self.inner_getitem_opt(key, vm) + .is_ok_and(|value| value.is_some()) + } + pub fn set_item( + &self, + key: &K, + value: PyObjectRef, + vm: &VirtualMachine, + ) -> PyResult<()> { + match self.as_mutable() { + Some(dict) => dict.set_item(key, value, vm), + None => Err(vm.new_type_error("frozendict object does not support item assignment")), + } + } + pub fn del_item(&self, key: &K, vm: &VirtualMachine) -> PyResult<()> { + match self.as_mutable() { + Some(dict) => dict.del_item(key, vm), + None => Err(vm.new_type_error("frozendict object does not support item deletion")), + } + } +} + +impl<'a> IntoIterator for &'a PyAnyDict { + type Item = (PyObjectRef, PyObjectRef); + type IntoIter = DictIter<'a>; + fn into_iter(self) -> Self::IntoIter { + DictIter::new(self.as_dict()) + } +} + // SAFETY: Traverse properly visits all owned PyObjectRefs unsafe impl Traverse for PyDict { fn traverse(&self, traverse_fn: &mut TraverseFn<'_>) { @@ -140,51 +393,51 @@ impl PyDict { self.entries.items_with_hashes() } - fn merge_object_with_override( + /// `dict_update_arg`. `mapping_name` names the target type in the + /// unhashable key error, as `dict_unhashable_type` does. + pub(crate) fn update_arg( &self, other: PyObjectRef, override_existing: bool, + mapping_name: &str, vm: &VirtualMachine, ) -> PyResult<()> { - let casted: Result, _> = other.downcast_exact(vm); - let other = match casted { - Ok(dict_other) => { - return self.merge_dict(&dict_other, override_existing, vm); - } - Err(other) => other, - }; + if let Some(other_dict) = PyAnyDict::from_object(&other) + && other_dict.uses_builtin_iter(vm) + { + return self.merge_dict(other_dict.as_dict(), override_existing, vm); + } let dict = &self.entries; // Use get_attr to properly invoke __getattribute__ for proxy objects - let keys_result = other.get_attr(vm.ctx.intern_str("keys"), vm); - let has_keys = match keys_result { - Ok(keys_method) => { - let keys = PyIter::try_from_object(vm, keys_method.call((), vm)?)?; - while let PyIterReturn::Return(key) = keys.next(vm)? { - let hash = Self::hash_or_unhashable(&*key, vm)?; - if !override_existing && dict.contains(vm, &*key, hash)? { - continue; - } - let val = other.get_item(&*key, vm)?; - dict.insert(vm, &*key, hash, val)?; - } - true + let keys_method = match other.get_attr(vm.ctx.intern_str("keys"), vm) { + Ok(keys_method) => keys_method, + Err(e) if e.fast_isinstance(vm.ctx.exceptions.attribute_error) => { + return self.merge_seq2(&other, override_existing, mapping_name, vm); } - Err(e) if e.fast_isinstance(vm.ctx.exceptions.attribute_error) => false, Err(e) => return Err(e), }; - if !has_keys { - return self.merge_from_seq2(&other, override_existing, vm); + let keys = PyIter::try_from_object(vm, keys_method.call((), vm)?)?; + while let PyIterReturn::Return(key) = keys.next(vm)? { + if !override_existing { + let hash = key_hash(&*key, mapping_name, vm)?; + if dict.contains(vm, &*key, hash)? { + continue; + } + } + let value = other.get_item(&*key, vm)?; + let hash = key_hash(&*key, mapping_name, vm)?; + dict.insert(vm, &*key, hash, value)?; } Ok(()) } // Used in update and ior. pub fn merge_object(&self, other: PyObjectRef, vm: &VirtualMachine) -> PyResult<()> { - self.merge_object_with_override(other, true, vm) + self.update_arg(other, true, "dict", vm) } pub fn merge_object_if_missing(&self, other: PyObjectRef, vm: &VirtualMachine) -> PyResult<()> { - self.merge_object_with_override(other, false, vm) + self.update_arg(other, false, "dict", vm) } fn add_update_sequence_note( @@ -262,6 +515,17 @@ impl PyDict { seq2: &PyObject, override_existing: bool, vm: &VirtualMachine, + ) -> PyResult<()> { + self.merge_seq2(seq2, override_existing, "dict", vm) + } + + /// `dict_merge_from_seq2` + fn merge_seq2( + &self, + seq2: &PyObject, + override_existing: bool, + mapping_name: &str, + vm: &VirtualMachine, ) -> PyResult<()> { let iter = seq2.get_iter(vm)?; let dict = &self.entries; @@ -269,7 +533,7 @@ impl PyDict { for (index, element) in iter.iter::(vm)?.enumerate() { let (key, value) = Self::update_sequence_pair(element?, index, vm)?; - let hash = Self::hash_or_unhashable(&*key, vm)?; + let hash = key_hash(&*key, mapping_name, vm)?; if !override_existing && dict.contains(vm, &*key, hash)? { continue; } @@ -280,7 +544,7 @@ impl PyDict { pub(crate) fn merge_dict( &self, - dict_other: &Py, + dict_other: &Self, override_existing: bool, vm: &VirtualMachine, ) -> PyResult<()> { @@ -314,22 +578,7 @@ impl PyDict { /// Hashing up front also means a `__hash__` that fails only intermittently /// is still reported. fn hash_or_unhashable(key: &K, vm: &VirtualMachine) -> PyResult { - match key.key_hash(vm) { - Ok(hash) => Ok(hash), - // Exact `TypeError` only: a `__hash__` raising a *subclass* of - // `TypeError` (e.g. a user `MyTypeError`) must propagate unchanged, - // matching CPython's exact-type check. - Err(cause) if cause.class().is(vm.ctx.exceptions.type_error) => { - let message = cause.as_object().str(vm)?; - let key = key.to_pyobject(vm); - let name = key.class().fully_qualified_name(vm)?; - let err = - vm.new_type_error(format!("cannot use '{name}' as a dict key ({message})")); - err.set_cause(Some(cause)); - Err(err) - } - Err(other) => Err(other), - } + key_hash(key, "dict", vm) } /// Set item variant which can be called with multiple @@ -409,6 +658,8 @@ impl PyDict { ) -> Option> { if let Some(dict) = obj.downcast_ref_if_exact::(vm) { Some(dict.entries.keys_with_hashes()) + } else if let Some(frozen) = obj.downcast_ref_if_exact::(vm) { + Some(frozen.dict.entries.keys_with_hashes()) } else { set::exact_set_keys_with_hashes(obj, vm) } @@ -426,13 +677,48 @@ struct DictGetArgs { #[derive(FromArgs)] struct FromKeysArgs { #[pyarg(positional)] - iterable: ArgIterable, + iterable: PyObjectRef, #[pyarg(positional, optional)] value: Option, } // Python dict methods: impl PyDict { + pub(crate) fn compare_entries( + &self, + other: &Self, + op: PyComparisonOp, + item: bool, + vm: &VirtualMachine, + ) -> PyResult { + if op == PyComparisonOp::Ne { + return Self::compare_entries(self, other, PyComparisonOp::Eq, item, vm) + .map(|x| x.map(|eq| !eq)); + } + if !op.eval_ord(self.__len__().cmp(&other.__len__())) { + return Ok(PyArithmeticValue::Implemented(false)); + } + let mut position = 0; + while let Some((next, k, v1, hash)) = self.entries.next_entry_with_hash(position) { + position = next; + let found = other.entries.get(vm, &*k, hash)?; + match found { + Some(v2) => { + if v1.is(&v2) { + continue; + } + if item && !vm.bool_eq(&v1, &v2)? { + return Ok(PyArithmeticValue::Implemented(false)); + } + } + None => { + return Ok(PyArithmeticValue::Implemented(false)); + } + } + } + Ok(PyArithmeticValue::Implemented(true)) + } + pub fn __len__(&self) -> usize { self.entries.len() } @@ -460,17 +746,6 @@ impl PyDict { self.entries.setdefault(vm, &*key, hash, || default) } - fn __or__(&self, other: PyObjectRef, vm: &VirtualMachine) -> PyResult { - // Accept dict subclasses that inherit this implementation. Subclasses with - // their own reflected slot, such as OrderedDict, take precedence in dispatch. - if let Ok(other) = other.downcast::() { - let self_cp = self.copy(); - self_cp.merge_dict(&other, true, vm)?; - return Ok(self_cp.into_pyobject(vm)); - } - Ok(vm.ctx.not_implemented()) - } - pub(crate) fn __sizeof__(&self) -> usize { core::mem::size_of::() + self.entries.sizeof() } @@ -519,88 +794,20 @@ impl PyDict {} #[pyclass] impl Py { - pub(crate) fn inner_cmp( - &self, - other: &Self, - op: PyComparisonOp, - item: bool, - vm: &VirtualMachine, - ) -> PyResult { - if op == PyComparisonOp::Ne { - return Self::inner_cmp(self, other, PyComparisonOp::Eq, item, vm) - .map(|x| x.map(|eq| !eq)); - } - if !op.eval_ord(self.__len__().cmp(&other.__len__())) { - return Ok(PyArithmeticValue::Implemented(false)); - } - let (superset, subset) = if self.__len__() < other.__len__() { - (other, self) - } else { - (self, other) - }; - for (k, v1) in subset { - match superset.get_item_opt(&*k, vm)? { - Some(v2) => { - if v1.is(&v2) { - continue; - } - if item && !vm.bool_eq(&v1, &v2)? { - return Ok(PyArithmeticValue::Implemented(false)); - } - } - None => { - return Ok(PyArithmeticValue::Implemented(false)); - } - } - } - Ok(PyArithmeticValue::Implemented(true)) - } - #[cfg_attr(feature = "flame-it", flame("PyDictRef"))] #[pymethod(coexist)] fn __getitem__(&self, key: PyObjectRef, vm: &VirtualMachine) -> PyResult { self.inner_getitem(&*key, vm) } - fn __ror__(&self, other: PyObjectRef, vm: &VirtualMachine) -> PyResult { - let other_dict = other.downcast::(); - if let Ok(other) = other_dict { - let other_cp = other.copy(); - other_cp.merge_dict(self, true, vm)?; - return Ok(other_cp.into_pyobject(vm)); - } - Ok(vm.ctx.not_implemented()) - } - #[pyclassmethod] fn fromkeys(class: PyTypeRef, args: FromKeysArgs, vm: &VirtualMachine) -> PyResult { - let FromKeysArgs { iterable, value } = args; - let value = value.unwrap_or_else(|| vm.ctx.none()); - let d = PyType::call(&class, ().into(), vm)?; - match d.downcast_exact::(vm) { - Ok(pydict) => { - if let Some(keys) = PyDict::fromkeys_known_hashes(iterable.as_object(), vm) { - if class.is(vm.ctx.types.dict_type) { - pydict.entries.reserve_for_empty(keys.len()); - } - for (key, hash) in keys { - pydict.entries.insert(vm, &*key, hash, value.clone())?; - } - } else { - for key in iterable.iter(vm)? { - let key = key?; - pydict.__setitem__(&key, value.clone(), vm)?; - } - } - Ok(pydict.into_pyref().into()) - } - Err(pyobj) => { - for key in iterable.iter(vm)? { - pyobj.set_item(&*key?, value.clone(), vm)?; - } - Ok(pyobj) - } - } + dict_fromkeys( + class, + args.iterable, + args.value.unwrap_or_else(|| vm.ctx.none()), + vm, + ) } #[pymethod] @@ -681,23 +888,23 @@ impl Py { #[pyclass] impl PyRef { #[pymethod] - const fn keys(self) -> PyDictKeys { - PyDictKeys::new(self) + fn keys(self) -> PyDictKeys { + PyDictKeys::new(self.into()) } #[pymethod] - const fn values(self) -> PyDictValues { - PyDictValues::new(self) + fn values(self) -> PyDictValues { + PyDictValues::new(self.into()) } #[pymethod] - const fn items(self) -> PyDictItems { - PyDictItems::new(self) + fn items(self) -> PyDictItems { + PyDictItems::new(self.into()) } #[pymethod] fn __reversed__(self) -> PyDictReverseKeyIterator { - PyDictReverseKeyIterator::new(self) + PyDictReverseKeyIterator::new(self.into()) } fn __ior__(self, other: PyObjectRef, vm: &VirtualMachine) -> PyResult { @@ -758,13 +965,7 @@ impl AsSequence for PyDict { impl AsNumber for PyDict { fn as_number() -> &'static PyNumberMethods { static AS_NUMBER: PyNumberMethods = PyNumberMethods { - or: Some(|a, b, vm| { - if let Some(a) = a.downcast_ref::() { - PyDict::__or__(a, b.to_pyobject(vm), vm) - } else { - Ok(vm.ctx.not_implemented()) - } - }), + or: Some(dict_or), inplace_or: Some(|a, b, vm| { if let Some(a) = a.downcast_ref::() { a.to_owned() @@ -787,16 +988,19 @@ impl Comparable for PyDict { op: PyComparisonOp, vm: &VirtualMachine, ) -> PyResult { - op.eq_only(|| { - let other = class_or_notimplemented!(Self, other); - zelf.inner_cmp(other, PyComparisonOp::Eq, true, vm) + op.eq_only(|| match PyAnyDict::from_object(other) { + Some(other) => { + zelf.payload + .compare_entries(other.as_dict(), PyComparisonOp::Eq, true, vm) + } + None => Ok(PyComparisonValue::NotImplemented), }) } } impl Iterable for PyDict { fn iter(zelf: PyRef, vm: &VirtualMachine) -> PyResult { - Ok(PyDictKeyIterator::new(zelf).into_pyobject(vm)) + Ok(PyDictKeyIterator::new(zelf.into()).into_pyobject(vm)) } } @@ -1165,6 +1369,292 @@ impl PyExact { } } +#[pyclass(module = false, name = "frozendict", traverse = "manual")] +#[derive(Debug)] +pub struct PyFrozenDict { + // Own the table directly, never a mutable Python dict visible through gc or a view. + pub(crate) dict: PyDict, + hash: PyAtomic, +} + +unsafe impl Traverse for PyFrozenDict { + fn traverse(&self, tracer: &mut TraverseFn<'_>) { + self.dict.traverse(tracer); + } + fn clear(&mut self, out: &mut Vec) { + Traverse::clear(&mut self.dict, out); + } +} + +impl PyPayload for PyFrozenDict { + fn class(ctx: &Context) -> &'static Py { + ctx.types.frozendict_type + } +} + +impl PyFrozenDict { + pub(crate) fn from_dict(dict: PyDict) -> Self { + Self { + dict, + hash: Radium::new(hash::SENTINEL), + } + } + + /// Hash a dictionary key, adding the frozendict context to an exact TypeError. + pub fn key_hash(key: &K, vm: &VirtualMachine) -> PyResult { + key_hash(key, "frozendict", vm) + } + + pub fn inner_getitem_opt( + &self, + key: &K, + vm: &VirtualMachine, + ) -> PyResult> { + let hash = Self::key_hash(key, vm)?; + self.dict.entries.get(vm, key, hash) + } +} + +impl Constructor for PyFrozenDict { + type Args = (OptionalArg, KwArgs); + fn py_new( + _cls: &Py, + (input, kwargs): Self::Args, + vm: &VirtualMachine, + ) -> PyResult { + // The table is built before the frozendict is published, as + // frozendict_new_untracked keeps it away from the GC. + let dict = PyDict::default(); + if let OptionalArg::Present(input) = input { + dict.update_arg(input, true, "frozendict", vm)?; + } + for (key, value) in kwargs { + dict.inner_setitem(&key, value, vm)?; + } + Ok(Self::from_dict(dict)) + } +} + +#[pyclass( + with( + Py, + PyRef, + Constructor, + Hashable, + Comparable, + Iterable, + AsMapping, + AsSequence, + AsNumber, + Representable + ), + flags(BASETYPE, MAPPING, _MATCH_SELF) +)] +impl PyFrozenDict { + fn __len__(&self) -> usize { + self.dict.__len__() + } + + #[pymethod] + fn get( + zelf: &Py, + key: PyObjectRef, + default: OptionalArg, + vm: &VirtualMachine, + ) -> PyResult { + Ok(zelf + .inner_getitem_opt(&*key, vm)? + .unwrap_or_else(|| default.unwrap_or_else(|| vm.ctx.none()))) + } + + #[pymethod] + fn __sizeof__(zelf: &Py) -> usize { + zelf.dict.__sizeof__() + core::mem::size_of::>() + } + + #[pyclassmethod] + fn fromkeys( + cls: PyTypeRef, + iterable: PyObjectRef, + value: OptionalArg, + vm: &VirtualMachine, + ) -> PyResult { + dict_fromkeys(cls, iterable, value.unwrap_or_else(|| vm.ctx.none()), vm) + } + + #[pyclassmethod] + fn __class_getitem__( + cls: PyTypeRef, + object: PyObjectRef, + vm: &VirtualMachine, + ) -> PyResult { + PyGenericAlias::from_args(cls, object, vm) + } +} + +#[pyclass] +impl Py { + #[pymethod(coexist)] + fn __getitem__(&self, key: PyObjectRef, vm: &VirtualMachine) -> PyResult { + if let Some(value) = self.inner_getitem_opt(&*key, vm)? { + return Ok(value); + } + if !self.class().is(vm.ctx.types.frozendict_type) + && let Some(method) = + vm.get_method(self.to_owned().into(), identifier!(vm, __missing__)) + { + return method?.call((key,), vm); + } + Err(vm.new_key_error(key)) + } + + #[pymethod] + fn copy(&self, vm: &VirtualMachine) -> PyResult> { + if self.class().is(vm.ctx.types.frozendict_type) { + Ok(self.to_owned()) + } else { + let dict = PyDict::default(); + if !self.dict.is_empty() { + dict.update_arg(self.to_owned().into(), true, "frozendict", vm)?; + } + Ok(PyFrozenDict::from_dict(dict).into_ref(&vm.ctx)) + } + } + + #[pymethod] + fn __getnewargs__(&self, vm: &VirtualMachine) -> PyResult { + let dict = vm + .ctx + .types + .dict_type + .as_object() + .call((self.to_owned(),), vm)?; + Ok(vm.new_tuple((dict,))) + } +} + +#[pyclass] +impl PyRef { + #[pymethod] + fn keys(self) -> PyDictKeys { + PyDictKeys::new(self.into()) + } + #[pymethod] + fn values(self) -> PyDictValues { + PyDictValues::new(self.into()) + } + #[pymethod] + fn items(self) -> PyDictItems { + PyDictItems::new(self.into()) + } + #[pymethod] + fn __reversed__(self) -> PyDictReverseKeyIterator { + PyDictReverseKeyIterator::new(self.into()) + } +} + +impl AsMapping for PyFrozenDict { + fn as_mapping() -> &'static PyMappingMethods { + static METHODS: PyMappingMethods = PyMappingMethods { + length: atomic_func!(|obj, _vm| Ok(PyFrozenDict::mapping_downcast(obj).__len__())), + subscript: atomic_func!( + |obj, key, vm| PyFrozenDict::mapping_downcast(obj).__getitem__(key.to_owned(), vm) + ), + ass_subscript: None, + }; + &METHODS + } +} +impl AsSequence for PyFrozenDict { + fn as_sequence() -> &'static PySequenceMethods { + static METHODS: LazyLock = LazyLock::new(|| PySequenceMethods { + contains: atomic_func!(|obj, key, vm| Ok(PyFrozenDict::sequence_downcast(obj) + .inner_getitem_opt(key, vm)? + .is_some())), + ..PySequenceMethods::NOT_IMPLEMENTED + }); + &METHODS + } +} +impl AsNumber for PyFrozenDict { + fn as_number() -> &'static PyNumberMethods { + static METHODS: PyNumberMethods = PyNumberMethods { + or: Some(frozendict_or), + ..PyNumberMethods::NOT_IMPLEMENTED + }; + &METHODS + } +} +impl Iterable for PyFrozenDict { + fn iter(zelf: PyRef, vm: &VirtualMachine) -> PyResult { + Ok(PyDictKeyIterator::new(zelf.into()).into_pyobject(vm)) + } +} +impl Comparable for PyFrozenDict { + fn cmp( + zelf: &Py, + other: &PyObject, + op: PyComparisonOp, + vm: &VirtualMachine, + ) -> PyResult { + op.eq_only(|| match PyAnyDict::from_object(other) { + Some(other) => zelf + .dict + .compare_entries(other.as_dict(), PyComparisonOp::Eq, true, vm), + None => Ok(PyComparisonValue::NotImplemented), + }) + } +} +impl Hashable for PyFrozenDict { + fn hash(zelf: &Py, vm: &VirtualMachine) -> PyResult { + let cached = zelf.hash.load(Ordering::Relaxed); + if cached != hash::SENTINEL { + return Ok(cached); + } + let mut hasher = hash::FrozenSetHash::new(zelf.__len__()); + let mut position = 0; + while let Some((next, _key, value, key_hash)) = + zelf.dict.entries.next_entry_with_hash(position) + { + position = next; + // next_entry releases the table lock before invoking arbitrary Python __hash__. + let pair_hash = hash::hash_tuple([Ok(key_hash), value.hash(vm)])?; + hasher.add(pair_hash); + } + let result = hasher.finish(); + zelf.hash.store(result, Ordering::Relaxed); + Ok(result) + } +} +impl Representable for PyFrozenDict { + fn repr(zelf: &Py, vm: &VirtualMachine) -> PyResult { + let name = zelf.class().name(); + if zelf.dict.is_empty() { + return Ok(vm.ctx.new_str(format!("{name}()"))); + } + let mut result = Wtf8Buf::from(format!("{name}(")); + if let Some(_guard) = ReprGuard::enter(vm, zelf.as_object()) { + result.push_str("{"); + for (index, (key, value)) in DictIter::new(&zelf.dict).enumerate() { + if index != 0 { + result.push_str(", "); + } + result.push_wtf8(key.repr(vm)?.as_wtf8()); + result.push_str(": "); + result.push_wtf8(value.repr(vm)?.as_wtf8()); + } + result.push_str("}"); + } else { + result.push_str("{...}"); + } + result.push_str(")"); + Ok(vm.ctx.new_str(result)) + } + fn repr_str(_zelf: &Py, _vm: &VirtualMachine) -> PyResult { + unreachable!("use repr instead") + } +} + // Implement IntoIterator so that we can easily iterate dictionaries from rust code. impl IntoIterator for PyDictRef { type Item = (PyObjectRef, PyObjectRef); @@ -1226,12 +1716,12 @@ impl ExactSizeIterator for DictIntoIter { } pub struct DictIter<'a> { - dict: &'a Py, + dict: &'a PyDict, position: usize, } impl<'a> DictIter<'a> { - pub const fn new(dict: &'a Py) -> Self { + pub const fn new(dict: &'a PyDict) -> Self { DictIter { dict, position: 0 } } } @@ -1261,7 +1751,7 @@ impl ExactSizeIterator for DictIter<'_> { trait DictView: PyPayload + PyClassDef + Iterable + Representable { type ReverseIter: PyPayload + core::fmt::Debug; - fn dict(&self) -> &Py; + fn dict(&self) -> &PyAnyDict; fn item(vm: &VirtualMachine, key: PyObjectRef, value: PyObjectRef) -> PyObjectRef; fn __len__(&self) -> usize { @@ -1287,14 +1777,14 @@ macro_rules! dict_view { $project_fn: expr, $result_fn: expr ) => { - #[pyclass(module = false, name = $class_name, unhashable = $unhashable)] + #[pyclass(module = false, name = $class_name, unhashable = $unhashable, traverse)] #[derive(Debug)] pub(crate) struct $name { - pub(crate) dict: PyDictRef, + pub(crate) dict: PyRef, } impl $name { - pub(crate) const fn new(dict: PyDictRef) -> Self { + pub(crate) fn new(dict: PyRef) -> Self { $name { dict } } } @@ -1302,7 +1792,7 @@ macro_rules! dict_view { impl DictView for $name { type ReverseIter = $reverse_iter_name; - fn dict(&self) -> &Py { + fn dict(&self) -> &PyAnyDict { &self.dict } @@ -1333,7 +1823,7 @@ macro_rules! dict_view { let s = if let Some(_guard) = ReprGuard::enter(vm, zelf.as_object()) { let mut result = Wtf8Buf::from(format!("{}([", Self::NAME)); let mut first = true; - for (key, value) in zelf.dict().clone() { + for (key, value) in zelf.dict() { if !first { result.push_str(", "); } @@ -1354,16 +1844,18 @@ macro_rules! dict_view { } } - #[pyclass(module = false, name = $iter_class_name)] + #[pyclass(module = false, name = $iter_class_name, traverse)] #[derive(Debug)] pub(crate) struct $iter_name { + #[pytraverse(skip)] pub(crate) size: dict_inner::DictSize, /// Whether the dict was found to have changed, which /// `dictiter_iternextkey()` records by writing a size no dict can /// have. Sticky: what it makes the iterator answer, it answers /// from then on. + #[pytraverse(skip)] changed: PyAtomic, - pub(crate) internal: PyMutex>, + pub(crate) internal: PyMutex>>, } impl PyPayload for $iter_name { @@ -1374,7 +1866,7 @@ macro_rules! dict_view { } impl $iter_name { - fn new(dict: PyDictRef) -> Self { + pub(crate) fn new(dict: PyRef) -> Self { $iter_name { size: dict.size(), changed: Radium::new(false), @@ -1411,7 +1903,7 @@ macro_rules! dict_view { let mut position = internal.position; let mut entries = Vec::new(); while let Some((next_position, key, value)) = - dict.entries.next_entry(position) + dict.as_dict().entries.next_entry(position) { entries.push(($result_fn)(vm, ($project_fn)(&key, &value))); position = next_position; @@ -1440,9 +1932,11 @@ macro_rules! dict_view { // raising. return (Err(mutated()), None); } - let entry = - dict.entries - .next_entry_checked(internal.position, &zelf.size, $project_fn); + let entry = dict.as_dict().entries.next_entry_checked( + internal.position, + &zelf.size, + $project_fn, + ); match entry { Err(dict_inner::DictChanged) => { zelf.changed.store(true, Ordering::Relaxed); @@ -1458,14 +1952,17 @@ macro_rules! dict_view { } } - #[pyclass(module = false, name = $reverse_iter_class_name)] + #[pyclass(module = false, name = $reverse_iter_class_name, traverse)] #[derive(Debug)] pub(crate) struct $reverse_iter_name { + #[pytraverse(skip)] pub(crate) size: dict_inner::DictSize, /// As in `$iter_name`. + #[pytraverse(skip)] changed: PyAtomic, + #[pytraverse(skip)] remaining: PyAtomic, - internal: PyMutex>, + internal: PyMutex>>, } impl PyPayload for $reverse_iter_name { @@ -1476,7 +1973,7 @@ macro_rules! dict_view { } impl $reverse_iter_name { - fn new(dict: PyDictRef) -> Self { + pub(crate) fn new(dict: PyRef) -> Self { let size = dict.size(); // usize::MAX represents the position before the first entry. let position = size.entries_size.wrapping_sub(1); @@ -1500,7 +1997,7 @@ macro_rules! dict_view { let mut position = internal.position; let mut entries = Vec::new(); while let Some((found_index, key, value)) = - dict.entries.prev_entry(position) + dict.as_dict().entries.prev_entry(position) { entries.push(($result_fn)(vm, ($project_fn)(&key, &value))); if found_index == 0 { @@ -1549,9 +2046,11 @@ macro_rules! dict_view { // raising. return (Err(mutated()), None); } - let entry = - dict.entries - .prev_entry_checked(internal.position, &zelf.size, $project_fn); + let entry = dict.as_dict().entries.prev_entry_checked( + internal.position, + &zelf.size, + $project_fn, + ); match entry { Err(dict_inner::DictChanged) => { zelf.changed.store(true, Ordering::Relaxed); @@ -1637,14 +2136,6 @@ trait ViewSetOps: DictView { op: PyComparisonOp, vm: &VirtualMachine, ) -> PyResult { - if let Some(dictview) = other.downcast_ref::() { - return zelf.dict().inner_cmp( - dictview.dict(), - op, - !zelf.class().is(vm.ctx.types.dict_keys_type), - vm, - ); - } if !is_set_or_dict_view(other) { return Ok(PyComparisonValue::NotImplemented); } @@ -1718,7 +2209,7 @@ impl PyDictKeys { #[pygetset] fn mapping(zelf: PyRef) -> PyMappingProxy { - PyMappingProxy::from(zelf.dict().to_owned()) + PyMappingProxy::from_any_dict(zelf.dict().to_owned()) } } @@ -1738,11 +2229,10 @@ impl AsSequence for PyDictKeys { static AS_SEQUENCE: LazyLock = LazyLock::new(|| PySequenceMethods { length: atomic_func!(|seq, _vm| Ok(PyDictKeys::sequence_downcast(seq).__len__())), contains: atomic_func!(|seq, target, vm| { - let hash = PyDict::hash_or_unhashable(target, vm)?; - PyDictKeys::sequence_downcast(seq) + Ok(PyDictKeys::sequence_downcast(seq) .dict - .entries - .contains(vm, target, hash) + .inner_getitem_opt(target, vm)? + .is_some()) }), ..PySequenceMethods::NOT_IMPLEMENTED }); @@ -1782,7 +2272,7 @@ impl PyDictItems { } #[pygetset] fn mapping(zelf: PyRef) -> PyMappingProxy { - PyMappingProxy::from(zelf.dict().to_owned()) + PyMappingProxy::from_any_dict(zelf.dict().to_owned()) } } @@ -1844,7 +2334,7 @@ impl AsNumber for PyDictItems { impl PyDictValues { #[pygetset] fn mapping(zelf: PyRef) -> PyMappingProxy { - PyMappingProxy::from(zelf.dict().to_owned()) + PyMappingProxy::from_any_dict(zelf.dict().to_owned()) } } @@ -1944,6 +2434,107 @@ pub(crate) fn set_inner_number_or(a: &PyObject, b: &PyObject, vm: &VirtualMachin set_inner_number_op(a, b, |a, b| a.union(b, vm), vm) } +/// `_PyDict_Or` +fn dict_or(a: &PyObject, b: &PyObject, vm: &VirtualMachine) -> PyResult { + let (Some(left), Some(_)) = (PyAnyDict::from_object(a), PyAnyDict::from_object(b)) else { + return Ok(vm.ctx.not_implemented()); + }; + // anydict_copy_untracked converts any frozendict to an exact frozendict. + let frozen = left.is_frozen(); + let mapping_name = if frozen { "frozendict" } else { "dict" }; + let dict = PyDict::default(); + if !left.is_empty() { + dict.update_arg(a.to_owned(), true, mapping_name, vm)?; + } + dict.update_arg(b.to_owned(), true, mapping_name, vm)?; + Ok(if frozen { + PyFrozenDict::from_dict(dict).into_pyobject(vm) + } else { + dict.into_pyobject(vm) + }) +} + +fn frozendict_or(a: &PyObject, b: &PyObject, vm: &VirtualMachine) -> PyResult { + if a.class().is(vm.ctx.types.frozendict_type) + && let Some(left) = PyAnyDict::from_object(a) + && let Some(right) = PyAnyDict::from_object(b) + { + // frozendict() | frozendict(...) => frozendict(...) + if left.is_empty() && b.class().is(vm.ctx.types.frozendict_type) { + return Ok(b.to_owned()); + } + // frozendict(...) | frozendict() => frozendict(...) + if right.is_empty() + && (b.class().is(vm.ctx.types.dict_type) || b.class().is(vm.ctx.types.frozendict_type)) + { + return Ok(a.to_owned()); + } + } + dict_or(a, b, vm) +} + +pub(crate) fn dict_fromkeys( + class: PyTypeRef, + iterable: PyObjectRef, + value: PyObjectRef, + vm: &VirtualMachine, +) -> PyResult { + let exact_frozen = class.is(vm.ctx.types.frozendict_type); + let created = if exact_frozen { + None + } else { + Some(PyType::call(&class, ().into(), vm)?) + }; + let frozen_source = created + .as_ref() + .and_then(|obj| obj.downcast_ref::()); + if exact_frozen || frozen_source.is_some() { + let dict = PyDict::default(); + if let Some(source) = frozen_source { + dict.update_arg(source.to_owned().into(), true, "frozendict", vm)?; + } + let iterable = ArgIterable::::try_from_object(vm, iterable)?; + if let Some(keys) = PyDict::fromkeys_known_hashes(iterable.as_object(), vm) { + for (key, hash) in keys { + dict.entries.insert(vm, &*key, hash, value.clone())?; + } + } else { + for key in iterable.iter(vm)? { + let key = key?; + let hash = PyFrozenDict::key_hash(&*key, vm)?; + dict.entries.insert(vm, &*key, hash, value.clone())?; + } + } + let frozen = PyFrozenDict::from_dict(dict).into_ref(&vm.ctx); + return if exact_frozen { + Ok(frozen.into()) + } else { + class.as_object().call((frozen,), vm) + }; + } + let iterable = ArgIterable::::try_from_object(vm, iterable)?; + let created = created.unwrap(); + if let Some(dict) = created.downcast_ref_if_exact::(vm) { + if let Some(keys) = PyDict::fromkeys_known_hashes(iterable.as_object(), vm) { + if class.is(vm.ctx.types.dict_type) { + dict.entries.reserve_for_empty(keys.len()); + } + for (key, hash) in keys { + dict.entries.insert(vm, &*key, hash, value.clone())?; + } + } else { + for key in iterable.iter(vm)? { + dict.inner_setitem(&*key?, value.clone(), vm)?; + } + } + } else { + for key in iterable.iter(vm)? { + created.set_item(&*key?, value.clone(), vm)?; + } + } + Ok(created) +} + fn vectorcall_dict( zelf_obj: &PyObject, args: Vec, @@ -1958,6 +2549,28 @@ fn vectorcall_dict( Ok(obj.into()) } +fn vectorcall_frozendict( + zelf: &PyObject, + args: Vec, + nargs: usize, + kwnames: Option<&[PyObjectRef]>, + vm: &VirtualMachine, +) -> PyResult { + // frozendict(frozendict) returns the same object unmodified + if nargs == 1 + && kwnames.is_none_or(<[_]>::is_empty) + && args[0].class().is(vm.ctx.types.frozendict_type) + && zelf.is(vm.ctx.types.frozendict_type) + { + return Ok(args[0].clone()); + } + PyFrozenDict::slot_new( + zelf.downcast_ref::().unwrap().to_owned(), + FuncArgs::from_vectorcall_owned(args, nargs, kwnames), + vm, + ) +} + pub(crate) fn init(context: &'static Context) { PyDict::extend_class(context, context.types.dict_type); context @@ -1975,4 +2588,86 @@ pub(crate) fn init(context: &'static Context) { PyDictItems::extend_class(context, context.types.dict_items_type); PyDictItemIterator::extend_class(context, context.types.dict_itemiterator_type); PyDictReverseItemIterator::extend_class(context, context.types.dict_reverseitemiterator_type); + PyFrozenDict::extend_class(context, context.types.frozendict_type); + context + .types + .frozendict_type + .slots + .vectorcall + .store(Some(vectorcall_frozendict)); +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn anydict_reference_layout_and_ownership() { + static_assertions::assert_not_impl_any!(PyAnyDict: PyPayload); + static_assertions::assert_type_eq_all!( + ::Target, + PyAnyDict + ); + static_assertions::assert_eq_size!(PyRef, PyObjectRef, Option>); + static_assertions::assert_eq_align!(PyRef, PyObjectRef); + static_assertions::assert_eq_size!(PyAnyDict, PyObject); + static_assertions::assert_eq_align!(PyAnyDict, PyObject); + #[cfg(feature = "threading")] + static_assertions::assert_impl_all!(PyRef: Send, Sync); + #[cfg(not(feature = "threading"))] + static_assertions::assert_not_impl_any!(PyRef: Send, Sync); + + let ctx = Context::genesis(); + let dict: PyRef = PyDict::default().into_ref(ctx).into(); + let frozen: PyRef = PyFrozenDict::from_dict(PyDict::default()) + .into_ref(ctx) + .into(); + for owner in [dict, frozen] { + let object = owner.as_object() as *const PyObject; + assert_eq!(owner.as_object().strong_count(), 1); + let borrowed = PyAnyDict::from_object(owner.as_object()).unwrap(); + assert!(core::ptr::eq(borrowed.as_object(), object)); + assert_eq!(owner.as_object().strong_count(), 1); + assert_eq!(borrowed.as_mutable().is_none(), borrowed.is_frozen()); + let copy = borrowed.to_owned(); + assert_eq!(owner.as_object().strong_count(), 2); + let clone = owner.clone(); + assert_eq!(owner.as_object().strong_count(), 3); + let erased = clone.into_object(); + assert_eq!(erased.strong_count(), 3); + assert!(core::ptr::eq(&*erased, object)); + let mut visited = 0; + owner.traverse(&mut |edge| { + assert!(core::ptr::eq(edge, object)); + visited += 1; + }); + assert_eq!(visited, 1); + drop(copy); + drop(owner); + assert_eq!(erased.strong_count(), 1); + } + } + + #[test] + fn anydict_checked_reference_conversion() { + crate::Interpreter::without_stdlib(Default::default()).enter(|vm| { + let dict = PyDict::default().into_pyobject(vm); + let frozen = PyFrozenDict::from_dict(PyDict::default()).into_pyobject(vm); + for object in [dict, frozen] { + let owner = PyRef::::try_from_object(vm, object.clone()).unwrap(); + assert_eq!(object.strong_count(), 2); + assert!(core::ptr::eq(owner.as_object(), &*object)); + let erased: PyObjectRef = owner.into(); + assert_eq!(object.strong_count(), 2); + assert!(core::ptr::eq(&*erased, &*object)); + drop(erased); + assert_eq!(object.strong_count(), 1); + } + let unrelated: PyObjectRef = vm.ctx.new_int(1000).into(); + assert!(PyAnyDict::from_object(&unrelated).is_none()); + let error = PyRef::::try_from_object(vm, unrelated.clone()).unwrap_err(); + assert!(error.fast_isinstance(vm.ctx.exceptions.type_error)); + assert_eq!(unrelated.strong_count(), 1); + }); + } } diff --git a/crates/vm/src/builtins/frame.rs b/crates/vm/src/builtins/frame.rs index 905cbb582a4..be8ddd9f391 100644 --- a/crates/vm/src/builtins/frame.rs +++ b/crates/vm/src/builtins/frame.rs @@ -2,7 +2,7 @@ */ -use super::{PyAsyncGen, PyCode, PyCoroutine, PyDictRef, PyGenerator, PyIntRef}; +use super::{PyAnyDict, PyAsyncGen, PyCode, PyCoroutine, PyGenerator, PyIntRef}; use crate::{ AsObject, Context, Py, PyObjectRef, PyPayload, PyRef, PyResult, VirtualMachine, class::PyClassImpl, @@ -668,7 +668,7 @@ impl Py { } #[pygetset] - pub fn f_globals(&self) -> PyDictRef { + pub fn f_globals(&self) -> PyRef { self.iframe().globals().to_owned() } diff --git a/crates/vm/src/builtins/function.rs b/crates/vm/src/builtins/function.rs index be4994a6204..40e51c6b31b 100644 --- a/crates/vm/src/builtins/function.rs +++ b/crates/vm/src/builtins/function.rs @@ -2,8 +2,8 @@ mod jit; use super::{ - PyAsyncGen, PyCode, PyCoroutine, PyDictRef, PyGenerator, PyList, PyModule, PyStr, PyStrRef, - PyTuple, PyTupleRef, PyType, object, + PyAnyDict, PyAsyncGen, PyCode, PyCoroutine, PyDictRef, PyGenerator, PyList, PyModule, PyStr, + PyStrRef, PyTuple, PyTupleRef, PyType, object, }; use crate::common::hash::PyHash; use crate::common::lock::PyMutex; @@ -66,7 +66,7 @@ fn format_missing_args( pub struct PyFunction { pub(crate) code: PyAtomicRef, #[pymember(name = "__globals__")] - pub(crate) globals: PyDictRef, + pub(crate) globals: PyRef, #[pymember(name = "__builtins__")] pub(crate) builtins: PyObjectRef, #[pymember(name = "__closure__")] @@ -179,15 +179,17 @@ impl PyFunction { #[inline] pub(crate) fn new( code: PyRef, - globals: PyDictRef, + globals: PyRef, vm: &VirtualMachine, ) -> PyResult { let name = PyMutex::new(code.obj_name.to_owned()); - let module = vm.unwrap_or_none(globals.get_item_opt(identifier!(vm, __name__), vm)?); - let builtins = globals.get_item("__builtins__", vm).unwrap_or_else(|_| { - // If not in globals, inherit from current execution context - crate::frame::current_builtins().unwrap_or_else(|| vm.builtins.dict().into()) - }); + let module = vm.unwrap_or_none(globals.inner_getitem_opt(identifier!(vm, __name__), vm)?); + let builtins = globals + .inner_getitem_opt(identifier!(vm, __builtins__), vm)? + .unwrap_or_else(|| { + // If not in globals, inherit from current execution context. + crate::frame::current_builtins().unwrap_or_else(|| vm.builtins.dict().into()) + }); // If builtins is a module, use its __dict__ instead let builtins = if let Some(module) = builtins.downcast_ref::() { module.dict().into() @@ -621,7 +623,7 @@ impl Py { } else if let Some(locals) = locals { Some(locals) } else { - Some(ArgMapping::from_dict_exact(self.globals.clone())) + Some(ArgMapping::from_anydict_exact(self.globals.clone())) }; let use_datastack = !is_gen && !is_coro && !is_async_gen; let frame = FrameObject::new_ref( @@ -654,7 +656,7 @@ impl Py { } else if let Some(locals) = locals { crate::frame::FrameLocals::with_locals(locals) } else { - crate::frame::FrameLocals::with_locals(crate::function::ArgMapping::from_dict_exact( + crate::frame::FrameLocals::with_locals(crate::function::ArgMapping::from_anydict_exact( self.globals.clone(), )) }; @@ -805,7 +807,7 @@ impl Py { let locals = if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) { None } else { - Some(ArgMapping::from_dict_exact(self.globals.clone())) + Some(ArgMapping::from_anydict_exact(self.globals.clone())) }; let frame = FrameObject::new_ref( @@ -862,7 +864,7 @@ impl Py { let locals = if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) { None } else { - Some(ArgMapping::from_dict_exact(self.globals.clone())) + Some(ArgMapping::from_anydict_exact(self.globals.clone())) }; // Heap-backed: the frame outlives the call that made it. @@ -897,7 +899,7 @@ impl Py { let locals = if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) { crate::frame::FrameLocals::lazy() } else { - crate::frame::FrameLocals::with_locals(ArgMapping::from_dict_exact( + crate::frame::FrameLocals::with_locals(ArgMapping::from_anydict_exact( self.globals.clone(), )) }; @@ -1318,7 +1320,7 @@ impl Constructor for PyFunction { None }; - let mut func = Self::new(args.code.clone(), args.globals.clone(), vm)?; + let mut func = Self::new(args.code.clone(), args.globals.clone().into(), vm)?; // Set function name if provided if let Some(name) = args.name.into_option() { *func.name.lock() = name; diff --git a/crates/vm/src/builtins/mappingproxy.rs b/crates/vm/src/builtins/mappingproxy.rs index 3e53f1729a5..e1e0283d189 100644 --- a/crates/vm/src/builtins/mappingproxy.rs +++ b/crates/vm/src/builtins/mappingproxy.rs @@ -68,6 +68,12 @@ impl Constructor for PyMappingProxy { } impl PyMappingProxy { + pub(crate) fn from_any_dict(dict: PyRef) -> Self { + Self { + mapping: MappingProxyInner::Mapping(ArgMapping::new(dict.into())), + } + } + pub fn from_object(mapping: PyObjectRef, vm: &VirtualMachine) -> PyResult { if mapping.mapping_unchecked().check() && !mapping.downcastable::() diff --git a/crates/vm/src/builtins/mod.rs b/crates/vm/src/builtins/mod.rs index a938b9b3dd1..438cd9e5092 100644 --- a/crates/vm/src/builtins/mod.rs +++ b/crates/vm/src/builtins/mod.rs @@ -20,7 +20,7 @@ pub use complex::PyComplex; pub(crate) mod coroutine; pub use coroutine::PyCoroutine; pub(crate) mod dict; -pub use dict::{PyDict, PyDictRef}; +pub use dict::{PyAnyDict, PyDict, PyDictRef, PyFrozenDict}; pub(crate) mod enumerate; pub use enumerate::PyEnumerate; pub(crate) mod filter; diff --git a/crates/vm/src/builtins/type.rs b/crates/vm/src/builtins/type.rs index a2449f136ef..153842056c1 100644 --- a/crates/vm/src/builtins/type.rs +++ b/crates/vm/src/builtins/type.rs @@ -1,6 +1,6 @@ use super::{ - PyClassMethod, PyDict, PyDictRef, PyList, PyStaticMethod, PyStr, PyStrInterned, PyStrRef, - PyTupleRef, PyUtf8StrRef, PyWeak, mappingproxy::PyMappingProxy, object, union_, + PyAnyDict, PyClassMethod, PyDict, PyDictRef, PyList, PyStaticMethod, PyStr, PyStrInterned, + PyStrRef, PyTupleRef, PyUtf8StrRef, PyWeak, mappingproxy::PyMappingProxy, object, union_, }; use crate::{ AsObject, Context, Py, PyAtomicRef, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, @@ -1798,25 +1798,15 @@ impl Constructor for PyType { })); } - let (name, bases, dict, kwargs): (PyStrRef, PyTupleRef, PyDictRef, KwArgs) = + let (name, bases, dict, kwargs): (PyStrRef, PyTupleRef, PyObjectRef, KwArgs) = args.clone().bind_for(vm, Self::NAME)?; - - // A mapping that is not an exact dict (e.g. OrderedDict) keeps its - // own iteration order; copy via the mapping protocol so that order - // lands in the type dict. - let dict = if args.args[2].class().is(vm.ctx.types.dict_type) { - dict - } else { - let copied = vm.ctx.new_dict(); - copied.merge_object(args.args[2].clone(), vm)?; - copied + let Some(dict) = PyAnyDict::from_object(&dict) else { + return Err(vm.new_type_error(format!( + "type.__new__() argument 3 must be dict or frozendict, not {}", + dict.class().fully_qualified_name(vm)? + ))); }; - if name.as_bytes().contains(&0) { - return Err(vm.new_value_error("type name must not contain null characters")); - } - let name = name.try_into_utf8(vm)?; - let (metatype, base, bases, base_is_type) = if bases.as_slice().is_empty() { let base = vm.ctx.types.object_type.to_owned(); let bases = PyTuple::new_ref_typed(vec![base.clone()], &vm.ctx); @@ -1856,6 +1846,25 @@ impl Constructor for PyType { (metatype, base.to_owned(), bases, base_is_type) }; + // Copy only after resolving the metaclass: its __new__ receives the + // original namespace. A subclass with a custom iterator (e.g. + // OrderedDict) keeps its own mapping order in the mutable type dict. + let dict = if let Some(dict) = dict.as_object().downcast_ref_if_exact::(vm) { + dict.to_owned() + } else { + let copied = vm.ctx.new_dict(); + // Like dict.copy(), an empty table bypasses subclass hooks. + if !dict.is_empty() { + copied.merge_object(dict.to_owned().into(), vm)?; + } + copied + }; + + if name.as_bytes().contains(&0) { + return Err(vm.new_value_error("type name must not contain null characters")); + } + let name = name.try_into_utf8(vm)?; + let qualname = dict .get_item_opt(identifier!(vm, __qualname__), vm)? .map(|obj| downcast_qualname(obj, vm)) diff --git a/crates/vm/src/frame.rs b/crates/vm/src/frame.rs index 2c9d6075106..a043838d49b 100644 --- a/crates/vm/src/frame.rs +++ b/crates/vm/src/frame.rs @@ -6,10 +6,10 @@ use crate::{ AsObject, Py, PyExact, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, PyStackRef, TryFromObject, VirtualMachine, builtins::{ - PyBaseException, PyBaseExceptionRef, PyBaseObject, PyBytes, PyCode, PyCoroutine, PyDict, - PyDictRef, PyFloat, PyFrozenSet, PyGenerator, PyInt, PyInterpolation, PyList, PyModule, - PyProperty, PySet, PySlice, PyStr, PyStrInterned, PyTemplate, PyTraceback, PyType, - PyUtf8Str, + PyAnyDict, PyBaseException, PyBaseExceptionRef, PyBaseObject, PyBytes, PyCode, PyCoroutine, + PyDict, PyDictRef, PyFloat, PyFrozenSet, PyGenerator, PyInt, PyInterpolation, PyList, + PyModule, PyProperty, PySet, PySlice, PyStr, PyStrInterned, PyTemplate, PyTraceback, + PyType, PyUtf8Str, builtin_func::PyNativeFunction, descriptor::{PyMemberDescriptor, PyMethodDescriptor}, frame::stack_analysis, @@ -108,7 +108,7 @@ pub fn current_thread_frame_materialize(vm: &VirtualMachine) -> Option Option { +pub fn current_globals() -> Option> { let ptr = crate::vm::thread::get_current_frame(); if ptr.is_null() { return None; @@ -923,7 +923,7 @@ pub struct InterpreterFrame { // Borrowed pointers — owned by FrameObject or by PyFunction on caller's stack. pub(crate) code: *const Py, pub(crate) func_obj: *const PyObject, // nullable - pub(crate) globals: *const Py, + pub(crate) globals: *const PyAnyDict, pub(crate) builtins: *const PyObject, /// Unified storage for local variables and evaluation stack. @@ -983,7 +983,7 @@ impl InterpreterFrame { #[inline(always)] pub(crate) unsafe fn new( code: &Py, - globals: &Py, + globals: &PyAnyDict, builtins: &PyObject, func_obj: Option<&PyObject>, localsplus: LocalsPlus, @@ -1027,7 +1027,7 @@ impl InterpreterFrame { Some(obj) => obj as *const PyObject, None => core::ptr::null(), }, - globals: globals as *const Py, + globals: globals as *const PyAnyDict, builtins: builtins as *const PyObject, localsplus, locals, @@ -1060,7 +1060,7 @@ impl InterpreterFrame { #[inline(always)] pub(crate) unsafe fn new_on_datastack<'a>( code: &Py, - globals: &Py, + globals: &PyAnyDict, builtins: &PyObject, func_obj: Option<&PyObject>, locals: FrameLocals, @@ -1266,7 +1266,7 @@ impl InterpreterFrame { // Create a full FrameObject with its own InterpreterFrame copy. // The FrameObject owns references to the same objects (code, globals, etc). let code: PyRef = self.code().to_owned(); - let globals: PyDictRef = self.globals().to_owned(); + let globals = self.globals().to_owned(); let builtins: PyObjectRef = self.builtins().to_owned(); let func_obj: Option = self.func_obj().map(|o| o.to_owned()); @@ -1362,7 +1362,7 @@ impl InterpreterFrame { #[cold] fn materialize_slow_chain(&self, vm: &VirtualMachine) -> FrameObjectRef { let code: PyRef = self.code().to_owned(); - let globals: PyDictRef = self.globals().to_owned(); + let globals = self.globals().to_owned(); let builtins: PyObjectRef = self.builtins().to_owned(); let func_obj: Option = self.func_obj().map(|o| o.to_owned()); @@ -1419,7 +1419,7 @@ impl InterpreterFrame { /// Borrowed globals dict. #[inline(always)] - pub fn globals(&self) -> &Py { + pub fn globals(&self) -> &PyAnyDict { // SAFETY: established by `new` / `init_iframe_ptrs`. unsafe { &*self.globals } } @@ -1534,7 +1534,7 @@ pub struct FrameObject { // raw pointers. Wrapped in Option so Traverse::clear can release them, // allowing GC cycle collection to reclaim referenced objects. pub(crate) owned_code: Option>, - pub(crate) owned_globals: Option, + pub(crate) owned_globals: Option>, pub(crate) owned_builtins: Option, pub(crate) owned_func_obj: Option, @@ -1817,7 +1817,7 @@ impl FrameObject { let locals = match scope.locals { Some(locals) => FrameLocals::with_locals(locals), None if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) => FrameLocals::lazy(), - None => FrameLocals::with_locals(ArgMapping::from_dict_exact(scope.globals.clone())), + None => FrameLocals::with_locals(ArgMapping::from_anydict_exact(scope.globals.clone())), }; // Pointers are initially set from owned fields' references but will be @@ -1853,7 +1853,7 @@ impl FrameObject { fn init_iframe_ptrs(self_: &Py) { let iframe = unsafe { self_.iframe_mut() }; iframe.code = &**self_.owned_code.as_ref().unwrap() as *const Py; - iframe.globals = &**self_.owned_globals.as_ref().unwrap() as *const Py; + iframe.globals = &**self_.owned_globals.as_ref().unwrap() as *const PyAnyDict; iframe.builtins = &**self_.owned_builtins.as_ref().unwrap() as *const PyObject; iframe.func_obj = match &self_.owned_func_obj { Some(obj) => &**obj as *const PyObject, @@ -2497,7 +2497,7 @@ impl Py { }, ) }; - let builtins_dict = if globals.class().is(vm.ctx.types.dict_type) { + let builtins_dict = if globals.as_object().class().is(vm.ctx.types.dict_type) { builtins .downcast_ref_if_exact::(vm) // SAFETY: downcast_ref_if_exact already verified exact type @@ -2739,7 +2739,7 @@ fn exec_iframe<'a>( }, ) }; - let builtins_dict = if globals.class().is(vm.ctx.types.dict_type) { + let builtins_dict = if globals.as_object().class().is(vm.ctx.types.dict_type) { builtins .downcast_ref_if_exact::(vm) .map(|d| unsafe { PyExact::ref_unchecked(d) }) @@ -2931,7 +2931,7 @@ pub(crate) struct ExecutingFrame<'a> { code: &'a Py, localsplus: &'a mut LocalsPlus, locals: &'a FrameLocals, - globals: &'a Py, + globals: &'a PyAnyDict, builtins: &'a PyObject, /// Cached downcast of builtins to PyDict for fast LOAD_GLOBAL. /// Only set when both globals and builtins are exact dict types (not @@ -4651,15 +4651,10 @@ impl ExecutingFrame<'_> { } Instruction::DeleteName { namei: idx } => { let name = self.code.names[idx.get(arg) as usize]; - let res = self.locals.mapping(vm).ass_subscript(name, None, vm); - - match res { - Ok(()) => {} - Err(e) if e.fast_isinstance(vm.ctx.exceptions.key_error) => { - return Err(name_error(name, vm)); - } - Err(e) => return Err(e), - } + self.locals + .mapping(vm) + .ass_subscript(name, None, vm) + .map_err(|_| name_error(name, vm))?; Ok(None) } Instruction::DeleteSubscr => self.execute_delete_subscript(vm), @@ -5254,7 +5249,7 @@ impl ExecutingFrame<'_> { match result { Ok(x) => self.push_value(x), Err(e) if e.fast_isinstance(vm.ctx.exceptions.key_error) => { - self.push_value(self.load_global_or_builtin(name, vm)?); + self.push_value(self.load_name_global_or_builtin(name, vm)?); } Err(e) => return Err(e), } @@ -8043,9 +8038,9 @@ impl ExecutingFrame<'_> { let cached_version = self.code.instructions.read_cache_u16(cache_base + 1); let cached_index = self.code.instructions.read_cache_u16(cache_base + 3); if cached_version != 0 - && let Some(x) = self - .globals - .get_item_by_index_and_keys_version(cached_version, cached_index) + && let Some(globals) = self.globals.as_mutable() + && let Some(x) = + globals.get_item_by_index_and_keys_version(cached_version, cached_index) { self.push_value(x); if (oparg & 1) != 0 { @@ -8069,7 +8064,8 @@ impl ExecutingFrame<'_> { let cached_index = self.code.instructions.read_cache_u16(cache_base + 3); if cached_globals_ver != 0 && cached_builtins_ver != 0 - && let Ok(current_globals_ver) = u16::try_from(self.globals.keys_version()) + && let Some(globals) = self.globals.as_mutable() + && let Ok(current_globals_ver) = u16::try_from(globals.keys_version()) && cached_globals_ver == current_globals_ver && let Some(builtins_dict) = self.builtins.downcast_ref_if_exact::(vm) && let Some(x) = builtins_dict @@ -8541,7 +8537,12 @@ impl ExecutingFrame<'_> { if let Some(builtins_dict) = self.builtins_dict { // Fast path: both globals and builtins are exact dicts // SAFETY: builtins_dict is only set when globals is also exact dict - let globals_exact = unsafe { PyExact::ref_unchecked(self.globals) }; + let globals = self + .globals + .as_object() + .downcast_ref::() + .expect("exact dict globals"); + let globals_exact = unsafe { PyExact::ref_unchecked(globals) }; globals_exact .get_chain_exact(builtins_dict, name, vm)? .ok_or_else(|| { @@ -8552,16 +8553,29 @@ impl ExecutingFrame<'_> { if let Some(value) = self.globals.get_item_opt(name, vm)? { return Ok(value); } - self.builtins.get_item(name, vm).map_err(|e| { - if e.fast_isinstance(vm.ctx.exceptions.key_error) { - vm.new_name_error(format!("name '{name}' is not defined"), name.to_owned()) - } else { - e - } - }) + self.load_builtin(name, vm) } } + fn load_name_global_or_builtin(&self, name: &Py, vm: &VirtualMachine) -> PyResult { + // LOAD_NAME reads the stored global entries even for dict subclasses; + // LOAD_GLOBAL instead honors their __getitem__ implementation. + if let Some(value) = self.globals.inner_getitem_opt(name, vm)? { + return Ok(value); + } + self.load_builtin(name, vm) + } + + fn load_builtin(&self, name: &Py, vm: &VirtualMachine) -> PyResult { + self.builtins.get_item(name, vm).map_err(|e| { + if e.fast_isinstance(vm.ctx.exceptions.key_error) { + vm.new_name_error(format!("name '{name}' is not defined"), name.to_owned()) + } else { + e + } + }) + } + #[cfg_attr(feature = "flame-it", flame("FrameObject"))] fn import(&mut self, vm: &VirtualMachine, module_name: Option<&Py>) -> PyResult<()> { let module_name = module_name.unwrap_or(vm.ctx.empty_str); @@ -11786,7 +11800,7 @@ impl ExecutingFrame<'_> { let locals = if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) { FrameLocals::lazy() } else { - FrameLocals::with_locals(crate::function::ArgMapping::from_dict_exact( + FrameLocals::with_locals(crate::function::ArgMapping::from_anydict_exact( func.globals.clone(), )) }; @@ -11852,7 +11866,7 @@ impl ExecutingFrame<'_> { let locals = if code.flags.contains(bytecode::CodeFlags::NEWLOCALS) { FrameLocals::lazy() } else { - FrameLocals::with_locals(crate::function::ArgMapping::from_dict_exact( + FrameLocals::with_locals(crate::function::ArgMapping::from_anydict_exact( func.globals.clone(), )) }; @@ -11938,8 +11952,11 @@ impl ExecutingFrame<'_> { ) { return; } + let Some(globals) = self.globals.as_mutable() else { + return; + }; let name = self.code.names[(oparg >> 1) as usize]; - let Ok(globals_version @ 1..) = u16::try_from(self.globals.assign_keys_version(vm)) else { + let Ok(globals_version @ 1..) = u16::try_from(globals.assign_keys_version(vm)) else { unsafe { self.code.instructions.write_adaptive_counter( cache_base, @@ -11951,7 +11968,7 @@ impl ExecutingFrame<'_> { return; }; - if let Ok(Some(globals_hint)) = self.globals.hint_for_key(name, vm) { + if let Ok(Some(globals_hint)) = globals.hint_for_key(name, vm) { unsafe { self.code .instructions diff --git a/crates/vm/src/function/protocol.rs b/crates/vm/src/function/protocol.rs index 5883a2a4691..c9d8df5b186 100644 --- a/crates/vm/src/function/protocol.rs +++ b/crates/vm/src/function/protocol.rs @@ -1,7 +1,7 @@ use super::IntoFuncArgs; use crate::{ - AsObject, PyObject, PyObjectRef, PyPayload, PyResult, TryFromObject, VirtualMachine, - builtins::{PyDictRef, iter::PySequenceIterator}, + AsObject, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, TryFromObject, VirtualMachine, + builtins::{PyAnyDict, PyDictRef, iter::PySequenceIterator}, convert::ToPyObject, object::{Traverse, TraverseFn}, protocol::{PyIter, PyIterIter, PyMapping}, @@ -155,6 +155,12 @@ impl ArgMapping { Self { obj: dict.into() } } + #[inline(always)] + #[must_use] + pub fn from_anydict_exact(dict: PyRef) -> Self { + Self { obj: dict.into() } + } + #[inline(always)] #[must_use] pub fn obj(&self) -> &PyObject { diff --git a/crates/vm/src/object/core.rs b/crates/vm/src/object/core.rs index 8dacc13d303..fa2e7ae648a 100644 --- a/crates/vm/src/object/core.rs +++ b/crates/vm/src/object/core.rs @@ -2666,6 +2666,7 @@ mod ref_target_sealed { impl Sealed for T {} impl Sealed for super::PyObject {} impl Sealed for crate::builtins::set::PyAnySet {} + impl Sealed for crate::builtins::PyAnyDict {} impl Sealed for super::PyTuple> {} } @@ -2709,6 +2710,11 @@ pub trait PyTypedRefTarget: PyRefTarget {} impl>> PyTypedRefTarget for T {} impl PyTypedRefTarget for crate::builtins::set::PyAnySet {} +impl PyTypedRefTarget for crate::builtins::PyAnyDict {} + +impl PyRefTarget for crate::builtins::PyAnyDict { + type Target = Self; +} impl PyRefTarget for crate::builtins::set::PyAnySet { type Target = Self; diff --git a/crates/vm/src/protocol/callable.rs b/crates/vm/src/protocol/callable.rs index 49172a15222..67a5354b131 100644 --- a/crates/vm/src/protocol/callable.rs +++ b/crates/vm/src/protocol/callable.rs @@ -78,6 +78,11 @@ impl<'a> PyCallable<'a> { pub fn invoke(&self, args: impl IntoFuncArgs, vm: &VirtualMachine) -> PyResult { let args = args.into_args(vm); + // _PyVectorcall_Call + if self.vectorcall.is_some() && args.kwargs.is_empty() { + let nargs = args.args.len(); + return self.invoke_vectorcall(args.args, nargs, None, vm); + } if !vm.use_tracing.get() { return (self.call)(self.obj, args, vm); } diff --git a/crates/vm/src/protocol/mapping.rs b/crates/vm/src/protocol/mapping.rs index e791757901e..997df878ff0 100644 --- a/crates/vm/src/protocol/mapping.rs +++ b/crates/vm/src/protocol/mapping.rs @@ -184,7 +184,7 @@ impl PyMapping<'_> { pub fn keys(self, vm: &VirtualMachine) -> PyResult { if let Some(dict) = self.obj.downcast_ref_if_exact::(vm) { - PyDictKeys::new(dict.to_owned()).to_pyresult(vm) + PyDictKeys::new(dict.to_owned().into()).to_pyresult(vm) } else { self.method_output_as_list(identifier!(vm, keys), vm) } @@ -192,7 +192,7 @@ impl PyMapping<'_> { pub fn values(self, vm: &VirtualMachine) -> PyResult { if let Some(dict) = self.obj.downcast_ref_if_exact::(vm) { - PyDictValues::new(dict.to_owned()).to_pyresult(vm) + PyDictValues::new(dict.to_owned().into()).to_pyresult(vm) } else { self.method_output_as_list(identifier!(vm, values), vm) } @@ -200,7 +200,7 @@ impl PyMapping<'_> { pub fn items(self, vm: &VirtualMachine) -> PyResult { if let Some(dict) = self.obj.downcast_ref_if_exact::(vm) { - PyDictItems::new(dict.to_owned()).to_pyresult(vm) + PyDictItems::new(dict.to_owned().into()).to_pyresult(vm) } else { self.method_output_as_list(identifier!(vm, items), vm) } diff --git a/crates/vm/src/scope.rs b/crates/vm/src/scope.rs index 772d5c176c8..976cf750772 100644 --- a/crates/vm/src/scope.rs +++ b/crates/vm/src/scope.rs @@ -1,10 +1,14 @@ -use crate::{VirtualMachine, builtins::PyDictRef, function::ArgMapping}; +use crate::{ + PyRef, VirtualMachine, + builtins::{PyAnyDict, PyDictRef}, + function::ArgMapping, +}; use alloc::fmt; #[derive(Clone)] pub struct Scope { pub locals: Option, - pub globals: PyDictRef, + pub globals: PyRef, } impl fmt::Debug for Scope { @@ -17,8 +21,11 @@ impl fmt::Debug for Scope { impl Scope { #[inline] #[must_use] - pub const fn new(locals: Option, globals: PyDictRef) -> Self { - Self { locals, globals } + pub fn new(locals: Option, globals: impl Into>) -> Self { + Self { + locals, + globals: globals.into(), + } } pub fn with_builtins( diff --git a/crates/vm/src/stdlib/_testinternalcapi.rs b/crates/vm/src/stdlib/_testinternalcapi.rs index caafcf5ab57..f378eed11c4 100644 --- a/crates/vm/src/stdlib/_testinternalcapi.rs +++ b/crates/vm/src/stdlib/_testinternalcapi.rs @@ -1125,11 +1125,16 @@ fn expandtabs(input: &str, tab_size: usize) -> String { expanded } -type CodeNamespaces = (PyRef, Option>, Option>); +type CodeNamespaces = ( + PyRef, + Option>, + Option>, +); fn code_or_function(obj: &PyObject, vm: &VirtualMachine) -> PyResult { if let Ok(func) = obj.to_owned().downcast::() { - let builtins = func.builtins.clone().downcast::().ok(); + let builtins = + crate::builtins::PyAnyDict::from_object(&func.builtins).map(ToOwned::to_owned); return Ok(( (*func.code).to_owned(), Some(func.globals.clone()), @@ -1142,19 +1147,22 @@ fn code_or_function(obj: &PyObject, vm: &VirtualMachine) -> PyResult>>( override_ns: OptionalArg, - default: Option>, + default: Option, name: &str, vm: &VirtualMachine, -) -> PyResult>> { +) -> PyResult> { match override_ns.into_option() { - Some(obj) => obj.downcast::().map(Some).map_err(|obj| { - vm.new_type_error(format!( - "expected a dict for \"{name}\", got {}", - obj.class().name() - )) - }), + Some(obj) => obj + .downcast::() + .map(|dict| Some(dict.into())) + .map_err(|obj| { + vm.new_type_error(format!( + "expected a dict for \"{name}\", got {}", + obj.class().name() + )) + }), None => Ok(default), } } @@ -1280,8 +1288,8 @@ fn set_unbound_var_counts( counts: &mut VarCounts, globalnames: Option, attrnames: Option, - globalsns: Option<&Py>, - builtinsns: Option<&Py>, + globalsns: Option<&crate::builtins::PyAnyDict>, + builtinsns: Option<&crate::builtins::PyAnyDict>, vm: &VirtualMachine, ) -> PyResult<()> { let globalnames = optional_set(globalnames, "globalnames", vm)?; @@ -1317,8 +1325,8 @@ fn identify_unbound_names( code: &Py, globalnames: Option>, attrnames: Option>, - globalsns: Option<&Py>, - builtinsns: Option<&Py>, + globalsns: Option<&crate::builtins::PyAnyDict>, + builtinsns: Option<&crate::builtins::PyAnyDict>, vm: &VirtualMachine, ) -> PyResult<(UnboundCounts, i32)> { let mut seen_globals = HashSet::new(); diff --git a/crates/vm/src/stdlib/builtins.rs b/crates/vm/src/stdlib/builtins.rs index 655f30f87d8..e4c21007a0b 100644 --- a/crates/vm/src/stdlib/builtins.rs +++ b/crates/vm/src/stdlib/builtins.rs @@ -11,7 +11,7 @@ mod builtins { AsObject, Py, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, TryFromObject, VirtualMachine, builtins::{ - PyByteArray, PyBytes, PyDictRef, PyStr, PyStrRef, PyTuple, PyTupleRef, PyType, + PyAnyDict, PyByteArray, PyBytes, PyStr, PyStrRef, PyTuple, PyTupleRef, PyType, PyUtf8StrRef, enumerate::PyReverseSequenceIterator, function::{PyCell, PyCellRef, PyFunction}, @@ -425,6 +425,26 @@ mod builtins { locals: Option, } + /// `_PyEval_EnsureBuiltins` + fn ensure_builtins(globals: &PyAnyDict, vm: &VirtualMachine) -> PyResult<()> { + if globals + .inner_getitem_opt(identifier!(vm, __builtins__), vm)? + .is_some() + { + return Ok(()); + } + // PyEval_GetBuiltins + let builtins = + crate::frame::current_builtins().unwrap_or_else(|| vm.builtins.dict().into()); + match globals.as_mutable() { + Some(dict) => dict.inner_setitem(identifier!(vm, __builtins__), builtins, vm), + None => globals + .as_object() + .set_item(identifier!(vm, __builtins__), builtins, vm) + .map_err(|_| vm.new_type_error("cannot assign __builtins__ to frozendict globals")), + } + } + impl ScopeArgs { fn make_scope( self, @@ -436,21 +456,23 @@ mod builtins { vm: &VirtualMachine, func_name: &'static str, ) -> PyResult<()> { - if !globals.fast_isinstance(vm.ctx.types.dict_type) { + if !globals.fast_isinstance(vm.ctx.types.dict_type) + && !globals.fast_isinstance(vm.ctx.types.frozendict_type) + { return Err(match func_name { "eval" => { let is_mapping = globals.mapping_unchecked().check(); vm.new_type_error(if is_mapping { - "globals must be a real dict; try eval(expr, {}, mapping)" + "globals must be a real dict or a frozendict; try eval(expr, {}, mapping)" } else { - "globals must be a dict" + "globals must be a dict or a frozendict" }) } "exec" => vm.new_type_error(format!( - "exec() globals must be a dict, not {}", + "exec() globals must be a dict or a frozendict, not {}", globals.class().name() )), - _ => vm.new_type_error("globals must be a dict"), + _ => vm.new_type_error("globals must be a dict or a frozendict"), }); } Ok(()) @@ -460,15 +482,11 @@ mod builtins { Some(globals) => { validate_globals_dict(&globals, vm, func_name)?; - let globals = PyDictRef::try_from_object(vm, globals)?; - if !globals.contains_key(identifier!(vm, __builtins__), vm) { - let builtins_dict = vm.builtins.dict().into(); - globals.set_item(identifier!(vm, __builtins__), builtins_dict, vm)?; - } + let globals = PyRef::::try_from_object(vm, globals)?; ( globals.clone(), self.locals - .unwrap_or_else(|| ArgMapping::from_dict_exact(globals.clone())), + .unwrap_or_else(|| ArgMapping::from_anydict_exact(globals.clone())), ) } None => ( @@ -481,7 +499,9 @@ mod builtins { ), }; - let scope = crate::scope::Scope::with_builtins(Some(locals), globals, vm); + ensure_builtins(&globals, vm)?; + + let scope = crate::scope::Scope::new(Some(locals), globals); Ok(scope) } } @@ -650,6 +670,7 @@ mod builtins { func: &str, closure: Option>>, ) -> PyResult { + let from_string = matches!(&source, Either::B(_)); // Determine code object: let code_obj = match source { #[cfg(feature = "rustpython-compiler")] @@ -675,6 +696,18 @@ mod builtins { ))); } + // run_eval_code_obj: set globals['__builtins__'] if it doesn't exist + if from_string + && scope + .globals + .inner_getitem_opt(identifier!(vm, __builtins__), vm)? + .is_none() + { + scope + .globals + .set_item(identifier!(vm, __builtins__), vm.builtins.dict().into(), vm)?; + } + // Run the code: vm.run_code_obj_with_closure(code_obj, scope, closure) } @@ -721,7 +754,7 @@ mod builtins { } #[pyfunction] - fn globals(vm: &VirtualMachine) -> PyDictRef { + fn globals(vm: &VirtualMachine) -> PyRef { vm.current_globals() } @@ -1490,6 +1523,7 @@ pub fn init_module(vm: &VirtualMachine, module: &Py) { "dict" => ctx.types.dict_type.to_owned(), "enumerate" => ctx.types.enumerate_type.to_owned(), "float" => ctx.types.float_type.to_owned(), + "frozendict" => ctx.types.frozendict_type.to_owned(), "frozenset" => ctx.types.frozenset_type.to_owned(), "filter" => ctx.types.filter_type.to_owned(), "int" => ctx.types.int_type.to_owned(), diff --git a/crates/vm/src/types/zoo.rs b/crates/vm/src/types/zoo.rs index fce1a1e9eb6..5291e677add 100644 --- a/crates/vm/src/types/zoo.rs +++ b/crates/vm/src/types/zoo.rs @@ -40,6 +40,7 @@ pub struct TypeZoo { pub float_type: &'static Py, pub frame_type: &'static Py, pub frame_locals_proxy_type: &'static Py, + pub frozendict_type: &'static Py, pub frozenset_type: &'static Py, pub generator_type: &'static Py, pub int_type: &'static Py, @@ -139,6 +140,7 @@ impl TypeZoo { dict_type: dict::PyDict::init_builtin_type(), enumerate_type: enumerate::PyEnumerate::init_builtin_type(), float_type: float::PyFloat::init_builtin_type(), + frozendict_type: dict::PyFrozenDict::init_builtin_type(), frozenset_type: set::PyFrozenSet::init_builtin_type(), filter_type: filter::PyFilter::init_builtin_type(), list_type: list::PyList::init_builtin_type(), diff --git a/crates/vm/src/vm/crossinterp.rs b/crates/vm/src/vm/crossinterp.rs index fbab68f153d..a72d34ea92a 100644 --- a/crates/vm/src/vm/crossinterp.rs +++ b/crates/vm/src/vm/crossinterp.rs @@ -340,7 +340,7 @@ impl SharedValue { // Stateless functions have no globals, so `__main__` is used, // just like for builtins such as exec(). let globals = vm.main_namespace()?; - Ok(PyFunction::new(code, globals, vm)?.into_pyobject(vm)) + Ok(PyFunction::new(code, globals.into(), vm)?.into_pyobject(vm)) } Self::Pickled(data) => pickle_loads(&data, vm), } @@ -763,7 +763,12 @@ pub(crate) fn walk_instructions( /// `_PyFunction_VerifyStateless`. fn verify_stateless_function(func: &Py, vm: &VirtualMachine) -> PyResult<()> { - // `__globals__` is a dict by construction, so only the builtins are checked. + let globals = func.globals.as_mutable().ok_or_else(|| { + vm.new_type_error(format!( + "unsupported globals {}", + render_repr(func.globals.as_object(), vm) + )) + })?; let builtins = func.builtins.downcast_ref::().ok_or_else(|| { vm.new_type_error(format!( "unsupported builtins {}", @@ -786,7 +791,7 @@ fn verify_stateless_function(func: &Py, vm: &VirtualMachine) -> PyRe { return Err(vm.new_value_error("closures not supported")); } - verify_stateless(&func.code, Some((&func.globals, builtins)), vm) + verify_stateless(&func.code, Some((globals, builtins)), vm) } /// `verify_script`: a script takes no arguments and returns only None. diff --git a/crates/vm/src/vm/mod.rs b/crates/vm/src/vm/mod.rs index 28d50958a6a..c3625828cc4 100644 --- a/crates/vm/src/vm/mod.rs +++ b/crates/vm/src/vm/mod.rs @@ -24,7 +24,7 @@ mod vm_ops; use crate::{ AsObject, Py, PyObject, PyObjectRef, PyPayload, PyRef, PyResult, builtins::{ - self, PyBaseExceptionRef, PyBaseObject, PyDict, PyDictRef, PyFrozenSet, PyInt, PyList, + self, PyAnyDict, PyBaseExceptionRef, PyBaseObject, PyDict, PyFrozenSet, PyInt, PyList, PyModule, PySet, PyStr, PyStrInterned, PyStrRef, PyTypeRef, PyUtf8Str, PyUtf8StrInterned, PyWeak, code::PyCode, @@ -3157,7 +3157,7 @@ impl VirtualMachine { .locals(self) } - pub fn current_globals(&self) -> PyDictRef { + pub fn current_globals(&self) -> PyRef { let ptr = crate::vm::thread::get_current_frame(); if !ptr.is_null() { return unsafe { (*ptr).globals().to_owned() }; @@ -3270,7 +3270,7 @@ impl VirtualMachine { // Locals fallback: use the heavy frame if available, otherwise // use globals as locals (light frame locals are on the data stack). let locals_mapping = self.current_frame().map_or_else( - || ArgMapping::from_dict_exact(globals.clone()), + || ArgMapping::from_anydict_exact(globals.clone()), |f| f.iframe().locals.clone_mapping(self), ); (Some(locals_mapping), Some(globals)) diff --git a/crates/vm/src/vm/vm_new.rs b/crates/vm/src/vm/vm_new.rs index 1a1696b7106..7c5d43d1643 100644 --- a/crates/vm/src/vm/vm_new.rs +++ b/crates/vm/src/vm/vm_new.rs @@ -83,15 +83,16 @@ impl VirtualMachine { } pub fn new_scope_with_main(&self) -> PyResult { - let scope = self.new_scope_with_builtins(); - let main_module = self.new_module("__main__", scope.globals.clone(), None); + let globals = self.ctx.new_dict(); + let scope = Scope::with_builtins(None, globals.clone(), self); + let main_module = self.new_module("__main__", globals.clone(), None); self.sys_module.get_attr("modules", self)?.set_item( "__main__", main_module.into(), self, )?; - self.set_main_builtin_importer(&scope.globals)?; + self.set_main_builtin_importer(&globals)?; Ok(scope) } diff --git a/crates/vm/src/warn.rs b/crates/vm/src/warn.rs index 14ae7199a9b..e369606f581 100644 --- a/crates/vm/src/warn.rs +++ b/crates/vm/src/warn.rs @@ -762,12 +762,12 @@ fn setup_context( d.downcast::() .map_err(|_| vm.new_type_error("sys.__dict__ is not a dictionary")) })?; - (globals, vm.ctx.intern_str(""), 0) + (globals.into(), vm.ctx.intern_str(""), 0) }; - let registry = match globals.get_item("__warningregistry__", vm) { - Ok(r) => r, - Err(_) => { + let registry = match globals.inner_getitem_opt("__warningregistry__", vm)? { + Some(r) => r, + None => { let r = vm.ctx.new_dict(); globals.set_item("__warningregistry__", r.clone().into(), vm)?; r.into() @@ -776,7 +776,8 @@ fn setup_context( // Setup module. let module = globals - .get_item("__name__", vm) - .unwrap_or_else(|_| vm.new_pyobj("")); + .inner_getitem_opt("__name__", vm)? + .filter(|module| vm.is_none(module) || module.downcastable::()) + .unwrap_or_else(|| vm.new_pyobj("")); Ok((filename.to_owned(), lineno, Some(module), registry)) } diff --git a/extra_tests/snippets/builtin_dict.py b/extra_tests/snippets/builtin_dict.py index e593b92fd87..3d7347a5041 100644 --- a/extra_tests/snippets/builtin_dict.py +++ b/extra_tests/snippets/builtin_dict.py @@ -873,3 +873,297 @@ def __hash__(self): # entirely when popping from an empty dict). {1: 1}.pop(CountingHash()) assert CountingHash.calls == 1, CountingHash.calls + + +# frozendict as a namespace, beyond what CPython's own tests cover. +import builtins +import gc +import sys +import types +import unittest + + +class FrozenDictDescriptorTests(unittest.TestCase): + def test_unrelated_descriptors_cannot_mutate_or_access(self): + frozen = frozendict(protected=23) + for method, args in ( + (dict.__init__, ({"replacement": 8},)), + (dict.__setitem__, ("protected", 99)), + (dict.__delitem__, ("protected",)), + (dict.update, ({"replacement": 8},)), + (dict.clear, ()), + (dict.get, ("protected",)), + ): + with self.subTest(method=method), self.assertRaises(TypeError): + method(frozen, *args) + with self.assertRaises(TypeError): + frozendict.get({"protected": 23}, "protected") + self.assertEqual(frozen, {"protected": 23}) + for name in ( + "clear", + "update", + "pop", + "popitem", + "setdefault", + "__setitem__", + "__delitem__", + "__ior__", + ): + self.assertFalse(hasattr(frozen, name), name) + + +class FrozenGlobalsTests(unittest.TestCase): + def setUp(self): + self.ns = frozendict( + __builtins__=builtins.__dict__, + __name__="frozen_test", + answer=42, + sys=sys, + ) + + def test_eval_and_implicit_scope(self): + self.assertIs(eval("globals()", self.ns), self.ns) + self.assertIs(eval("locals()", self.ns), self.ns) + self.assertEqual(eval("eval('answer')", self.ns), 42) + self.assertEqual(eval("(lambda: answer)()", self.ns), 42) + + def test_function_and_materialized_frame(self): + local = {} + exec("def f():\n return sys._getframe(), globals(), answer\n", self.ns, local) + f = local["f"] + self.assertIs(f.__globals__, self.ns) + self.assertEqual(f.__module__, "frozen_test") + for _ in range(300): + frame, ns, answer = f() + self.assertIs(frame.f_globals, self.ns) + self.assertIs(ns, self.ns) + self.assertEqual(answer, 42) + gc.collect() + self.assertIs(frame.f_globals, self.ns) + + def test_generator_frames(self): + local = {} + exec( + "def gen():\n yield sys._getframe(), globals(), answer\n yield answer\n", + self.ns, + local, + ) + gen = local["gen"]() + self.assertIs(gen.gi_frame.f_globals, self.ns) + frame, ns, answer = next(gen) + self.assertIs(frame.f_globals, self.ns) + self.assertIs(ns, self.ns) + self.assertEqual(answer, 42) + self.assertEqual(next(gen), 42) + with self.assertRaises(StopIteration): + next(gen) + gc.collect() + self.assertIs(frame.f_globals, self.ns) + + def test_coroutine_frames(self): + local = {} + exec( + "async def coro():\n return sys._getframe(), globals(), answer\n", + self.ns, + local, + ) + coro = local["coro"]() + self.assertIs(coro.cr_frame.f_globals, self.ns) + with self.assertRaises(StopIteration) as stopped: + coro.send(None) + frame, ns, answer = stopped.exception.value + self.assertIs(frame.f_globals, self.ns) + self.assertIs(ns, self.ns) + self.assertEqual(answer, 42) + + def test_async_generator_frames(self): + local = {} + exec( + "async def agen():\n yield sys._getframe(), globals(), answer\n", + self.ns, + local, + ) + agen = local["agen"]() + self.assertIs(agen.ag_frame.f_globals, self.ns) + with self.assertRaises(StopIteration) as stopped: + agen.asend(None).send(None) + frame, ns, answer = stopped.exception.value + self.assertIs(frame.f_globals, self.ns) + self.assertIs(ns, self.ns) + self.assertEqual(answer, 42) + with self.assertRaises(StopIteration): + agen.aclose().send(None) + + def test_local_namespace_and_imports(self): + local = {} + exec( + "import math\nclass C:\n value = answer\ndef f():\n import math\n return math.sqrt(answer)\n", + self.ns, + local, + ) + self.assertEqual(local["C"].value, 42) + self.assertEqual(local["C"].__module__, "frozen_test") + self.assertEqual(local["f"](), local["math"].sqrt(42)) + self.assertNotIn("math", self.ns) + + def test_missing_builtins(self): + for func, source in [(eval, ""), (exec, "")]: + with self.assertRaises(TypeError) as caught: + func(source, frozendict()) + self.assertEqual( + str(caught.exception), + "cannot assign __builtins__ to frozendict globals", + ) + + def test_writes_and_deletes(self): + for source, error in [ + ("answer = 0", "'frozendict' object does not support item assignment"), + ( + "global answer; answer = 0", + "frozendict object does not support item assignment", + ), + ( + "global answer; del answer", + "frozendict object does not support item deletion", + ), + ( + "global missing; del missing", + "frozendict object does not support item deletion", + ), + ]: + with self.assertRaises(TypeError) as caught: + exec(source, self.ns) + self.assertEqual(str(caught.exception), error) + self.assertEqual(self.ns["answer"], 42) + with self.assertRaises(NameError): + exec("del answer", self.ns) + + def test_explicit_function_constructor_stays_dict_only(self): + with self.assertRaises(TypeError): + types.FunctionType((lambda: 42).__code__, self.ns) + + def test_subclass_name_and_global_lookup(self): + class F(frozendict): + def __getitem__(self, key): + if key == "answer": + return 81 + return super().__getitem__(key) + + ns = F(self.ns) + self.assertEqual(eval("answer", ns), 81) + self.assertEqual(eval("(lambda: answer)()", ns), 81) + self.assertEqual(eval("answer", ns, {}), 42) + + def test_subclass_function_metadata_reads_storage(self): + class F(frozendict): + def __getitem__(self, key): + if key == "__name__": + return "wrong_name" + if key == "__builtins__": + return {"len": lambda x: 99} + return super().__getitem__(key) + + ns = F(self.ns) + f = eval("lambda: len([])", ns) + self.assertEqual(f.__module__, "frozen_test") + self.assertIs(f.__globals__, ns) + self.assertEqual(f(), 0) + + def test_subclass_builtins_assignment_hook(self): + calls = [] + + class F(frozendict): + def __setitem__(self, key, value): + calls.append((key, value)) + + for operation, source, mode, expected in ( + (eval, "1", "eval", 1), + (exec, "pass", "exec", None), + ): + namespace = F() + calls.clear() + self.assertEqual( + operation(compile(source, "", mode), namespace), + expected, + ) + self.assertEqual(calls, [("__builtins__", builtins.__dict__)]) + self.assertNotIn("__builtins__", namespace) + calls.clear() + with self.assertRaisesRegex( + TypeError, "frozendict object does not support item assignment" + ): + operation(source, namespace) + self.assertEqual(calls, [("__builtins__", builtins.__dict__)]) + + def test_implicit_frozen_builtins_namespace_counts(self): + import _testinternalcapi + + function = eval( + "lambda: len(())", + frozendict(__builtins__=frozendict(len=len)), + ) + counts = _testinternalcapi.get_code_var_counts(function) + self.assertEqual( + counts["unbound"]["globals"], + {"total": 1, "numglobal": 0, "numbuiltin": 1, "numunknown": 0}, + ) + for argument in ("globalsns", "builtinsns"): + with self.assertRaises(TypeError): + _testinternalcapi.get_code_var_counts( + function, **{argument: frozendict()} + ) + + def test_warning_context_reads_frozen_storage(self): + import warnings + + class F(frozendict): + def __getitem__(self, key): + if key in ("__warningregistry__", "__name__"): + raise AssertionError("warning context must read stored entries") + return super().__getitem__(key) + + for has_registry in (False, True): + entries = dict(self.ns, warn=warnings.warn) + if has_registry: + entries["__warningregistry__"] = {} + function = eval("lambda: warn('probe')", F(entries)) + with warnings.catch_warnings(record=True) as caught: + warnings.simplefilter("always") + if has_registry: + function() + self.assertEqual(len(caught), 1) + self.assertEqual(str(caught[0].message), "probe") + else: + with self.assertRaisesRegex( + TypeError, + "frozendict object does not support item assignment", + ): + function() + self.assertEqual(caught, []) + + def test_frozen_builtins(self): + ns = frozendict(__builtins__=frozendict({"len": len})) + self.assertEqual(eval("(lambda: len([1, 2]))()", ns), 2) + + def test_shared_code_does_not_keep_mutable_global_specialization(self): + code = compile("lambda: answer", "", "eval") + mutable = {"__builtins__": builtins.__dict__, "answer": 19} + f = eval(code, mutable) + for _ in range(300): + self.assertEqual(f(), 19) + frozen_f = eval(code, self.ns) + for _ in range(300): + self.assertEqual(frozen_f(), 42) + self.assertEqual(f(), 19) + mutable["answer"] = 23 + self.assertEqual(f(), 23) + self.assertEqual(frozen_f(), 42) + + +frozen_suite = unittest.TestSuite( + unittest.defaultTestLoader.loadTestsFromTestCase(case) + for case in (FrozenDictDescriptorTests, FrozenGlobalsTests) +) +frozen_result = unittest.TextTestRunner(verbosity=2).run(frozen_suite) +assert frozen_result.wasSuccessful() +assert not frozen_result.skipped diff --git a/scripts/update_lib/__main__.py b/scripts/update_lib/__main__.py index 49399db6f43..7eb91970f2f 100644 --- a/scripts/update_lib/__main__.py +++ b/scripts/update_lib/__main__.py @@ -8,6 +8,7 @@ python scripts/update_lib migrate cpython/Lib/test/test_foo.py python scripts/update_lib patches --from Lib/test/foo.py --to cpython/Lib/test/foo.py python scripts/update_lib auto-mark Lib/test/test_foo.py + python scripts/update_lib copy-test SOURCE DESTINATION --test ClassName.method_name --source-sha256 HASH --destination-sha256 HASH """ import argparse @@ -49,6 +50,11 @@ def main(argv: list[str] | None = None) -> int: help="Copy library file/directory from CPython (delete existing first)", add_help=False, ) + subparsers.add_parser( + "copy-test", + help="Copy one complete canonical test from hash-pinned inputs", + add_help=False, + ) subparsers.add_parser( "deps", help="Show dependency information for a module", @@ -72,6 +78,11 @@ def main(argv: list[str] | None = None) -> int: return copy_lib_main(remaining) + if args.command == "copy-test": + from update_lib.cmd_copy_test import main as copy_test_main + + return copy_test_main(remaining) + if args.command == "migrate": from update_lib.cmd_migrate import main as migrate_main diff --git a/scripts/update_lib/cmd_copy_test.py b/scripts/update_lib/cmd_copy_test.py new file mode 100644 index 00000000000..5e14959bb02 --- /dev/null +++ b/scripts/update_lib/cmd_copy_test.py @@ -0,0 +1,134 @@ +"""Copy one complete canonical test definition without refreshing its module.""" + +import argparse +import ast +import hashlib +import json +from pathlib import Path + +from update_lib.patch_spec import _iter_class_methods, _iter_module_classes + + +def _definition(contents: str, qualified_name: str): + parts = qualified_name.split(".") + if len(parts) != 2 or not all(part.isidentifier() for part in parts): + raise ValueError("test name must be ClassName.method_name") + tree = ast.parse(contents) + classes = [node for node in _iter_module_classes(tree) if node.name == parts[0]] + if len(classes) != 1: + raise ValueError(f"Expected one class named {parts[0]}, found {len(classes)}") + methods = [ + node for node in _iter_class_methods(classes[0]) if node.name == parts[1] + ] + if len(methods) != 1: + raise ValueError( + f"Expected one method named {qualified_name}, found {len(methods)}" + ) + method = methods[0] + parents = { + child: parent + for parent in ast.walk(tree) + for child in ast.iter_child_nodes(parent) + } + guards = [] + child = method + while child in parents: + parent = parents[child] + if isinstance(parent, ast.If): + branch = "body" if child in parent.body else "orelse" + guards.append((ast.dump(parent.test), branch)) + elif not isinstance(parent, (ast.Module, ast.ClassDef)): + raise ValueError("Only module/class scopes and if guards are supported") + child = parent + start = min([method.lineno, *(node.lineno for node in method.decorator_list)]) - 1 + return method, guards, start, method.end_lineno + + +def copy_test(source: str, destination: str, qualified_name: str) -> str: + """Replace an unambiguous whole definition, preserving all surrounding bytes. + + Guard, signature, decorator and indentation differences require a broader + migration rather than silently changing the selected test's execution context. + """ + src, src_guards, src_start, src_end = _definition(source, qualified_name) + dst, dst_guards, dst_start, dst_end = _definition(destination, qualified_name) + if src_guards != dst_guards: + raise ValueError("Source and destination test guards differ") + + def signature(node): + return ( + ast.dump(node.args), + ast.dump(node.returns) if node.returns is not None else None, + [ast.dump(param) for param in getattr(node, "type_params", [])], + ) + + if type(src) is not type(dst) or signature(src) != signature(dst): + raise ValueError("Source and destination test signatures differ") + if [ast.dump(node) for node in src.decorator_list] != [ + ast.dump(node) for node in dst.decorator_list + ]: + raise ValueError( + "Source and destination decorators differ; preserve markers explicitly" + ) + if src.col_offset != dst.col_offset: + raise ValueError("Source and destination indentation differs") + source_lines = source.splitlines(keepends=True) + destination_lines = destination.splitlines(keepends=True) + # AST equality omits comments, including project marker rationales. Require + # the complete decorator block to match before replacing the definition. + if ( + source_lines[src_start : src.lineno - 1] + != destination_lines[dst_start : dst.lineno - 1] + ): + raise ValueError( + "Source and destination decorator text differs; preserve markers explicitly" + ) + definition = source_lines[src_start:src_end] + result = "".join( + destination_lines[:dst_start] + definition + destination_lines[dst_end:] + ) + _, _, result_start, result_end = _definition(result, qualified_name) + if result.splitlines(keepends=True)[result_start:result_end] != definition: + raise ValueError("Copied definition is not byte-identical to the source") + return result + + +def main(argv: list[str] | None = None) -> int: + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument("source", type=Path) + parser.add_argument("destination", type=Path) + parser.add_argument("--test", required=True, help="Exact ClassName.method_name") + parser.add_argument("--source-sha256", required=True) + parser.add_argument("--destination-sha256", required=True) + args = parser.parse_args(argv) + try: + source = args.source.read_bytes() + destination = args.destination.read_bytes() + for label, contents, expected in ( + ("source", source, args.source_sha256), + ("destination", destination, args.destination_sha256), + ): + if hashlib.sha256(contents).hexdigest() != expected: + raise ValueError(f"{label} SHA256 does not match the pinned input") + result = copy_test( + source.decode("utf-8"), destination.decode("utf-8"), args.test + ).encode("utf-8") + # Recheck the destination immediately before writing so stale invocations + # fail instead of overwriting an intervening edit. + if args.destination.read_bytes() != destination: + raise ValueError("Destination changed during test preparation") + args.destination.write_bytes(result) + except (OSError, UnicodeError, SyntaxError, ValueError) as exc: + parser.exit(1, f"Error: {exc}\n") + print( + json.dumps( + { + "test": args.test, + "source_sha256": args.source_sha256, + "destination_before_sha256": args.destination_sha256, + "destination_after_sha256": hashlib.sha256(result).hexdigest(), + }, + sort_keys=True, + ) + ) + return 0 diff --git a/scripts/update_lib/patch_spec.py b/scripts/update_lib/patch_spec.py index 996b78bfa18..994fc53b261 100644 --- a/scripts/update_lib/patch_spec.py +++ b/scripts/update_lib/patch_spec.py @@ -197,7 +197,10 @@ def visit_ClassDef(self, node: ast.ClassDef): patch = patch._replace(test_name="__self__") self.patches.append(patch) - self.generic_visit(node) + # Keep conditional methods in this namespace, but do not enter + # nested classes or the bodies of methods. + for item in _iter_class_methods(node): + self.visit(item) class PatchEntry(typing.NamedTuple): @@ -228,22 +231,35 @@ def iter_patch_entries( yield from visitor.patches -def iter_tests( - tree: ast.Module, -) -> "Iterator[tuple[ast.ClassDef, ast.FunctionDef | ast.AsyncFunctionDef]]": - for key, nodes in ast.iter_fields(tree): - if key != "body": - continue +def _iter_module_classes(node: ast.AST) -> "Iterator[ast.ClassDef]": + """Find module-scope classes, including conditional definitions. - for cls_node in nodes: - if not isinstance(cls_node, ast.ClassDef): - continue + Function-local and class-nested definitions have different namespaces and + cannot be addressed by the patch format's unqualified class names. + """ + for child in ast.iter_child_nodes(node): + if isinstance(child, ast.ClassDef): + yield child + elif not isinstance(child, (ast.FunctionDef, ast.AsyncFunctionDef, ast.Lambda)): + yield from _iter_module_classes(child) - for fn_node in cls_node.body: - if not isinstance(fn_node, (ast.FunctionDef, ast.AsyncFunctionDef)): - continue - yield (cls_node, fn_node) +def _iter_class_methods( + node: ast.AST, +) -> "Iterator[ast.FunctionDef | ast.AsyncFunctionDef]": + for child in ast.iter_child_nodes(node): + if isinstance(child, (ast.FunctionDef, ast.AsyncFunctionDef)): + yield child + elif not isinstance(child, (ast.ClassDef, ast.Lambda)): + yield from _iter_class_methods(child) + + +def iter_tests( + tree: ast.Module, +) -> "Iterator[tuple[ast.ClassDef, ast.FunctionDef | ast.AsyncFunctionDef]]": + for cls_node in _iter_module_classes(tree): + for fn_node in _iter_class_methods(cls_node): + yield (cls_node, fn_node) def iter_patches(contents: str) -> "Iterator[PatchEntry]": @@ -286,24 +302,28 @@ def _iter_patch_lines( # Track class bases for inherited async method lookup class_bases: dict[str, list[str]] = {} all_classes = set() - all_class_nodes = [] - for node in tree.body: - if not isinstance(node, ast.ClassDef): - continue - + all_class_nodes = list(_iter_module_classes(tree)) + for node in all_class_nodes: + if node.name in all_classes: + raise ValueError(f"Ambiguous module-scope class name: {node.name}") all_classes.add(node.name) - all_class_nodes.append(node) + method_counts = collections.Counter( + item.name for item in _iter_class_methods(node) + ) + for method in patches.get(node.name, {}): + if method_counts[method] > 1: + raise ValueError(f"Ambiguous method name: {node.name}.{method}") - for node in tree.body: + for node in all_class_nodes: if isinstance(node, ast.ClassDef): - cache[node.name] = node.end_lineno + cache[node.name] = (node.end_lineno, node.col_offset + len(DEFAULT_INDENT)) class_bases[node.name] = [ base.id for base in node.bases if isinstance(base, ast.Name) and base.id in all_classes ] cls_async: set[str] = set() - for item in node.body: + for item in _iter_class_methods(node): if isinstance(item, ast.AsyncFunctionDef): cls_async.add(item.name) if cls_async: @@ -319,10 +339,11 @@ def _iter_patch_lines( # Phase 2: Iterate and mark inherited tests for cls_name, tests in sorted(patches.items()): - lineno = cache.get(cls_name) - if not lineno: + location = cache.get(cls_name) + if not location: print(f"WARNING: {cls_name} does not exist in remote file", file=sys.stderr) continue + lineno, indent = location for test_name, specs in sorted(tests.items()): if test_name == "__self__": @@ -356,7 +377,7 @@ async def {test_name}(self): def {test_name}(self): {DEFAULT_INDENT}return super().{test_name}() """.rstrip() - yield (lineno, textwrap.indent(patch_lines, DEFAULT_INDENT)) + yield (lineno, textwrap.indent(patch_lines, " " * indent)) # Phase 3: Mark the class itself for cls_node in all_class_nodes: diff --git a/scripts/update_lib/tests/test_copy_test.py b/scripts/update_lib/tests/test_copy_test.py new file mode 100644 index 00000000000..5dc61452c90 --- /dev/null +++ b/scripts/update_lib/tests/test_copy_test.py @@ -0,0 +1,206 @@ +"""Tests for bounded, hash-pinned canonical test copying.""" + +import ast +import contextlib +import hashlib +import io +import tempfile +import unittest +from pathlib import Path + +from update_lib.cmd_copy_test import copy_test, main + + +class TestCopyTest(unittest.TestCase): + destination = '''# Destination-only header +class Finder: + if int.__doc__: + def example(): + r"""Count is below 800. π\\n""" + + def untouched(self): + return "destination" +''' + source = '''# Canonical header must not be copied +class Finder: + if int.__doc__: + def example(): + r"""Count is below 850. π\\n""" + + def untouched(self): + return "canonical" +''' + + def test_complete_definition_and_surroundings(self): + result = copy_test(self.source, self.destination, "Finder.example") + self.assertEqual(result, self.destination.replace("below 800", "below 850")) + self.assertIn('return "destination"', result) + source_method = ast.parse(self.source).body[0].body[0].body[0] + result_method = ast.parse(result).body[0].body[0].body[0] + self.assertEqual( + ast.get_source_segment(self.source, source_method), + ast.get_source_segment(result, result_method), + ) + + def test_guard_mismatch_is_rejected(self): + with self.assertRaisesRegex(ValueError, "guards differ"): + copy_test( + self.source.replace("if int.__doc__", "if enabled"), + self.destination, + "Finder.example", + ) + + def test_missing_or_ambiguous_definition_is_rejected(self): + for source in ("class Other: pass\n", self.source + self.source): + with self.subTest(source=source): + with self.assertRaisesRegex(ValueError, "Expected one class"): + copy_test(source, self.destination, "Finder.example") + duplicate = self.source.replace( + " def untouched(self):", + " def example():\n pass\n\n def untouched(self):", + ) + with self.assertRaisesRegex(ValueError, "Expected one method"): + copy_test(duplicate, self.destination, "Finder.example") + + def test_decorator_and_signature_changes_are_rejected(self): + decorated = self.source.replace( + " def example():", " @decorator\n def example():" + ) + with self.assertRaisesRegex(ValueError, "decorators differ"): + copy_test(decorated, self.destination, "Finder.example") + with self.assertRaisesRegex(ValueError, "signatures differ"): + copy_test( + self.source.replace("def example():", "def example(arg):"), + self.destination, + "Finder.example", + ) + + def test_nested_local_scope_is_not_selected(self): + local = "def factory():\n" + "".join( + " " + line for line in self.source.splitlines(True) + ) + with self.assertRaisesRegex(ValueError, "Expected one class"): + copy_test(local, self.destination, "Finder.example") + + def test_decorator_comments_must_match_exactly(self): + for source_marker, destination_marker in ( + ( + "@unittest.expectedFailure", + "@unittest.expectedFailure # TODO: RUSTPYTHON; reason", + ), + ( + "@unittest.expectedFailure # original reason", + "@unittest.expectedFailure # changed reason", + ), + ( + "@unittest.expectedFailure\n @other", + "@unittest.expectedFailure\n # marker rationale\n @other", + ), + ( + "@unittest.expectedFailureIf(\n condition,\n 'reason',\n )", + "@unittest.expectedFailureIf(\n condition, # project limitation\n 'reason',\n )", + ), + ): + with self.subTest(destination_marker=destination_marker): + source = self.source.replace( + " def example():", + f" {source_marker}\n def example():", + ) + destination = self.destination.replace( + " def example():", + f" {destination_marker}\n def example():", + ) + with self.assertRaisesRegex(ValueError, "decorator text differs"): + copy_test(source, destination, "Finder.example") + + def test_identical_decorator_comments_are_preserved(self): + prefix = ( + " @unittest.expectedFailure # TODO: RUSTPYTHON; reason\n" + " # Keep this rationale with the marker.\n" + " def example():" + ) + for declaration in ("def example():", "async def example():"): + with self.subTest(declaration=declaration): + source = self.source.replace(" def example():", prefix) + destination = self.destination.replace(" def example():", prefix) + source = source.replace("def example():", declaration) + destination = destination.replace("def example():", declaration) + self.assertEqual( + copy_test(source, destination, "Finder.example"), + destination.replace("below 800", "below 850"), + ) + + def test_cli_does_not_write_on_decorator_comment_mismatch(self): + with tempfile.TemporaryDirectory() as temp: + source = Path(temp) / "source.py" + destination = Path(temp) / "destination.py" + source.write_text( + self.source.replace( + " def example():", + " @unittest.expectedFailure\n def example():", + ) + ) + destination.write_text( + self.destination.replace( + " def example():", + " @unittest.expectedFailure # TODO: RUSTPYTHON; reason\n" + " def example():", + ) + ) + before = destination.read_bytes() + with ( + contextlib.redirect_stderr(io.StringIO()) as errors, + self.assertRaises(SystemExit), + ): + main( + [ + str(source), + str(destination), + "--test", + "Finder.example", + "--source-sha256", + hashlib.sha256(source.read_bytes()).hexdigest(), + "--destination-sha256", + hashlib.sha256(before).hexdigest(), + ] + ) + self.assertIn("decorator text differs", errors.getvalue()) + self.assertEqual(destination.read_bytes(), before) + + def test_return_annotation_and_type_parameters_are_rejected(self): + for declaration in ("def example() -> int:", "def example[T]():"): + with self.subTest(declaration=declaration): + source = self.source.replace("def example():", declaration) + with self.assertRaisesRegex(ValueError, "signatures differ"): + copy_test(source, self.destination, "Finder.example") + + def test_cli_hashes_and_no_write_on_stale_input(self): + with tempfile.TemporaryDirectory() as temp: + source = Path(temp) / "source.py" + destination = Path(temp) / "destination.py" + source.write_bytes(self.source.encode()) + destination.write_bytes(self.destination.encode()) + args = [ + str(source), + str(destination), + "--test", + "Finder.example", + "--source-sha256", + hashlib.sha256(source.read_bytes()).hexdigest(), + "--destination-sha256", + "0" * 64, + ] + with ( + contextlib.redirect_stderr(io.StringIO()), + self.assertRaises(SystemExit), + ): + main(args) + self.assertEqual(destination.read_bytes(), self.destination.encode()) + args[-1] = hashlib.sha256(destination.read_bytes()).hexdigest() + with contextlib.redirect_stdout(io.StringIO()) as output: + self.assertEqual(main(args), 0) + self.assertIn('"source_sha256"', output.getvalue()) + self.assertEqual( + destination.read_text(), + self.destination.replace("below 800", "below 850"), + ) diff --git a/scripts/update_lib/tests/test_patch_spec.py b/scripts/update_lib/tests/test_patch_spec.py index 92a0a35a153..2c6bb99ff84 100644 --- a/scripts/update_lib/tests/test_patch_spec.py +++ b/scripts/update_lib/tests/test_patch_spec.py @@ -67,6 +67,38 @@ async def test_async(self): class TestExtractPatches(unittest.TestCase): """Tests for extract_patches function.""" + def test_class_conditional_method_markers_are_preserved(self): + code = """class Owner: + if enabled: + @unittest.expectedFailure # TODO: RUSTPYTHON + def test_value(self): + return 42 +""" + patches = extract_patches(code) + self.assertIn("test_value", patches["Owner"]) + clean = code.replace( + " @unittest.expectedFailure # TODO: RUSTPYTHON\n", "" + ) + result = apply_patches(clean, patches) + self.assertIn(" @unittest.expectedFailure", result) + method = list(iter_tests(ast.parse(result)))[0][1] + self.assertEqual(method.col_offset, 8) + self.assertEqual(ast.dump(method.body[0]), "Return(value=Constant(value=42))") + + def test_nested_class_markers_do_not_target_module_names(self): + code = """class Container: + class Target: + @unittest.expectedFailure # TODO: RUSTPYTHON + def test_value(self): + pass +def factory(): + class Target: + @unittest.expectedFailure # TODO: RUSTPYTHON + def test_value(self): + pass +""" + self.assertEqual(extract_patches(code), {}) + def test_extract_expected_failure(self): """Test extracting @unittest.expectedFailure decorator.""" code = f""" @@ -177,6 +209,106 @@ def test_one(self): class TestApplyPatches(unittest.TestCase): """Tests for apply_patches function.""" + def test_conditional_class_existing_method(self): + code = """import unittest +if enabled: + class TestConditional: + def test_value(self): + assert 1 + 1 == 2 +""" + result = apply_patches( + code, + { + "TestConditional": { + "test_value": [ + PatchSpec( + UtMethod.ExpectedFailureIf, "condition", "known limitation" + ) + ] + } + }, + ) + original = list(iter_tests(ast.parse(code)))[0][1] + updated = list(iter_tests(ast.parse(result)))[0][1] + self.assertEqual(ast.dump(original.body[0]), ast.dump(updated.body[0])) + self.assertEqual(updated.col_offset, 8) + self.assertIn(" @unittest.expectedFailureIf", result) + + def test_conditional_class_inherited_override(self): + code = """import unittest +class Parent: + async def test_value(self): + return 42 +if enabled: + class Child(Parent): + pass +""" + result = apply_patches( + code, {"Child": {"test_value": [PatchSpec(UtMethod.ExpectedSuccess)]}} + ) + methods = list(iter_tests(ast.parse(result))) + self.assertEqual([cls.name for cls, _ in methods], ["Parent", "Child"]) + self.assertIsInstance(methods[1][1], ast.AsyncFunctionDef) + self.assertIn(" return await super().test_value()", result) + self.assertIn(" @unittest.expectedSuccess", result) + + def test_duplicate_conditional_class_names_are_rejected(self): + code = """if enabled: + class Duplicate: + pass +else: + class Duplicate: + pass +""" + with self.assertRaisesRegex(ValueError, "Ambiguous.*Duplicate"): + apply_patches( + code, + {"Duplicate": {"test_value": [PatchSpec(UtMethod.ExpectedFailure)]}}, + ) + + def test_local_class_names_do_not_shadow_module_class(self): + code = """import unittest +class Target: + def test_value(self): + return 1 +class Container: + class Target: + def test_value(self): + return 2 +def factory(): + class Target: + def test_value(self): + return 3 +""" + result = apply_patches( + code, {"Target": {"test_value": [PatchSpec(UtMethod.ExpectedFailure)]}} + ) + tree = ast.parse(result) + targets = [ + node + for node in ast.walk(tree) + if isinstance(node, ast.ClassDef) and node.name == "Target" + ] + self.assertEqual( + [len(node.body[0].decorator_list) for node in targets], [1, 0, 0] + ) + + def test_duplicate_conditional_method_names_are_rejected(self): + code = """class Owner: + if enabled: + def test_value(self): + return 1 + else: + def test_value(self): + return 2 +""" + with self.assertRaisesRegex( + ValueError, "Ambiguous method name: Owner.test_value" + ): + apply_patches( + code, {"Owner": {"test_value": [PatchSpec(UtMethod.ExpectedFailure)]}} + ) + def test_apply_expected_failure(self): """Test applying @unittest.expectedFailure.""" code = """import unittest diff --git a/src/shell/helper.rs b/src/shell/helper.rs index 15fc75563dc..dfd3571bfe4 100644 --- a/src/shell/helper.rs +++ b/src/shell/helper.rs @@ -1,14 +1,14 @@ #![cfg_attr(target_arch = "wasm32", allow(dead_code))] use rustpython_vm::{ - AsObject, PyResult, TryFromObject, VirtualMachine, - builtins::{PyDictRef, PyStrRef}, + AsObject, PyRef, PyResult, TryFromObject, VirtualMachine, + builtins::{PyAnyDict, PyStrRef}, function::ArgIterable, identifier, }; pub(super) struct ShellHelper<'vm> { vm: &'vm VirtualMachine, - globals: PyDictRef, + globals: PyRef, } fn reverse_string(s: &mut String) { @@ -54,7 +54,7 @@ fn split_idents_on_dot(line: &str) -> Option<(usize, Vec)> { } impl<'vm> ShellHelper<'vm> { - pub(super) const fn new(vm: &'vm VirtualMachine, globals: PyDictRef) -> Self { + pub(super) const fn new(vm: &'vm VirtualMachine, globals: PyRef) -> Self { ShellHelper { vm, globals } }