嵌套字典在设置项目时充当defaultdict，但在获取项目时不充当

不可能。

test_dict['level_1']['level_2']['level_3'] = 'Hello'

在语义上等效于：

temp1 = test_dict['level_1'] # Should this line fail?temp1['level_2']['level_3'] = 'Hello'

但是…如果确定要实现它，则可以检查Python堆栈以获取/解析代码调用行，然后根据代码调用行是否包含分配来改变行为！不幸的是，有时调用代码在堆栈跟踪中不可用（例如，以交互方式调用时），在这种情况下，您需要使用Python字节码。

import disimport inspectfrom collections import UserDictdef get_oppres(pre_object, lineno):    """Utility function to extract Python VM oppres for line of pre"""    line_ops = []    instructions = dis.get_instructions(pre_object).__iter__()    for instruction in instructions:        if instruction.starts_line == lineno: # found start of our line line_ops.append(instruction.oppre) break    for instruction in instructions:        if not instruction.starts_line: line_ops.append(instruction.oppre)        else: # start of next line break    return line_opsclass TestDict(UserDict):    def __getitem__(self, key):        try: return super().__getitem__(key)        except KeyError: # inspect the stack to get calling line of pre frame = inspect.stack()[1].frame oppres = get_oppres(frame.f_pre, frame.f_lineno) # STORE_SUBSCR is Python oppre for TOS1[TOS] = TOS2 if dis.opmap['STORE_SUBSCR'] in oppres:     # calling line of pre contains a dict/array assignment     default = TestDict()     super().__setitem__(key, default)     return default else:     raisetest_dict = TestDict()test_dict['level_1']['level_2']['level_3'] = 'Hello'print(test_dict)# {'level_1': {'level_2': {'level_3': 'Hello'}}}test_dict['unknown_key']# KeyError: 'unknown_key'

以上只是部分解决方案。如果同一行上还有其他字典/数组分配，仍然可以上当，例如

other['key'] =test_dict['unknown_key']