我在这里做了一个简单的例子,但是文档仍然让我有些困惑。
以下是示例代码:
class A(yaml.YAMLObject):
yaml_tag = u'!A'
def __init__(self, val):
self.val = val
if __name__ == '__main__':
t = datetime.time()
a = A(t)
print yaml.dump(a)
print yaml.load(yaml.dump(a)).val == t
输出
!一个val:!! python / object / apply:datetime.time [" \ 0 \ 0 \ 0 \ 0 \ 0 \ 0"]
真
因此,它似乎是忠实的d- /序列化,但默认的时间对象格式留下了一些需要。如何在保留我强大的转储/装载能力的同时让它更漂亮?
由于
答案 0 :(得分:11)
您可以使用PyYAML 'representers' and 'constructors'为文档中的任何datetime.time
个实例定义自己的序列化格式。
import datetime
import yaml
time_format = '%H:%M:%S'
def time_representer(dumper, data):
return dumper.represent_scalar(u'!time', data.strftime(time_format))
def time_constructor(loader, node):
value = loader.construct_scalar(node)
return datetime.datetime.strptime(value, time_format).time()
yaml.add_representer(datetime.time, time_representer)
yaml.add_constructor(u'!time', time_constructor)
class A(yaml.YAMLObject):
yaml_tag = u'!A'
def __init__(self, val):
self.val = val
if __name__ == '__main__':
t = datetime.time()
a = A(t)
print yaml.dump(a, default_flow_style=False)
print yaml.load(yaml.dump(a)).val == t
将输出:
!A
val: !time '00:00:00'
True
您还可以定义特定类如何序列化和反序列化。这样您就可以只更改A
的时间表示方式。 Docs for YAMLObject
import datetime
import yaml
time_format = '%H:%M:%S'
class A(yaml.YAMLObject):
yaml_tag = u'!A'
def __init__(self, val):
self.val = val
@classmethod
def to_yaml(cls, dumper, data):
dict_representation = {
'val':data.val.strftime(time_format)
}
node = dumper.represent_mapping(u'!A', dict_representation)
return node
@classmethod
def from_yaml(cls, loader, node):
dict_representation = loader.construct_mapping(node)
val = datetime.datetime.strptime(
dict_representation['val'], time_format).time()
return A(val)
if __name__ == '__main__':
t = datetime.time()
a = A(t)
print yaml.dump([t,a], default_flow_style=False)
loaded = yaml.load(yaml.dump([t,a]))
print loaded[0] == t
print loaded[1].val == t
将输出:
- !!python/object/apply:datetime.time
- "\0\0\0\0\0\0"
- !A
val: 00:00:00
True
True