曾与蒿藜同雨露,한때 잡초와 쑥과 함께 비와 이슬을 나누던 곳이 이제는 소나무와 삼나무와 함께 서리와 눈을 견뎌내고 있다.终随松柏到冰霜.かつては雑草やヨモギと共に雨や露を分かち合っていたが、今では松やヒノキと共に霜や雪に耐えている。曾与蒿藜同雨露,Once sharing rain and dew with weeds and wormwood, now enduring frost and snow with pines and cypresses.终随松柏到冰霜.曾与蒿藜同雨露한때 잡초와 쑥과 함께 비와 이슬을 나누던 곳이 이제는 소나무와 삼나무와 함께 서리와 눈을 견뎌내고 있다.,终随松柏到冰霜.譖セ荳手珍阯懷酔髮ィ髴イ�檎サ磯囂譚セ譟丞芦蜀ー髴�曾与蒿藜同雨露,鏇句笌钂胯棞鍚岄洦闇诧紝缁堥殢鏉炬煆鍒板啺闇�终随松柏到冰霜.曾与蒿藜同雨露,한때 잡초와 쑥과 함께 비와 이슬을 나누던 곳이 이제는 소나무와 삼나무와 함께 서리와 눈을 견뎌내고 있다.终随松柏到冰霜.曾与蒿藜同雨露,终随松柏到冰霜.
"""Support Eiffel-style preconditions and postconditions.
For example,
class C:
def m1(self, arg):
require arg > 0
return whatever
ensure Result > arg
can be written (clumsily, I agree) as:
class C(Eiffel):
def m1(self, arg):
return whatever
def m1_pre(self, arg):
assert arg > 0
def m1_post(self, Result, arg):
assert Result > arg
Pre- and post-conditions for a method, being implemented as methods
themselves, are inherited independently from the method. This gives
much of the same effect of Eiffel, where pre- and post-conditions are
inherited when a method is overridden by a derived class. However,
when a derived class in Python needs to extend a pre- or
post-condition, it must manually merge the base class' pre- or
post-condition with that defined in the derived class', for example:
class D(C):
def m1(self, arg):
return arg**2
def m1_post(self, Result, arg):
C.m1_post(self, Result, arg)
assert Result < 100
This gives derived classes more freedom but also more responsibility
than in Eiffel, where the compiler automatically takes care of this.
In Eiffel, pre-conditions combine using contravariance, meaning a
derived class can only make a pre-condition weaker; in Python, this is
up to the derived class. For example, a derived class that takes away
the requirement that arg > 0 could write:
def m1_pre(self, arg):
pass
but one could equally write a derived class that makes a stronger
requirement:
def m1_pre(self, arg):
require arg > 50
It would be easy to modify the classes shown here so that pre- and
post-conditions can be disabled (separately, on a per-class basis).
A different design would have the pre- or post-condition testing
functions return true for success and false for failure. This would
make it possible to implement automatic combination of inherited
and new pre-/post-conditions. All this is left as an exercise to the
reader.
"""
from Meta import MetaClass, MetaHelper, MetaMethodWrapper
class EiffelMethodWrapper(MetaMethodWrapper):
def __init__(self, func, inst):
MetaMethodWrapper.__init__(self, func, inst)
# Note that the following causes recursive wrappers around
# the pre-/post-condition testing methods. These are harmless
# but inefficient; to avoid them, the lookup must be done
# using the class.
try:
self.pre = getattr(inst, self.__name__ + "_pre")
except AttributeError:
self.pre = None
try:
self.post = getattr(inst, self.__name__ + "_post")
except AttributeError:
self.post = None
def __call__(self, *args, **kw):
if self.pre:
apply(self.pre, args, kw)
Result = apply(self.func, (self.inst,) + args, kw)
if self.post:
apply(self.post, (Result,) + args, kw)
return Result
class EiffelHelper(MetaHelper):
__methodwrapper__ = EiffelMethodWrapper
class EiffelMetaClass(MetaClass):
__helper__ = EiffelHelper
Eiffel = EiffelMetaClass('Eiffel', (), {})
def _test():
class C(Eiffel):
def m1(self, arg):
return arg+1
def m1_pre(self, arg):
assert arg > 0, "precondition for m1 failed"
def m1_post(self, Result, arg):
assert Result > arg
x = C()
x.m1(12)
## x.m1(-1)
if __name__ == '__main__':
_test()