如何获取GAE多态代码的所有类属性（在Python中），包括通过@property定义的那些

Question

I am using the Google App Engine polymodel to model data that can have more than one instance of a property - eg a contact could have multiple phone numbers. Say this is my setup:

class Foo(polymodel.PolyModel):
    some_prop = ndb.StringProperty()

    @property
    def bar(self):
        return Bar.query(Bar.foo == self.key)

class Bar(ndb.Model):
    foo = ndb.KeyProperty(kind = Foo)
    other_prop= ndb.StringProperty()

(I got this approach after reading this GAE article on data modeling: https://developers.google.com/appengine/articles/modeling )

Now when I do:

Foo._properties

I only get access to the following:

{'some_prop': StringProperty('some_prop'), 
 'class': _ClassKeyProperty('class', repeated=True)}

Is there any way to access to ALL properties, including those defined with "@property"?

Many thanks for any help or or insight on where I'm going wrong. - Lee

UPDATE: Based on @FastTurle's great answer, I've now added a class method that returns both class properties as well as methods tagged as properties via @property:

def props(self):
    return dict(self._properties.items() +  \
                   {attr_name:getattr(self,attr_name) for \
                    attr_name, attr_value in \
                    Foo.__dict__.iteritems() if \
                    isinstance(attr_value,property)}.items())

Answer 1

Performing Foo._properties should get you access to any attributes you define on your polymodel.PolyModel that inherit from google.appengine.ext.db.Property .

This means that KeyProperty , StringProperty , etc... will show up in Foo._properties , so I'm assuming you now need to find all the methods decorated by property .

Fortunately, this isn't that hard. First, a quick sidetrack into decorators (forgive me if you know this already).

In python decorators are just syntactic sugar . For example, these two methods result in the same thing:

@property
def bar(self):
    pass

def bar(self):
    pass
bar = property(bar)

Fortunately for us, property(obj) returns a new property object. This also means that @property returns a property object as well. You can kind of think of property as a class! And finally can use isinstance(bar, property) to see if bar is a property.

Finally, we can put this into use by examining each of Foo 's attributes and selecting only those that are property instances.

for attr_name, attr_value in Foo.__dict__.iteritems():
    if isinstance(attr_value, property):
        print attr_name, attr_value

# bar, <property object at ...>