将一个自定义类的序列映射到另一个自定义类的Python方法是什么？

Question

(...or, alternatively, what is the Pythonic version of C#'s Select(...) method? )

Given a list l of a custom class A what is the (most?) Pythonic way to map each element of l to a different custom class B ?

for example, the following code will do it, but is it the most Pythonic way of doing it? Note, the real types have many properties.

l = [A('Greg', 33), A('John', 39)]

def map_to_type_b(the_list):
    new_list = []
    for item in the_list:
        new_list.append(B(item.name, item.age))

    return new_list

l2 = map_to_type_b(l)

I'm coming from a C# background, where I would use LinQ select or the Select() extensions method to project from the source sequence to a new sequence of type B .

Answer 1

I would say it's part of the job of the B class to determine how an instance of some arbitrary other class should be transformed to an instance of B , so I would use the class method alternate constructor approach, eg as follows:

class A(object):

    def __init__(self, name, age):
        self.name = name
        self.age = age

    def __repr__(self):
        return 'A({0.name!r}, {0.age!r})'.format(self)


class B(A):

    def __repr__(self):
        return 'B({0.name!r}, {0.age!r})'.format(self)

    @classmethod
    def from_A(cls, inst):
        return cls(inst.name, inst.age)

You can then use a simple list comprehension or even map to convert a list of one class to another, eg:

>>> l = [A('Greg', 33), A('John', 39)]
>>> l
[A('Greg', 33), A('John', 39)]

>>> map(B.from_A, l)  # will look different, but is more memory-efficient, in 3.x
[B('Greg', 33), B('John', 39)]

>>> [B.from_A(a) for a in l]  # works (nearly) identically in 2.x and 3.x
[B('Greg', 33), B('John', 39)]

Answer 2

Writing data-only objects is frowned upon not only in Python but in most OO-based languages. Probably the most Pythonic way would be pass flat data around, lets say, a dict or list of dicts:

{'Greg': 33, 'John': 39}

[{'name': 'Greg', 'age': 33}, {'name': 'John', 'age': 39}]

That said, suppose you have classes A and B and you want to instantiate new Bs from existing A instances:

class A(object):
    def __init__(self, name, age):
        self.name = name
        self.age = age

    def __repr__(self):
        return '<{cls} name={s.name}, age={s.age}>'.format(
            cls=self.__class__.__name__,
            s=self
        )

class B(A):
    def __init__(self, name, age, born_as='male'):
        super(B, self).__init__(name, age)
        self.born_as = born_as

data = {'Greg': 33, 'John': 39}
list_of_a = [A(k, v) for k, v in data.items()]

You can keep it simple and just be explicit:

>>> list_of_a
[<A name=Greg, age=33>, <A name=John, age=39>]

>>> [B(a.name, a.age) for a in list_of_a]
[<B name=Greg, age=33>, <B name=John, age=39>]

If there are a lot of attributes involved, this can get a bit verbose. Lets teach B how to clone an A:

class B(A):
    def __init__(self, name, age, born_as='male'):
        super(B, self).__init__(name, age)
        self.born_as = born_as

    @classmethod
    def clone(cls, instance, *args, **kwargs):
        return cls(instance.name, instance.age, *args, **kwargs)

Since B now knows how to clone A:

>>> [B.clone(a) for a in list_of_a]
[<B name=Greg, age=33>, <B name=John, age=39>]        

It can get tedious to write clone methods for all B-like classes. Introspection is very Pythonic, so lets not repeat ourselves:

class CloneFromInstanceMixin(object):
    @classmethod
    def clone(cls, instance, **kwargs):
        constructor_args = inspect.getargspec(instance.__init__).args
        for attr_name in constructor_args:
            if attr_name in kwargs:
                continue # overrides instance attribute
            try:
                kwargs[attr_name] = getattr(instance, attr_name)
            except AttributeError:
                pass
        return cls(**kwargs)

class B(CloneFromInstanceMixin, A):
    def __init__(self, name, age, born_as='male'):
        super(B, self).__init__(name, age)
        self.born_as = born_as

>>> [B.clone(a) for a in list_of_a]
[<B name=Greg, age=33>, <B name=John, age=39>]   

I probably have too much free time.