Dynamically create a type with Python

Python is a dynamic language, it means you can create new types at runtime. I had a case where I needed to create a new class depending of what a remote server could do.

The use case was:

• Ask the server what class / methods are available, for example:

  ('/cat/meow',
   '/cat/eat',
   '/cat/sleep',
   '/dog/bark',
   '/dog/eat',
   '/dog/sleep')
  

• Create dynamically the class with the associated methods to call the corresponding remote methods.

This means I couldn’t use the well known syntax:

class A(object):
    ...

class B(A):
    ...

I could have by-passed the problem by using exec:

exec('class %s(object): pass' % 'Cat')

But that’s not the right way to do it. Code generation is prone to breakage and easy to get wrong. Using type() is a better option, it is simpler to use and faster to execute: no need to parse a string.

new_class = type('Cat', (object,), {'meow': remote_call('meow'),
                                    'eat': remote_call('eat'),
                                    'sleep': remote_call('sleep')})

new_class is a regular class:

>>> type(object)
<type 'type'>
>>> type(new_class)
<type 'type'>
>>> new_class.__name__
'Cat'
>>> new_class.__bases__
(<type 'object'>,)

It’s useful when you have to build a class from something external like a database schema, or a web-service.