My data structure is like this:
user = UserTuple(
name=u'Anakin', surname=u'Skywalker', birthdate=datetime.date(1981, 7, 25),
profile=ProfileTuple(avatar=u'http://localhost/profile.jpg')
)
And I want to pack this data with msgpack-python module. But msgpack converts namedtuples to lists. It can be possible pack data like this with msgpack and preserve namedtuples, just like pickle/cpickle?
You need to have the latest msgpack-python version. v0.4.7 doesn't work. (Currently have to install from master branch).
import msgpack
from collections import namedtuple
User = namedtuple('User', ['name', 'profile'])
Profile = namedtuple('Profile', ['avatar'])
def ext_pack(x):
if isinstance(x, User):
return msgpack.ExtType(1, msgpack.packb([x[0], x[1]], default=ext_pack, strict_types=True))
elif isinstance(x, Profile):
return msgpack.ExtType(2, msgpack.packb(x[0], default=ext_pack, strict_types=True))
return x
def ext_unpack(code, data):
if code == 1:
name, profile = msgpack.unpackb(data, ext_hook=ext_unpack)
return User(name, profile)
elif code == 2:
avatar = msgpack.unpackb(data, ext_hook=ext_unpack)
return Profile(avatar)
return msgpack.ExtType(code, data)
x = User("me", Profile(234))
s = msgpack.packb([x, [1, x]], default=ext_pack, strict_types=True)
msgpack.unpackb(s, ext_hook=ext_unpack)
>>> [User(name='me', profile=Profile(avatar=234)),
[1, User(name='me', profile=Profile(avatar=234))]]
Here we tag User, Profile as type code 1, 2 respectively. Alternatively, you can treat all namedtuple as the same type code, and store the actual types in the data fields.
msgpack supports custom object packing and unpacking. You just need to create methods for your named tuples.
See the documentation here: https://github.com/msgpack/msgpack-python#packingunpacking-of-custom-data-type
