pyjrpc
A bare bones JSONRPC 2.0 implementation in python (that you can flesh out)
from pyjrpc import JSONRPC
import operator
methods = {
"add": operator.add,
"sub": operator.sub,
}
jsonrpc_handler = JSONRPC(methods)
print jsonrpc_handler('{"jsonrpc":"2.0", "id":"A", "method":"add", "params":[2,3]}')
# {"jsonrpc": "2.0", "id": "A", "result": 5}
print jsonrpc_handler('{"jsonrpc":"2.0", "id":"B", "method":"sub", "params":[3]}')
# {"jsonrpc": "2.0", "id": "B", "error": {"message": "Invalid parameters", "code": -32602}}
print jsonrpc_handler('{"jsonrpc":"2.0", "id":"C", "method":"div", "params":[2,3]}')
# {"jsonrpc": "2.0", "id": "C", "error": {"message": "Method not found", "code": -32601}}
print jsonrpc_handler(
'[{"jsonrpc":"2.0", "id":"D1", "method":"add", "params":[1,1]},'
'{"jsonrpc":"2.0", "id":"D2", "method":"sub", "params":[63,21]}]'
)
# [{"jsonrpc": "2.0", "id": "D1", "result": 2},
# {"jsonrpc": "2.0", "id": "D2", "result": 42}]
Only the spec
pyjrpc aims to be a bare implementation of the JSON-RPC 2.0 specification in python, with minimum dependancies.
pyjrpc aims to be easy to use, easy to plug and easy to extend.
Server not included
pyjrpc does not implement the transport layer, it doesn't even consider it.
In it's simplest interpretation, the JSONRPC protocol only deals with requests and responses, not their transportation. This is why pyjrpc won't do anything about it.
However, shall you have the means to transport a string, pyjrpc is dead easy to plug : its main object is a callable which takes a string request parameter and applies the protocol upon it. It might return a string response if you're gentle enough.
Say my name
pyjrpc does not implement function registering magic. Keep it simple, stupid.
Names are identified by their method, or maybe the other way. Therefore,
pyjrpc only assumes that the method map passed as a parameter to the
JSONRPC()
constructor implements __getitem__
. Yup, you can use a
dict.
If you really WANT to implement the magic, feel free to do it. just
implement __getitem__
so that methods[name]
will return the
function you want. Yeah, service introspection, namespaces, fuzzy
naming, you can do it all... wait, fuzzy naming ? I hope you're not
serious about that but, hey, whatever.
Cascading handlers
TODO : A tale about handler decoration, and all the amazing things you can do with that.
Client POV
raise NotImplementedError
Compatibility
pyjrpc has been seen working live on python 2.7.
Tests
Run python setup.py test
Moar docz, moar examplz
Soon.