RPC4Django Documentation

Version: 0.1.5

4 October 2009

Table of Contents

Overview

RPC4Django is an XMLRPC and JSONRPC server for Django powered projects. Simply plug it into any existing Django project and you can make your methods available via XMLRPC and JSONRPC. In addition, it can display nice documentation about the methods it makes available in a more customizable way than DocXMLRPCServer. RPC4Django is not affiliated with the Django Project.

Features

  • Detects request type (JSONRPC or XMLRPC) based on content
  • Easy identification of RPC methods via a decorator
  • Pure python and requires no external modules except Django
  • Customizable RPC method documentation including reST
  • Supports XMLRPC and JSONRPC introspection
  • Supports method signatures (unlike SimpleXMLRPCServer)
  • Easy installation and integration with existing Django projects
  • Licensed for inclusion in open source and commercial software
  • Ties in with Django's authentication and authorization

License

RPC4Django is licensed under the BSD license.

Copyright (c) 2009, David Fischer

All rights reserved.

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

THIS SOFTWARE IS PROVIDED BY David Fischer ''AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL David Fischer BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.


Installation

Prerequisites

RPC4Django has been tested on Mac OS, Linux and Windows.

Source Install Method

This is the preferred method for installing RPC4Django.

$> tar xvfz rpc4django-x.y.z.tar.gz
$> cd rpc4django-x.y.z
$> python setup.py install

Easy Install Method

$> easy_install rpc4django

Pip Install Method

$> pip install rpc4django

No Installation Method

This method installs RPC4Django only for one specific django project but does not require any special system access.

$> tar xvfz rpc4django-x.y.z.tar.gz
$> cd rpc4django-x.y.z
$> cp -r rpc4django YOUR_DJANGO_PROJECT_DIRECTORY

Configuration

Usual Setup

  1. First, you need to add new url pattern to your root urls.py file. This file should be the one pointed to by ROOT_URLCONF in settings.py. You can replace r'^RPC2$' with anything you like.

    # urls.py
    #...
    urlpatterns = patterns('',
        # if installed via no install method
        #(r'^RPC2$', 'YOURPROJECT.rpc4django.views.serve_rpc_request'),
    
        # if installed via source or easy_install
        (r'^RPC2$', 'rpc4django.views.serve_rpc_request'),
    )
    
  2. Second, add RPC4Django to the list of installed applications in your settings.py.

    # settings.py
    #...
    INSTALLED_APPS = (
        # if installed via no install
        #'YOURPROJECT.rpc4django',
    
        # if installed via source or easy_install
        'rpc4django',
    )
    
  3. Lastly, you need to let RPC4Django know which methods to make available. This is done with the decorator @rpcmethod. RPC4Django imports all the apps in INSTALLED_APPS and makes any methods importable via __init__.py with the @rpcmethod decorator available as RPC methods. You can always write your RPC methods in another module and simply import it in __init__.py.

    # testapp/__init__.py
    from rpc4django import rpcmethod
    
    # This imports another method to be made available as an RPC method
    # This method should also have the @rpcmethod decorator
    # from mymodule import myrpcmethod
    
    # The doc string supports reST if docutils is installed
    @rpcmethod(name='mynamespace.add', signature=['int', 'int', 'int'])
    def add(a, b):
        '''Adds two numbers together
        >>> add(1, 2)
        3
        '''
    
        return a+b
    

    The decorator @rpcmethod accepts three optional parameters:

    • name - is the name under which the method will be made externally available. It defaults to the method's real name.
    • signature - is the signature of the method and will be returned by calls to the XMLRPC introspection method system.methodSignature. It is of the form: [return_value, arg1, arg2, arg3, ...]. All of the types should be XMLRPC types (eg. struct, int, array, etc. -- see the XMLRPC spec for details). Since python is relatively loosely typed, all types are set to object by default.
    • permission - the Django permission required to execute this method [new in 0.1.5]

Remotely Calling Methods

XMLRPC

XMLRPC libraries exist in virtually every major programming language. The usage varies from library to library, but I'll give a python example using the xmlrpclib standard library:

from xmlrpclib import ServerProxy
s = ServerProxy('http://localhost:8000')
s.system.listMethods()

JSONRPC

JSONRPC is less mature than XMLRPC and libraries are not as common. However, one of the main benefits of JSONRPC is that it can be used more easily from Javascript or Actionscript since JSON is their native format. There is an older JSONRPC library for python on json-rpc.org. Here is an example using that library:

from jsonrpc import ServiceProxy
s = ServiceProxy('http://localhost:8000')
s.rpc4django.mytestmethod(1,2,3)

For an example of JSONRPC being used from Javascript running in a browser, please see the demo site.


Additional Information and Settings

Running the Unit Tests

The unit tests can be run using manage.py

$> python manage.py test rpc4django

How RPC4Django Handles Requests

  • If the request is a GET request, return an HTML method summary
  • A POST request with content-type header set to text/xml or application/xml will be processed as XMLRPC
  • A POST request with content-type header containing json or javascript will be processed as JSONRPC
  • Try to parse the POST data as xml. If it parses successfully, process it as XMLRPC
  • Process it as JSONRPC

Method Summary

  • The method summary displays docstrings, signatures, and names from methods marked with the @rpcmethod decorator.
  • The method summary allows testing of methods via JSONRPC (unless it is disabled)
  • The summary is served (unless it is disabled) from a template rpc4django/rpcmethod_summary.html and can be customized in a similar way to the django admin.
  • The method summary supports reST in docstrings if the docutils library is installed. Plain text is used otherwise. ReST warnings and errors are not reported in the output.

Optional Settings

These are settings that can go in your project's settings.py:

RPC4DJANGO_LOG_REQUESTS_RESPONSES=True|False (default True)
By default RPC4Django will log (using the python logging module) all requests and responses. This can be disabled by setting RPC4DJANGO_LOG_REQUESTS_RESPONSES=False.
RPC4DJANGO_RESTRICT_INTROSPECTION=True|False (default False)
By default RPC4Django registers the standard XMLRPC and JSONRPC introspection functions. This can be disabled by setting RPC4DJANGO_RESTRICT_INTROSPECTION=True.
RPC4DJANGO_RESTRICT_JSONRPC=True|False (default False)
If RPC4DJANGO_RESTRICT_JSONRPC=True, RPC4Django will never serve a JSONRPC request. Instead, either XMLRPC will be tried or status code 404 will be returned.
RPC4DJANGO_RESTRICT_XMLRPC=True|False (default False)
If RPC4DJANGO_RESTRICT_XMLRPC=True, RPC4Django will never serve an XMLRPC request. Instead, either JSONRPC will be tried or status code 404 will be returned.
RPC4DJANGO_RESTRICT_METHOD_SUMMARY=True|False (default False)
If RPC4DJANGO_RESTRICT_METHOD_SUMMARY=True, status code 404 will be returned instead of serving the method summary as a response to a GET request.
RPC4DJANGO_RESTRICT_RPCTEST=True|False (default False)
If RPC4DJANGO_RESTRICT_RPCTEST=True, the method summary will not allow testing via JSONRPC.
RPC4DJANGO_RESTRICT_REST=True|False (default False) [new in 0.1.4]
If RPC4DJANGO_RESTRICT_REST=True, RPC4Django does not attempt to convert any of the method summary docstrings to restructured text.

Authentication

RPC4Django can be used with authenticated HTTP(s) requests and Django's authentication and authorization. [new in 0.1.5, requires Django 1.1+] It uses RemoteUserMiddleware or another derived middleware which is only built-in to Django 1.1 and higher. It is possible to use it with Django versions prior to 1.1, but it would require the RemoteUserMiddleware and the RemoteUserBackend to be added manually.

Setup

Firstly, the webserver should be configured to use basic HTTP authentication or some sort of single sign on (SSO) solution.

In settings.py, the following changes need to be made:

MIDDLEWARE_CLASSES = (
    # ...
    # Must be enabled for RPC4Django authenticated method calls
    'django.contrib.auth.middleware.AuthenticationMiddleware',

    # Required for RPC4Django authenticated method calls
    # Requires Django 1.1+
    'django.contrib.auth.middleware.RemoteUserMiddleware',
)

# Required for RPC4Django authenticated method calls
# Requires Django 1.1+
AUTHENTICATION_BACKENDS = (
    'django.contrib.auth.backends.RemoteUserBackend',
)

Usage

To protect a method, it needs to be defined with the @rpcmethod decorator and the permission parameter.

from rpc4django import rpcmethod

@rpcmethod(name='rpc4django.secret', signature=['string'], permission='auth.add_group')
def secret():
    return "Successfully called a protected method"

To call an authenticated method from the Python command prompt, use the following:

from xmlrpclib import ServerProxy
s = ServerProxy('https://username:password@example.com')
s.rpc4django.secret()

Planned Features


Known Issues


Changelog

Version 0.1.5 (4 October 2009)

Version 0.1.4 (31 August 2009)

Version 0.1.3 (15 July 2009)

Version 0.1.2 (13 July 2009)

Version 0.1.1 (11 July 2009)

Version 0.1.0 (6 July 2009)