Self-describing JSON-RPC services made easy

Overview

ReflectRPC

Build Status Documentation Status PyPI

Self-describing JSON-RPC services made easy

Contents

What is ReflectRPC?

ReflectRPC is a Python library implementing an RPC client and server using the JSON-RPC 1.0 protocol. What sets it apart from most other such implementations is that it allows the client to get a comprehensive description of the functions exposed by the server. This includes type information of parameters and return values as well as human readable JavaDoc-like descriptions of all fields. To retrieve this information the client only has to call the special RPC function __describe_functions and it will get a data structure containing the whole description of all RPC functions provided by the server.

This ability to use reflection is utilized by the included JSON-RPC shell rpcsh. It can connect to every JSON-RPC server serving line terminated JSON-RPC 1.0 over a plain socket and can be used to call RPC functions on the server and display the results. If the server implements the __describe_functions interface it can also list all RPC functions provided by the server and show a description of the functions and their parameters.

ReflectRPC does not change the JSON-RPC 1.0 protocol in any way and strives to be as compatible as possible. It only adds some special builtin RPC calls to your service to make it self-describing. That way any JSON-RPC 1.0 compliant client can talk to it while a client aware of ReflectRPC can access the extra features it provides.

Example

Write a function and register it (including its documentation):

import reflectrpc
import reflectrpc.simpleserver

def add(a, b):
    return int(a) + int(b)

rpc = reflectrpc.RpcProcessor()

add_func = reflectrpc.RpcFunction(add, 'add', 'Adds two numbers', 'int',
        'Sum of the two numbers')
add_func.add_param('int', 'a', 'First int to add')
add_func.add_param('int', 'b', 'Second int to add')
rpc.add_function(add_func)

server = reflectrpc.simpleserver.SimpleJsonRpcServer(rpc, 'localhost', 5500)
server.run()

Connect to the server:

rpcsh localhost 5500

rpcsh

Now you can get a list of RPC functions available on the server:

List remote functions

You can take a look at the documentation of a function and its parameters:

Show documentation of remote function

You can call it from rpcsh:

Execute remote function

Or send a literal JSON-RPC request to the server:

Send raw JSON-RPC request to server

To get an overview of what rpcsh can do just type help:

Help

Installation

ReflectRPC is available in the Python Package Index. Therefore you can easily install it with a single command:

pip install reflectrpc

Features

  • JSON-RPC 1.0 (it doesn't get any more simple than that)
  • Registration and documentation of RPC calls is done in one place
  • Type checking
  • Special RPC calls allow to get descriptions of the service, available functions, and custom types
  • Interactive shell (rpcsh) to explore an RPC service and call its functions
  • Baseclass for exceptions that are to be serialized and replied to the caller while all other exceptions are suppressed as internal errors
  • Custom types enum and named hashes (like structs in C)
  • Protocol implementation is easily reusable in custom servers
  • Twisted-based server that supports TCP and UNIX Domain Sockets, line-based plain sockets, HTTP, HTTP Basic Auth, TLS, and TLS client auth
  • Client that supports TCP and UNIX Domain Sockets, line-based plain sockets, HTTP, HTTP Basic Auth, TLS, and TLS client auth
  • Create HTML documentation from a running RPC service by using the program rpcdoc
  • Create documented client code from a running RPC service with the program rpcgencode

Datatypes

ReflectRPC supports the following basic datatypes:

Type Description
bool true or false
int integer number
float floating point number
string string
array JSON array with arbitrary content
hash JSON hash with arbitrary content
base64 Base64 encoded binary data
array<type> Typed array. Only elements of the given type are allowed. E.g. array<int>, array<string> etc. Custom types are also supported as elements.

Custom Datatypes

There are two types of custom datatypes you can define: Enums and named hashes. For that you have to create an instance of the class JsonEnumType or JsonHashType, respectively. This object is filled similarly to RpcProcessor and then registered to your RpcProcessor by calling the add_custom_type method.

But lets look at an example:

phone_type_enum = reflectrpc.JsonEnumType('PhoneType', 'Type of a phone number')
phone_type_enum.add_value('HOME', 'Home phone')
phone_type_enum.add_value('WORK', 'Work phone')
phone_type_enum.add_value('MOBILE', 'Mobile phone')
phone_type_enum.add_value('FAX', 'FAX number')

address_hash = reflectrpc.JsonHashType('Address', 'Street address')
address_hash.add_field('firstname', 'string', 'First name')
address_hash.add_field('lastname', 'string', 'Last name')
address_hash.add_field('street1', 'string', 'First address line')
address_hash.add_field('street2', 'string', 'Second address line')
address_hash.add_field('zipcode', 'string', 'Zip code')
address_hash.add_field('city', 'string', 'City')

rpc = reflectrpc.RpcProcessor()
rpc.add_custom_type(phone_type_enum)
rpc.add_custom_type(address_hash)

This creates an enum named PhoneType and a named hash type to hold street addresses which is named Address and registers them to an RpcProcessor. These new types can now be used with all RPC functions that are to be added to this RpcProcessor simply by using their instead of one of the basic datatype names. All custom type names have to start with an upper-case letter.

Custom types can be inspected in rpcsh with the type command:

Inspecting custom datatypes in rpcsh

Returning Errors

A common problem when writing RPC services is returning errors to the user. On the one hand you want to report as much information about a problem to the user to make life as easy as possible for him. On the other hand you have to hide internal errors for security reasons and only make errors produced by the client visible outside because otherwise you make life easy for people who want to break into your server.

Therefore when an RPC function is called ReflectRPC catches all exceptions and returns only a generic "internal error" in the JSON-RPC reply. To return more information about an error to the user you can derive custom exception classes from JsonRpcError. All exceptions that are of this class or a subclass are serialized and returned to the client.

This allows to serialize exceptions and return them to the user but at the same time gives you fine-grained control over what error information actually leaves the server.

Example

We can define two RPC functions named internal_error() and json_error() to demonstrate this behaviour. The first function raises a ValueError. Internal exceptions like this must not be visible to the client. The function json_error() on the other hand raises an exception of type JsonRpcError. Since this exception is specially defined for the sole purpose of being returned to the client it will be serialized and returned as a JSON-RPC error object.

def internal_error():
    raise ValueError("This should not be visible to the client")

def json_error():
    raise reflectrpc.JsonRpcError("User-visible error")

rpc = reflectrpc.RpcProcessor()

error_func1 = reflectrpc.RpcFunction(internal_error, 'internal_error', 'Produces internal error',
        'bool', '')
error_func2 = reflectrpc.RpcFunction(json_error, 'json_error', 'Raises JsonRpcError',
        'bool', '')

rpc.add_function(error_func1)
rpc.add_function(error_func2)

Now a call to internal_error() will yield the following response from the server:

{"result": null, "error": {"name": "InternalError", "message": "Internal error"}, "id": 1}

While the result of json_error() will look like this:

{"result": null, "error": {"name": "JsonRpcError", "message": "User error"}, "id": 2}

Both results are as expected. You can send back your own errors over JSON-RPC in a controlled manner but internal errors are hidden from the client.

Serving RPCs

When you build an RPC service you want to serve it over a network of course. To make this as easy as possible ReflectRPC already comes with two different server implementations. The first one is named SimpleJsonRpcServer and if you've read the first example section of this document you've already seen some example code. SimpleJsonRpcServer is a very simple server that serves JSON-RPC requests over a plain TCP socket, with each JSON message being delimited by a linebreak.

That's how it is used:

import reflectrpc
import reflectrpc.simpleserver

# create an RpcProcessor object and register your functions
...

server = reflectrpc.simpleserver.SimpleJsonRpcServer(rpc, 'localhost', 5500)
server.run()

Since this server only handles one client at a time you only want to use it for testing purposes. For production use there is a concurrent server implementation that is also much more feature rich. It is based on the Twisted framework.

The following example creates a TwistedJsonRpcServer that behaves exactly as the SimpleJsonRpcServer and serves line-delimited JSON-RPC messages over a plain TCP socket:

import reflectrpc
import reflectrpc.twistedserver

# create an RpcProcessor object and register your functions
...

server = reflectrpc.twistedserver.TwistedJsonRpcServer(rpc, 'localhost', 5500)
server.run()

Of course it is powered by Twisted and can handle more than one connection at a time. This server also support TLS encryption, TLS client authentication and HTTP as an alternative to line-delimited messages.

The following example code creates a TwistedJsonRpcServer that serves JSON-RPC over HTTP (JSON-RPC message are to be sent as POST requests to '/rpc'). The connection is encrypted with TLS and the client has to present a valid certificate that is signed by the CA certificate in the file clientCA.crt:

import reflectrpc
import reflectrpc.twistedserver

# create an RpcProcessor object and register your functions
...

jsonrpc = rpcexample.build_example_rpcservice()
server = reflectrpc.twistedserver.TwistedJsonRpcServer(jsonrpc, 'localhost', 5500)
server.enable_tls('server.pem')
server.enable_client_auth('clientCA.crt')
server.enable_http()
server.run()

Custom Servers

If you have custom requirements and want to write your own server that is no problem at all. All you have to do is pass the request string you receive from your client to the process_request method of an RpcProcessor object. It will the reply as a dictionary or None in case of a JSON-RPC notification. If you get a dictionary you encode it as JSON and send it back to the client.

# create an RpcProcessor object and register your functions
...

reply = rpc.process_request(line)

# in case of a notification request process_request returns None
# and we send no reply back
if reply:
    reply_line = json.dumps(reply)
    send_data(reply_line.encode("utf-8"))

Authentication

Some protocols like e.g. TLS with client authentication allow to authenticate the client. Normally, your RPC functions have no idea about in what context they are called so they also know nothing about authentication. You can change this by calling the method require_rpcinfo on your RpcFunction object. Your function will then be called with a Python dict called rpcinfo as its first parameter which provides your RPC function with some context information:

def whoami(rpcinfo):
    if rpcinfo['authenticated']:
        return 'Username: ' + rpcinfo['username']

    return 'Not logged in'

func = RpcFunction(whoami, 'whoami', 'Returns login information',
        'string', 'Login information')
func.require_rpcinfo()

Of course your function has to declare an additional parameter for the rpcinfo dict.

You can also use rpcinfo in a custom server to pass your own context information. Just call process_request with your custom rpcinfo dict as a second parameter:

rpcinfo = {
    'authenticated': False,
    'username': None,
    'mydata': 'SOMEUSERDATA'
}

reply = rpc.process_request(line, rpcinfo)

This dict will then be passed to every RPC function that declared that it wants to get the rpcinfo dict while all other RPC functions will know nothing about it.

Generating Documentation

To generate HTML documentation for a running service just call rpcdoc from the commandline and tell it which server to connect to and where to write its output:

rpcdoc localhost 5500 doc.html

It will output some formatted HTML documentation for your service:

HTML Documentation

Generating Client Code

It is nice to have a generic JSON-RPC client like the one in reflectrpc.client.RpcClient. But it is even nicer to have a client library that is specifically made for your particular service. Such a client library should expose all the RPC calls of your service and have docstrings with the description of your functions and their parameters, as well as the typing information.

Such a client can be generated with the following command:

rpcgencode localhost 5500 client.py

And it will look something like this:

Generated Client

Supported Python Versions

ReflectRPC supports the following Python versions:

  • CPython 2.7
  • CPython 3.3
  • CPython 3.4
  • CPython 3.5

Current versions of PyPy should also work.

License

ReflectRPC is licensed under the MIT license

How to Contribute

Pull requests are always welcome.

If you create a pull request for this project you agree that your code will be released under the terms of the MIT license.

Ideas for improvements can be found in the TODO file.

Contact

Andreas Heck <[email protected]>

You might also like...
Regression Metrics Calculation Made easy for tensorflow2 and scikit-learn

Regression Metrics Installation To install the package from the PyPi repository you can execute the following command: pip install regressionmetrics I

Deep Learning with PyTorch made easy 🚀 !

Deep Learning with PyTorch made easy 🚀 ! Carefree? carefree-learn aims to provide CAREFREE usages for both users and developers. It also provides a c

Locally Enhanced Self-Attention: Rethinking Self-Attention as Local and Context Terms
Locally Enhanced Self-Attention: Rethinking Self-Attention as Local and Context Terms

LESA Introduction This repository contains the official implementation of Locally Enhanced Self-Attention: Rethinking Self-Attention as Local and Cont

This repository contains the code for "Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP".

Self-Diagnosis and Self-Debiasing This repository contains the source code for Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based

The Self-Supervised Learner can be used to train a classifier with fewer labeled examples needed using self-supervised learning.
The Self-Supervised Learner can be used to train a classifier with fewer labeled examples needed using self-supervised learning.

Published by SpaceML • About SpaceML • Quick Colab Example Self-Supervised Learner The Self-Supervised Learner can be used to train a classifier with

Automatic self-diagnosis program (python required)Automatic self-diagnosis program (python required)

auto-self-checker 자동으로 자가진단 해주는 프로그램(python 필요) 중요 이 프로그램이 실행될때에는 절대로 마우스포인터를 움직이거나 키보드를 건드리면 안된다(화면인식, 마우스포인터로 직접 클릭) 사용법 프로그램을 구동할 폴더 내의 cmd창에서 pip

Json2Xml tool will help you convert from json COCO format to VOC xml format in Object Detection Problem.

JSON 2 XML All codes assume running from root directory. Please update the sys path at the beginning of the codes before running. Over View Json2Xml t

Import Python modules from dicts and JSON formatted documents.

Paker Paker is module for importing Python packages/modules from dictionaries and JSON formatted documents. It was inspired by httpimporter. Important

Fast image augmentation library and easy to use wrapper around other libraries. Documentation:  https://albumentations.ai/docs/ Paper about library: https://www.mdpi.com/2078-2489/11/2/125
Fast image augmentation library and easy to use wrapper around other libraries. Documentation: https://albumentations.ai/docs/ Paper about library: https://www.mdpi.com/2078-2489/11/2/125

Albumentations Albumentations is a Python library for image augmentation. Image augmentation is used in deep learning and computer vision tasks to inc

Releases(v0.7.6)
Owner
Andreas Heck
Andreas Heck
The deployment framework aims to provide a simple, lightweight, fast integrated, pipelined deployment framework that ensures reliability, high concurrency and scalability of services.

savior是一个能够进行快速集成算法模块并支持高性能部署的轻量开发框架。能够帮助将团队进行快速想法验证(PoC),避免重复的去github上找模型然后复现模型;能够帮助团队将功能进行流程拆解,很方便的提高分布式执行效率;能够有效减少代码冗余,减少不必要负担。

Tao Luo 125 Dec 22, 2022
banditml is a lightweight contextual bandit & reinforcement learning library designed to be used in production Python services.

banditml is a lightweight contextual bandit & reinforcement learning library designed to be used in production Python services. This library is developed by Bandit ML and ex-authors of Facebook's applied reinforcement learning platform, Reagent.

Bandit ML 51 Dec 22, 2022
Rayvens makes it possible for data scientists to access hundreds of data services within Ray with little effort.

Rayvens augments Ray with events. With Rayvens, Ray applications can subscribe to event streams, process and produce events. Rayvens leverages Apache

CodeFlare 32 Dec 25, 2022
Public scripts, services, and configuration for running a smart home K3S network cluster

makerhouse_network Public scripts, services, and configuration for running MakerHouse's home network. This network supports: TODO features here For mo

Scott Martin 1 Jan 15, 2022
This program uses trial auth token of Azure Cognitive Services to do speech synthesis for you.

??️ aspeak A simple text-to-speech client using azure TTS API(trial). ?? TL;DR: This program uses trial auth token of Azure Cognitive Services to do s

Levi Zim 359 Jan 5, 2023
A Lighting Pytorch Framework for Recommendation System, Easy-to-use and Easy-to-extend.

Torch-RecHub A Lighting Pytorch Framework for Recommendation Models, Easy-to-use and Easy-to-extend. 安装 pip install torch-rechub 主要特性 scikit-learn风格易用

Mincai Lai 67 Jan 4, 2023
Model Serving Made Easy

The easiest way to build Machine Learning APIs BentoML makes moving trained ML models to production easy: Package models trained with any ML framework

BentoML 4.4k Jan 8, 2023
NLP made easy

GluonNLP: Your Choice of Deep Learning for NLP GluonNLP is a toolkit that helps you solve NLP problems. It provides easy-to-use tools that helps you l

Distributed (Deep) Machine Learning Community 2.5k Jan 4, 2023
InferPy: Deep Probabilistic Modeling with Tensorflow Made Easy

InferPy: Deep Probabilistic Modeling Made Easy InferPy is a high-level API for probabilistic modeling written in Python and capable of running on top

PGM-Lab 141 Oct 13, 2022
PyTorch deep learning projects made easy.

PyTorch Template Project PyTorch deep learning project made easy. PyTorch Template Project Requirements Features Folder Structure Usage Config file fo

Victor Huang 3.8k Jan 1, 2023