WARNING: this library is a prototype, under active development.
This project is a pure-Python gRPC implementation, based on hyper-h2 project, requires Python >= 3.5.
Motivation: grpclib
is intended to implement gRPC protocol in Python once
and natively support all concurrency models. However, currently grpclib
supports only asyncio
library and only with async/await
syntax, because
this is the most modern for Python stack, which was not supported yet.
Note: Python 2.7 support is not planned, but you can use official grpcio library for projects with such requirements.
Currently you can install grpclib
only from the GitHub:
$ pip3 install git+https://github.com/vmagamedov/grpclib.git
For the code generation you will also need a protoc
compiler, which can be
installed with protobuf
package:
$ brew install protobuf # example for macOS users
Or you can use protoc
compiler from the grpcio-tools
Python package:
$ pip3 install grpcio-tools
Note: grpcio
and grpcio-tools
packages are not required in
runtime, grpcio-tools
package will be used only during code generation.
In order to use this library you will have to generate special stub files using plugin provided, which can be used like this:
$ python3 -m grpc_tools.protoc -I. --python_out=. --python_grpc_out=. helloworld.proto
This command will generate helloworld_pb2.py
and helloworld_grpc.py
files.
Plugin, which implements --python_grpc_out
option is available for
protoc
compiler as protoc-gen-python_grpc
executable, which will be
installed by setuptools
into your PATH
during installation of the
grpclib
library.
See example
directory for a full example of the helloworld
service.
example/README.rst
contains instructions about how to generate
helloworld_pb2.py
and helloworld_grpc.py
files and how to run example.
Example basically looks like this:
import asyncio
from grpclib.server import Server
from grpclib.client import Channel
from .helloworld_pb2 import HelloRequest, HelloReply
from .helloworld_grpc import GreeterBase, GreeterStub
loop = asyncio.get_event_loop()
# Server
class Greeter(GreeterBase):
async def SayHello(self, stream):
request = await stream.recv_message()
message = 'Hello, {}!'.format(request.name)
await stream.send_message(HelloReply(message=message))
server = Server([Greeter()], loop=loop)
loop.run_until_complete(server.start('127.0.0.1', 50051))
# Client
channel = Channel(loop=loop)
stub = GreeterStub(channel)
async def make_request():
response = await stub.SayHello(HelloRequest(name='World'))
assert response.message == 'Hello, World!'
# Test request
loop.run_until_complete(make_request())
# Shutdown
server.close()
loop.run_until_complete(server.wait_closed())
loop.close()
Where helloworld.proto
contains:
syntax = "proto3";
package helloworld;
service Greeter {
rpc SayHello (HelloRequest) returns (HelloReply) {}
}
message HelloRequest {
string name = 1;
}
message HelloReply {
string message = 1;
}
0.2.0
- complete rewrite, pure-Python, based on hyper-h20.1.0
– workaround implemented, only server implementation available and only for unary calls