我个人对GRPC是比较感兴趣的，最近在玩通过前端调用GRPC。通过前端调用GRPC业界有两种方式：GRPC Web和GRPC JSON转码。

GRPC Web

通过JS或者Blazor WASM调用GRPC，微软在这方面做的还是很好的，从.NET Core3.0之后就提供了两种实现GRPC Web的方式（Grpc.AspNetCore.Web与Envoy）。我在之前的一篇里也写过如何通过Blazor WASM调用GRPC Web。

GRPC JSON

通过Restful api调用一个代理服务，代理服务将数据转发到GRPC Server就是GRPC JSON。微软从.NET7开始也正式提供了GRPC JSON转码的方式。

既然有了GRPC Web与GRPC Json，那我为啥还要再造这么一个轮子？

原因是有位同行看了如何通过Blazor WASM调用GRPC Web 这篇文章后，告诉我微信小程序目前没办法通过这种方式调用GRPC。我当时觉得很奇怪，微信小程序也属于前端，为啥不能调用GRPC呢？

GRPC Web+小程序遇到的问题

只是听说还不能确认，要自己试一试，于是我用GRPC Web的方式让小程序调用GRPC，首先需要生成GRPC JS Client代码：

protoc.exe -I=. test.proto --js_out=import_style=commonjs:.\grpcjs\ --plugin=protoc-gen-grpc=.\protoc-gen-grpc-web.exe --grpc-web_out=import_style=commonjs,mode=grpcwebtext:.\grpcjs\

然后将生成的代码引入小程序端，发现确实有问题，微信小程序编译后无法正常识别GRPC的namespace，会报以下错误：

proto is not defined

去查了下原因，应该是因为小程序目前不支持protobuf序列化。然后我通过一种取巧的方式手动在生成的GRPC JS中添加了proto变量

var proto = {}

再次尝试，虽然proto能找到，但是又找不到其他对象，并且最主要的是GRPC JS Client是通过proto工具生成的，每次生成手动定义proto变量也不现实。

GRPC Web+小程序遇到问题总结：

1. 小程序目前不支持protobuf序列化
2. 手动修改生成的GRPC JS Client不友好

既然小程序通过GRPC Web方式调用GRPC失败，那还有GRPC Json。

GRPC JSON+Envoy+小程序遇到的问题

我使用了Envoy来充当restful代理，调用GRPC。我在之前有一篇通过Envoy JSON代理GRPC的帖子。按这个帖子来了一遍。

计划通过docker-compose方式运行GRPC Server和Envoy代理。

既然用GRPC，那肯定用http2/http2，在docker里运行.net core必然需要证书，没有证书就自己搞一个自签证书。

openssl req -newkey rsa:2048 -nodes -keyout server.key -x509 -days 365 -out server.cer

openssl pkcs12 -export -in server.cer -inkey server.key -out server.pfx

证书有了，在GRPC里配置https

builder.WebHost.ConfigureKestrel(o =>  
{

    o.ListenAnyIP(1111, p =>  
    {  
        p.Protocols = HttpProtocols.Http2;  
        p.UseHttps("/app/server.pfx", "123456");  
    });  
});

然后就开始配置envoy

首先生成grpc proto描述符

protoc.exe -I=. --descriptor_set_out=.\test.pb --include_imports .\test.proto --proto_path=.

然后定义envoy配置文件

admin:
address:
socket_address: {address: 0.0.0.0, port_value: 9901}

static_resources:
listeners:

• name: listener1 address: socket_address: {address: 0.0.0.0, port_value: 10000} filter_chains:
  • filters:
    • name: envoy.filters.network.http_connection_manager typed_config: "@type": type.googleapis.com/envoy.extensions.filters.network.http_connection_manager.v3.HttpConnectionManager stat_prefix: grpc_json codec_type: AUTO route_config: name: local_route virtual_hosts:
      • name: local_service domains: ["*"] routes:
        • match: {prefix: "/test"}
          route:
          cluster: grpc
          http_filters:
      • name: envoy.filters.http.grpc_json_transcoder
        typed_config:
        "@type": type.googleapis.com/envoy.extensions.filters.http.grpc_json_transcoder.v3.GrpcJsonTranscoder
        proto_descriptor: "/etc/envoy/test.pb"
        services: ["test"]
        print_options:
        add_whitespace: true
        always_print_primitive_fields: true
        always_print_enums_as_ints: false
        preserve_proto_field_names: false
        auto_mapping: true
      • name: envoy.filters.http.router
        typed_config:
        "@type": type.googleapis.com/envoy.extensions.filters.http.router.v3.Router
        clusters:
• name: grpc type: static connect_timeout: 15s lb_policy: ROUND_ROBIN dns_lookup_family: V4_ONLY typed_extension_protocol_options: envoy.extensions.upstreams.http.v3.HttpProtocolOptions: "@type": type.googleapis.com/envoy.extensions.upstreams.http.v3.HttpProtocolOptions explicit_http_config: http2_protocol_options: {} load_assignment: cluster_name: grpc endpoints:
  • lb_endpoints:
    • endpoint:
      address:
      socket_address:
      address: 某ip
      port_value: 1111

下面就定义envoy的dockerfile，主要是信任自签证书

#See https://aka.ms/containerfastmode to understand how Visual Studio uses this Dockerfile to build your images for faster debugging.

FROM envoyproxy/envoy-dev:e834c24e061b710348ffd72016d5d1069698b4ff

COPY ["server.crt","/usr/local/share/ca-certificates/"]

RUN ["update-ca-certificates"]

最后就是定义docker-compsoe.yaml

version: '3.4'

services:
myenvoy:
image: myenvoy
container_name: myenvoy
command: "-c /etc/envoy/envoy.yaml --log-level debug"
build:
context: .
dockerfile: GrpcServer/DockerfileEnvoy
volumes:
- "grpcpbs/:/etc/envoy/"
- "grpcpbs/logs:/logs"
ports:
- "9901:9901"
- "10000:10000"
depends_on:
- grpcserver
networks:
- mynetwork
grpcserver:
image: grpcserver
container_name: grpcserver
networks:
- mynetwork
build:
context: .
dockerfile: GrpcServer/Dockerfile
ports:
- "1111:1111"

networks:
mynetwork:

最后通过docker-compsoe up -d运行，但是postman调用的时候，envoy与grpcserver的通信连接成功了，但是数据传输时总是被 connection reset，去github上找原因也没找到。至此grpc json+envoy又失败了。

GRPC JSON+Envoy+小程序遇到问题总结：

1. 数据传输时connection 被莫名reset

既然envoy走不通不行，那就自己造一个吧。

开始造轮子

GRPC JSON的形式，原理就是通过一个web api接收restful请求，将请求数据转发到GRPC Server。

首先创建一个web api命名为GrpcGateway，并引入proto文件，生成grpc client代码

runtime; build; native; contentfiles; analyzers; buildtransitive all

然后创建一个控制器去接受restful请求，而grpc client可采用反射来创建。

[ApiController]
[Route("[controller]")]
public class ProcessGrpcRequestController : ControllerBase
{
private readonly ILogger _logger;
private readonly Func _getGrpcClient;
public ProcessGrpcRequestController(ILogger logger, Func getGrpcClient)
{
_logger = logger;
_getGrpcClient = getGrpcClient;
}

    /// <summary>  
    /// 调用grpc  
    /// </summary>  
    /// <param name="serviceName">Grpc Service Name 从proto文件中查询</param>  
    /// <param name="method">Grpc Method Name 从proto文件中查询</param>  
    /// <returns></returns>  
    \[HttpPost("serviceName/{serviceName}/method/{method}")\]  
    public async Task<IActionResult> ProcessAsync(string serviceName, string method)  
    {  
        try  
        {  
            if (string.IsNullOrEmpty(serviceName))  
            {  
                return BadRequest("serviceName不能为空");  
            }  
            if (string.IsNullOrEmpty(method))  
            {  
                return BadRequest("method不能为空");  
            }

            using var sr = new StreamReader(Request.Body, leaveOpen: true, encoding: Encoding.UTF8);  
            var paramJson = await sr.ReadToEndAsync();  
            if (string.IsNullOrEmpty(paramJson))  
            {  
                return BadRequest("参数不能为空");  
            }  
            var client = \_getGrpcClient(serviceName);  
            if (client == null)  
            {  
                return NotFound();  
            }

            Type t = client.GetType();  
            var processMethod = t.GetMethods().Where(e => e.Name == method).FirstOrDefault();  
            if (processMethod == null)  
            {  
                return NotFound();  
            }

            var parameters = processMethod.GetParameters();  
            if (parameters == null)  
            {  
                return NotFound();  
            }

            var param = JsonConvert.DeserializeObject(paramJson, parameters\[0\].ParameterType);  
            if (param == null)  
            {  
                return BadRequest("参数不能为空");  
            }  
            var pt = param.GetType();  
            var headers = new Metadata();

            if (Request.Headers.Keys.Contains("Authorization"))  
            {  
                headers.Add("Authorization", Request.Headers\["Authorization"\]);  
            }

            var result = processMethod.Invoke(client, new object\[\] { param, headers, null, null });

            return Ok(result);  
        }  
        catch(Exception ex) when (  
        ex.InnerException !=null && ex.InnerException !=null && ex.InnerException is RpcException &&  
        ((ex.InnerException as RpcException).StatusCode == Grpc.Core.StatusCode.Unauthenticated ||  
        ((ex.InnerException as RpcException).StatusCode == Grpc.Core.StatusCode.PermissionDenied)))  
        {  
            \_logger.LogError(ex, ex.ToString());  
            return Unauthorized();  
        }  
        catch (Exception ex)  
        {  
            \_logger.LogError(ex, ex.Message);  
            return BadRequest(ex.ToString());  
        }

    }  
}

然后注入动态反射创建grpc client的方法

        services.AddScoped(p => {  
            Func<string, ClientBase> func = serviceName =>  
            {  
                var channel = GrpcChannel.ForAddress(grpcServerAddress);

                var parentClassName = $"{serviceName}";

                var assembly = Assembly.Load("你的dll名字");  
                var parentType = assembly.GetType(parentClassName);

                var clientType=  parentType.GetNestedType($"{serviceName}Client");  
                if (clientType == null)  
                {  
                    throw new Exception($"serviceName:{serviceName}不存在");  
                }  
                var client = Activator.CreateInstance(clientType, new object\[\] { channel });  
                return (ClientBase)client;  
            };  
            return func;  
        });

然后定义grpc gateway dockerfile ，最主要需要信任证书

#See https://aka.ms/containerfastmode to understand how Visual Studio uses this Dockerfile to build your images for faster debugging.

FROM mcr.microsoft.com/dotnet/aspnet:6.0 AS base
WORKDIR /app
EXPOSE 16666

FROM mcr.microsoft.com/dotnet/sdk:6.0 AS build
WORKDIR /src
COPY ["MyGateway/MyGateway.csproj", "MyGateway/"]
COPY . .
WORKDIR "/src/MyGateway"

FROM build AS publish
RUN dotnet publish "MyGateway.csproj" -c Release -o /app/publish

FROM base AS final
COPY ["server.crt","/usr/local/share/ca-certificates/"]
RUN ["update-ca-certificates"]
WORKDIR /app
COPY --from=publish /app/publish .
ENTRYPOINT ["dotnet", "MyGateway.dll"]

最后通过定义docker-compose

version: '3.4'

services:
mygateway:
image: mygateway
container_name: mygateway
networks:
- mynetwork
build:
context: .
dockerfile: MyGateway/Dockerfile
ports:
- "2222:2222"
grpcserver:
image: grpcserver
container_name: grpcserver
networks:
- mynetwork
build:
context: .
dockerfile: GrpcServer/Dockerfile
ports:
- "1111:1111"

networks:
mynetwork:

通过docker-compsoe up -d 启动

通过postman调用，看到200状态码，终于成功了，最后试了下小程序也能通过这种方式调用后端GRPC了，整个人都舒服了…