API

一个API做了两件事

• 客户端发起请求Request
• 服务端作出响应Response

REST是什么

REST（Representational State Transfer）：表现层状态传输，是一种设计风格，通常将 HTTP API 称为 RESTful API、RESTful 服务或 REST 服务

• 资源由URL决定
• 通过GET、POST、PUT、DELETE、PATCH、OPTIONS、HEAD、TRACE方法来操作资源
• 操作资源的表现形式是XML、HTML、JSON等格式
• 由客户端保存状态

RPC是什么

RPC（Remote Procedure Call）：远程过程调用，像调用本地函数一样调用远程函数，grpc是RPC的一种实现

Tonic是什么

tonic基于HTTP/2构建，grpc的Rust实现

Protocol Buffers

Protocol Buffers协议缓冲区

服务方法

允许定义四种服务方法

• 一元RPC：客户端向服务端发送单个请求并得到单个响应，就像调用普通函数一样
• 服务端流式RPC：客户端向服务端发送请求并获取流以读取一系列消息，客户端从返回的流中读取直到没有消息
• 客户端流式RPC：客户端使用流编写一系列消息发送到服务端，等待服务端做出响应
• 双向流式RPC：双方使用读写流发送一系列消息，两个流独立运行

Protocol Buffers语法

• 文件名以.proto结尾
• 类似于json，体积更小、速度更快，会生成本机语言绑定

message：定义数据传递格式

• 一个proto文件可以定义多个消息，消息可以嵌套

• required：proto2必填、proto3不需要填

• optional：可选字段

• repeate：可重复字段

• 标识号：每个字段必须要有一个唯一的标识号，范围1~2^29 -1（19000-19999保留标识号不能用）
  service：RPC服务接口

认证

• SSL/TLS认证
  • TLS（Transport layerSecurity）安全传输层，建立在TCP协议上，前身是SSL（Secure Socket Layer）安全套接字层，将应用层的报文进行加密后再交由TCP传输
• 基于Token认证
• 自定义认证

安装protobuf

https://github.com/protocolbuffers/protobuf/releases/tag/v28.2

解压后将路径加入PATH

验证

protoc --version

Tonic实战

• 需要安装protobuf
• 项目地址：https://github.com/VCCICCV/tonic-demo

创建项目

cargo new tonic-server

cargo new tonic-client

两个项目都添加依赖

[dependencies]
tonic = "0.12.2"
tokio = { version = "1.40.0", features = ["macros", "rt-multi-thread"] }
# grpc编码器
prost = "0.13.2"
# 我们只在构建的时候需要它
[build-dependencies]
tonic-build = "0.12.2"

在两个项目同级新建proto\order.proto

syntax = "proto3";// 指定proto版本
package order;// 指定包名
// 定义服务
service OrderService {
    rpc GetOrder (OrderRequest) returns (OrderResponse);
    rpc SetOrder (Order) returns (OrderResponse);
}
// 定义message类型
message OrderRequest {
    int32 id = 1;
}

message OrderResponse {
    int32 id = 1;
    string description = 2;
    double price = 3;
}

message Order {
    int32 id = 1;
    string description = 2;
    double price = 3;
}

两个项目根目录新建build.rs，用于编译客户端和服务端的代码

// 告诉tonic-build，需要编译哪些文件
fn main() -> Result<(), Box<dyn std::error::Error>> {
    tonic_build::compile_protos("../proto/order.proto")?;// 注意导入的是项目外也就是项目同级的文件夹
    Ok(())
}

此时两个项目编译

cargo build

生成的rust代码在target\debug\build\tonic-server-****\out\order.rs

server项目新建src\order_service.rs

use tonic::{ Request, Response, Status };
use order::order_service_server::{ OrderService, OrderServiceServer };
use order::{ OrderRequest, OrderResponse, Order };

pub mod order {
    tonic::include_proto!("order");// 使用宏引入proto文件
}

#[derive(Debug, Default)] // 实现打印、默认值
pub struct MyOrderService;

#[tonic::async_trait] // 异步方法
impl OrderService for MyOrderService {
    // 获取订单
    async fn get_order(
        &self,
        request: Request<OrderRequest>
    ) -> Result<Response<OrderResponse>, Status> {
        let req = request.into_inner();
        let response = OrderResponse {
            id: req.id,
            description: format!("Order {}", req.id),
            price: 100.0,
        };
        Ok(Response::new(response))
    }
    // 创建订单
    async fn set_order(&self, request: Request<Order>) -> Result<Response<OrderResponse>, Status> {
        let order = request.into_inner();
        let response = OrderResponse {
            id: order.id,
            description: order.description,
            price: order.price,
        };
        Ok(Response::new(response))
    }
}
// 启动grpc服务
pub fn create_server() -> OrderServiceServer<MyOrderService> {
    OrderServiceServer::new(MyOrderService::default())
}

server项目main.rs

mod order_service;
use order_service::create_server;
use tonic::transport::Server;
#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // 监听端口
    let addr = "[::1]:50051".parse()?;
    // 创建grpc服务
    let order_service = create_server();
    println!("OrderService listening on {}", addr);
    // 启动服务
    Server::builder().add_service(order_service).serve(addr).await?;
    Ok(())
}

此时启动server项目

cargo run

使用Apifox创建grpc请求

设置端口[::1]:50051

导入proto文件

此时调用即可响应请求

{
    "id": 0,
    "description": "Order 0",
    "price": 100
}

client项目main.rs

use order::order_service_client::OrderServiceClient;
use order::{ OrderRequest, Order };

pub mod order {
    tonic::include_proto!("order");
}
#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // 连接到gRPC服务端
    let mut client = OrderServiceClient::connect("http://[::1]:50051").await?;
    // 创建请求
    let request = tonic::Request::new(OrderRequest { id: 1 });
    // 发送请求并等待响应
    let response = client.get_order(request).await?;
    println!("RESPONSE={:?}", response);

    let order = Order {
        id: 1,
        description: "New Order".to_string(),
        price: 150.0,
    };
    // 创建请求
    let request = tonic::Request::new(order);
    // 发送请求并等待响应
    let response = client.set_order(request).await?;
    println!("RESPONSE={:?}", response);

    Ok(())
}

运行即可调用服务端的方法

cargo run

响应数据

RESPONSE=Response { metadata: MetadataMap { headers: {"content-type": "application/grpc", "date": "Fri, 20 Sep 2024 17:35:55 GMT", "grpc-status": "0"} }, message: OrderResponse { id: 1, description: "Order 1", price: 100.0 }, extensions: Extensions }
RESPONSE=Response { metadata: MetadataMap { headers: {"content-type": "application/grpc", "date": "Fri, 20 Sep 2024 tent-type": "application/grpc", "date": "Fri, 20 Sep 2024 17:35:55 GMT", "grpc-status": "0"} }, message: OrderResponse { id: 1, description: "Order 1", price: 100.0 }, extensions: Extensions }
RESPONSE=Response { metadata: MetadataMap { headers: {"content-type": "application/grpc", "date": "Fri, 20 Sep 2024 17:35:55 GMT", "grpc-status": "0"} }, message: OrderResponions: Extensions }
RESPONSE=Response { metadata: MetadataMap { headers: {"content-type": "application/grpc", "date": "Fri, 20 Sep 2024 17:35:55 GMT", "grpc-status": "0"} }, message: OrderRespontent-type": "application/grpc", "date": "Fri, 20 Sep 2024 17:35:55 GMT", "grpc-status": "0"} }, message: OrderResponse { id: 1, description: "New Order", price: 150.0 }, extensions: Extensions }

一元RPC、服务端流式RPC、客户端流式RPC、双向流式RPC

example.proto

syntax = "proto3";

package example;

service ExampleService {
  // Unary RPC
  rpc UnaryCall(RequestMessage) returns (ResponseMessage);

  // Server-side streaming RPC
  rpc ServerStream(RequestMessage) returns (stream ResponseMessage);

  // Client-side streaming RPC
  rpc ClientStream(stream RequestMessage) returns (ResponseMessage);

  // Bidirectional streaming RPC
  rpc BidiStream(stream RequestMessage) returns (stream ResponseMessage);
}

message RequestMessage {
  string message = 1;
}

message ResponseMessage {
  string message = 1;
}

server项目build.rs

// 告诉tonic-build，需要编译哪些文件
fn main() -> Result<(), Box<dyn std::error::Error>> {
    tonic_build::compile_protos("../proto/example.proto")?;// 注意导入的是项目外也就是项目同级的文件夹
    Ok(())
}

server项目cargo.toml

[dependencies]
tonic = "0.12.2"
tokio = { version = "1.40.0", features = ["full"] }
prost = "0.13.2"
futures-util = "0.3.30"
tokio-stream = "0.1.14"
tonic-reflection = "0.12.2"
[build-dependencies]
tonic-build = "0.12.2"

server项目main.rs

use tonic::{ transport::Server, Request, Response, Status };
use futures_util::Stream; // 使用 futures_util 提供的 Stream trait
use tokio_stream::wrappers::ReceiverStream; // 引入 tokio_stream
use tokio::sync::mpsc;
use std::pin::Pin;
use example::example_service_server::{ ExampleService, ExampleServiceServer };
use example::{ RequestMessage, ResponseMessage };

pub mod example {
    tonic::include_proto!("example");
}

#[derive(Default)]
pub struct MyExampleService {}

#[tonic::async_trait]
impl ExampleService for MyExampleService {
    //    1. 一元 RPC 调用
    async fn unary_call(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<ResponseMessage>, Status> {
        println!("一元调用");
        let message = request.into_inner().message;
        Ok(
            Response::new(ResponseMessage {
                message: format!("Hello from Unary: {}", message),
            })
        )
    }

    // 2. 服务端流式 RPC 调用
    type ServerStreamStream = Pin<Box<dyn Stream<Item = Result<ResponseMessage, Status>> + Send>>;

    async fn server_stream(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<Self::ServerStreamStream>, Status> {
        println!("服务端流式调用");
        let message = request.into_inner().message;

        let (tx, rx) = mpsc::channel(4);
        tokio::spawn(async move {
            for i in 1..=5 {
                tx.send(
                    Ok(ResponseMessage {
                        message: format!("Stream {}: {}", i, message),
                    })
                ).await.unwrap();
            }
            println!("流结束");
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx))))
    }

    // 3. 客户端流式 RPC 调用
    async fn client_stream(
        &self,
        request: Request<tonic::Streaming<RequestMessage>>
    ) -> Result<Response<ResponseMessage>, Status> {
        println!("客户端流式调用");
        let mut stream = request.into_inner();
        let mut messages = vec![];

        while let Some(req) = stream.message().await? {
            messages.push(req.message);
        }

        Ok(
            Response::new(ResponseMessage {
                message: format!("Received: {:?}", messages),
            })
        )
    }

    // 4. 双向流式 RPC 调用
    type BidiStreamStream = Pin<Box<dyn Stream<Item = Result<ResponseMessage, Status>> + Send>>;

    async fn bidi_stream(
        &self,
        request: Request<tonic::Streaming<RequestMessage>>
    ) -> Result<Response<Self::BidiStreamStream>, Status> {
        println!("双向流式调用");
        let mut stream = request.into_inner();
        let (tx, rx) = mpsc::channel(4);

        tokio::spawn(async move {
            while let Some(req) = stream.message().await.unwrap() {
                tx.send(
                    Ok(ResponseMessage {
                        message: format!("Echo: {}", req.message),
                    })
                ).await.unwrap();
            }
            println!("流结束");
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx))))
    }
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let addr = "[::1]:50051".parse()?;
    let example_service = MyExampleService::default();

    println!("Server listening on {}", addr);

    Server::builder().add_service(ExampleServiceServer::new(example_service)).serve(addr).await?;

    Ok(())
}

此时cargo build、cargo run即可用apifox发送rpc请求

注意我们更改了proto文件，需要重新指定

一元rpc

服务端流式 RPC

客户端流式 RPC

双向流式 RPC 调用

client项目main.rs

use tonic::Request;
use example::example_service_client::ExampleServiceClient;
use example::{ RequestMessage };

pub mod example {
    tonic::include_proto!("example");
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // 连接到服务端
    let mut client = ExampleServiceClient::connect("http://[::1]:50051").await?;

    // 1. 一元 RPC 调用
    let request = Request::new(RequestMessage {
        message: String::from("Unary Hello"),
    });
    let response = client.unary_call(request).await?;
    println!("Unary Response: {:?}", response.into_inner());

    // 2. 服务器流式 RPC 调用
    let request = Request::new(RequestMessage {
        message: String::from("Stream Hello"),
    });
    let mut stream = client.server_stream(request).await?.into_inner();
    while let Some(res) = stream.message().await? {
        println!("Server Stream Response: {:?}", res);
    }
    Ok(())
}

由于proto生成的rust方法在不同的位置，例如server生成的代码在server/target***下，客户端无法调用server提供的方法，我们需要将代码生成的目录设置为两个项目的公共目录

改造为gRPC微服务通信

• 反射（Reflection）：允许客户端在运行时发现 gRPC 服务所提供的方法、消息类型等信息，而无需在编译时就知道这些详细信息，但是它只提供了一种动态发现服务的能力，你可以用postman、apifox的反射调用而不需要导入proto文件，在编写客户端代码时仍然需要导入生成后的代码

反射仅仅是为了让客户端在运行时查询服务定义，而不是为了消除客户端对服务定义的依赖

项目结构

创建公共目录proto
proto\example.proto

syntax = "proto3";

package example;

service ExampleService {
  // Unary RPC
  rpc UnaryCall(RequestMessage) returns (ResponseMessage);

  // Server-side streaming RPC
  rpc ServerStream(RequestMessage) returns (stream ResponseMessage);

  // Client-side streaming RPC
  rpc ClientStream(stream RequestMessage) returns (ResponseMessage);

  // Bidirectional streaming RPC
  rpc BidiStream(stream RequestMessage) returns (stream ResponseMessage);
}

message RequestMessage {
  string message = 1;
}

message ResponseMessage {
  string message = 1;
}

两个项目的cargo.toml

[dependencies]
tonic = "0.12.2"
tokio = { version = "1.40.0", features = ["full"] }
prost = "0.13.3"
tokio-stream = "0.1.14"
tonic-reflection = "0.12.2"
[build-dependencies]
tonic-build = "0.12.2"

server\build.rs

use std::path::PathBuf;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let out_dir = PathBuf::from("../proto");
    // let out_dir = PathBuf::from(env::var("OUT_DIR").unwrap()); // 指定输出目录在target下
    tonic_build
        ::configure()
        .build_server(true)// 指定是否生成服务端代码
        .out_dir("src/generated")// 指定rust文件输出目录
        .file_descriptor_set_path(out_dir.join("example_descriptor.bin")) // 生成文件描述符集
        .compile(&["../proto/example.proto"], &["../proto"])
        .unwrap();
    Ok(())
}

client\build.rs

use std::path::PathBuf;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let out_dir = PathBuf::from("../proto");
    // let out_dir = PathBuf::from(env::var("OUT_DIR").unwrap()); // 指定输出目录在target下
    tonic_build
        ::configure()
        .build_client(true)// 指定是否生成客户端端代码
        .out_dir("src/generated")// 指定rust文件输出目录
        .file_descriptor_set_path(out_dir.join("example_descriptor.bin")) // 生成文件描述符集
        .compile(&["../proto/example.proto"], &["../proto"])
        .unwrap();
    Ok(())
}

• .build_client(true)：生成客户端Rust代码
• .build_server(true)：生成服务端Rust代码
• .out_dir(“src/generated”)：指定生成的Rust代码的目录，cargo build你会发现生成的example.rs内容是一样的，因为我们定义的请求与响应的proto文件内容是一样的
• .file_descriptor_set_path：生成文件描述符集，用于创建反射服务
• .compile(&[“…/proto/example.proto”], &[“…/proto”])：指定编译的proto文件及其目录

server\src\main.rs

mod server_lib;
#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    server_lib::start_server().await?;
    Ok(())
}

server\src\server_lib.rs 这里我们添加了反射

use tonic::{ Request, Response, Status, Streaming };
use tokio_stream::{ wrappers::ReceiverStream, StreamExt };
use std::pin::Pin;
use tokio::sync::mpsc;

pub mod example {
    // 1. 导入文件描述符集
    pub const FILE_DESCRIPTOR_SET: &[u8] = include_bytes!("../../proto/example_descriptor.bin");
    // tonic::include_proto!("example");// 你可以ctrl点击查看源码，它是一个宏，添加了env的编译目录，我们这里指定生成的rust目录，所以直接导入即可
    include!("./generated/example.rs");
}

use example::example_service_server::{ ExampleService, ExampleServiceServer };
use example::{ RequestMessage, ResponseMessage };

#[derive(Debug, Default)]
pub struct MyExampleService;

#[tonic::async_trait]
impl ExampleService for MyExampleService {
    // 一元RPC调用
    async fn unary_call(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<ResponseMessage>, Status> {
        let response = ResponseMessage {
            message: format!("Unary response to {}", request.into_inner().message),
        };
        Ok(Response::new(response))
    }

    type ServerStreamStream = Pin<
        Box<dyn tokio_stream::Stream<Item = Result<ResponseMessage, Status>> + Send>
    >;
    // 服务端流式RPC调用
    async fn server_stream(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<Self::ServerStreamStream>, Status> {
        let message = request.into_inner().message;
        let (tx, rx) = mpsc::channel(4);

        tokio::spawn(async move {
            for i in 0..3 {
                let response = ResponseMessage {
                    message: format!("Stream response {} to {}", i, message),
                };
                tx.send(Ok(response)).await.unwrap();
            }
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx)) as Self::ServerStreamStream))
    }
    // 客户端流式RPC调用
    async fn client_stream(
        &self,
        request: Request<Streaming<RequestMessage>>
    ) -> Result<Response<ResponseMessage>, Status> {
        let mut stream = request.into_inner();
        let mut messages = Vec::new();

        while let Some(req) = stream.next().await {
            let req = req?;
            messages.push(req.message);
        }

        let response = ResponseMessage {
            message: format!("Received messages: {:?}", messages.join(", ")),
        };
        Ok(Response::new(response))
    }

    type BidiStreamStream = Pin<
        Box<dyn tokio_stream::Stream<Item = Result<ResponseMessage, Status>> + Send>
    >;
    // 双向流式RPC调用
    async fn bidi_stream(
        &self,
        request: Request<Streaming<RequestMessage>>
    ) -> Result<Response<Self::BidiStreamStream>, Status> {
        let mut stream = request.into_inner();
        let (tx, rx) = mpsc::channel(4);

        tokio::spawn(async move {
            while let Some(req) = stream.next().await {
                let req = req.unwrap();
                let response = ResponseMessage {
                    message: format!("Echo: {}", req.message),
                };
                tx.send(Ok(response)).await.unwrap();
            }
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx)) as Self::BidiStreamStream))
    }
}

pub async fn start_server() -> Result<(), Box<dyn std::error::Error>> {
    let addr = "[::1]:50051".parse()?;
    let example_service = MyExampleService::default();

    println!("Server listening on {}", addr);
    // 2. 创建反射服务
    let reflection_service = tonic_reflection::server::Builder
        ::configure()
        .register_encoded_file_descriptor_set(example::FILE_DESCRIPTOR_SET)
        .build_v1()?;

    tonic::transport::Server
        ::builder()
        .add_service(ExampleServiceServer::new(example_service))
        .add_service(reflection_service) // 3. 添加反射服务
        .serve(addr).await?;

    Ok(())
}

tonic-client\src\main.rs你可以在生成的client\src\generated\example.rs查看client端提供的方法

use tonic::transport::Channel;
use example::example_service_client::ExampleServiceClient;
use example::RequestMessage;
use tokio_stream::StreamExt;

pub mod example {
    // tonic::include_proto!("example");
    include!("./generated/example.rs");
}
// 一元RPC调用
async fn unary_call(
    client: &mut ExampleServiceClient<Channel>
) -> Result<(), Box<dyn std::error::Error>> {
    let request = tonic::Request::new(RequestMessage {
        message: "Hello from unary".into(),
    });
    let response = client.unary_call(request).await?;
    println!("Unary response: {:?}", response.into_inner());
    Ok(())
}
// 服务端流式RPC调用
async fn server_stream(
    client: &mut ExampleServiceClient<Channel>
) -> Result<(), Box<dyn std::error::Error>> {
    let request = tonic::Request::new(RequestMessage {
        message: "Hello from stream".into(),
    });
    let mut response = client.server_stream(request).await?.into_inner();
    while let Some(res) = response.next().await {
        println!("Server stream response: {:?}", res?);
    }
    Ok(())
}
// 客户端流式RPC调用
async fn client_stream(
    client: &mut ExampleServiceClient<Channel>
) -> Result<(), Box<dyn std::error::Error>> {
    let request = tokio_stream::iter(
        vec![
            RequestMessage { message: "Hello".into() },
            RequestMessage { message: "from".into() },
            RequestMessage { message: "client".into() }
        ]
    );
    let response = client.client_stream(request).await?;
    println!("Client stream response: {:?}", response.into_inner());
    Ok(())
}
// 双向流式RPC调用
async fn bidi_stream(
    client: &mut ExampleServiceClient<Channel>
) -> Result<(), Box<dyn std::error::Error>> {
    let request = tokio_stream::iter(
        vec![
            RequestMessage { message: "Hello".into() },
            RequestMessage { message: "from".into() },
            RequestMessage { message: "bidi".into() }
        ]
    );
    let mut response = client.bidi_stream(request).await?.into_inner();
    while let Some(res) = response.next().await {
        println!("Bidi stream response: {:?}", res?);
    }
    Ok(())
}

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // 连接到gRPC服务器
    let mut client = ExampleServiceClient::connect("http://[::1]:50051").await?;

    unary_call(&mut client).await?;
    server_stream(&mut client).await?;
    client_stream(&mut client).await?;
    bidi_stream(&mut client).await?;

    Ok(())
}

此时cargo run两个服务即可看到响应信息，你也可以用postman、apifox请求

tonic超时处理、认证看官方例子就可以了https://github.com/hyperium/tonic/tree/master/examples/src

bin crate实现client\server（可选了解）

• bin：二进制可执行文件
• lib：库crate
  一般来说是用两个服务，两个服务间通信用rpc，bin crate的方法直接调用就好了，很多教程都是用的bin，包括官方例子也是用的bin

这里给出的是在main中启动服务端和客户端，可能你在嵌入式中使用gRPC？

如果你想作为bin使用gRPC，你可以在cargo.toml里添加[[bin]]指定crate，bin crate会编译为单独的二进制文件

[[bin]]
name="client"
path="src/example_client.rs"
[[bin]]
name="server"
path="src/example_server.rs"

在client、server编写main函数

然后cargo run --bin server、cargo run --bin client

cargo.toml

[dependencies]
tonic = "0.12.2"
tokio = { version = "1.40.0", features = ["full"] }
prost = "0.13.3"
tokio-stream = "0.1.14"
tonic-reflection = "0.12.2"
[build-dependencies]
tonic-build = "0.12.2"

../proto/example.proto

syntax = "proto3";

package example;

service ExampleService {
  // Unary RPC
  rpc UnaryCall(RequestMessage) returns (ResponseMessage);

  // Server-side streaming RPC
  rpc ServerStream(RequestMessage) returns (stream ResponseMessage);

  // Client-side streaming RPC
  rpc ClientStream(stream RequestMessage) returns (ResponseMessage);

  // Bidirectional streaming RPC
  rpc BidiStream(stream RequestMessage) returns (stream ResponseMessage);
}

message RequestMessage {
  string message = 1;
}

message ResponseMessage {
  string message = 1;
}

build.rs

fn main() -> Result<(), Box<dyn std::error::Error>> {
    tonic_build
        ::configure()
        .compile(&["../proto/example.proto"], &["../proto"])
        .unwrap();
    Ok(())
}

src/example_client.rs

use example::example_service_client::ExampleServiceClient;
use example::RequestMessage;
use tokio_stream::StreamExt;

pub mod example {
    tonic::include_proto!("example");
}
// 一元RPC
pub async fn unary_call() -> Result<(), Box<dyn std::error::Error>> {
    let mut client = ExampleServiceClient::connect("http://[::1]:50051").await?;
    let request = tonic::Request::new(RequestMessage {
        message: "Hello from unary".into(),
    });
    let response = client.unary_call(request).await?;
    println!("Unary response: {:?}", response.into_inner());
    Ok(())
}
// 服务端流式RPC
pub async fn server_stream() -> Result<(), Box<dyn std::error::Error>> {
    let mut client = ExampleServiceClient::connect("http://[::1]:50051").await?;
    let request = tonic::Request::new(RequestMessage {
        message: "Hello from stream".into(),
    });
    let mut response = client.server_stream(request).await?.into_inner();
    while let Some(res) = response.next().await {
        println!("Server stream response: {:?}", res?);
    }
    Ok(())
}
// 客户端流式RPC
pub async fn client_stream() -> Result<(), Box<dyn std::error::Error>> {
    let mut client_stream = ExampleServiceClient::connect("http://[::1]:50051").await?;
    let request = tokio_stream::iter(
        vec![
            RequestMessage { message: "Hello".into() }, 
            RequestMessage { message: "from".into() },
            RequestMessage { message: "client".into() }
        ]
    );
    let response = client_stream.client_stream(request).await?;
    println!("Client stream response: {:?}", response.into_inner());
    Ok(())
}
// 双向流式RPC
pub async fn bidi_stream() -> Result<(), Box<dyn std::error::Error>> {
    let mut client = ExampleServiceClient::connect("http://[::1]:50051").await?;
    let request = tokio_stream::iter(
        vec![
            RequestMessage { message: "Hello".into() },
            RequestMessage { message: "from".into() },
            RequestMessage { message: "bidi".into() }
        ]
    );
    let mut response = client.bidi_stream(request).await?.into_inner();
    while let Some(res) = response.next().await {
        println!("Bidi stream response: {:?}", res?);
    }
    Ok(())
}

src/example_server.rs

use tonic::{ transport::Server, Request, Response, Status, Streaming };
use tokio_stream::{ wrappers::ReceiverStream, StreamExt };
use std::pin::Pin;
use tokio::sync::mpsc;

pub mod example {
    tonic::include_proto!("example");
}

use example::example_service_server::{ ExampleService, ExampleServiceServer };
use example::{ RequestMessage, ResponseMessage };

#[derive(Debug, Default)]
pub struct MyExampleService;

#[tonic::async_trait]
impl ExampleService for MyExampleService {
    async fn unary_call(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<ResponseMessage>, Status> {
        let response = ResponseMessage {
            message: format!("Unary response to {}", request.into_inner().message),
        };
        Ok(Response::new(response))
    }

    type ServerStreamStream = Pin<
        Box<dyn tokio_stream::Stream<Item = Result<ResponseMessage, Status>> + Send>
    >;

    async fn server_stream(
        &self,
        request: Request<RequestMessage>
    ) -> Result<Response<Self::ServerStreamStream>, Status> {
        let message = request.into_inner().message;
        let (tx, rx) = mpsc::channel(4);

        tokio::spawn(async move {
            for i in 0..3 {
                let response = ResponseMessage {
                    message: format!("Stream response {} to {}", i, message),
                };
                tx.send(Ok(response)).await.unwrap();
            }
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx)) as Self::ServerStreamStream))
    }

    async fn client_stream(
        &self,
        request: Request<Streaming<RequestMessage>>
    ) -> Result<Response<ResponseMessage>, Status> {
        let mut stream = request.into_inner();
        let mut messages = Vec::new();

        while let Some(req) = stream.next().await {
            let req = req?;
            messages.push(req.message);
        }

        let response = ResponseMessage {
            message: format!("Received messages: {:?}", messages.join(", ")),
        };
        Ok(Response::new(response))
    }

    type BidiStreamStream = Pin<
        Box<dyn tokio_stream::Stream<Item = Result<ResponseMessage, Status>> + Send>
    >;

    async fn bidi_stream(
        &self,
        request: Request<Streaming<RequestMessage>>
    ) -> Result<Response<Self::BidiStreamStream>, Status> {
        let mut stream = request.into_inner();
        let (tx, rx) = mpsc::channel(4);

        tokio::spawn(async move {
            while let Some(req) = stream.next().await {
                let req = req.unwrap();
                let response = ResponseMessage {
                    message: format!("Echo: {}", req.message),
                };
                tx.send(Ok(response)).await.unwrap();
            }
        });

        Ok(Response::new(Box::pin(ReceiverStream::new(rx)) as Self::BidiStreamStream))
    }
}

pub async fn start_server() -> Result<(), Box<dyn std::error::Error>> {
    let addr = "[::1]:50051".parse()?;
    let example_service = MyExampleService::default();

    println!("Server listening on {}", addr);

    Server::builder().add_service(ExampleServiceServer::new(example_service)).serve(addr).await?;

    Ok(())
}

main.rs

use tokio::task;
mod example_client;
mod example_server;
use std::error::Error;

#[tokio::main]
async fn main() -> Result<(), Box<dyn Error>> {
    // 启动服务端 
    task::spawn(async {
        example_server::start_server().await.unwrap();
    });

    // 延时
    tokio::time::sleep(tokio::time::Duration::from_secs(1)).await;

    // 调用客户端
    example_client::unary_call().await?;
    example_client::server_stream().await?;
    example_client::client_stream().await?;
    example_client::bidi_stream().await?;

    Ok(())
}

cargo build然后cargo run即可看到调用结果

Unary response: ResponseMessage { message: "Unary response to Hello from unary" }
Server stream response: ResponseMessage { message: "Stream response 0 to Hello from stream" }
Server stream response: ResponseMessage { message: "Stream response 1 to Hello from stream" }
Server stream response: ResponseMessage { message: "Stream response 2 to Hello from stream" }
Client stream response: ResponseMessage { message: "Received messages: \"Hello, from, client\"" }
Bidi stream response: ResponseMessage { message: "Echo: Hello" }       
Bidi stream response: ResponseMessage { message: "Echo: from" }        
Bidi stream response: ResponseMessage { message: "Echo: bidi" }