.net core grpc单元测试 - 服务器端
前言
gRPC凭借其严谨的接口定义、高效的传输效率、多样的调用方式等优点,在微服务开发方面占据了一席之地。dotnet core正式支持gRPC也有一段时间了,官方文档也对如何使用gRPC进行了比较详细的说明,但是关于如何对gRPC的服务器和客户端进行单元测试,却没有描述。经过查阅官方代码,找到了一些解决方法,总结在此,供大家参考。
本文重点介绍gRPC服务器端代码的单元测试,包括普通调用、服务器端流、客户端流等调用方式的单元测试,另外,引入sqlite的内存数据库模式,对数据库相关操作进行测试。
准备gRPC服务端项目
使用dotnet new grpc命令创建一个gRPC服务器项目。
修改protos/greeter.proto, 添加两个接口方法:
- //服务器流
- rpc SayHellos (HelloRequest) returns (stream HelloReply);
- //客户端流
- rpc Sum (stream HelloRequest) returns (HelloReply);
- using System;
- using System.Collections.Generic;
- using System.Linq;
- using System.Threading.Tasks;
- using Grpc.Core;
- using GrpcTest.Server.Models;
- using Microsoft.Extensions.Logging;
- namespace GrpcTest.Server
- {
- public class GreeterService : Greeter.GreeterBase
- {
- private readonly ILogger<GreeterService> _logger;
- private readonly ApplicationDbContext _db;
- public GreeterService(ILogger<GreeterService> logger,
- ApplicationDbContext db)
- {
- _logger = logger;
- _db = db;
- }
- public override Task<HelloReply> SayHello(HelloRequest request, ServerCallContext context)
- {
- return Task.FromResult(new HelloReply
- {
- Message = "Hello " + request.Name
- });
- }
- public override async Task SayHellos(HelloRequest request,
- IServerStreamWriter<HelloReply> responseStream,
- ServerCallContext context)
- {
- foreach (var student in _db.Students)
- {
- if (context.CancellationToken.IsCancellationRequested)
- break;
- var message = student.Name;
- _logger.LogInformation($"Sending greeting {message}.");
- await responseStream.WriteAsync(new HelloReply { Message = message });
- }
- }
- public override async Task<HelloReply> Sum(IAsyncStreamReader<HelloRequest> requestStream, ServerCallContext context)
- {
- var sum = 0;
- await foreach (var request in requestStream.ReadAllAsync())
- {
- if (int.TryParse(request.Name, out var number))
- sum += number;
- else
- throw new ArgumentException("参数必须是可识别的数字");
- }
- return new HelloReply { Message = $"sum is {sum}" };
- }
- }
- }
SayHello: 简单的返回一个文本消息。
SayHellos: 从数据库的表中读取所有数据,并且使用服务器端流的方式返回。
Sum:从客户端流获取输入数据,并计算所有数据的和,如果输入的文本无法转换为数字,抛出异常。
单元测试
新建xunit项目,并引用刚才建立的gRPC项目,引入如下包:
- <ItemGroup>
- <PackageReference Include="Grpc.Core.Testing" Version="2.28.1" />
- <PackageReference Include="Microsoft.EntityFrameworkCore.Sqlite" Version="3.1.3" />
- <PackageReference Include="Microsoft.NET.Test.Sdk" Version="16.5.0" />
- <PackageReference Include="moq" Version="4.14.1" />
- <PackageReference Include="xunit" Version="2.4.0" />
- <PackageReference Include="xunit.runner.visualstudio" Version="2.4.0" />
- <PackageReference Include="coverlet.collector" Version="1.2.0" />
- </ItemGroup>
伪造Logger
使用sqlite inmemory的DbContext
- public static ApplicationDbContext CreateDbContext(){
- var db = new ApplicationDbContext(new DbContextOptionsBuilder<ApplicationDbContext>()
- .UseSqlite(CreateInMemoryDatabase()).Options);
- db.Database.EnsureCreated();
- return db;
- }
- private static DbConnection CreateInMemoryDatabase()
- {
- var connection = new SqliteConnection("Filename=:memory:");
- connection.Open();
- return connection;
- }
重点:虽然是内存模式,数据库也必须是open的,并且需要运行EnsureCreated,否则调用数据库功能是会报告找不到表。
伪造ServerCallContext
使用如下代码伪造:
- public static ServerCallContext CreateTestContext(){
- return TestServerCallContext.Create("fooMethod",
- null,
- DateTime.UtcNow.AddHours(1),
- new Metadata(),
- CancellationToken.None,
- "127.0.0.1",
- null,
- null,
- (metadata) => TaskUtils.CompletedTask,
- () => new WriteOptions(),
- (writeOptions) => { });
- }
里面的具体参数要依据实际测试需要进行调整,比如测试客户端取消操作时,修改CancellationToken参数。
普通调用的测试
- [Fact]
- public void SayHello()
- {
- var service = new GreeterService(logger, null);
- var request = new HelloRequest{Name="world"};
- var response = service.SayHello(request, scc).Result;
- var expected = "Hello world";
- var actual = response.Message;
- Assert.Equal(expected, actual);
- }
其中scc = 伪造的ServerCallContext,如果被测方法中没有实际使用它,也可以直接传入null。
服务器端流的测试
服务器端流的方法包含一个IServerStreamWriter<HelloReply>类型的参数,该参数被用于将方法的计算结果逐个返回给调用方,可以创建一个通用的类实现此接口,将写入的消息存储为一个list,以便测试。
- public class TestServerStreamWriter<T> : IServerStreamWriter<T>
- {
- public WriteOptions WriteOptions { get; set; }
- public List<T> Responses { get; } = new List<T>();
- public Task WriteAsync(T message)
- {
- this.Responses.Add(message);
- return Task.CompletedTask;
- }
- }
测试时,向数据库表中插入两条记录,然后测试对比,看接口方法是否返回两条记录。
- public async Task SayHellos(){
- var db = TestTools.CreateDbContext();
- var students = new List<Student>{
- new Student{Name="1"},
- new Student{Name="2"}
- };
- db.AddRange(students);
- db.SaveChanges();
- var service = new GreeterService(logger, db);
- var request = new HelloRequest{Name="world"};
- var sw = new TestServerStreamWriter<HelloReply>();
- await service.SayHellos(request, sw, scc);
- var expected = students.Count;
- var actual = sw.Responses.Count;
- Assert.Equal(expected, actual);
- }
客户端流的测试
与服务器流类似,客户端流方法也有一个参数类型为IAsyncStreamReader<HelloRequest>,简单实现一个类用于测试。
该类通过直接将客户端要传入的数据通过IEnumable<T>参数传入,模拟客户端的流式请求多个数据。
- public class TestStreamReader<T> : IAsyncStreamReader<T>
- {
- private readonly IEnumerator<T> _stream;
- public TestStreamReader(IEnumerable<T> list){
- _stream = list.GetEnumerator();
- }
- public T Current => _stream.Current;
- public Task<bool> MoveNext(CancellationToken cancellationToken)
- {
- return Task.FromResult(_stream.MoveNext());
- }
- }
正常流程测试代码
- [Fact]
- public void Sum_NormalInput_ReturnSum()
- {
- var service = new GreeterService(null, null);
- var data = new List<HelloRequest>{
- new HelloRequest{Name="1"},
- new HelloRequest{Name="2"},
- };
- var stream = new TestStreamReader<HelloRequest>(data);
- var response = service.Sum(stream, scc).Result;
- var expected = "sum is 3";
- var actual = response.Message;
- Assert.Equal(expected, actual);
- }
参数错误的测试代码
- [Fact]
- public void Sum_BadInput_ThrowException()
- {
- var service = new GreeterService(null, null);
- var data = new List<HelloRequest>{
- new HelloRequest{Name="1"},
- new HelloRequest{Name="abc"},
- };
- var stream = new TestStreamReader<HelloRequest>(data);
- Assert.ThrowsAsync<ArgumentException>(async () => await service.Sum(stream, scc));
- }
总结
以上代码,通过对gRPC服务依赖的关键资源进行mock或简单实现,达到了单元测试的目的。