# 说明

gRPC暴露了两个拦截器接口，分别是：

* `grpc.UnaryServerInterceptor`服务端拦截器
* `grpc.UnaryClientInterceptor`客户端拦截器

基于两个拦截器可以针对性的定制公共模块的封装代码，比如`warden/logging.go`是通用日志逻辑。

# 分析

## 服务端拦截器

让我们先看一下`grpc.UnaryServerInterceptor`的声明，[官方代码位置](https://github.com/grpc/grpc-go/blob/master/interceptor.go)：

```go
// UnaryServerInfo consists of various information about a unary RPC on
// server side. All per-rpc information may be mutated by the interceptor.
type UnaryServerInfo struct {
	// Server is the service implementation the user provides. This is read-only.
	Server interface{}
	// FullMethod is the full RPC method string, i.e., /package.service/method.
	FullMethod string
}

// UnaryHandler defines the handler invoked by UnaryServerInterceptor to complete the normal
// execution of a unary RPC. If a UnaryHandler returns an error, it should be produced by the
// status package, or else gRPC will use codes.Unknown as the status code and err.Error() as
// the status message of the RPC.
type UnaryHandler func(ctx context.Context, req interface{}) (interface{}, error)

// UnaryServerInterceptor provides a hook to intercept the execution of a unary RPC on the server. info
// contains all the information of this RPC the interceptor can operate on. And handler is the wrapper
// of the service method implementation. It is the responsibility of the interceptor to invoke handler
// to complete the RPC.
type UnaryServerInterceptor func(ctx context.Context, req interface{}, info *UnaryServerInfo, handler UnaryHandler) (resp interface{}, err error)
```

看起来很简单包括：

* 一个`UnaryServerInfo`结构体用于`Server`和`FullMethod`字段传递，`Server`为`gRPC server`的对象实例，`FullMethod`为请求方法的全名
* 一个`UnaryHandler`方法用于传递`Handler`，就是基于`proto`文件`service`内声明而生成的方法
* 一个`UnaryServerInterceptor`用于拦截`Handler`方法，可在`Handler`执行前后插入拦截代码

为了更形象的说明拦截器的执行过程，请看基于`proto`生成的以下代码[代码位置](https://github.com/go-kratos/kratos-demo/blob/master/api/api.pb.go)：

```go
func _Demo_SayHello_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
	in := new(HelloReq)
	if err := dec(in); err != nil {
		return nil, err
	}
	if interceptor == nil {
		return srv.(DemoServer).SayHello(ctx, in)
	}
	info := &grpc.UnaryServerInfo{
		Server:     srv,
		FullMethod: "/demo.service.v1.Demo/SayHello",
	}
	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
		return srv.(DemoServer).SayHello(ctx, req.(*HelloReq))
	}
	return interceptor(ctx, in, info, handler)
}
```

这个`_Demo_SayHello_Handler`方法是关键，该方法会被包装为`grpc.ServiceDesc`结构，被注册到gRPC内部，具体可在生成的`pb.go`代码内查找`s.RegisterService(&_Demo_serviceDesc, srv)`。

* 当`gRPC server`收到一次请求时，首先根据请求方法从注册到`server`内的`grpc.ServiceDesc`找到该方法对应的`Handler`如：`_Demo_SayHello_Handler`并执行
* `_Demo_SayHello_Handler`执行过程请看上面具体代码，当`interceptor`不为`nil`时，会将`SayHello`包装为`grpc.UnaryHandler`结构传递给`interceptor`

这样就完成了`UnaryServerInterceptor`的执行过程。那么`_Demo_SayHello_Handler`内的`interceptor`是如何注入到`gRPC server`内，则看下面这段代码[官方代码位置](https://github.com/grpc/grpc-go/blob/master/server.go)：

```go
// UnaryInterceptor returns a ServerOption that sets the UnaryServerInterceptor for the
// server. Only one unary interceptor can be installed. The construction of multiple
// interceptors (e.g., chaining) can be implemented at the caller.
func UnaryInterceptor(i UnaryServerInterceptor) ServerOption {
	return func(o *options) {
		if o.unaryInt != nil {
			panic("The unary server interceptor was already set and may not be reset.")
		}
		o.unaryInt = i
	}
}
```

请一定注意这方法的注释！！！

> Only one unary interceptor can be installed. The construction of multiple interceptors (e.g., chaining) can be implemented at the caller.

`gRPC`本身只支持一个`interceptor`，想要多`interceptors`需要自己实现~~所以`warden`基于`grpc.UnaryClientInterceptor`实现了`interceptor chain`，请看下面代码[代码位置](https://github.com/go-kratos/kratos/blob/master/pkg/net/rpc/warden/server.go)：

```go
// Use attachs a global inteceptor to the server.
// For example, this is the right place for a rate limiter or error management inteceptor.
func (s *Server) Use(handlers ...grpc.UnaryServerInterceptor) *Server {
	finalSize := len(s.handlers) + len(handlers)
	if finalSize >= int(_abortIndex) {
		panic("warden: server use too many handlers")
	}
	mergedHandlers := make([]grpc.UnaryServerInterceptor, finalSize)
	copy(mergedHandlers, s.handlers)
	copy(mergedHandlers[len(s.handlers):], handlers)
	s.handlers = mergedHandlers
	return s
}

// interceptor is a single interceptor out of a chain of many interceptors.
// Execution is done in left-to-right order, including passing of context.
// For example ChainUnaryServer(one, two, three) will execute one before two before three, and three
// will see context changes of one and two.
func (s *Server) interceptor(ctx context.Context, req interface{}, args *grpc.UnaryServerInfo, handler grpc.UnaryHandler) (interface{}, error) {
	var (
		i     int
		chain grpc.UnaryHandler
	)

	n := len(s.handlers)
	if n == 0 {
		return handler(ctx, req)
	}

	chain = func(ic context.Context, ir interface{}) (interface{}, error) {
		if i == n-1 {
			return handler(ic, ir)
		}
		i++
		return s.handlers[i](ic, ir, args, chain)
	}

	return s.handlers[0](ctx, req, args, chain)
}
```

很简单的逻辑：

* `warden server`使用`Use`方法进行`grpc.UnaryServerInterceptor`的注入，而`func (s *Server) interceptor`本身就实现了`grpc.UnaryServerInterceptor`
* `func (s *Server) interceptor`可以根据注册的`grpc.UnaryServerInterceptor`顺序从前到后依次执行

而`warden`在初始化的时候将该方法本身注册到了`gRPC server`，在`NewServer`方法内可以看到下面代码：

```go
opt = append(opt, keepParam, grpc.UnaryInterceptor(s.interceptor))
s.server = grpc.NewServer(opt...)
```

如此完整的服务端拦截器逻辑就串联完成。

## 客户端拦截器


让我们先看一下`grpc.UnaryClientInterceptor`的声明，[官方代码位置](https://github.com/grpc/grpc-go/blob/master/interceptor.go)：

```go
// UnaryInvoker is called by UnaryClientInterceptor to complete RPCs.
type UnaryInvoker func(ctx context.Context, method string, req, reply interface{}, cc *ClientConn, opts ...CallOption) error

// UnaryClientInterceptor intercepts the execution of a unary RPC on the client. invoker is the handler to complete the RPC
// and it is the responsibility of the interceptor to call it.
// This is an EXPERIMENTAL API.
type UnaryClientInterceptor func(ctx context.Context, method string, req, reply interface{}, cc *ClientConn, invoker UnaryInvoker, opts ...CallOption) error
```

看起来和服务端拦截器并没有什么太大的区别，比较简单包括：

* 一个`UnaryInvoker`表示客户端具体要发出的执行方法
* 一个`UnaryClientInterceptor`用于拦截`Invoker`方法，可在`Invoker`执行前后插入拦截代码

具体执行过程，请看基于`proto`生成的下面代码[代码位置](https://github.com/go-kratos/kratos-demo/blob/master/api/api.pb.go)：

```go
func (c *demoClient) SayHello(ctx context.Context, in *HelloReq, opts ...grpc.CallOption) (*google_protobuf1.Empty, error) {
	out := new(google_protobuf1.Empty)
	err := grpc.Invoke(ctx, "/demo.service.v1.Demo/SayHello", in, out, c.cc, opts...)
	if err != nil {
		return nil, err
	}
	return out, nil
}
```

当客户端调用`SayHello`时可以看到执行了`grpc.Invoke`方法，并且将`fullMethod`和其他参数传入，最终会执行下面代码[官方代码位置](https://github.com/grpc/grpc-go/blob/master/call.go)：

```go
// Invoke sends the RPC request on the wire and returns after response is
// received.  This is typically called by generated code.
//
// All errors returned by Invoke are compatible with the status package.
func (cc *ClientConn) Invoke(ctx context.Context, method string, args, reply interface{}, opts ...CallOption) error {
	// allow interceptor to see all applicable call options, which means those
	// configured as defaults from dial option as well as per-call options
	opts = combine(cc.dopts.callOptions, opts)

	if cc.dopts.unaryInt != nil {
		return cc.dopts.unaryInt(ctx, method, args, reply, cc, invoke, opts...)
	}
	return invoke(ctx, method, args, reply, cc, opts...)
}
```

其中的`unaryInt`即为客户端连接创建时注册的拦截器，使用下面代码注册[官方代码位置](https://github.com/grpc/grpc-go/blob/master/dialoptions.go)：

```go
// WithUnaryInterceptor returns a DialOption that specifies the interceptor for
// unary RPCs.
func WithUnaryInterceptor(f UnaryClientInterceptor) DialOption {
	return newFuncDialOption(func(o *dialOptions) {
		o.unaryInt = f
	})
}
```

需要注意的是客户端的拦截器在官方`gRPC`内也只能支持注册一个，与服务端拦截器`interceptor chain`逻辑类似`warden`在客户端拦截器也做了相同处理，并且在客户端连接时进行注册，请看下面代码[代码位置](https://github.com/go-kratos/kratos/blob/master/pkg/net/rpc/warden/client.go)：

```go
// Use attachs a global inteceptor to the Client.
// For example, this is the right place for a circuit breaker or error management inteceptor.
func (c *Client) Use(handlers ...grpc.UnaryClientInterceptor) *Client {
	finalSize := len(c.handlers) + len(handlers)
	if finalSize >= int(_abortIndex) {
		panic("warden: client use too many handlers")
	}
	mergedHandlers := make([]grpc.UnaryClientInterceptor, finalSize)
	copy(mergedHandlers, c.handlers)
	copy(mergedHandlers[len(c.handlers):], handlers)
	c.handlers = mergedHandlers
	return c
}

// chainUnaryClient creates a single interceptor out of a chain of many interceptors.
//
// Execution is done in left-to-right order, including passing of context.
// For example ChainUnaryClient(one, two, three) will execute one before two before three.
func (c *Client) chainUnaryClient() grpc.UnaryClientInterceptor {
	n := len(c.handlers)
	if n == 0 {
		return func(ctx context.Context, method string, req, reply interface{},
			cc *grpc.ClientConn, invoker grpc.UnaryInvoker, opts ...grpc.CallOption) error {
			return invoker(ctx, method, req, reply, cc, opts...)
		}
	}

	return func(ctx context.Context, method string, req, reply interface{},
		cc *grpc.ClientConn, invoker grpc.UnaryInvoker, opts ...grpc.CallOption) error {
		var (
			i            int
			chainHandler grpc.UnaryInvoker
		)
		chainHandler = func(ictx context.Context, imethod string, ireq, ireply interface{}, ic *grpc.ClientConn, iopts ...grpc.CallOption) error {
			if i == n-1 {
				return invoker(ictx, imethod, ireq, ireply, ic, iopts...)
			}
			i++
			return c.handlers[i](ictx, imethod, ireq, ireply, ic, chainHandler, iopts...)
		}

		return c.handlers[0](ctx, method, req, reply, cc, chainHandler, opts...)
	}
}
```

如此完整的客户端拦截器逻辑就串联完成。

# 实现自己的拦截器

以服务端拦截器`logging`为例：

```go
// serverLogging warden grpc logging
func serverLogging() grpc.UnaryServerInterceptor {
	return func(ctx context.Context, req interface{}, info *grpc.UnaryServerInfo, handler grpc.UnaryHandler) (interface{}, error) {
        // NOTE: handler执行之前的拦截代码：主要获取一些关键参数，如耗时计时、ip等
        // 如果自定义的拦截器只需要在handler执行后，那么可以直接执行handler

		startTime := time.Now()
		caller := metadata.String(ctx, metadata.Caller)
		if caller == "" {
			caller = "no_user"
		}
		var remoteIP string
		if peerInfo, ok := peer.FromContext(ctx); ok {
			remoteIP = peerInfo.Addr.String()
		}
		var quota float64
		if deadline, ok := ctx.Deadline(); ok {
			quota = time.Until(deadline).Seconds()
		}

		// call server handler
		resp, err := handler(ctx, req) // NOTE: 以具体执行的handler为分界线！！！

        // NOTE: handler执行之后的拦截代码：主要进行耗时计算、日志记录
        // 如果自定义的拦截器在handler执行后不需要逻辑，这可直接返回

		// after server response
		code := ecode.Cause(err).Code()
		duration := time.Since(startTime)

		// monitor
		statsServer.Timing(caller, int64(duration/time.Millisecond), info.FullMethod)
		statsServer.Incr(caller, info.FullMethod, strconv.Itoa(code))
		logFields := []log.D{
			log.KVString("user", caller),
			log.KVString("ip", remoteIP),
			log.KVString("path", info.FullMethod),
			log.KVInt("ret", code),
			// TODO: it will panic if someone remove String method from protobuf message struct that auto generate from protoc.
			log.KVString("args", req.(fmt.Stringer).String()),
			log.KVFloat64("ts", duration.Seconds()),
			log.KVFloat64("timeout_quota", quota),
			log.KVString("source", "grpc-access-log"),
		}
		if err != nil {
			logFields = append(logFields, log.KV("error", err.Error()), log.KV("stack", fmt.Sprintf("%+v", err)))
		}
		logFn(code, duration)(ctx, logFields...)
		return resp, err
	}
}
```

# 内置拦截器

## 自适应限流拦截器

更多关于自适应限流的信息，请参考：[kratos 自适应限流](/doc/wiki-cn/ratelimit.md)

```go
package grpc

import (
	pb "kratos-demo/api"
	"kratos-demo/internal/service"
	"github.com/go-kratos/kratos/pkg/conf/paladin"
	"github.com/go-kratos/kratos/pkg/net/rpc/warden"
	"github.com/go-kratos/kratos/pkg/net/rpc/warden/ratelimiter"
)

// New new a grpc server.
func New(svc *service.Service) *warden.Server {
	var rc struct {
		Server *warden.ServerConfig
	}
	if err := paladin.Get("grpc.toml").UnmarshalTOML(&rc); err != nil {
		if err != paladin.ErrNotExist {
			panic(err)
		}
	}
	ws := warden.NewServer(rc.Server)
	
	// 挂载自适应限流拦截器到 warden server，使用默认配置
	limiter := ratelimiter.New(nil)
	ws.Use(limiter.Limit())
	
	// 注意替换这里：
	// RegisterDemoServer方法是在"api"目录下代码生成的
	// 对应proto文件内自定义的service名字，请使用正确方法名替换
	pb.RegisterDemoServer(ws.Server(), svc)
	
	ws, err := ws.Start()
	if err != nil {
		panic(err)
	}
	return ws
}
```

# 扩展阅读

[warden快速开始](warden-quickstart.md) [warden基于pb生成](warden-pb.md) [warden负载均衡](warden-balancer.md) [warden服务发现](warden-resolver.md)

-------------

[文档目录树](summary.md)