Spring Cloud Gateway 如何修改HTTP响应信息

Gateway 修改HTTP响应信息

实践Spring Cloud的过程中，使用Gateway作为路由组件，并且基于Gateway实现权限的验证、拦截、过滤，对于下游微服务的响应结果，我们总会有需要修改以统一数据格式，或者修改过滤用户没有权限看到的数据信息，这时候就需要有一个能够修改响应体的Filter。

Spring Cloud Gateway 版本为2.1.0

在当前版本，ModifyRequestBodyGatewayFilterFactory是官方提供的修改响应体的参考类，This filter is BETA and may be subject to change in a future release.，类的注释中说明这个类在以后版本中会改进，实际使用可以参考实现功能，但是性能影响较大，不过没有别的选择还是得选择这个。

官方文档：

实现

最终代码

先贴最终代码

public class ResponseDecryptionGlobalFilter implements GlobalFilter, Ordered {
    private static Logger log = LoggerFactory.getLogger(ResponseDecryptionGlobalFilter.class);
    @Override
    public int getOrder() {
    	// 控制在NettyWriteResponseFilter后执行
        return NettyWriteResponseFilter.WRITE_RESPONSE_FILTER_ORDER - 1;
    }
    @Override
    public Mono<Void> filter(ServerWebExchange exchange, GatewayFilterChain chain) {
        return processResponse(exchange, chain);
    }
    private Mono<Void> processResponse(ServerWebExchange exchange, GatewayFilterChain chain) {
        // 路由中如果不需要过滤则不进行过滤
        if (!BooleanUtils.isTrue()) {
            return chain.filter(exchange);
        }
        ServerHttpResponseDecorator responseDecorator = new ServerHttpResponseDecorator(exchange.getResponse()) {
            @Override
            public Mono<Void> writeWith(Publisher<? extends DataBuffer> body) {
                String originalResponseContentType = exchange.getAttribute(ORIGINAL_RESPONSE_CONTENT_TYPE_ATTR);
                HttpHeaders httpHeaders = new HttpHeaders();
                httpHeaders.add(HttpHeaders.CONTENT_TYPE, originalResponseContentType);
                ResponseAdapter responseAdapter = new ResponseAdapter(body, httpHeaders);
                DefaultClientResponse clientResponse = new DefaultClientResponse(responseAdapter, ExchangeStrategies.withDefaults());
                Mono<String> rawBody = clientResponse.bodyToMono(String.class).map(s -> s);
                BodyInserter<Mono<String>, ReactiveHttpOutputMessage> bodyInserter = BodyInserters.fromPublisher(rawBody, String.class);
                CachedBodyOutputMessage outputMessage = new CachedBodyOutputMessage(exchange, exchange.getResponse().getHeaders());
                return bodyInserter.insert(outputMessage, new BodyInserterContext())
                        .then(Mono.defer(() -> {
                            Flux<DataBuffer> messageBody = outputMessage.getBody();
                            Flux<DataBuffer> flux = messageBody.map(buffer -> {
                                CharBuffer charBuffer = StandardCharsets.UTF_8.decode(buffer.asByteBuffer());
                                DataBufferUtils.release(buffer);
								// 将响应信息转化为字符串
                                String responseStr = charBuffer.toString();
                                if (StringUtils.isNotBlank(responseStr)) {
                                    try {
                                        JSONObject result = JSONObject.parseObject(responseStr);
                                        System.out.println(dataFilter(result));
                                        if (result.containsKey("data")) {
                                            responseStr = dataFilter(result);
                                        } else {
                                            log.error("响应结果序列化异常:{}", responseStr);
                                        }
                                    } catch (JSONException e) {
                                        log.error("响应结果序列化异常:{}", responseStr);
                                    }
                                }
                                return getDelegate().bufferFactory().wrap(responseStr.getBytes(StandardCharsets.UTF_8));
                            });
                            HttpHeaders headers = getDelegate().getHeaders();
                            // 修改响应包的大小，不修改会因为包大小不同被浏览器丢掉
                            flux = flux.doOnNext(data -> headers.setContentLength(data.readableByteCount()));
                            return getDelegate().writeWith(flux);
                        }));
            }
        };
        return chain.filter(exchange.mutate().response(responseDecorator).build());
    }
    /**
     * 权限数据过滤
     *
     * @param result
     * @return
     */
    private String dataFilter(JSONObject result) {
        Object data = result.get("data");
        return result.toJSONString();
    }
    private class ResponseAdapter implements ClientHttpResponse {
        private final Flux<DataBuffer> flux;
        private final HttpHeaders headers;
        @SuppressWarnings("unchecked")
        private ResponseAdapter(Publisher<? extends DataBuffer> body, HttpHeaders headers) {
            this.headers = headers;
            if (body instanceof Flux) {
                flux = (Flux) body;
            } else {
                flux = ((Mono) body).flux();
            }
        }
        @Override
        public Flux<DataBuffer> getBody() {
            return flux;
        }
        @Override
        public HttpHeaders getHeaders() {
            return headers;
        }
        @Override
        public HttpStatus getStatusCode() {
            return null;
        }
        @Override
        public int getRawStatusCode() {
            return 0;
        }
        @Override
        public MultiValueMap<String, ResponseCookie> getCookies() {
            return null;
        }
    }
}

踩过的坑

• 响应体报文过大: 起初直接读取buffer的响应信息，包小的情况没有问题，但是包大了会抛出json无法转换异常，因为没能读取完整的响应内容，参考ModifyRequestBodyGatewayFilter，等待buffer全部读完再转为数组，然后执行处理。本质原因是底层的Reactor-Netty的数据块读取大小限制导致获取到的DataBuffer实例里面的数据是不完整的。
• 修改响应信息后，响应的ContentLength会发生变化，忘记修改response中的Content-Length长度，导致前端请求无法获取修改后的响应结果。

flux = flux.doOnNext(data -> headers.setContentLength(data.readableByteCount()));

• order值必须小于-1，因为覆盖返回响应体，自定义的GlobalFilter必须比NettyWriteResponseFilter处理完后执行。order越小越早进行处理，越晚处理响应结果。

理解ServerWebExchange

先看ServerWebExchange的注释：

Contract for an HTTP request-response interaction. Provides access to the HTTP request and response and also exposes additional server-side processing related properties and features such as request attributes.

翻译一下大概是：

ServerWebExchange是一个**HTTP请求-响应交互的契约。**提供对HTTP请求和响应的访问，并公开额外的服务器端处理相关属性和特性，如请求属性。

ServerWebExchange有点像Context的角色，我把它理解为http请求信息在Filter透传的容器，之所以称之为容器，因为它可以存储我们像放进去的数据。

注意：

ServerHttpRequest是一个只读类，因此需要通过下面例子的方法来进行修改，对于读多写少的场景，这种设计模式是值得借鉴的

ServerHttpRequest newRequest = request.mutate().headers("key","value").path("/myPath").build();
ServerWebExchange newExchange = exchange.mutate().response(responseDecorator).build();

Gateway 修改返回的响应体

问题描述：

在gateway中修改返回的响应体，在全局Filter中添加如下代码：

import org.springframework.core.Ordered;
import org.springframework.cloud.gateway.filter.GlobalFilter;
import org.springframework.cloud.gateway.filter.GatewayFilterChain;
import org.springframework.stereotype.Component;
import org.springframework.web.server.ServerWebExchange;
import reactor.core.publisher.Mono;
@Component
public class RequestGlobalFilter implements GlobalFilter, Ordered {
 //...

 @Override
 public Mono<Void> filter(ServerWebExchange exchange, GatewayFilterChain chain) {
  //...
  ResponseDecorator decorator = new ResponseDecorator(exchange.getResponse());
  return chain.filter(exchange.mutate().response(decorator).build());
 }
 @Override
 public int getOrder() {
  return -1000;
 }
}

通过.response(decorator)设置一个响应装饰器（自定义），以下是装饰器具体实现：

import cn.hutool.json.JSONObject;
import org.reactivestreams.Publisher;
import org.springframework.core.io.buffer.DataBuffer;
import org.springframework.core.io.buffer.DataBufferFactory;
import org.springframework.core.io.buffer.DataBufferUtils;
import org.springframework.core.io.buffer.DefaultDataBufferFactory;
import org.springframework.http.server.reactive.ServerHttpResponse;
import org.springframework.http.server.reactive.ServerHttpResponseDecorator;
import reactor.core.publisher.Flux;
import reactor.core.publisher.Mono;
import java.nio.charset.Charset;
/**
 * @author visy.wang
 * @desc 响应装饰器（重构响应体）
 */
public class ResponseDecorator extends ServerHttpResponseDecorator{
 public ResponseDecorator(ServerHttpResponse delegate){
  super(delegate);
 }
 @Override
 @SuppressWarnings(value = "unchecked")
 public Mono<Void> writeWith(Publisher<? extends DataBuffer> body) {
  if(body instanceof Flux) {
   Flux<DataBuffer> fluxBody = (Flux<DataBuffer>) body;
   return super.writeWith(fluxBody.buffer().map(dataBuffers -> {
    DataBufferFactory dataBufferFactory = new DefaultDataBufferFactory();
    DataBuffer join = dataBufferFactory.join(dataBuffers);
    byte[] content = new byte[join.readableByteCount()];
    join.read(content);
    DataBufferUtils.release(join);// 释放掉内存

    String bodyStr = new String(content, Charset.forName("UTF-8"));
                //修改响应体
    bodyStr = modifyBody(bodyStr);
    getDelegate().getHeaders().setContentLength(bodyStr.getBytes().length);
    return bufferFactory().wrap(bodyStr.getBytes());
   }));
  }
  return super.writeWith(body);
 }
    //重写这个函数即可
 private String modifyBody(String jsonStr){
  JSONObject json = new JSONObject(jsonStr);
        //TODO...修改响应体
  return json.toString();
 }
}

以上为个人经验，希望能给大家一个参考，也希望大家多多支持我们。