Spring Cloud Gateway用法详解

• T1 - 简介

  Spring Cloud Gateway提供了一个基于Spring生态系统的API网关，是依赖于webflux的响应式框架,不能在传统的Servlet容器中工作。网关的常见作用包括路由转发、鉴权、限流、过滤等，以下为Spring Cloud Gateway的工作流程图:

  
  主要组件：

  • Route:路由网关的基本构建块，由ID、目标URI、一系列Predicates和Filters组成。路由根据predicate返回是否为true进行批评。
  • Predicate:Java8的函数式接口，输入类型是Spring的ServerWebExchange。Predicate可用于匹配HTTP请求的各种内容，如头部和参数。
  • Filter:使用特定工厂构建的实例，请求和响应都可以在被发送到下游请求之前或之后被修改

  流程：request->RouteLocator->RoutePredicateHandlerMapping(predicate)->FilteringWebHandler(filter)->Service->filter
  1.RouteLocatorBuilder.build()生成RouteLocator
  2.RoutePredicateHandlerMapping.lookupRoute()通过RouteLocator获取路由映射调用配置的predicate
  3.FilteringWebHandler.handle()调用filter

  gateway的rout、predicate、filter都可通过实例Bean或配置文件进行配置，且都是基于COC(Convention Over Configuration-约定由于配置)的设计方式，在gateway中显示为根据特定命名方式设计、实例化Bean。

• T2 - Predicate用法

  gateway的所有Predicate都由XxxRoutePredicateFactory工厂类进行实例化，在配置文件中只需设置{Xxx=paramA[, paramB…]}即可完成一个Predicate类的实例化。下图为PredicateFactory的类层级图：
  
  predicate与filter都可通过bean或spring配置文件进行实例化，Demo如下：

  • java bean配置

    @Bean
    public RouteLocator pathRouteLocator(RouteLocatorBuilder builder) {
        return builder.routes()
                .route(p -> p
                        .path("/**")
                        .and()
                        .header("token", "\\w+")
                        .uri("lb://user-consumer"))
                .build();
    }     
    

  • application.yml配置

    server:
      port: 8999
    spring:
      application:
        name: gateway-server
      cloud:
        gateway:
          discovery:
            locator:
              enabled: false
              lower-case-service-id: true
          routes:
            - uri: lb://user-consumer
              predicates:
                - Header=token, \w+
                - Path=/**
    

  以上例子配置了HeaderRoutePredicateFactory与PathRoutePredicateFactory,Header Predicate设置若header中包含符合\w+的token header,请求将路由到user-consumer服务。无论是通过bean还是application.yml配置，都是通过predicate的前缀名进行bean的配置，具体应用为bean route()中调用PredicateSpec.path()实现PathRoutePredicate，调用header方法实现HeaderRoutePredicate。使用配置文件的配置格式为:{Predicate-prefix}=param1[,param2,param3…]，参数顺序以XxxRoutePredicateFactory.shortcutFieldOrder()的顺序为准,以HeaderRoutePredicateFactory为例,列表中第一个为header,第二个为regexp,则配置格式为Header={header},{regexp}。HeaderRoutePredicateFactory部分源码：

  /**
   * Header key.
   */
  public static final String HEADER_KEY = "header";
  
  /**
   * Regexp key.
   */
  public static final String REGEXP_KEY = "regexp";
  
  public HeaderRoutePredicateFactory() {
    super(Config.class);
  }
  
  @Override
  public List shortcutFieldOrder() {
    return Arrays.asList(HEADER_KEY, REGEXP_KEY);
  }
  

• T2 - Filter用法

  
  上图为filter的所在包与依赖树，其配置方式与predicate类似，配置Demo如下:

  • java bean配置

    @Bean
    public RouteLocator pathRouteLocator(RouteLocatorBuilder builder) {
        return builder.routes()
                .route(p -> p
                        .path("/**")
                        .and()
                        .header("token", "\\w+")
                        .filters(f -> f.stripPrefix(1)
                                .addResponseHeader("response-header", "head-val"))
                        .uri("lb://user-consumer"))
                .build();
    } 
    

  • application.yml配置

    server:
      port: 8999
    spring:
      application:
        name: gateway-server
      cloud:
        gateway:
          discovery:
            locator:
              enabled: false
              lower-case-service-id: true
          routes:
            - uri: lb://user-consumer
              predicates:
                - Header=token, \w+
                - Path=/**
              filters:
                - StripPrefix=1
                - AddResponseHeader=response-header, head-val
    

  以上例子基于predicate的基础上添加了StripPrefixGatewayFilterFactory与AddResponseHeaderGatewayFilterFactory的过滤器，StripPrefixGatewayFilterFactory为截取路径中的前i个路径(如i=2,请求路径为/a/b/hi,则由网关请求到service的路径为/hi)，AddResponseHeaderGatewayFilterFactory则对service返回结果进行加工添加header，然后再返回结果。这2个体现了filter可对请求到service前|后进行参数、header、路径、响应的处理,即官方文档中的"pre"与"post"两种filter,对响应的处理(post)Filter工厂类一般名称中包含Response，配置文件的配置方式同predicate。

• T3 - Eureka服务注册发现

  当没有配置注册中心时，每次添加新服务都需在网关添加新的服务配置，请求流程为Web->Gateway->Service；当添加注册中心后，网关便可通过注册中心访问中心上的服务，无论服务新增还是减少都无需重新配置网关路由，请求流程转变为Web->Gateway->Registry Center->Service。集成注册中心eureka配置Demo:

  eureka:  
    client:  
      service-url:  
        defaultZone: http://localhost:8000/eureka/  
  logging:  
    level:  
      org.springframework.cloud.gateway: debug  
  spring:  
    profiles: gateway-eureka  
    cloud:  
      gateway:  
        discovery:  
          locator:  
            # 启用DiscoveryClient网关集成,通过网关地址与serviceName访问注册中心的service  
            enabled: true  
            # 服务名转小写  
            lower-case-service-id: true
  

  配置注册中心后便可通过 http://{ip}.{gatewayPort}/{serviceName}/{api} 访问注册中心上的服务接口，由T2可知gateway可通过Predicate将gateway接收到的请求路由到其它url，但配置注册中心后会发现我们无需配置具体服务路由gateway就可将请求分发到相应的service上，由此我联想到了以下2个问题：

  1. 那么gateway是怎么分发这些路由的呢？
  2. 没有配置注册中心时是在routes中配置predicate与filter的，那配置注册中心后该如何配置呢？

  • T3.1 - gateway分发路由规则

    以下是gateway的自动化配置类GatewayDiscoveryClientAutoConfiguration.java：

    @Configuration  
    @ConditionalOnProperty(name = "spring.cloud.gateway.enabled", matchIfMissing = true)  
    @AutoConfigureBefore(GatewayAutoConfiguration.class)  
    @AutoConfigureAfter(CompositeDiscoveryClientAutoConfiguration.class)  
    @ConditionalOnClass({ DispatcherHandler.class, DiscoveryClient.class })  
    @EnableConfigurationProperties  
    public class GatewayDiscoveryClientAutoConfiguration {  
    
      public static List initPredicates() {  
          ArrayList definitions = new ArrayList<>();  
        // TODO: add a predicate that matches the url at /serviceId?  
        // add a predicate that matches the url at /serviceId/**  
        PredicateDefinition predicate = new PredicateDefinition();  
        predicate.setName(normalizeRoutePredicateName(PathRoutePredicateFactory.class));  
        predicate.addArg(PATTERN_KEY, "'/'+serviceId+'/**'");  
        definitions.add(predicate);  
        return definitions;  
      }  
    
       public static List initFilters() {  
          ArrayList definitions = new ArrayList<>();  
         // add a filter that removes /serviceId by default  
         FilterDefinition filter = new FilterDefinition();  
         filter.setName(normalizeFilterFactoryName(RewritePathGatewayFilterFactory.class));  
         String regex = "'/' \+ serviceId + '/(?.*)'";  
         String replacement = "'/${remaining}'";  
         filter.addArg(REGEXP_KEY, regex);  
         filter.addArg(REPLACEMENT_KEY, replacement);  
         definitions.add(filter);  
         return definitions;  
      }  
    
      @Bean  
      @ConditionalOnBean(DiscoveryClient.class)  
      @ConditionalOnProperty(name = "spring.cloud.gateway.discovery.locator.enabled")  
      public DiscoveryClientRouteDefinitionLocator discoveryClientRouteDefinitionLocator(  
             DiscoveryClient discoveryClient, DiscoveryLocatorProperties properties) {  
        return new DiscoveryClientRouteDefinitionLocator(discoveryClient, properties);  
      }  
    
      @Bean  
      public DiscoveryLocatorProperties discoveryLocatorProperties() {  
          DiscoveryLocatorProperties properties = new DiscoveryLocatorProperties();  
        properties.setPredicates(initPredicates());  
        properties.setFilters(initFilters());  
        return properties;  
      }  
    }
    

    由以上类可看出gateway默认配置了匹配/serviceId/**的PathRoutePredicate与RewritePathGatewayFilter,spring.cloud.gateway.discovery.locator.enabled为true时会将predicate和filter配置到类DiscoveryClientRouteDefinitionLocator,DiscoveryClientRouteDefinitionLocator会以SpEL解析predicate、filter配置，serviceId会被解析成对应的service，当有服务注册到注册中心时,gateway会检测并将predicate与filter应用到服务，如下图所示，问题1解决。

    

  • T3.2 - 统一配置服务predicate与filter

    配置注册中心后，可用过spring.cloud.gateway.discovery.locator.predicates|filters对注册中心中的服务进行predicate与filter的统一配置，但会覆盖默认的predicate与filter配置，此时可添加PathRoutePredicate与RewritePathGatewayFilter默认配置保持默认功能，配置Demo如下图所示:

    spring:
      profiles: gateway-eureka
      cloud:
        gateway:
          discovery:
            locator:
              # 启用DiscoveryClient网关集成,可通过网关地址与serviceName访问注册中心的service
              enabled: true
              # 服务名转小写
              lower-case-service-id: true
              predicates:
                - name: Path
                  args:
                    patterns: "'/'+serviceId+'/**'"
                - name: Header
                  args:
                    header: "'token'"
                    regexp: "'\\w+'"
              filters:
                - name: RewritePath
                  args:
                    replacement: "'/${remaining}'"
                    regexp: "'/' + serviceId + '/(?.*)'"
                - name: AddResponseHeader
                  args:
                    name: "'response-test-head'"
                    value: "'response-test'"
    

• T4 - 自定义全局Filter

  当配置注册中心后，可能就会出现各服务统一Token鉴权、统一日志记录、超时设置等需求，这些可以统一在gateway进行处理。gateway全局过滤器都需实现GlobalFilter，一般还需实现Ordered接口控制Bean实例化顺序，以一个简单token全局过滤器为例：

  @Slf4j
  @Component
  @ConditionalOnProperty(value = "spring.cloud.gateway.discovery.locator.global-token-filter-enabled", havingValue = "true")
  public class GlobalTokenFilter implements GlobalFilter, Ordered {
      @PostConstruct
      public void init() {
          log.info("GlobalTokenFilter init finish");
      }
  
      @Override
      public Mono filter(ServerWebExchange exchange, GatewayFilterChain chain) {
          Map headers = exchange.getRequest().getHeaders().toSingleValueMap();
          boolean hasToken = headers.containsKey("token") && StringUtils.isNotBlank(headers.get("token"));
          if (!hasToken) {
              exchange.getResponse().setStatusCode(HttpStatus.UNAUTHORIZED);
              return exchange.getResponse().setComplete();
          }
          log.info("token value:" + headers.get("token"));
          return chain.filter(exchange);
      }
  
      @Override
      public int getOrder() {
          return -100;
      }
  }
  

  请求的各种信息都可通过ServerWebExchange获取，当有请求到达gateway时,GlobalTokenFilter检测是否含token的头部，没有的话gateway将返回404。我们也可以通过继承AbstractGatewayFilterFactory来自定义过滤器，如简单的请求记录过滤器：

  @Slf4j
  @Component
  public class RequestRecordGatewayFilterFactory extends AbstractGatewayFilterFactory {
  
      public static final String PARTS_KEY = "print";
  
      public RequestRecordGatewayFilterFactory() {
          super(Config.class);
      }
  
      @Override
      public List shortcutFieldOrder() {
          return Arrays.asList(PARTS_KEY);
      }
  
      @Override
      public GatewayFilter apply(Config config) {
          return (exchange, chain) -> {
              if (config.print != null && config.print) {
                  // 打印相对路径
                  log.info("requestRecord1:{},参数:{}", exchange.getRequest().getPath().value(), exchange.getRequest().getQueryParams().toSingleValueMap());
                  // 打印绝对路径
                  log.info("requestRecord1:{},参数:{}", exchange.getAttribute(GATEWAY_ORIGINAL_REQUEST_URL_ATTR), exchange.getRequest().getQueryParams().toSingleValueMap());
              }
              return chain.filter(exchange);
          };
      }
  
      @Override
      public GatewayFilter apply(Consumer consumer) {
          Config config = newConfig();
          consumer.accept(config);
          return apply(config);
      }
  
      @ToString
      public static class Config {
          private Boolean print;
  
          public boolean isPrint() {
              return print;
          }
  
          public void setPrint(boolean print) {
              this.print = print;
          }
      }
  }
  

  同时application.yml添加RequestRecordFilter配置:

  spring:
    profiles: gateway-eureka
    cloud:
      gateway:
        discovery:
          locator:
            # 启用DiscoveryClient网关集成,可通过网关地址与serviceName访问注册中心的service
            enabled: true
            # 服务名转小写
            lower-case-service-id: true
            predicates:
              - name: Path
                args:
                  patterns: "'/'+serviceId+'/**'"
              - name: Header
                args:
                  header: "'token'"
                  regexp: "'\\w+'"
            filters:
              - name: RewritePath
                args:
                  replacement: "'/${remaining}'"
                  regexp: "'/' + serviceId + '/(?.*)'"
              - name: AddResponseHeader
                args:
                  name: "'response-test-head'"
                  value: "'response-test'"
              - name: RequestRecord
                args:
                  print: true
  

  请求的参数日志也可通过一个GlobalFilter去实现，区别是GlobalFilter的ServerWebExchange尚未把原始请求的url设置到属性中(Attrubute)，所以打印出的绝对路径为null，且继承AbstractGatewayFilterFactory获取的相对路径是截掉了serviceId的，因为在其它过滤器操作执行前路径已被RewritePath过滤器进行了路径处理，具体效果图如下：
  

  若要自定义Predicate只需继承AbstractRoutePredicateFactory类根据业务需求重写相应方法，具体可参考AbstractRoutePredicateFactory的子类，与自定义Fitler大同小异，这里便不唠叨了。

• T5 - 总结

  该文章主要讲gateway的主要用法并根据个人的思考扩展了一些gateway的用例，大部分知识点都是基于官方文档进行总结，且由以上功能Demo可看出gateway的处理流程并不复杂，由Web->Gateway(Predicate->GlobalFilter->Pre Filter)->Service->Gateway(Post Gateway)->Web