WebFlux 全异步及分页支持

WebFlux 分页支持

webflux是spring框架构建响应式web新框架，但是后端只能使用非关系型数据库，包括：redis，mongo，Cassandra，elasticsearch，此文档中演示通过elasticsearch构建全异步支持的日志服务，重点在对分页支持上。

基础组件：

• spring-boot 2.2.1
• spring-boot-starter-data-elasticsearch
• spring-boot-starter-webflux

实体类

@AllArgsConstructor
@NoArgsConstructor
@Document(indexName = "application-log*", type = "doc")
public class AppLog implements Serializable {
    @Id
  private String id;

  @Field
 @JsonProperty("app")
    private String app;
  @Field
  private String logger;
  @Field
  private String thread;
  @Field
  private String level;
  @Field
  private String tid;

  @Field(name = "@timestamp" ,type = FieldType.Date,fielddata = true)
    @DateTimeFormat(iso = DateTimeFormat.ISO.DATE_TIME)
    @JsonProperty("timestamp")
    @JsonFormat(pattern = "yyyy-MM-dd HH:mm:ss,SSS")
    private Date timestamp;
  @Field
  private String details;
}

Repository类

@Repository
public interface AppLogRepository extends ReactiveElasticsearchRepository<AppLog,String> {
    Flux<AppLog> findByApp(String app,Pageable page);
    Mono<Long> countByApp(String app);

Service类

@Service
@Slf4j
public class AppLogServiceImpl implements AppLogService {
	public static final String APPLOG_INDEX = "application-log";
	private final AppLogRepository appLogRepository;
	public AppLogServiceImpl(AppLogRepository appLogRepository){
		this.appLogRepository = appLogRepository;
	}

	@Override
	public Flux<AppLog> findByApp(String app, Pageable page) {
		return appLogRepository.findByApp(app,page);
	}

	@Override
	public Mono<Long> count(String app) {
		return appLogRepository.countByApp(app);
	}

RestController类

这里是重点，此处分页采用在Response Header中增加总数，通过通过Link 规范（RFC）输出上下页信息。

@GetMapping("/app/{app}")
public Mono<ResponseEntity> getApplog(@PathVariable String app,Pageable page){
    Mono<List<AppLog>> applogs = appLogService.findByApp(app,page).collectList();
  Mono<Long> total = appLogService.count(app);
 return applogs.zipWith(total,(log,size) ->        ResponseEntity.ok().headers(PaginationUtil.generatePaginationHttpHeaders(page,log,size,"base")).body(applogs));
}

此处：

1. Pageable为Spring的分页类，可以直接引入
2. 调用count是因为findByApp返回的是Flux类型，没有分页信息
3. zipWith是Mono的一个函数，用于合并两个Publisher对象，此处将分页后的数据和总数合并到一个Tuple对象中，以便在构造ResponseEntity时使用
4. generatePaginationHttpHeaders构造了输出头

public static <T> HttpHeaders generatePaginationHttpHeaders(Pageable pageable, List list, Long totalNumber, String baseUrl) {

    HttpHeaders headers = new HttpHeaders();
 int totalPages = Math.toIntExact(totalNumber/pageable.getPageSize()+1);

  headers.add("X-Total-Count", Long.toString(totalNumber));
  String link = "";
 if ((pageable.getPageNumber() + 1) < totalPages) {
        link = "<" + PaginationUtil.generateUri(baseUrl, pageable.getPageNumber() + 1, list.size()) + ">; rel=\"next\",";
  }
    // prev link
  if ((pageable.getPageNumber()) > 0) {
        link += "<" + PaginationUtil.generateUri(baseUrl, pageable.getPageNumber() - 1, list.size()) + ">; rel=\"prev\",";
  }
    // last and first link
  int lastPage = 0;
 if (totalPages > 0) {
        lastPage = totalPages - 1;
  }
    link += "<" + PaginationUtil.generateUri(baseUrl, lastPage, list.size()) + ">; rel=\"last\",";
  link += "<" + PaginationUtil.generateUri(baseUrl, 0, list.size()) + ">; rel=\"first\"";
  headers.add(HttpHeaders.LINK, link);
 return headers;
}

static String generateUri(String baseUrl, int page, int size) {
    return UriComponentsBuilder.fromUriString(baseUrl).queryParam("page", page).queryParam("size", size).toUriString();
}