使用 Java 注解自动化处理对应关系实现注释代码化

概述

假设我们要从一个 ES 索引（相当于一张DB表）查询数据，ES表有 order_no, order_type, state 等字段， 而应用对象则有属性 orderNo, orderType, state等。这样，就会面临“将应用对象的属性与ES字段对应起来”的问题。

固然可以通过注释来说明，不过这样显得比较生硬。因为注释并不起实际作用，代码里还得写一套映射关系，就会存在注释与代码不一致的情况。 那么，是否可以将这种对应关系的注释用代码形式来解决呢？ Java 注解可以解决这个问题。

实现

定义注解

首先定义注解类。注解类需要提供对应的ES字段名 name、类型 type 以及是否必传 required。

• @Retention 指明注解在何时起作用，这里是在运行时。
• @Target 指明注解应用于何种对象，这里应用于字段。

@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.FIELD)
@Documented
public @interface EsField {

  /**
   * 对应的ES字段名
   */
  String name();

  /**
   * 对应的ES字段的值类型
   * @return
   */
  String type() default "";

  /**
   * 是否必传
   */
  boolean required() default false;

}

应用领域对象

接着，将注解应用到应用领域对象。为简洁，应用领域对象只有四个字段。

@Data
public class CustomerDomain implements DomainSearch {

  /** 店铺ID */
  @EsField(name="shop_id", required = true)
  private Long shopId;

  /** 订单编号 */
  @EsField(name="order_no")
  private String orderNo;

  /** 订单状态 */
  @EsField(name="state", type="list")
  private List<Integer> state;

  /** 订单类型 */
  @EsField(name="order_type", type="list")
  private List<Integer> orderType;

}

注解解析器

接着，需要提供注解解析器，将对应的映射关系转成ES查询对象的一部分。

• 使用接口的默认方法来实现，是为了支持不同的业务类自动可以转化为ES查询串；
• 注解解析器需要使用Java反射机制，来获取相应的字段，以及字段上的注解定义，然后根据字段的类型、值、注解定义来做相应处理；
• 使用反射来处理字段时，由于字段一般是私有的，因此必须先设置为可访问的，处理完成后还原为不可访问；
• EsField field = f.getAnnotation(EsField.class) 用来获取字段上的注解信息（name, type, required）；Object value = f.get(customerDomain) 用来获取字段的值；字段的其他类型信息可以通过 Field 的方法拿到。

public interface DomainSearch {

  Log logger = LogFactory.getLog(DomainSearch.class);

  default String toEsQuery() {
    Object customerDomain = this;
    EsQuery esQuery = new EsQuery();
    Field[] fields = this.getClass().getDeclaredFields();
    for (Field f: fields) {
      try {
        if (Modifier.isStatic(f.getModifiers())) {
          continue;
        }

        f.setAccessible(true);

        Object value = f.get(customerDomain);

        if (f.getAnnotation(EsField.class) != null) {
          EsField field = f.getAnnotation(EsField.class);

          if (field.required() && value == null) {
            throw new RuntimeException("field '" + field + "' is required. value is null");
          }

          if (isNeedOmitted(value)) {
            f.setAccessible(false);
            continue;
          }

          if ((value instanceof List) && ((List)value).size() == 1) {
            // 针对 List 中单个值做优化查询
            esQuery = esQuery.addTermFilter(field.name(), ((List)value).get(0));
          }
          else {
            esQuery = esQuery.addTermFilter(field.name(), value);
          }
        }

        f.setAccessible(false);

      } catch (Exception ex) {
        logger.error("failed to build es query for field: " + f.getName(), ex);
        throw new RuntimeException(ex.getCause());
      }
    }
    return esQuery.toJsonString();
  }

  /**
   * 判断是否需要忽略该字段的查询
   * @param value 字段值
   * @return 是否要忽略
   */
  default boolean isNeedOmitted(Object value) {
    if (value == null) {
      return true;
    }

    // 空字符串搜索值忽略
    if ((value instanceof String) && StringUtils.isBlank(value.toString())) {
      return true;
    }

    // 空列表串忽略
    if ((value instanceof List) && ((List)value).isEmpty()) {
      return true;
    }
    return false;
  }

}

查询对象

ES查询对象将所有生成的查询条件转化为ES可以接受的查询字符串。

public class EsQuery {

  private static int DEFAULT_SIZE = 100;

  private final Map<String, Object> termFilter;
  private final Map<String, Range> rangeFilter;
  private final Map<String, Match> matchFilter;
  private int size;
  private String orderBy = null;
  private String order = null;

  // query 查询语法, 是否需要 filtered, filter 这两层
  // 5.x 版本不再需要这两层
  private boolean isNeedFilterLayer = true;

  private Integer from;

  private final Map<String, Object> mustNotTermFilter;

  private final Map<String,Object> shouldTermFilter;
  private Integer shouldMatchMinimum;

  private List<String> includes;
  private List<String> excludes;

  public EsQuery() {
    this.termFilter = new HashMap<>();
    this.rangeFilter = new HashMap();
    this.matchFilter = new HashMap();
    this.mustNotTermFilter = new HashMap<>();
    this.shouldTermFilter = new HashedMap();
    this.size = DEFAULT_SIZE;
    this.includes = new ArrayList<>();
    this.excludes = new ArrayList<>();
  }

  public EsQuery addTermFilter(String key, Object value) {
    this.termFilter.put(key, value);
    return this;
  }

  public EsQuery addMustNotTermFilter(String key, Object value) {
    this.mustNotTermFilter.put(key, value);
    return this;
  }

  public EsQuery addAllMustNotTermFilter(Map<String,Object> mustNot) {
    if (mustNot != null && !mustNot.isEmpty()) {
      this.mustNotTermFilter.putAll(mustNot);
    }
    return this;
  }

  public EsQuery addShouldTermFilter(String key, Object value) {
    this.shouldTermFilter.put(key, value);
    return this;
  }

  public EsQuery addAllShouldTermFilter(Map<String,Object> should) {
    if (should != null && !should.isEmpty()) {
      this.shouldTermFilter.putAll(should);
    }
    return this;
  }

  public EsQuery addRangeFilter(String key, long gte, long lte){
    this.rangeFilter.put(key, new Range(gte, lte));
    return this;
  }

  public EsQuery addMatchFilter(String key, Match value) {
    this.matchFilter.put(key, value);
    return this;
  }

  public EsQuery addIncludeFields(List<String> includes) {
    this.includes.addAll(includes);
    return this;
  }

  public EsQuery addExcludeFields(List<String> excludes) {
    this.excludes.addAll(excludes);
    return this;
  }


  @Override
  public String toString() {
    return toJsonString();
  }

  public String toJsonString() {
    Map<String, Object> finalQuery = new HashMap<>();
    Map<String, Object> queryMap = new HashMap<>();
    Map<String, Object> filteredMap = new HashMap<>();
    Map<String, Object> filterMap = new HashMap<>();
    Map<String, Object> boolMap = new HashMap<>();

    List<Object> mustList = obtainTermFilterList(this.termFilter);

    List<Object> mustNotList = obtainTermFilterList(this.mustNotTermFilter);

    List<Object> shouldList = obtainTermFilterList(this.shouldTermFilter);

    if(!this.rangeFilter.isEmpty()){
      for(Map.Entry<String, Range> e: this.rangeFilter.entrySet()){
        Map<String, Object> rangeMap = new HashMap<>();
        Map<String, Object> rangeEntityMap = new HashMap<>();
        rangeEntityMap.put(e.getKey(), e.getValue().toMap());
        rangeMap.put(Constant.range, rangeEntityMap);
        mustList.add(rangeMap);
      }
    }

    if(!this.matchFilter.isEmpty()){
      this.matchFilter.forEach(
          (key, match) -> {
            Map<String, String> matchEntityMap = new HashMap<>();
            Map<String, Map> matchMap = new HashMap<>();
            Map<String, Map> subMatchMap = new HashMap<>();
            matchEntityMap.put(Constant.query, match.getQuery());
            matchEntityMap.put(Constant.should_minum, match.getMinimumShouldMatch());
            matchMap.put(key, matchEntityMap);
            subMatchMap.put(Constant.match, matchMap);
            mustList.add(subMatchMap);
          });
    }

    boolMap.put(Constant.must, mustList);
    if (!mustNotList.isEmpty())
      boolMap.put(Constant.mustNot, mustNotList);
    if (!shouldList.isEmpty()) {
      // 有 minimum_should_match 不带过滤器
      boolMap.put(Constant.should, shouldList);
      boolMap.put(Constant.should_minum, shouldMatchMinimum);
      queryMap.put(Constant.bool, boolMap);
    }
    else {
      if (isNeedFilterLayer) {
        filterMap.put(Constant.bool, boolMap);
        filteredMap.put(Constant.filter, filterMap);
        queryMap.put(Constant.filtered, filteredMap);
      }
      else {
        queryMap.put(Constant.bool, boolMap);
      }
    }
    finalQuery.put(Constant.query, queryMap);

    Map<String, Object> orderMap = new HashMap<>();
    Map<String, Object> orderItem = new HashMap<>();

    if(order != null && orderBy != null){
      orderItem.put(Constant.order, this.order);
      orderMap.put(this.orderBy, orderItem);
      finalQuery.put(Constant.sort, orderMap);
    }

    Map<String,Object> source = new HashMap<>();
    if (!includes.isEmpty()) {
      source.put(Constant.includes, this.includes);
    }
    if (!excludes.isEmpty()) {
      source.put(Constant.excludes, this.excludes);
    }
    if (!source.isEmpty()) {
      finalQuery.put(Constant.source, source);
    }

    finalQuery.put(Constant.size, this.size);
    if (from != null) {
      finalQuery.put(Constant.from, from.intValue());
    }
    return JSON.toJSONString(finalQuery);
  }

  public List<Object> obtainTermFilterList(Map<String, Object> termFilter) {
    List<Object> termFilterList = new ArrayList<>();
    for (Map.Entry<String, Object> e: termFilter.entrySet()){
      Map<String, Object> termMap = new HashMap<>();
      Map<String, Object> itemMap = new HashMap<>();
      itemMap.put(e.getKey(), e.getValue());
      if(e.getValue() instanceof List){
        termMap.put(Constant.terms, itemMap);
      }else{
        termMap.put(Constant.term, itemMap);
      }
      termFilterList.add(termMap);
    }
    return termFilterList;
  }

  public String getOrderBy() {
    return orderBy;
  }

  public void setOrderBy(String orderBy) {
    this.orderBy = orderBy;
  }

  public String getOrder() {
    return order;
  }

  public void setOrder(String order) {
    this.order = order;
  }

  public int getSize() {
    return size;
  }

  public void setSize(int size) {
    this.size = size;
  }

  public Integer getFrom() {
    return from;
  }

  public void setFrom(Integer from) {
    this.from = from;
  }

  public Map<String, Object> getTermFilter() {
    return Collections.unmodifiableMap(termFilter);
  }

  public Map<String, Range> getRangeFilter() {
    return Collections.unmodifiableMap(rangeFilter);
  }

  public Map<String, Object> getMustNotTermFilter() {
    return Collections.unmodifiableMap(mustNotTermFilter);
  }

  public Map<String, Object> getShouldTermFilter() {
    return Collections.unmodifiableMap(shouldTermFilter);
  }

  public Map<String, Match> getMatchFilter() {
    return matchFilter;
  }

  public void setShouldMatchMinimum(Integer shouldMatchMinimum) {
    this.shouldMatchMinimum = shouldMatchMinimum;
  }

  public Integer getShouldMatchMinimum() {
    return shouldMatchMinimum;
  }

  public Map<String, Object> getRangeMap(String key) {
    return Collections.unmodifiableMap(rangeFilter.get(key).toMap());
  }

  public List<String> getIncludes() {
    return Collections.unmodifiableList(includes);
  }

  public boolean isNeedFilterLayer() {
    return isNeedFilterLayer;
  }

  public void setNeedFilterLayer(boolean needFilterLayer) {
    isNeedFilterLayer = needFilterLayer;
  }

  @Override
  public boolean equals(Object o) {
    // for you to write
  }

  @Override
  public int hashCode() {
    // for you to write
}

小结

通过ES搜索示例，展示了如何运用注解自动化处理领域对象属性与底层ES存储字段之间的对应关系。实际上，如果想为应用对象或组件添加某种说明或注释，不妨先想想是否可以通过注解自动化处理。注解亦可用于框架自动处理对象与组件的集成。Spring框架的Resource, Component, AOP，以及 Plugin 化设计思想等都是好的应用例子。