关系图谱服务的技术方案设计(下篇)

453 阅读15分钟

在二手车业务线,现阶段无法实现车辆、人、车商信息的在业务审核流程中的数据查重应用,因此业务方为了达成这一目标,基于数据采集和数据查询,应运而生了关系图谱服务。

本文由于字数太多分为上下篇完成。

关系图谱服务的技术方案设计-上篇

3.2、数据查重

数据查重

上图是整个代码设计的UML类图,从上述图可以看出:

  • 1、第一层的"API"层,对外提供HTTP接口。
  • 2、第二层的"查询执行器",包装API层的请求,并交由查询处理器处理。
  • 3、第三层的"查询处理器",抽象类仅派生一种业务场景多条件查询处理器MultiSearchHandler
  • 3、第四层的"命中查询处理器",按照业务场景派生出三个子类PersonInfoHitQuerierVehicleInfoHitQuerierDealerInfoHitQuerier,同时提供了两种数据命中模式ExactSearchModeSimilarSearchMode

3.2.1、MultiSearchRequest

/**
 * @description: 多条件查重查询条件DTO
 * @Date : 2020/12/15 3:17 PM
 * @Author : 石冬冬-Seig Heil
 */
@Data
@Builder
@NoArgsConstructor
@AllArgsConstructor
public class MultiSearchRequest {

    @ApiModelProperty(value="请求ID(可以使用UUID生成)",dataType="java.lang.String")
    @NotNull(message = "请求ID[requestId]非空")
    private String requestId;

    @ApiModelProperty(value="查询条件",dataType="List<Condition>")
    @NotNull(message = "查询条件[conditions]非空")
    private List<Condition> conditions;

    /**
     * @description: 查询条件
     * @Date : 2020/12/17 3:42 PM
     * @Author : 石冬冬-Seig Heil
     */
    @Data
    @Builder
    @NoArgsConstructor
    @AllArgsConstructor
    public static class Condition{

        @ApiModelProperty(value="查询字段类型",dataType="SourceTypeEnum")
        private SourceTypeEnum sourceType;

        @ApiModelProperty(value="查询字段名称",dataType="java.lang.String")
        @NotNull(message = "查询字段名称[searchFieldName]不能为空!")
        @NotEmpty(message = "查询字段名称[searchFieldName]不能为空!")
        private String searchFieldName;

        @ApiModelProperty(value="模糊查询分数区间",dataType="java.lang.Double")
        @Size(max = 2, message = "查询字段名称[scoreRange]列表长度应该0-2")
        private List<Double> scoreRange;

        @ApiModelProperty(value="查询字段输入值",dataType="java.lang.String")
        @NotNull(message = "查询字段输入值[searchFieldValue]不能为空!")
        @NotEmpty(message = "查询字段输入值[searchFieldValue]不能为空!")
        private String searchFieldValue;

        @ApiModelProperty(value="查询字段描述",dataType="java.lang.String")
        @NotNull(message = "查询字段描述[searchFieldDesc]不能为空!")
        @NotEmpty(message = "查询字段描述[searchFieldDesc]不能为空!")
        private String searchFieldDesc;
    }

}

/**
 * @description:  来源类型(1-主贷人;2-销售;3-销售主管;4-配偶;5-担保人;6-二手车卖方;7-紧急联系人1;8-紧急联系人2;9-车商法人;10-车商负责人;11-车商收款人;12-车辆;)
 * @Date : 2020/11/27 4:08 PM
 * @Author : 石冬冬-Seig Heil
 */
public enum SourceTypeEnum implements EnumValue {

	CREDITOR_INFO(1,"主贷人"),
	SELLER_INFO(2,"销售"),
	SALES_MANAGER_INFO(3,"销售主管"),
	MATE(4,"配偶"),
	GUARANTOR_INFO(5,"担保人"),
	SELLER_INFO_OF_OLD_CAR(6,"二手车卖方"),
	EMERGENCY_CONTACT_ONE(7,"紧急联系人1"),
	EMERGENCY_CONTACT_TWO(8,"紧急联系人2"),
	CAR_DEALER_LEGAL_PERSON(9,"车商法人"),
	CAR_DEALER_PRINCIPAL(10,"车商负责人"),
	CAR_DEALER_PAYEE(11,"车商收款人"),
	VEHICLE_INFO(12,"车辆"),
	;


	private int index;
	private String value;

	SourceTypeEnum(int index, String value ){
		this.value = value;
		this.index = index;
	}

	@Override
	public int getIndex() {
		return index;
	}

	@Override
	public String getName() {
		return value;
	}

	/**
	 * 根据索引获取对象
	 * @param index
	 * @return
	 */
	public static SourceTypeEnum getByIndex(int index){
		return Stream.of(SourceTypeEnum.values()).filter(each -> each.getIndex() == index).findFirst().get();
	}
	/**
	 * 根据索引获取名称
	 * @param index
	 * @return
	 */
	public static String getNameByIndex(int index){
		SourceTypeEnum find = Stream.of(SourceTypeEnum.values()).filter(each -> each.getIndex() == index).findFirst().get();
		return null == find ? "" : find.getName();
	}
}

3.2.2、SearchResultRe

/**
 * @description: 查重结果对象
 * @Date : 2020/12/15 3:29 PM
 * @Author : 石冬冬-Seig Heil
 */
@Data
@Builder
@NoArgsConstructor
@AllArgsConstructor
public class SearchResultRe<E extends SearchResultRe.Record>{

    @ApiModelProperty(value="提示信息",dataType="java.lang.String")
    private String message;

    @ApiModelProperty(value="查询字段名称",dataType="java.lang.String")
    private String searchFieldName;

    @ApiModelProperty(value="查询字段输入值",dataType="java.lang.String")
    private String searchFieldValue;

    @ApiModelProperty(value="查询字段描述",dataType="java.lang.String")
    private String searchFieldDesc;

    @ApiModelProperty(value="命中条数",dataType="java.lang.Integer")
    private int hitCount;

    @ApiModelProperty(value="命中记录",dataType="List<Record>")
    private List<E> hitRecords;


    /**
     * 累加命中条数
     * @param delta
     */
    public synchronized void increaseHitCount(int delta){
        this.hitCount += delta;
    }

    /**
     * 累加命中记录
     * @param hits
     */
    public synchronized void addHits(List<E> hits){
        if(!isEmpty(hits)){
            if(isEmpty(this.hitRecords)){
                this.hitRecords = new ArrayList<>(hits.size());
            }
            this.hitRecords.addAll(hits);
        }
    }

    /**
     * 判断集合元素是否为空
     * @param collection
     * @return
     */
    boolean isEmpty(Collection<?> collection){
        return null == collection || collection.isEmpty();
    }

    /**
     * @description: 查重结果对象
     * @Date : 2020/12/15 3:29 PM
     * @Author : 石冬冬-Seig Heil
     */
    @Data
    @SuperBuilder
    @NoArgsConstructor
    @AllArgsConstructor
    public static class Record{

        @ApiModelProperty(value="主键ID",dataType="java.lang.Integer")
        private Integer id;

        @ApiModelProperty(value="外部数据id",dataType="java.lang.String")
        private String externalId;

        @ApiModelProperty(value="业务单号",dataType="java.lang.String")
        private String appCode;

        @ApiModelProperty(value="数据单号",dataType="java.lang.String")
        private String dataCode;

        /**
         * {@link SceneEnum#getIndex()}
         */
        @ApiModelProperty(value="场景",dataType="java.lang.Integer")
        private Integer scene;
        private String sceneDesc;
        /**
         * {@link SourceTypeEnum#getIndex()}
         */
        @ApiModelProperty(value="来源类型",dataType="java.lang.Integer")
        private Integer sourceType;
        private String sourceTypeDesc;

    }

    /**
     * @description: 自然人信息
     * @Date : 2020/12/15 3:29 PM
     * @Author : 石冬冬-Seig Heil
     */
    @Data
    @SuperBuilder
    @NoArgsConstructor
    public static class PersonInfoRecord extends Record{

        @ApiModelProperty(value="姓名",dataType="java.lang.String")
        private String name;

        @ApiModelProperty(value="身份证号",dataType="java.lang.String")
        private String idNo;

        @ApiModelProperty(value="手机号",dataType="java.lang.String")
        private String mobile;

        @ApiModelProperty(value="银行卡号",dataType="java.lang.String")
        private String creditCardNo;

        @ApiModelProperty(value="省",dataType="java.lang.String")
        private String provinceAddress;

        @ApiModelProperty(value="市",dataType="java.lang.String")
        private String cityAddress;

        @ApiModelProperty(value="区",dataType="java.lang.String")
        private String districtAddress;

        @ApiModelProperty(value="详细地址",dataType="java.lang.String")
        private String detailAddress;

    }


    /**
     * @description: 车辆信息
     * @Date : 2020/12/15 3:29 PM
     * @Author : 石冬冬-Seig Heil
     */
    @Data
    @SuperBuilder
    @NoArgsConstructor
    public static class VehicleInfoRecord extends Record{

        @ApiModelProperty(value="VIN码",dataType="java.lang.String")
        private String vin;

        @ApiModelProperty(value="公里数",dataType="java.lang.Integer")
        private Integer mileage;

        @ApiModelProperty(value="评估备注",dataType="java.lang.String")
        private String evaluateRemark;

    }
}

3.2.3、SearchController

/**
 * @description: 查重数据
 * @Date : 2020/12/15 4:03 PM
 * @Author : 石冬冬-Seig Heil
 */
@RestController
@Api(description = "查重数据", tags = "查重数据")
@RequestMapping("/search")
public class SearchController {

    @Autowired
    SearchQueryExecutor searchQueryExecutor;

    /**
     * 多条件查询
     * @return
     */
    @PostMapping("/multiQuery")
    @ApiOperation(value = "多条件查询", notes = "多条件查询")
    @NoAuthRequired
    @OvalValidator(value = "多条件查询[multiQuery]")
    public Result<List<SearchResultRe>> multiQuery(@RequestBody MultiSearchRequest request) {
        return searchQueryExecutor.execute(SearchQueryExecutor.Type.MULTI,request);
    }
}

3.2.4、SearchQueryExecutor

/**
 * @description: 查询执行器
 * @Date : 2020/12/16 11:27 AM
 * @Author : 石冬冬-Seig Heil
 */
@Component
public class SearchQueryExecutor {

    final Map<Type,AbstractSearchHandler> HANDLER_MAP = Maps.newHashMap();

    @Resource
    MultiSearchHandler multiSearchHandler;

    @PostConstruct
    void init(){
        HANDLER_MAP.put(Type.MULTI, multiSearchHandler);
    }

    public enum Type{
        /**
         * 单项查询
         */
        SIMPLE,
        /**
         * 多项
         */
        MULTI
    }

    /**
     * 执行
     * @param type
     * @param request
     * @param <T>
     */
    public <T> Result<List<SearchResultRe>> execute(Type type, T request){
        if(Type.MULTI == type){
            SearchQueryContext<MultiSearchRequest> context = SearchQueryContext.<MultiSearchRequest>builder().param((MultiSearchRequest)request).build();
            HANDLER_MAP.get(type).execute(context);
            return Result.suc(context.getResults());
        }
        return Result.suc();
    }
}

3.2.5、SearchQueryContext

/**
 * @description: 数据查重上下文对象
 * @Date : 2020/12/16 11:06 AM
 * @Author : 石冬冬-Seig Heil
 */
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class SearchQueryContext<T> {
    /**
     * 请求ID
     */
    private String requestId;
    /**
     * 查询条件
     */
    private T param;
    /**
     * 查询结果
     */
    private List<SearchResultRe> results;
    /**
     * 查询是否成功
     */
    private boolean success;
    /**
     * 业务信息
     */
    private String message;


    public SearchQueryContext withSuccess(boolean success,String message){
        this.setSuccess(success);
        this.setMessage(message);
        return this;
    }
}

3.2.6、AbstractSearchHandler

/**
 * @description: 抽象查询处理器
 * @Date : 2020/12/16 11:08 AM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
public abstract class AbstractSearchHandler<T> {
    /**
     * 上下文容器
     */
    final ThreadLocal<SearchQueryContext<T>> CONTEXT = new ThreadLocal<>();

    @Resource
    DiamondConfigProxy diamondConfigProxy;

    @Resource
    HitQuerierManager hitQuerierManager;

    @Resource
    ExecutorService searchQueryPoolExecutor;


    /**
     * 外部调用执行方法
     * @param context
     */
    public void execute(SearchQueryContext<T> context){
        try {
            CONTEXT.set(context);
            doQuery(context);
        } catch (Exception e) {
            log.error("[execute],requestId={},ctx={}",context.getRequestId(),JSONObject.toJSONString(context),e);
        }finally {
            CONTEXT.remove();
        }
    }

    /**
     * 执行查询
     * @param context
     */
    abstract void doQuery(SearchQueryContext<T> context);

    /**
     * 构建一个异常的查询结果
     * @param condition
     * @param message
     * @return
     */
    SearchResultRe buildExceptionSearchResult(MultiSearchRequest.Condition condition,String message){
        return SearchResultRe.builder()
                .searchFieldName(condition.getSearchFieldName())
                .searchFieldValue(condition.getSearchFieldValue())
                .searchFieldDesc(condition.getSearchFieldDesc())
                .message(message)
                .hitRecords(Collections.emptyList())
                .build();
    }
}

3.2.7、MultiSearchHandler

多条件查询处理器,这里采用多线程,以条件维度,以子线程处理,主线程阻塞等待所有子线程处理,并把命中数据结果统一封装返回。

/**
 * @description: 抽象查询处理器
 * @Date : 2020/12/16 11:08 AM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class MultiSearchHandler extends AbstractSearchHandler<MultiSearchRequest> {

    @Override
    void doQuery(SearchQueryContext<MultiSearchRequest> context) {
        context.setRequestId(context.getParam().getRequestId());
        String requestId = context.getRequestId();
        //条件列表
        List<MultiSearchRequest.Condition> conditions = context.getParam().getConditions();
        //结果列表
        List<SearchResultRe> results = Lists.newArrayListWithExpectedSize(conditions.size());
        //Future任务列表
        List<Future<HitQuerierContext>> futureList = new ArrayList<>(conditions.size());

        for(MultiSearchRequest.Condition condition : conditions){
            //参数验证
            if(!checkCondition(condition)){
                results.add(super.buildExceptionSearchResult(condition, MessageFormat.format("查询异常|参数非法,field={0}",condition.getSearchFieldName())));
                continue;
            }
            Future<HitQuerierContext> future = searchQueryPoolExecutor.submit(() -> {
                HitQuerierContext hitQuerierContext = HitQuerierContext.builder()
                        .requestId(requestId)
                        .condition(condition)
                        .build();
                hitQuerierManager.execute(hitQuerierContext);
                return hitQuerierContext;
            });
            futureList.add(future);
        }
        int count = 0;
        for (Future<HitQuerierContext> f : futureList) {
            try {
                HitQuerierContext hitQuerierContext = f.get();
                results.add(hitQuerierContext.getHitResult());
            } catch (InterruptedException | ExecutionException e) {
                log.error("[doQuery][InterruptedException|ExecutionException],requestId={},context={}",requestId, JSONObject.toJSON(context),e);
                results.add(super.buildExceptionSearchResult(conditions.get(count), MessageFormat.format("查询异常|运行时异常,message={0}",e.getMessage())));
                continue;
            }
            count++;
        }

        context.setResults(results);
        context.withSuccess(Boolean.TRUE,"查询成功");
    }

    /**
     * 参数验证
     * @param condition
     */
    boolean checkCondition(MultiSearchRequest.Condition condition){
        Set<String> configFields = diamondConfigProxy.searchFieldConfig().keySet();
        log.info("[checkCondition],configFields={}", configFields.toArray());
        String searchFieldName = condition.getSearchFieldName();
        return configFields.contains(searchFieldName);
    }
}

3.2.8、HitQuerierManager

命中查询管理器,由于一个查询条件会去多张表进行数据查询,然后再数据合并,因此这里吧查询器统一注册给hitList,每个请求则统一调用execute方法循环迭代所有查询器处理。

/**
 * @description: 命中查询管理器
 * @Date : 2020/12/17 4:21 PM
 * @Author : 石冬冬-Seig Heil
 */
@Component
public class HitQuerierManager {

    static final List<AbstractHitQuerier> hitList = new ArrayList<>();

    @Resource
    PersonInfoHitQuerier personInfoHitQuerier;

    @Resource
    DealerInfoHitQuerier dealerInfoHitQuerier;

    @Resource
    VehicleInfoHitQuerier vehicleInfoHitQuerier;

    @PostConstruct
    void init(){
        hitList.add(personInfoHitQuerier);
        hitList.add(dealerInfoHitQuerier);
        hitList.add(vehicleInfoHitQuerier);
    }

    /**
     * 执行查询
     * @param context
     */
    public void execute(HitQuerierContext context){
        MultiSearchRequest.Condition condition = context.getCondition();
        SearchResultRe hitResult = SearchResultRe.builder()
                .searchFieldValue(condition.getSearchFieldValue())
                .searchFieldName(condition.getSearchFieldName())
                .searchFieldDesc(condition.getSearchFieldDesc())
                .message("查询成功")
                .build();
        for(AbstractHitQuerier querier : hitList){
            SearchResultRe hit = querier.execute(context);
            hitResult.increaseHitCount(hit.getHitCount());
            hitResult.addHits(hit.getHitRecords());
        }
        context.setHitResult(hitResult);
    }
}

3.2.9、HitQuerierContext

/**
 * @description: 数据查询器上下文
 * @Date : 2020/12/17 3:38 PM
 * @Author : 石冬冬-Seig Heil
 */
@Data
@Builder
@AllArgsConstructor
@NoArgsConstructor
public class HitQuerierContext {
    /**
     * 请求ID
     */
    private String requestId;
    /**
     * 查询条件
     */
    private MultiSearchRequest.Condition condition;
    /**
     * 查询结果
     */
    private SearchResultRe hitResult;
}

3.2.10、AbstractHitQuerier

AbstractHitQuerier

/**
 * @description: 抽象命中查询器
 * @Date : 2020/12/17 3:38 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
public abstract class AbstractHitQuerier<E extends SearchResultRe.Record> {
    /**
     * 处理查询table
     */
    protected final SearchFieldConfig.Table table;

    @Resource
    DiamondConfigProxy diamondConfigProxy;

    public AbstractHitQuerier(SearchFieldConfig.Table table) {
        this.table = table;
    }

    /**
     * 查询
     * @param context
     * @return
     */
    abstract SearchResultRe<E> doQuery(HitQuerierContext context);

    /**
     * 外部执行方法
     * @param context
     * @return
     */
    public SearchResultRe execute(HitQuerierContext context){
        SearchResultRe searchResultRe;
        try {
            if(!executeCurrent(context)){
                searchResultRe = buildDefaultSearchResult(context.getCondition());
                return searchResultRe;
            }
            searchResultRe = doQuery(context);
        } catch (Exception e) {
            log.error("[execute],requestId={},ctx={}",context.getRequestId(),JSONObject.toJSONString(context),e);
            searchResultRe = buildDefaultSearchResult(context.getCondition());
            searchResultRe.setMessage("查询异常|message=" + e.getMessage());
        }
        return searchResultRe;
    }

    /**
     * 是否执行当前查询器
     * @param context
     * @return
     */
    protected boolean executeCurrent(HitQuerierContext context){
        return getSearchFieldConfig(context.getCondition().getSearchFieldName()).getTables().contains(table);
    }

    /**
     * 构建默认查询结果
     * @param condition
     * @return
     */
    SearchResultRe buildDefaultSearchResult(MultiSearchRequest.Condition condition){
        String fieldName = condition.getSearchFieldName();
        String fieldValue = condition.getSearchFieldValue();
        SearchResultRe hitResult = SearchResultRe.builder()
                .searchFieldName(fieldName)
                .searchFieldValue(fieldValue)
                .hitCount(0)
                .hitRecords(Collections.emptyList())
                .build();
        return hitResult;
    }

    /**
     * 获取配置字段
     * @param fieldName
     * @return
     */
    SearchFieldConfig getSearchFieldConfig(String fieldName){
        Map<String,SearchFieldConfig> mapping = diamondConfigProxy.searchFieldConfig();
        SearchFieldConfig searchFieldConfig = mapping.get(fieldName);
        log.info("[getSearchFieldConfig],fieldName={},config={}",fieldName, JSONObject.toJSONString(searchFieldConfig));
        return searchFieldConfig;
    }

    /**
     * 处理命中数据
     * @param context
     * @param queryForm
     * @param recordsCaller
     * @return
     */
    List<E> hitRecords(HitQuerierContext context, Object queryForm, BiFunction<String,Object,List<E>> recordsCaller){
        MultiSearchRequest.Condition condition = context.getCondition();
        SearchFieldConfig searchFieldConfig = getSearchFieldConfig(condition.getSearchFieldName());
        SearchFieldConfig.SearchMode searchMode = searchFieldConfig.getSearchMode();

        return AbstractSearchMode.getAbstractSearchMode(searchMode).execute(searchFieldConfig, context, queryForm, recordsCaller);
    }

}


/**
 * @description: 车商信息命中查询
 * @Date : 2020/12/17 3:51 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class DealerInfoHitQuerier extends AbstractHitQuerier<SearchResultRe.PersonInfoRecord> {

    public DealerInfoHitQuerier() {
        super(SearchFieldConfig.Table.DEALER_INFO);
    }

    @Resource
    DealerInfoService dealerInfoService;

    @Override
    SearchResultRe<SearchResultRe.PersonInfoRecord> doQuery(HitQuerierContext context) {
        MultiSearchRequest.Condition condition = context.getCondition();
        SearchResultRe hitResult = super.buildDefaultSearchResult(condition);
        DealerInfoForm queryForm = DealerInfoForm.builder().dataStatus(0).build();
        List<SearchResultRe.PersonInfoRecord> hitRecords = super.hitRecords(context,queryForm,(String fieldName,Object searchForm) -> {
            List<DealerInfo> recordList = dealerInfoService.queryList((DealerInfoForm)searchForm);
            SearchFieldConfig config = getSearchFieldConfig(condition.getSearchFieldName());
            if(SearchFieldConfig.SearchMode.SIMILAR == config.getSearchMode() && condition.getSearchFieldName().contains("Address")){
                String provinceAddressFiledName = config.getMapping().get(config.getMapping().size()-1);
                String cityAddressFiledName = config.getMapping().get(config.getMapping().size()-1);
                String districtAddressFiledName = config.getMapping().get(config.getMapping().size()-1);
                String detailAddressFiledName = config.getMapping().get(config.getMapping().size()-1);
                return BeanConverter.convertFromDealer(recordList,
                        each -> BeanTool.getObjectValue(each, provinceAddressFiledName),
                        each -> BeanTool.getObjectValue(each, cityAddressFiledName),
                        each -> BeanTool.getObjectValue(each, districtAddressFiledName),
                        each -> BeanTool.getObjectValue(each, detailAddressFiledName));
            }
            return BeanConverter.convertFromDealer(recordList,
                    null, null, null,null);
        });
        hitResult.setHitRecords(hitRecords);
        hitResult.setHitCount(hitRecords.size());
        return hitResult;
    }
}



/**
 * @description: 自然人信息命中查询
 * @Date : 2020/12/17 3:51 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class PersonInfoHitQuerier extends AbstractHitQuerier<SearchResultRe.PersonInfoRecord> {
    /**
     * 手机号字段输入
     */
    static final String MOBILE_FIELD = "mobile";
    /**
     * 地址类字段命名后缀
     */
    static final String ADDRESS_FIELD_SUFFIX = "Address";

    public PersonInfoHitQuerier() {
        super(SearchFieldConfig.Table.PERSON_INFO);
    }

    @Resource
    PersonInfoService personInfoService;

    @Override
    SearchResultRe<SearchResultRe.PersonInfoRecord> doQuery(HitQuerierContext context) {
        MultiSearchRequest.Condition condition = context.getCondition();
        SearchResultRe hitResult = super.buildDefaultSearchResult(condition);
        PersonInfoForm queryForm = PersonInfoForm.builder().dataStatus(0).build();
        List<SearchResultRe.PersonInfoRecord> hitRecords = super.hitRecords(context,queryForm,(String fieldName,Object searchForm) -> {
            List<PersonInfo> recordList = personInfoService.queryList((PersonInfoForm)searchForm);
            if(MOBILE_FIELD.equals(condition.getSearchFieldName())){
                return BeanConverter.convertFromPerson(recordList, each -> BeanTool.getObjectValue(each, fieldName),
                        null, null, null, null);
            }
            SearchFieldConfig config = getSearchFieldConfig(condition.getSearchFieldName());
            if(SearchFieldConfig.SearchMode.SIMILAR == config.getSearchMode() && condition.getSearchFieldName().contains(ADDRESS_FIELD_SUFFIX)){
                String provinceAddressFiledName = config.getMapping().get(0);
                String cityAddressFiledName = config.getMapping().get(1);
                String districtAddressFiledName = config.getMapping().get(2);
                String detailAddressFiledName = config.getMapping().get(3);
                return BeanConverter.convertFromPerson(recordList, null,
                        each -> BeanTool.getObjectValue(each, provinceAddressFiledName),
                        each -> BeanTool.getObjectValue(each, cityAddressFiledName),
                        each -> BeanTool.getObjectValue(each, districtAddressFiledName),
                        each -> BeanTool.getObjectValue(each, detailAddressFiledName));
            }
            return BeanConverter.convertFromPerson(recordList);
        });
        hitResult.setHitRecords(hitRecords);
        hitResult.setHitCount(hitRecords.size());
        return hitResult;
    }
}

/**
 * @description: 车辆信息命中查询
 * @Date : 2020/12/17 3:51 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class VehicleInfoHitQuerier extends AbstractHitQuerier<SearchResultRe.VehicleInfoRecord> {

    public VehicleInfoHitQuerier() {
        super(SearchFieldConfig.Table.VEHICLE_INFO);
    }

    @Resource
    VehicleInfoService vehicleInfoService;

    @Override
    SearchResultRe<SearchResultRe.VehicleInfoRecord> doQuery(HitQuerierContext context) {
        MultiSearchRequest.Condition condition = context.getCondition();
        SearchResultRe hitResult = super.buildDefaultSearchResult(condition);
        VehicleInfoForm queryForm = VehicleInfoForm.builder().dataStatus(0).build();
        List<SearchResultRe.VehicleInfoRecord> hitRecords = super.hitRecords(context,queryForm,(String fieldName,Object searchForm) -> {
            List<VehicleInfo> recordList = vehicleInfoService.queryList((VehicleInfoForm)searchForm);
            return BeanConverter.convertFromVehicle(recordList);
        });
        hitResult.setHitRecords(hitRecords);
        hitResult.setHitCount(hitRecords.size());
        return hitResult;
    }
}

3.2.11、AbstractSearchMode

AbstractSearchMode

/**
 * @description: 命中查询方式
 * @see com.creditease.horus.core.search.querier.mode.impl.ExactSearchMode
 * @see com.creditease.horus.core.search.querier.mode.impl.SimilarSearchMode
 * @Date : 2020/12/24 3:27 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
public abstract class AbstractSearchMode {

    /**
     * 策略集合
     */
    public static Map<SearchFieldConfig.SearchMode, AbstractSearchMode> abstractSearchModeMap = new HashMap<>();

    /**
     * 选择策略
     * @param modeEnum
     * @return
     */
    public static AbstractSearchMode getAbstractSearchMode(SearchFieldConfig.SearchMode modeEnum){
        return abstractSearchModeMap.get(modeEnum);
    }

    /**
     * 策略抽象方法
     * @param searchFieldConfig
     * @param context
     * @param queryForm
     * @param recordsCaller
     * @param <E>
     * @return
     */
    public abstract <E extends SearchResultRe.Record> List<E> execute(SearchFieldConfig searchFieldConfig,
                                                                      HitQuerierContext context,
                                                                      Object queryForm,
                                                                      BiFunction<String, Object, List<E>> recordsCaller);

    /**
     * 拷贝配置的扩展参数到实体类
     * @param searchFieldConfig
     * @param queryForm
     */
    public void copyExclude (SearchFieldConfig searchFieldConfig, Object queryForm){
        Map<String,Object> exclude = searchFieldConfig.getExclude();
        if(MapUtils.isNotEmpty(exclude)){
            BeanTool.copyFromOneMap(exclude,queryForm);
        }
    }

}



/**
 * @description: 命中查询方式-精确处理方式
 * @Date : 2020/12/24 3:28 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class ExactSearchMode extends AbstractSearchMode {

    {
        abstractSearchModeMap.put(SearchFieldConfig.SearchMode.EXACT, this);
    }

    /**
     * 精准查询
     * @param searchFieldConfig
     * @param context
     * @param queryForm
     * @param recordsCaller
     * @return
     */
    @Override
    public <E extends SearchResultRe.Record> List<E> execute(SearchFieldConfig searchFieldConfig,
                                                             HitQuerierContext context,
                                                             Object queryForm,
                                                             BiFunction<String, Object, List<E>> recordsCaller) {
        //拷贝配置的扩展参数到实体类
        copyExclude (searchFieldConfig, queryForm);
        //入参条件
        String requestId = context.getRequestId();
        String fieldName = context.getCondition().getSearchFieldName();
        String fieldValue = context.getCondition().getSearchFieldValue();
        //结果集
        List<E> hitRecords;
        //有效查询参数
        Map<String,Object> sourceValues = Maps.newHashMap();
        //入参一个参数映射成查询两个参数(入参mobile对应数据库primaryMobile,SecondMobile)
        List<String> mapping = searchFieldConfig.getMapping();
        if(null != mapping && !mapping.isEmpty()){
            hitRecords = Lists.newArrayList();
            for(String fieldNameAlias : mapping){
                sourceValues.put(fieldNameAlias,fieldValue);
                //map键值对拷贝到实体类
                BeanTool.copyFromOneMap(sourceValues,queryForm);
                log.info("[execute][hitRecords],requestId={},queryForm={}",requestId, JSONObject.toJSONString(queryForm));
                List<E> hitRecordsTemp = recordsCaller.apply(fieldNameAlias,queryForm);
                log.info("[execute][hitRecords],requestId={},hitRecordsFromDB={}",requestId,hitRecordsTemp.size());
                if(CollectionsTools.isNotEmpty(hitRecordsTemp)){
                    hitRecords.addAll(hitRecordsTemp);
                }
                //抹掉本次参数(本次参数设置为null,并拷贝到查询实体类)
                sourceValues.put(fieldNameAlias,null);
            }
        }else{
            sourceValues.put(fieldName,fieldValue);
            BeanTool.copyFromOneMap(sourceValues,queryForm);
            log.info("[execute][hitRecords],requestId={},queryForm={}",requestId,JSONObject.toJSONString(queryForm));
            hitRecords = recordsCaller.apply(fieldName,queryForm);
            log.info("[execute][hitRecords],requestId={},hitRecordsFromDB={}",requestId,hitRecords.size());
        }
        return hitRecords;
    }

}



/**
 * @description: 命中查询方式-相似度处理方式
 * @Date : 2020/12/24 3:28 PM
 * @Author : 石冬冬-Seig Heil
 */
@Slf4j
@Component
public class SimilarSearchMode extends AbstractSearchMode {
    /**
     * 相似度切分四级详细字段
     */
    final String SIMILAR_DETAIL_FIELD = "detail";
    /**
     *  输出模型,地址字段名称
     */
    final String OUT_MODEL_PROVINCE_ADDRESS = "provinceAddress";
    final String OUT_MODEL_CITY_ADDRESS = "cityAddress";
    final String OUT_MODEL_DISTRICT_ADDRESS = "districtAddress";
    final String OUT_MODEL_DETAIL_ADDRESS = "detailAddress";
    /**
     * 相似度查询类型
     */
    final String SIMILAR_TYPE = "m:organization.organization.name";


    {
        abstractSearchModeMap.put(SearchFieldConfig.SearchMode.SIMILAR, this);
    }

    /**
     * 相似度查询
     * @param context
     * @param queryForm
     * @param recordsCaller
     * @return
     */
    @Override
    public <E extends SearchResultRe.Record> List<E> execute(SearchFieldConfig searchFieldConfig,
                                                             HitQuerierContext context,
                                                             Object queryForm,
                                                             BiFunction<String, Object, List<E>> recordsCaller) {
        //拷贝配置的扩展参数到实体类
        copyExclude (searchFieldConfig, queryForm);
        //入参条件
        String requestId = context.getRequestId();
        String fieldName = context.getCondition().getSearchFieldName();
        String fieldValue = context.getCondition().getSearchFieldValue();
        //入参条件拆分(地址拆分为省、市、区、详细)
        List<String> mapping = searchFieldConfig.getMapping();
        String[] fieldValues = fieldValue.split("\\|");
        //需要分词比较的详细地址(具体字段名称)
        String similarField = mapping.get(mapping.size()-1);
        //用户传过来的详细地址
        String addressDetail = fieldValues[fieldValues.length-1];
        HsmmAddressNormalizer anm = new HsmmAddressNormalizer();
        String addressDetailFormat = fieldValues[0] + fieldValues[1] + fieldValues[2]
                + ( (HashMap<String,String>)anm.splitAddress(addressDetail) ).get(SIMILAR_DETAIL_FIELD);
        //有效查询参数
        Map<String,Object> sourceValues;
        //结果集
        List<E> hitRecords = Collections.emptyList();
        if(CollectionUtils.isEmpty(mapping)){
            log.warn("[execute]diamond mapping is null,requestId={},fieldName={}",requestId,fieldName);
            return hitRecords;
        }
        sourceValues = Maps.newHashMap();
        String[] mappingValues = mapping.toArray(new String[mapping.size()]);
        //最后一项不作为查询条件
        for (int i = 0; i < mappingValues.length - 1; i++) {
            sourceValues.put(mappingValues[i],fieldValues[i]);
        }
        BeanTool.copyFromOneMap(sourceValues,queryForm);
        log.info("[execute][hitRecords],requestId={},queryForm={}",requestId, JSONObject.toJSONString(queryForm));
        List<E> hitRecordsTemp = recordsCaller.apply(fieldName,queryForm);
        log.debug("[execute][hitRecords],requestId={},hitRecordsFromDB={}",requestId,hitRecordsTemp.size());
        if(CollectionsTools.isEmpty(hitRecordsTemp)){
            return hitRecords;
        }
        //匹配度分数区间,长度限制0-2【分数区间为空-没有分数要求;分数区间长度为1-最低分要求;分数区间长度为2-分数区间要求】
        List<Double> scoreRangeList = context.getCondition().getScoreRange();
        Double[] scoreRange = CollectionUtils.isEmpty(scoreRangeList) ? new Double[0] : scoreRangeList.toArray(new Double[scoreRangeList.size()]);
        if(scoreRange.length == 0){
            //分数区间为空-没有分数要求
            return hitRecordsTemp;
        }else{
            String provinceAddressFromDb;
            String cityAddressFromDb;
            String districtAddressFromDb;
            String detailAddressFromDb;
            String detailAddressFromDbFormat;
            Double score;
            hitRecords = Lists.newArrayListWithExpectedSize(hitRecordsTemp.size());
            for (E item : hitRecordsTemp) {
                //相似分数符合条件的添加到结果集
                provinceAddressFromDb = BeanTool.getObjectValue(item, OUT_MODEL_PROVINCE_ADDRESS);
                cityAddressFromDb = BeanTool.getObjectValue(item, OUT_MODEL_CITY_ADDRESS);
                districtAddressFromDb = BeanTool.getObjectValue(item, OUT_MODEL_DISTRICT_ADDRESS);
                detailAddressFromDb = BeanTool.getObjectValue(item, OUT_MODEL_DETAIL_ADDRESS);
                detailAddressFromDbFormat = provinceAddressFromDb + cityAddressFromDb + districtAddressFromDb
                        + ( (HashMap<String,String>)anm.splitAddress(detailAddressFromDb) ).get(SIMILAR_DETAIL_FIELD);
                score = NLPUtil.getUtil().similarity(SIMILAR_TYPE, detailAddressFromDbFormat, addressDetailFormat);
                log.info("[execute][hitRecords][similarScore],requestId={},addressDetailFormat={},addressDetailFromDb={},score={}",
                        requestId, addressDetailFormat, detailAddressFromDbFormat, score);
                //分数区间长度为1-最低分要求
                if(scoreRange.length == 1 && score >= scoreRange[0]){
                    hitRecords.add(item);
                }
                //分数区间长度为2-分数区间要求
                if(scoreRange.length == 2 && score >= scoreRange[0] && score <= scoreRange[1]){
                    hitRecords.add(item);
                }
            }
        }
        return hitRecords;
    }

}

4、扩展部分

4.1、查重服务请求参数

请求参数示例

{
    "requestId":"1",
    "conditions": [{
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "主贷人身份证号",
        "searchFieldName": "idNo",
        "searchFieldValue": "350****8114118"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "主贷人手机号",
        "searchFieldName": "mobile",
        "searchFieldValue": "186****2901"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "销售手机号",
        "searchFieldName": "mobile",
        "searchFieldValue": "182****4023"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "二手车卖方身份证号",
        "searchFieldName": "idNo",
        "searchFieldValue": "3522****2138"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "紧急联系人1手机号",
        "searchFieldName": "mobile",
        "searchFieldValue": "139****603"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "紧急联系人2手机号",
        "searchFieldName": "mobile",
        "searchFieldValue": "18****637"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "担保人身份证号",
        "searchFieldName": "idNo",
        "searchFieldValue": "350****38"
    }, {
        "sourceType": "CREDITOR_INFO",
        "searchFieldDesc": "担保人手机号",
        "searchFieldName": "mobile",
        "searchFieldValue": "15****020"
    }, {
        "sourceType": "VEHICLE_INFO",
        "searchFieldDesc": "车辆VIN",
        "searchFieldName": "vin",
        "searchFieldValue": "LFV****3721"
    }]
}

响应结果示例

部分数据脱敏展示了,比如手机号、银行卡号、身份证号。

{
    "code": 0,
    "data": [
        {
            "hitCount": 1,
            "hitRecords": [
                {
                    "appCode": "F2009111915000180101",
                    "creditCardNo": "-",
                    "dataCode": "P20121700551635",
                    "externalId": "100021309",
                    "id": 243677,
                    "idNo": "350122**4118",
                    "mobile": "186**01",
                    "name": "郑**",
                    "scene": 1003,
                    "sourceType": 1
                }
            ],
            "message": "查询成功",
            "searchFieldDesc": "主贷人身份证号",
            "searchFieldName": "idNo",
            "searchFieldValue": "3501**14118"
        },
        {
            "hitCount": 2,
            "hitRecords": [
                {
                    "appCode": "F2009151915000180101",
                    "creditCardNo": "-",
                    "dataCode": "P20121700538028",
                    "externalId": "100022351",
                    "id": 219723,
                    "idNo": "-",
                    "mobile": "186**901",
                    "name": "郑**",
                    "scene": 1002,
                    "sourceType": 8
                },
                {
                    "appCode": "F2009111915000180101",
                    "creditCardNo": "-",
                    "dataCode": "P20121700551635",
                    "externalId": "100021309",
                    "id": 243677,
                    "idNo": "3501**118",
                    "mobile": "186**01",
                    "name": "郑**",
                    "scene": 1003,
                    "sourceType": 1
                }
            ],
            "message": "查询成功",
            "searchFieldDesc": "主贷人手机号",
            "searchFieldName": "mobile",
            "searchFieldValue": "186**2901"
        },
        {
            "hitCount": 1,
            "hitRecords": [
                {
                    "appCode": "F2009151915000180106",
                    "creditCardNo": "-",
                    "dataCode": "P20121700537447",
                    "externalId": "100022605",
                    "id": 218697,
                    "idNo": "3505**5550",
                    "mobile": "182**4023",
                    "name": "欧**",
                    "scene": 1003,
                    "sourceType": 1
                }
            ],
            "message": "查询成功",
            "searchFieldDesc": "销售手机号",
            "searchFieldName": "mobile",
            "searchFieldValue": "182**023"
        },
        {
            "hitCount": 6,
            "hitRecords": [
                {
                    "appCode": "F2011091915000180104",
                    "creditCardNo": "-",
                    "dataCode": "P20121700417419",
                    "externalId": "100038555",
                    "id": 7575,
                    "idNo": "35223**38",
                    "mobile": "-",
                    "name": "阮**",
                    "scene": 1002,
                    "sourceType": 6
                },
                 //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "二手车卖方身份证号",
            "searchFieldName": "idNo",
            "searchFieldValue": "3522**32138"
        },
        {
            "hitCount": 3,
            "hitRecords": [
                {
                    "appCode": "F2010191915000180107",
                    "creditCardNo": "-",
                    "dataCode": "P20121700481894",
                    "externalId": "100032121",
                    "id": 121031,
                    "idNo": "-",
                    "mobile": "139**603",
                    "name": "陈**",
                    "scene": 1002,
                    "sourceType": 7
                },
                 //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "紧急联系人1手机号",
            "searchFieldName": "mobile",
            "searchFieldValue": "139**603"
        },
        {
            "hitCount": 3,
            "hitRecords": [
                {
                    "appCode": "F2007281915000180103",
                    "creditCardNo": "-",
                    "dataCode": "P20121700447417",
                    "externalId": "100010141",
                    "id": 60345,
                    "idNo": "-",
                    "mobile": "1810**37",
                    "name": "林**",
                    "scene": 1002,
                    "sourceType": 8
                },
                 //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "紧急联系人2手机号",
            "searchFieldName": "mobile",
            "searchFieldValue": "181**637"
        },
        {
            "hitCount": 3,
            "hitRecords": [
                {
                    "appCode": "F2009281915000180104",
                    "creditCardNo": "-",
                    "dataCode": "P20121700442684",
                    "externalId": "100026661",
                    "id": 52033,
                    "idNo": "35012**938",
                    "mobile": "152**20",
                    "name": "许**",
                    "scene": 1003,
                    "sourceType": 1
                },
                 //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "担保人身份证号",
            "searchFieldName": "idNo",
            "searchFieldValue": "35012****195938"
        },
        {
            "hitCount": 4,
            "hitRecords": [
                {
                    "appCode": "F2009281915000180104",
                    "creditCardNo": "-",
                    "dataCode": "P20121700442684",
                    "externalId": "100026661",
                    "id": 52033,
                    "idNo": "3501**8",
                    "mobile": "152050**",
                    "name": "许**",
                    "scene": 1003,
                    "sourceType": 1
                },
                //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "担保人手机号",
            "searchFieldName": "mobile",
            "searchFieldValue": "152**"
        },
        {
            "hitCount": 6,
            "hitRecords": [
                {
                    "appCode": "F2011091915000180104",
                    "dataCode": "V20121700417457",
                    "evaluateRemark": "正常。1.备胎槽照片重新拍摄,要求完整清晰。\n2.补充左右后叶子板流水槽照片\n3.补充左右前纵梁照片\n4.补充主副驾驶座椅滑轨照片\n5.有补领记录,补充车架拓印号照片(铁皮上的)",
                    "externalId": "100038555",
                    "id": 1045,
                    "mileage": 136029,
                    "scene": 1002,
                    "sourceType": 12,
                    "vin": "LFV4A24F7A30837**"
                },
                //省略其他
            ],
            "message": "查询成功",
            "searchFieldDesc": "车辆VIN",
            "searchFieldName": "vin",
            "searchFieldValue": "LFV4A24F7A3083**"
        }
    ],
    "msg": "操作成功",
    "success": true
}

4.2、数据查重字段配置

为了提高数据查重接口的扩展性,基于配置化的元数据配置。

value包含如下参数

  • desc:参数描述,无业务逻辑;仅仅作为字段说明使用。
  • searchMode:查询方式,目前仅支持两种 EXACT(精准查询)、SIMILAR(相似度查询)。
  • tables:json字符串数组,适用于该域查询的表,目前表共三个(PERSON_INFODEALER_INFOVEHICLE_INFO)。
  • mapping:查询字段映射,字符串数据,查询形式以或作为条件,结果集会进行合并。
  • exclude:查询过滤条件,查询结果集以该配置参数作为过滤条件。字段key作为查询条件field,value作为条件。
{
  "idNo": {
    "desc": "身份证号",
    "searchMode": "EXACT",
    "mapping": [],
    "tables": [
      "PERSON_INFO",
      "DEALER_INFO"
    ]
  },
  "name": {
    "desc": "姓名",
    "searchMode": "EXACT",
    "mapping": [],
    "tables": [
      "PERSON_INFO",
      "DEALER_INFO"
    ]
  },
  "mobile": {
    "desc": "手机号(primaryMobile,SecondMobile)",
    "searchMode": "EXACT",
    "mapping": [
      "primaryMobile",
      "secondMobile"
    ],
    "tables": [
      "PERSON_INFO",
      "DEALER_INFO"
    ],
    "exclude": {
      "sourceTypeScopeExclude": [
        2,
        3
      ]
    }
  },
  "creditCardNo": {
    "desc": "银行卡号",
    "searchMode": "EXACT",
    "mapping": [],
    "tables": [
      "PERSON_INFO",
      "DEALER_INFO"
    ]
  },
  "companyAddress": {
    "desc": "单位地址",
    "searchMode": "SIMILAR",
    "mapping": [
      "companyAddressProvince",
      "companyAddressCity",
      "companyAddressDistrict",
      "companyAddressDetail"
    ],
    "tables": [
      "PERSON_INFO",
      "DEALER_INFO"
    ]
  },
  "censusAddress": {
    "desc": "户籍地址",
    "searchMode": "SIMILAR",
    "mapping": [
      "censusAddressProvince",
      "censusAddressCity",
      "censusAddressDistrict",
      "censusAddressDetail"
    ],
    "tables": [
      "PERSON_INFO"
    ]
  },
  "residenceAddress": {
    "desc": "居住地址",
    "searchMode": "SIMILAR",
    "mapping": [
      "residenceAddressProvince",
      "residenceAddressCity",
      "residenceAddressDistrict",
      "residenceAddressDetail"
    ],
    "tables": [
      "PERSON_INFO"
    ]
  },
  "vin": {
    "desc": "车辆VIN",
    "searchMode": "EXACT",
    "mapping": [],
    "tables": [
      "VEHICLE_INFO"
    ]
  }
}

5、总结

总体设计上运用了相关设计模式,并分成了多个模块,每个模块负责各自的业务逻辑职责。其中在数据查重接口设计上,考虑查询数据量比较多,基于输入的多个条件,运用并行处理,并把多个处理器的查询结果再进行合并,从而提高接口的性能。