Hibernate Validator -集合對象驗證（三）（可能是東半球最全的講解了）

Tips

在上線到測試環境之后，關於Spring boot的國際化問題，煩擾了我整整一天的時間，大家可以參考這篇文章去處理國際化的問題

前情提示

前兩篇文章已經介紹了Hibernate Validator的對象基礎驗證和對象分組驗證，沒有看過的童鞋可以先去回顧一下，本章節主要解決第一章節所說的具體的業務需求

業務需求

最近在做和Excel導入、導出相關的需求，需求是對用戶上傳的整個Excel的數據進行驗證，如果不符合格式要求，在表頭最后增加一列“錯誤信息”描述此行數據錯誤原因。例如：代理人列不能為空；代理人手機號列如果填寫有值肯定是需要符合手機號格式;結算方式列只允許填寫“全保費、凈保費”字段...
Excel校驗模板

解決思路

1.根據模板分析，大體有以下幾種校驗規則

• 列數據唯一校驗
• 字段為空校驗（表頭中帶*號的都需要非空判斷）
• 日期列格式校驗
• 數值列格式校驗 （金額保留最多保留2為小數，比例最多保留6為小數）
• 手機號格式校驗
• 身份證號格式校驗
• 具體的業務邏輯（某些列只能填寫固定的值；保單類型為批單時，批單號必須有值...）
  2.算法思想
• 首先將Excel中的行數據全部讀進來，轉化為List<Map<String, Object>>，某些校驗是無法在bean實體類上去校驗的，實體類上只能做單個對象或者單個屬性的校驗，像序號列唯一性校驗，這種需要依賴於整個Excel所有行數據去進行判斷是否重復，只能在List<Map<String, Object>>去處理
• 所以校驗會分為兩個層次，一層是在List層次的校驗，一層是Bean層的數據校驗
  基本就是這些吧，具體設計到業務邏輯的我就不說了，下面拿一種財務數據上傳場景去看代碼實現

代碼實現

//將Excel中數據全部讀進來轉化指定的列；比如序號，rowNo:1
List<Map<String, Object>> mapList = fileInfoService.getExcelMaps(template.getId(), file, null, false);
if (CollectionUtils.isEmpty(mapList)) {
      return RestResponse.failedMessage("當前文件中數據為空");
}
//調取校驗接口，並在Excel中增加一列錯誤信息列
RestResponse validateExcel = validateService.validMapListAndWriteFile(fileInfoService.getById(fileId), mapList, sourceCodeEnum);
if (!validateExcel.isSuccess()) {
    return validateExcel;
}

public RestResponse validMapListAndWriteFile(FileInfo fileInfo, List<Map<String, Object>> mapList, SourceCodeEnum sourceCode) {
        RestResponse restResponse = validMapList(mapList, sourceCode, SourceCodeModel.SOURCE_CODE_BEAN_CLASS_MAP.get(sourceCode));
        if (!restResponse.isSuccess()) {
            try {
                ValidationErrorWriter.errorWriteToExcel(fileInfo, (Map<Integer, List<ValidationErrorResult>>) restResponse.getRestContext());
                return RestResponse.failedMessage("表格數據校驗未通過，請點擊源文件下載");
            } catch (Exception e) {
                e.printStackTrace();
            }
        }
        return restResponse;
    }

其中SourceCodeModel.SOURCE_CODE_BEAN_CLASS_MAP.get(sourceCode)根據每一個上傳模板獲取對應的實體類信息

public static final Map<SourceCodeEnum, Class> SOURCE_CODE_BEAN_CLASS_MAP = new ImmutableMap.Builder<SourceCodeEnum, Class>()
            .put(SourceCodeEnum.BUSINESS_AUTO, DataPoolAutoValidModel.class)
            .put(SourceCodeEnum.BUSINESS_UNAUTO, DataPoolUnAutoValidModel.class)
            .put(SourceCodeEnum.BUSINESS_AUTO_SUPPLY, BusinessSupplyAutoModel.class)
            .put(SourceCodeEnum.BUSINESS_UNAUTO_SUPPLY, BusinessSupplyUnAutoModel.class)
            .put(SourceCodeEnum.COMMISSION_AUTO, CommissionApplyAutoModel.class)
            .put(SourceCodeEnum.COMMISSION_UNAUTO, CommissionApplyUnAutoModel.class)
            .put(SourceCodeEnum.SETTLEMENT_AUTO, SettlementApplyAutoModel.class)
            .put(SourceCodeEnum.SETTLEMENT_UNAUTO, SettlementApplyUnAutoModel.class)
            .put(SourceCodeEnum.COMMISSION_DETAIL_AUTO, CommissionBackAutoModel.class)
            .put(SourceCodeEnum.COMMISSION_DETAIL_UNAUTO, CommissionBackUnAutoModel.class)
            .put(SourceCodeEnum.BACK_AUTO, SettlementPayBackAutoModel.class)
            .put(SourceCodeEnum.BACK_UNAUTO, SettlementPayBackUnAutoModel.class)
            .put(SourceCodeEnum.COMMISSION_RESULT, CommissionResultBackModel.class)
            .build();

最終校驗方法

public RestResponse validMapList(List<Map<String, Object>> mapList, SourceCodeEnum sourceCode, Class beanClazz) {
        if (CollectionUtils.isEmpty(mapList)) {
            return RestResponse.success();
        }

        Map<String, String> refMap = LoadTemplateInfoMap.getMappingInfoBySourceCode(sourceCode);
        Map<Integer, List<ValidationErrorResult>> errorMap;
        List beanClassList = null;
        try {
            //是否需要在Map層校驗數據是否重復
            boolean businessImport = SourceCodeModel.businessImport(sourceCode);
            //車險非車險
            DataAutoType dataAutoType = SourceCodeModel.businessDataAutoType(sourceCode);
            //Map層校驗
            Map<Integer, List<ValidationErrorResult>> validMap = mapValidationService.valid(mapList, businessImport, dataAutoType);
            //Bean層次校驗
            beanClassList = CommonUtils.transMap2BeanForList(mapList, beanClazz);
            Map<Integer, List<ValidationErrorResult>> validBean = beanValidationService.validBeanList(beanClassList, beanClazz);
            if (haveErrorMessage(validMap) || haveErrorMessage(validBean)) {
                errorMap = makeValidErrorMap(validMap, validBean);
                setCorrectColumnName(errorMap, refMap);
                return RestResponse.failed(errorMap);
            }
        } catch (InterruptedException e) {
            e.printStackTrace();
        }
        return RestResponse.success(beanClassList);
    }

Map層包含序號重復的校驗，日期格式、數字格式和枚舉類型的校驗，具體的實現邏輯就不展現了，這不是重點要講的，附一張代碼格式的截圖吧

重點：Bean層次的校驗

• 實體類

@Data
@EqualsAndHashCode(callSuper=false)
@CommissionValidType(groups = DataValidGroup.class)
@FinanceMatchType(groups = MatchGroup.class)
@MatchBusinessOrder(groups = MatchBusinessGroup.class)
@CommissionApplyTimes(groups = LimitTimesGroup.class)
public class CommissionApplyAutoModel extends Commission {

    @NotNull
    private Long rowNo;

    @NotNull
    private OrderType orderType;

    @NotBlank
    private String insuranceType;

    @NotBlank
    private String applicant;

    @NotNull
    private BigDecimal premium;

    @NotBlank
    private String policyNo;

    @NotBlank
    private String agent;

    @NotBlank
    private String agentBankNo;

    private String beneficiary;

    private BigDecimal commissionRate;

    private BigDecimal downstreamCommissionRate;

    private CommissionSettlementType coSettlementType;

    @Pattern(regexp = "^[1][3-9]\\d{9}$", message = "代理人手機號格式不正確")
    private String agentPhone;

    @NotBlank
    @Size(min = ValidateUtils.MIN_LICENSEPLATENO_LENGTH, max = ValidateUtils.MAX_LICENSEPLATENO_LENGTH,
            message = "車牌號長度在" + ValidateUtils.MIN_LICENSEPLATENO_LENGTH + "~" + ValidateUtils.MAX_LICENSEPLATENO_LENGTH + "之間")
    private String licensePlateNo;

}

• 校驗邏輯：校驗實體類上的每一層 group，如果數據不符合規則，直接返回，某些group需要單獨處理，eg:UniqueGroup、BusinessSupplyGroup這些有些單獨的處理場景，所以單列出來

public <T> Map<Integer, List<ValidationErrorResult>> validBeanList(List<T> dataList, Class clazz) throws InterruptedException {
        Map<Integer, List<ValidationErrorResult>> errorMap;
        //通過反射獲取當前實體類上所有的group
        List<Class> groupList = getGroupClassFromBean(clazz);
        for (Class groupClass : groupList) {
            //去重校驗
            if (groupClass == UniqueGroup.class) {
                errorMap = uniqueAndInsert(dataList, groupClass);
                if (ValidateService.haveErrorMessage(errorMap)) {
                    return errorMap;
                }
            } else if (groupClass == BusinessSupplyGroup.class) {
                //業務數據補足
                dataPoolService.updateBatchById((List<DataPool>) dataList, 1000);
            } else {
                //普通校驗
                errorMap = valid(dataList, groupClass);
                if (ValidateService.haveErrorMessage(errorMap)) {
                    return errorMap;
                }
            }
        }
        return Collections.emptyMap();
    }

• 關於group 都是一些空接口，上一章已經講過，主要用來標注校驗順序的，沒有其他深意

public interface BusinessSupplyGroup {
}

、

• eg:注解@FinanceMatchType(groups = MatchGroup.class)

@Target({ElementType.TYPE, ElementType.ANNOTATION_TYPE})
@Retention(RetentionPolicy.RUNTIME)
@Constraint(validatedBy = FinanceMatchTypeValidator.class)
@Documented
public @interface FinanceMatchType {

    String message() default "";

    Class<?>[] groups() default {};

    Class<? extends Payload>[] payload() default { };

    @Target({ElementType.TYPE, ElementType.FIELD, ElementType.ANNOTATION_TYPE})
    @Retention(RUNTIME)
    @Documented
    @interface List {
        FinanceMatchType[] value();
    }

}

public class FinanceMatchTypeValidator extends PubValidator<FinanceMatchValidator, BusinessFinanceModel>
        implements ConstraintValidator<FinanceMatchType, BusinessFinanceModel> {
}

• 多行數據校驗過程中，使用多線程去校驗

@Component
@Log4j2
public abstract class PubValidator<T extends Validator, E> {

    @Autowired
    private List<T> validatorList;
    @Autowired
    ThreadPool runThreadPool;

    public boolean isValid(E value, ConstraintValidatorContext context) {
        boolean result = true;
        List<Boolean> resultList = null;
        try {
            resultList = runThreadPool.submitWithResult(validatorList, validator -> validator.isValid(value, context));
        } catch (ExecutionException | InterruptedException e) {
            log.error("數據校驗錯誤", e);
        }

        if (resultList != null && resultList.size() > 0) {
            result = resultList.stream().allMatch(i -> i);
        }
        return result;
    }

}

public interface FinanceMatchValidator extends Validator<BusinessFinanceModel> {
}

• 當前層次有兩個類型需要校驗
• 險種類型校驗，車險業務，險種類型只能為{交強險，商業險};非車險業務，險種類別和性質不能為空
• 保單類型校驗，如果為批單，批單號不能為空
  對應到兩個實現類

/**
 * author:Java
 * Date:2020/6/1 16:02
 * 校驗：車險險種驗證
 *      車險：險種是否為交強商業
 *      非車險：險種類別、性質
 */
@Component
public class InsuranceTypeValidateErrorMessage implements MatchGroupValidator, CommisssionDetailValidator, SettlementPayBackValidator, FinanceMatchValidator {

    @Override
    public boolean isValid(BusinessFinanceModel bfm, ConstraintValidatorContext context) {
        if (bfm.getDataAutoType() == DataAutoType.AUTO) {
            if (!validAutoInsuranceType(bfm.getInsuranceType(), bfm)) {
                setErrorMessage("insuranceType", "車險險種名稱不能為空，只能填寫：[交強險，商業險]", context);
                return false;
            }
        }
        return true;
    }

    private boolean validAutoInsuranceType(String insuranceType, BusinessFinanceModel bfm) {
        Long insuranceTypeId = null;
        InsuranceCategory insuranceCategory = null;
        for (Map.Entry<String, Long> entry : BusinessConstants.INSTRANCETYPE_ID.entrySet()) {
            if (insuranceType.contains(entry.getKey())) {
                insuranceTypeId = entry.getValue();
                insuranceCategory = insuranceTypeId == 1 ? InsuranceCategory.TRAFFIC_INSURANCE : InsuranceCategory.COMMERCIAL_INSURANCE;
                break;
            }
        }
        if (insuranceTypeId != null) {
            bfm.setInsuranceTypeId(insuranceTypeId);
            bfm.setInsuranceCategory(insuranceCategory);
            return true;
        }
        return false;
    }

}

/**
 * author:WangZhaoliang
 * Date:2020/6/1 15:48
 * 校驗：保單類型不能為空
 *      保單類型為保單 || 被沖正保單 || 沖正保單時：保單號不能為空
 *      保單類型為被沖正批單 || 沖正批單 || 批單時：批單號不能為空
 */

@Component
public class OrderTypeValidateErrorMessage implements MatchGroupValidator, FinanceMatchValidator {

    @Override
    public boolean isValid(BusinessFinanceModel value, ConstraintValidatorContext context) {
        OrderType orderType = value.getOrderType();
        if (orderType == null) {
            setErrorMessage("orderType", "保單類型必填並且只能填寫" + OrderType.getAllText(), context);
            return false;
        }

        // 保單、被沖正保單、沖正保單
        if (orderType == OrderType.POLICY || orderType == OrderType.REVERSED_POLICY || orderType == OrderType.CORRECTION_POLICY) {
            if (StringUtils.isBlank(value.getPolicyNo())) {
                setErrorMessage("policyNo", "保單類型為[保單、被沖正保單、沖正保單]時,保單號必須有值", context);
                return false;
            }
        } else if (OrderType.isBatchNo(orderType)) {
            if (StringUtils.isBlank(value.getBatchNo())) {
                setErrorMessage("batchNo", "保單類型為[批單、被沖正批單、沖正批單]時,批單號必須有值", context);
                return false;
            }
        }
        return true;
    }
}

/**
 * author:WangZhaoliang
 * Date:2020/6/3 11:34
 */
public interface ValidatorErrorMessage {

    default void setErrorMessage(String property, String message, ConstraintValidatorContext context) {
        context.disableDefaultConstraintViolation();
        context.buildConstraintViolationWithTemplate(message)
                .addPropertyNode(property)
                .addBeanNode()
                .addConstraintViolation();
    }

}

將所有實現了FinanceMatchValidator接口的實現類，都會加入到 PubValidator類中的List<T> validatorList中去，再調用 PubValidator類的isValid(E value, ConstraintValidatorContext context)方法去校驗，將錯誤信息加入到ConstraintValidatorContext中
每一層group都是如此校驗，代碼邏輯較多，就不再一一贅述了，最終將錯誤信息追加到Excel最后一列，反饋給用戶
這樣完成第一章節中提到的需求：校驗整個Excel中的信息；看一下最終反饋到Excel中的錯誤信息

產品大佬終於露出了滿意的笑臉...

這塊目前只是剛上線了最初版本，后面還有需要優化的點，未完待續...
Java is the best language in the world