users/lib/data-dictionary.ts

// 数据字典和映射规则管理
// 根据需求文档中的要求，创建动态维护的数据字典

export interface DataField {
  name: string
  type: "string" | "number" | "boolean" | "date" | "json" | "array"
  description: string
  isRequired: boolean
  isPII: boolean // 个人身份信息标识
  isIdentityKey: boolean // 身份识别关键字段
  isAIFeature: boolean // 用于AI分析的特征字段
  isDistributionKey: boolean // 用于分销/返点计算的关联字段
}

export interface SourceMapping {
  sourceSystem: string
  sourceField: string
  targetField: string
  transformRule?: string
  validationRule?: string
}

export interface DataDictionary {
  coreFields: Record<string, DataField>
  unifiedTags: string[]
  unifiedAttributes: Record<string, DataField>
  sourceMappings: Record<string, SourceMapping[]>
}

// 核心数据字典定义
export const DATA_DICTIONARY: DataDictionary = {
  coreFields: {
    userId: {
      name: "userId",
      type: "string",
      description: "全局唯一用户ID",
      isRequired: true,
      isPII: false,
      isIdentityKey: true,
      isAIFeature: false,
      isDistributionKey: true,
    },
    username: {
      name: "username",
      type: "string",
      description: "用户名",
      isRequired: false,
      isPII: true,
      isIdentityKey: true,
      isAIFeature: true,
      isDistributionKey: false,
    },
    email: {
      name: "email",
      type: "string",
      description: "邮箱地址",
      isRequired: false,
      isPII: true,
      isIdentityKey: true,
      isAIFeature: false,
      isDistributionKey: false,
    },
    phone: {
      name: "phone",
      type: "string",
      description: "手机号码",
      isRequired: false,
      isPII: true,
      isIdentityKey: true,
      isAIFeature: false,
      isDistributionKey: false,
    },
    fullName: {
      name: "fullName",
      type: "string",
      description: "真实姓名",
      isRequired: false,
      isPII: true,
      isIdentityKey: true,
      isAIFeature: true,
      isDistributionKey: false,
    },
    gender: {
      name: "gender",
      type: "string",
      description: "性别",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
    birthDate: {
      name: "birthDate",
      type: "date",
      description: "出生日期",
      isRequired: false,
      isPII: true,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
    city: {
      name: "city",
      type: "string",
      description: "所在城市",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
    province: {
      name: "province",
      type: "string",
      description: "所在省份",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
  },
  unifiedTags: [
    "高价值客户",
    "新用户",
    "活跃用户",
    "流失风险",
    "科技爱好者",
    "内容创作者",
    "90后",
    "00后",
    "北京地区",
    "上海地区",
    "游戏爱好者",
    "旅游达人",
    "美食家",
  ],
  unifiedAttributes: {
    totalSpend: {
      name: "totalSpend",
      type: "number",
      description: "总消费金额",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: true,
    },
    lastActiveDays: {
      name: "lastActiveDays",
      type: "number",
      description: "最后活跃天数",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
    contentPreference: {
      name: "contentPreference",
      type: "array",
      description: "内容偏好",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: true,
      isDistributionKey: false,
    },
    distributionLevel: {
      name: "distributionLevel",
      type: "number",
      description: "分销层级",
      isRequired: false,
      isPII: false,
      isIdentityKey: false,
      isAIFeature: false,
      isDistributionKey: true,
    },
  },
  sourceMappings: {
    douyin: [
      { sourceSystem: "douyin", sourceField: "openid", targetField: "userId", transformRule: "prefix_dy_" },
      { sourceSystem: "douyin", sourceField: "nickname", targetField: "username" },
      { sourceSystem: "douyin", sourceField: "avatar", targetField: "avatarUrl" },
    ],
    xiaohongshu: [
      { sourceSystem: "xiaohongshu", sourceField: "openid", targetField: "userId", transformRule: "prefix_xhs_" },
      { sourceSystem: "xiaohongshu", sourceField: "nickname", targetField: "username" },
    ],
    cunkebao_form: [
      { sourceSystem: "cunkebao_form", sourceField: "name", targetField: "fullName" },
      { sourceSystem: "cunkebao_form", sourceField: "phone", targetField: "phone" },
      { sourceSystem: "cunkebao_form", sourceField: "email", targetField: "email" },
    ],
    touchkebao_call: [
      { sourceSystem: "touchkebao_call", sourceField: "phone_number", targetField: "phone" },
      { sourceSystem: "touchkebao_call", sourceField: "call_time", targetField: "lastContactTime" },
    ],
  },
}

// 获取字段映射规则
export function getFieldMapping(sourceSystem: string, sourceField: string): SourceMapping | undefined {
  const mappings = DATA_DICTIONARY.sourceMappings[sourceSystem]
  return mappings?.find((mapping) => mapping.sourceField === sourceField)
}

// 获取身份识别关键字段
export function getIdentityKeyFields(): string[] {
  return Object.entries(DATA_DICTIONARY.coreFields)
    .filter(([_, field]) => field.isIdentityKey)
    .map(([name, _]) => name)
}

// 获取AI特征字段
export function getAIFeatureFields(): string[] {
  return Object.entries(DATA_DICTIONARY.coreFields)
    .filter(([_, field]) => field.isAIFeature)
    .map(([name, _]) => name)
}

// 获取分销关联字段
export function getDistributionKeyFields(): string[] {
  return Object.entries(DATA_DICTIONARY.coreFields)
    .filter(([_, field]) => field.isDistributionKey)
    .map(([name, _]) => name)
}