NIbaaa.DataCleaning 1.0.0

using System; using System.Collections.Generic;

namespace DataCleaning.Entity { class Program { static void Main(string[] args) { // 创建清洗器 var cleaner = new EntityCleaner();

        // ========== 配置清洗规则 ==========
        var cleanRules = new List<FieldCleanConfig>
        {
            new FieldCleanConfig
            {
                FieldName = "StudentName",
                RuleType = CleanRuleType.Trim,
                IsRequired = true,
                ErrorMessage = "学生姓名不能为空"
            },
            new FieldCleanConfig
            {
                FieldName = "StudentNo",
                RuleType = CleanRuleType.Trim,
                IsRequired = true
            },
            new FieldCleanConfig
            {
                FieldName = "PhoneNumber",
                RuleType = CleanRuleType.FormatPhone,
                DefaultValue = null
            },
            new FieldCleanConfig
            {
                FieldName = "Email",
                RuleType = CleanRuleType.FormatEmail,
                DefaultValue = null
            },
            new FieldCleanConfig
            {
                FieldName = "Age",
                RuleType = CleanRuleType.RangeCheck,
                MinValue = 0,
                MaxValue = 120,
                DefaultValue = 0
            },
            new FieldCleanConfig
            {
                FieldName = "Address",
                RuleType = CleanRuleType.RemoveSpecialChars
            },
            new FieldCleanConfig
            {
                FieldName = "ClassName",
                RuleType = CleanRuleType.Trim,
                IsRequired = true,
                DefaultValue = "默认班级"
            }
        };

        cleaner.ConfigureCleanRules(cleanRules);

        // ========== 配置匿名化规则 ==========
        var anonymizeRules = new List<FieldAnonymizeConfig>
        {
            new FieldAnonymizeConfig
            {
                FieldName = "StudentName",
                Type = AnonymizeType.Pseudonymize,
                IsSensitive = true
            },
            new FieldAnonymizeConfig
            {
                FieldName = "PhoneNumber",
                Type = AnonymizeType.Mask,
                MaskPattern = "****",
                MaskStart = 3,
                MaskEnd = 4
            },
            new FieldAnonymizeConfig
            {
                FieldName = "IdCard",
                Type = AnonymizeType.Mask,
                MaskPattern = "********"
            },
            new FieldAnonymizeConfig
            {
                FieldName = "Email",
                Type = AnonymizeType.Hash,
                Salt = "student_salt"
            },
            new FieldAnonymizeConfig
            {
                FieldName = "Age",
                Type = AnonymizeType.Generalization,
                GeneralizationInterval = 10
            },
            new FieldAnonymizeConfig
            {
                FieldName = "BirthDate",
                Type = AnonymizeType.DateOffset,
                DateOffsetDays = 30
            },
            new FieldAnonymizeConfig
            {
                FieldName = "Score",
                Type = AnonymizeType.Perturbation,
                PerturbationRate = 0.05
            }
        };

        cleaner.ConfigureAnonymizeRules(anonymizeRules);

        // ========== 测试数据 ==========
        var student = new StudentEntity
        {
            Id = 1,
            StudentName = "  张三  ",
            StudentNo = "2021001",
            PhoneNumber = "13812345678",
            Email = "zhangsan@qq.com",
            Age = 20,
            Address = "北京市朝阳区建国路88号",
            ClassName = "计算机1班",
            BirthDate = new DateTime(1990, 3, 7),
            IdCard = "11010119900307663X",
            Score = 85.5m
        };

        Console.WriteLine("========== 原始数据 ==========");
        PrintStudent(student);

        // ========== 只清洗 ==========
        var cleanResult = cleaner.Clean(student);
        Console.WriteLine("\n========== 清洗结果 ==========");
        Console.WriteLine($"成功: {cleanResult.Success}");
        Console.WriteLine($"清洗字段: {string.Join(", ", cleanResult.CleanedFields)}");
        PrintStudent(student);

        // ========== 清洗+匿名化 ==========
        var student2 = new StudentEntity
        {
            Id = 2,
            StudentName = "  李四  ",
            StudentNo = "2021002",
            PhoneNumber = "13912345678",
            Email = "lisi@test.com",
            Age = 25,
            Address = "上海市浦东新区",
            ClassName = "计算机2班",
            BirthDate = new DateTime(1995, 5, 10),
            IdCard = "31010119950510123X",
            Score = 92.0m
        };

        var fullResult = cleaner.CleanAndAnonymize(student2, student2.Id.ToString());
        Console.WriteLine("\n========== 清洗+匿名化结果 ==========");
        Console.WriteLine($"成功: {fullResult.Success}");
        Console.WriteLine($"清洗字段: {string.Join(", ", fullResult.CleanedFields)}");
        Console.WriteLine($"匿名化字段: {string.Join(", ", fullResult.AnonymizedFields)}");
        PrintStudent(student2);
    }

    static void PrintStudent(StudentEntity student)
    {
        Console.WriteLine($"ID: {student.Id}");
        Console.WriteLine($"姓名: {student.StudentName}");
        Console.WriteLine($"学号: {student.StudentNo}");
        Console.WriteLine($"手机: {student.PhoneNumber}");
        Console.WriteLine($"邮箱: {student.Email}");
        Console.WriteLine($"年龄: {student.Age}");
        Console.WriteLine($"地址: {student.Address}");
        Console.WriteLine($"班级: {student.ClassName}");
        Console.WriteLine($"生日: {student.BirthDate:yyyy-MM-dd}");
        Console.WriteLine($"身份证: {student.IdCard}");
        Console.WriteLine($"分数: {student.Score}");
    }
}

}