我尝试序列化从实体数据模型.edmx自动生成的POCO类,当我使用

JsonConvert.SerializeObject 

我得到了以下错误:

为System.data.entity类型检测到自我引用循环错误。

我怎么解决这个问题?


当前回答

我也面临着同样的问题,我尝试使用JsonSetting来忽略自引用错误,它的某种工作,直到我得到一个类,自引用非常深入,我的。net进程挂在Json写入值。

我的问题

    public partial class Company : BaseModel
{
    public Company()
    {
        CompanyUsers = new HashSet<CompanyUser>();
    }

    public string Name { get; set; }

    public virtual ICollection<CompanyUser> CompanyUsers { get; set; }
}

public partial class CompanyUser
{
    public int Id { get; set; }
    public int CompanyId { get; set; }
    public int UserId { get; set; }

    public virtual Company Company { get; set; }

    public virtual User User { get; set; }
}

public partial class User : BaseModel
{
    public User()
    {
        CompanyUsers = new HashSet<CompanyUser>();
    }

    public string DisplayName { get; set; }
    public virtual ICollection<CompanyUser> CompanyUsers { get; set; }

}

你可以在User类中看到问题,它引用了CompanyUser类,这是一个自引用。

现在,我正在调用包含所有关系属性的GetAll方法。

cs.GetAll("CompanyUsers", "CompanyUsers.User");

在这个阶段,我的DotNetCore进程挂着执行JsonResult,写入值…永远不要来。在Startup.cs中,我已经设置了JsonOption。出于某种原因,EFCore包含了嵌套属性,我没有要求Ef给出。

    options.SerializerSettings.ReferenceLoopHandling = Newtonsoft.Json.ReferenceLoopHandling.Ignore;

预期的行为应该是这样的

嘿,EfCore,你能把“公司用户”数据也包括在我的 公司类,以便我可以很容易地访问数据。

then

嘿,EfCore,你能把“公司用户”也包括进来吗?用户数据为 这样我就可以像这样轻松地访问数据 .User.DisplayName Company.CompanyUsers.First ()

在这个阶段,我应该只得到这个"Company.CompanyUsers.First(). user。DisplayName"它不应该给我company。companyusers。first (). user。导致自引用问题的CompanyUsers;从技术上讲,它不应该给我User。CompanyUsers作为CompanyUsers是一个导航属性。EfCore非常兴奋,给了我user。companyusers。

所以,我决定为属性写一个扩展方法来排除对象(它实际上不是排除它只是设置属性为null)。不仅如此,它还适用于数组属性。下面是我还将为其他用户导出nuget包的代码(不确定这是否甚至有助于某人)。原因很简单,因为我懒得写.Select(n => new {n.p1, n.p2});我只是不想写选择语句只排除1个属性!

这不是最好的代码(我会在某个阶段更新),因为我写得很匆忙,尽管这可能会帮助那些想要排除(设置null)对象数组的人。

    public static class PropertyExtensions
{
    public static void Exclude<T>(this T obj, Expression<Func<T, object>> expression)
    {
        var visitor = new PropertyVisitor<T>();
        visitor.Visit(expression.Body);
        visitor.Path.Reverse();
        List<MemberInfo> paths = visitor.Path;
        Action<List<MemberInfo>, object> act = null;

        int recursiveLevel = 0;
        act = (List<MemberInfo> vPath, object vObj) =>
        {

            // set last propert to null thats what we want to avoid the self-referencing error.
            if (recursiveLevel == vPath.Count - 1)
            {
                if (vObj == null) throw new ArgumentNullException("Object cannot be null");

                vObj.GetType().GetMethod($"set_{vPath.ElementAt(recursiveLevel).Name}").Invoke(vObj, new object[] { null });
                return;
            }

            var pi = vObj.GetType().GetProperty(vPath.ElementAt(recursiveLevel).Name);
            if (pi == null) return;
            var pv = pi.GetValue(vObj, null);
            if (pi.PropertyType.IsArray || pi.PropertyType.Name.Contains("HashSet`1") || pi.PropertyType.Name.Contains("ICollection`1"))
            {
                var ele = (IEnumerator)pv.GetType().GetMethod("GetEnumerator").Invoke(pv, null);

                while (ele.MoveNext())
                {
                    recursiveLevel++;
                    var arrItem = ele.Current;

                    act(vPath, arrItem);

                    recursiveLevel--;
                }

                if (recursiveLevel != 0) recursiveLevel--;
                return;
            }
            else
            {
                recursiveLevel++;
                act(vPath, pv);
            }

            if (recursiveLevel != 0) recursiveLevel--;

        };

        // check if the root level propert is array
        if (obj.GetType().IsArray)
        {
            var ele = (IEnumerator)obj.GetType().GetMethod("GetEnumerator").Invoke(obj, null);
            while (ele.MoveNext())
            {
                recursiveLevel = 0;
                var arrItem = ele.Current;

                act(paths, arrItem);
            }
        }
        else
        {
            recursiveLevel = 0;
            act(paths, obj);
        }

    }

    public static T Explode<T>(this T[] obj)
    {
        return obj.FirstOrDefault();
    }

    public static T Explode<T>(this ICollection<T> obj)
    {
        return obj.FirstOrDefault();
    }
}

上面的扩展类将使您能够将属性设置为null,以避免自引用循环甚至数组。

表达式构建器

    internal class PropertyVisitor<T> : ExpressionVisitor
{
    public readonly List<MemberInfo> Path = new List<MemberInfo>();

    public Expression Modify(Expression expression)
    {
        return Visit(expression);
    }


    protected override Expression VisitMember(MemberExpression node)
    {
        if (!(node.Member is PropertyInfo))
        {
            throw new ArgumentException("The path can only contain properties", nameof(node));
        }

        Path.Add(node.Member);
        return  base.VisitMember(node);
    }
}

Usages:

模型类

    public class Person
{
    public string Name { get; set; }
    public Address AddressDetail { get; set; }
}

public class Address
{
    public string Street { get; set; }
    public Country CountryDetail { get; set; }
    public Country[] CountryDetail2 { get; set; }
}

public class Country
{
    public string CountryName { get; set; }
    public Person[] CountryDetail { get; set; }
}

虚拟数据

           var p = new Person
        {
            Name = "Adeel Rizvi",
            AddressDetail = new Address
            {
                Street = "Sydney",
                CountryDetail = new Country
                {
                    CountryName = "AU"
                }
            }
        };

        var p1 = new Person
        {
            Name = "Adeel Rizvi",
            AddressDetail = new Address
            {
                Street = "Sydney",
                CountryDetail2 = new Country[]
                {
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A1" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A2" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A3" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A4" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A5" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A6" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A7" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A8" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A9" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A1" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A2" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A3" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A4" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A5" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A6" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A7" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A8" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },
                    new Country{ CountryName = "AU", CountryDetail = new Person[]{ new Person { Name = "A9" }, new Person { Name = "A1" }, new Person { Name = "A1" } } },

                }
            }
        };

例:

情况1:只排除没有任何数组的属性

p.Exclude(n => n.AddressDetail.CountryDetail.CountryName);

案例2:排除带有1个数组的属性

p1.Exclude(n => n.AddressDetail.CountryDetail2.Explode().CountryName);

案例3:排除带有2个嵌套数组的属性

p1.Exclude(n => n.AddressDetail.CountryDetail2.Explode().CountryDetail.Explode().Name);

案例4:包含EF GetAll查询

var query = cs.GetAll("CompanyUsers", "CompanyUsers.User").ToArray();
query.Exclude(n => n.Explode().CompanyUsers.Explode().User.CompanyUsers);
return query;

你已经注意到,explosion()方法也是一个扩展方法,用于我们的表达式构建器从数组属性中获取属性。只要有数组属性,就使用. explosion()。YourPropertyToExclude或. explosion (). property1 . myarrayproperty . explosion (). mystupidproperty。上面的代码帮助我避免像我想要的那样深的自我引用。现在我可以使用GetAll并排除我不想要的属性!

感谢你阅读这篇文章!

其他回答

我继承了一个数据库应用程序,为web页面提供数据模型。默认情况下,序列化将尝试遍历整个模型树,这里的大多数答案都是如何防止这种情况的良好开端。

一个尚未探索的选项是使用接口来提供帮助。我将从之前的例子中窃取:

public partial class CompanyUser
{
    public int Id { get; set; }
    public int CompanyId { get; set; }
    public int UserId { get; set; }

    public virtual Company Company { get; set; }

    public virtual User User { get; set; }
}

public interface IgnoreUser
{
    [JsonIgnore]
    User User { get; set; }
}

public interface IgnoreCompany
{
    [JsonIgnore]
    User User { get; set; }
}

public partial class CompanyUser : IgnoreUser, IgnoreCompany
{
}

在上述解决方案中,没有Json设置受到损害。将LazyLoadingEnabled和或ProxyCreationEnabled设置为false会影响所有的后端编码,并阻止ORM工具的一些真正好处。LazyLoading/ProxyCreation设置可以在不手动加载的情况下阻止导航属性的加载,这取决于您的应用程序。

这里有一个更好的解决方案,以防止导航属性序列化,它使用标准的json功能: 我怎么能做JSON序列化器忽略导航属性?

简单地放置Configuration。ProxyCreationEnabled = false;在上下文文件中;这样问题就解决了。

public demEntities()
    : base("name=demEntities")
{
    Configuration.ProxyCreationEnabled = false;
}

人们已经讨论过将[JsonIgnore]添加到类中的虚拟属性,例如:

[JsonIgnore]
public virtual Project Project { get; set; }

我还将分享另一个选项,[JsonProperty(NullValueHandling = NullValueHandling. ignore)],仅当它为空时才从序列化中省略该属性:

[JsonProperty(NullValueHandling = NullValueHandling.Ignore)]
public virtual Project Project { get; set; }

使用JsonSerializerSettings

ReferenceLoopHandling。如果遇到引用循环,Error(默认)将出错。这就是为什么会出现异常。 ReferenceLoopHandling。如果对象嵌套但不是无限的,序列化是有用的。 ReferenceLoopHandling。如果对象是自身的子对象,Ignore将不会序列化该对象。

例子:

JsonConvert.SerializeObject(YourPOCOHere, Formatting.Indented, 
new JsonSerializerSettings 
{ 
        ReferenceLoopHandling = ReferenceLoopHandling.Serialize
});

如果你必须序列化一个无限嵌套的对象,你可以使用PreserveObjectReferences来避免StackOverflowException。

例子:

JsonConvert.SerializeObject(YourPOCOHere, Formatting.Indented, 
new JsonSerializerSettings 
{ 
        PreserveReferencesHandling = PreserveReferencesHandling.Objects
});

选择对要序列化的对象有意义的内容。

参考http://james.newtonking.com/json/help/

当你有循环问题时,序列化使用NEWTONSOFTJSON,在我的情况下,我不需要修改全局。野蔷薇或野蔷薇。我只是使用JsonSerializesSettings忽略循环处理。

JsonSerializerSettings jss = new JsonSerializerSettings();
jss.ReferenceLoopHandling = ReferenceLoopHandling.Ignore;
var lst = db.shCards.Where(m => m.CardID == id).ToList();
string json = JsonConvert.SerializeObject(lst, jss);