In the previous post we ended up with entities that hide the join entity from the public surface. However, it was not possible to add or removed entities through this public surface. To enable this we need an ICollection implementation that acts as a true facade over the real join entity collection and delegates all responsibilities to that collection.

The collection implementation

Here’s one possible implementation of such a collection:

public class JoinCollectionFacade<T, TJoin> : ICollection<T>
{
private readonly ICollection<TJoin> _collection;
private readonly Func<TJoin, T> _selector;
private readonly Func<T, TJoin> _creator; public JoinCollectionFacade(
ICollection<TJoin> collection,
Func<TJoin, T> selector,
Func<T, TJoin> creator)
{
_collection = collection;
_selector = selector;
_creator = creator;
} public IEnumerator<T> GetEnumerator()
=> _collection.Select(e => _selector(e)).GetEnumerator(); IEnumerator IEnumerable.GetEnumerator()
=> GetEnumerator(); public void Add(T item)
=> _collection.Add(_creator(item)); public void Clear()
=> _collection.Clear(); public bool Contains(T item)
=> _collection.Any(e => Equals(_selector(e), item)); public void CopyTo(T[] array, int arrayIndex)
=> this.ToList().CopyTo(array, arrayIndex); public bool Remove(T item)
=> _collection.Remove(
_collection.FirstOrDefault(e => Equals(_selector(e), item))); public int Count
=> _collection.Count; public bool IsReadOnly
=> _collection.IsReadOnly;
}

The idea is pretty simple–operations on the facade are translated into operations on the underlying collection. Where needed, a “selector” delegate is used to extract the desired target entity from the join entity. Likewise, a “creator” delegate creates a new join entity instance from the target entity when a new relationship is added.

实际上我觉得改成下面这样会更好,另外Remove和Contains方法我觉得没什么用,所以暂时就先放的抛出NotSupportedException异常:

public class ActionCollection<T, TJoin> : ICollection<T>
{
protected readonly Func<T, TJoin> creator;
protected readonly Func<TJoin, T> selector;
protected readonly Func<ICollection<TJoin>> collectionSelector; public ActionCollection(Func<ICollection<TJoin>> collectionSelector, Func<T, TJoin> creator, Func<TJoin, T> selector)
{
this.collectionSelector = collectionSelector;
this.creator = creator;
this.selector = selector;
} public int Count => collectionSelector().Count; public bool IsReadOnly => collectionSelector().IsReadOnly; public void Add(T item)
{
collectionSelector().Add(creator(item));
} public void Clear()
{
collectionSelector().Clear();
} public bool Contains(T item)
{
throw new NotSupportedException("Contains is not supported");
} public void CopyTo(T[] array, int arrayIndex)
{
List<T> list = new List<T>(); foreach (var tJoin in collectionSelector())
{
list.Add(selector(tJoin));
} list.CopyTo(array, arrayIndex);
} public IEnumerator<T> GetEnumerator()
{
return this.collectionSelector().Select(tj => this.selector(tj)).GetEnumerator();
} public bool Remove(T item)
{
throw new NotSupportedException("Remove is not supported");
} IEnumerator IEnumerable.GetEnumerator()
{
return this.GetEnumerator();
}
}

Updating the model

We need to initialize instances of this collection in our entities:

public class Post
{
public Post()
=> Tags = new JoinCollectionFacade<Tag, PostTag>(
PostTags,
pt => pt.Tag,
t => new PostTag { Post = this, Tag = t }); public int PostId { get; set; }
public string Title { get; set; } private ICollection<PostTag> PostTags { get; } = new List<PostTag>(); [NotMapped]
public ICollection<Tag> Tags { get; }
} public class Tag
{
public Tag()
=> Posts = new JoinCollectionFacade<Post, PostTag>(
PostTags,
pt => pt.Post,
p => new PostTag { Post = p, Tag = this }); public int TagId { get; set; }
public string Text { get; set; } private ICollection<PostTag> PostTags { get; } = new List<PostTag>(); [NotMapped]
public ICollection<Post> Posts { get; }
}

Using the ICollection navigations

Notice how Tags and Posts are now ICollection properties instead of IEnumerable properties. This means we can add and remove entities from the many-to-many collections without using the join entity directly. Here’s the test application updated to show this:

public class Program
{
public static void Main()
{
using (var context = new MyContext())
{
context.Database.EnsureDeleted();
context.Database.EnsureCreated(); var tags = new[]
{
new Tag { Text = "Golden" },
new Tag { Text = "Pineapple" },
new Tag { Text = "Girlscout" },
new Tag { Text = "Cookies" }
}; var posts = new[]
{
new Post { Title = "Best Boutiques on the Eastside" },
new Post { Title = "Avoiding over-priced Hipster joints" },
new Post { Title = "Where to buy Mars Bars" }
}; posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]);
posts[].Tags.Add(tags[]); context.AddRange(tags);
context.AddRange(posts); context.SaveChanges();
} using (var context = new MyContext())
{
var posts = LoadAndDisplayPosts(context, "as added"); posts.Add(context.Add(new Post { Title = "Going to Red Robin" }).Entity); var newTag1 = new Tag { Text = "Sweet" };
var newTag2 = new Tag { Text = "Buzz" }; foreach (var post in posts)
{
var oldTag = post.Tags.FirstOrDefault(e => e.Text == "Pineapple");
if (oldTag != null)
{
post.Tags.Remove(oldTag);
post.Tags.Add(newTag1);
}
post.Tags.Add(newTag2);
} context.SaveChanges();
} using (var context = new MyContext())
{
LoadAndDisplayPosts(context, "after manipulation");
}
} private static List<Post> LoadAndDisplayPosts(MyContext context, string message)
{
Console.WriteLine($"Dumping posts {message}:"); var posts = context.Posts
.Include("PostTags.Tag")
.ToList(); foreach (var post in posts)
{
Console.WriteLine($" Post {post.Title}");
foreach (var tag in post.Tags)
{
Console.WriteLine($" Tag {tag.Text}");
}
} Console.WriteLine(); return posts;
}
}

Notice that:

  • When seeding the database, we add Tags directly to the Tags collection on Post.
  • When finding and removing existing tags, we can search directly for the Tag and remove it from the Post.Tags collection without needing to use the join entity.

It’s worth calling out again that, just like in the previous post, we still can’t use Tags directly in any query. For example, using it for Include won’t work:

var posts = context.Posts
.Include(e => e.Tags) // Won't work
.ToList();

EF has no knowledge of “Tags”–it is not mapped. EF only knows about the private PostTags navigation property.

Functionally, this is about as far as we can go without starting to mess with the internals of EF. However, in one last post I’ll show how to abstract out the collection and join entity a bit more so that it is easier reuse for different types.

原文链接

Many-to-many relationships in EF Core 2.0 – Part 3: Hiding as ICollection的更多相关文章

  1. Many-to-many relationships in EF Core 2.0 – Part 2: Hiding as IEnumerable

    In the previous post we looked at how many-to-many relationships can be mapped using a join entity. ...

  2. Many-to-many relationships in EF Core 2.0 – Part 1: The basics

    转载这个系列的文章,主要是因为EF Core 2.0在映射数据库的多对多关系时,并不像老的EntityFramework那样有原生的方法进行支持,希望微软在以后EF Core的版本中加入原生支持多对多 ...

  3. Many-to-many relationships in EF Core 2.0 – Part 4: A more general abstraction

    In the last few posts we saw how to hide use of the join entity from two entities with a many-to-man ...

  4. EF Core 1.0 和 SQLServer 2008 分页的问题

    EF Core 1.0 在sqlserver2008分页的时候需要指定用数字分页. EF Core1.0 生成的分页语句中使用了 Featch Next.这个语句只有在SqlServer2012的时候 ...

  5. ASP.NET Core 开发-Entity Framework (EF) Core 1.0 Database First

    ASP.NET Core 开发-Entity Framework Core 1.0 Database First,ASP.NET Core 1.0 EF Core操作数据库. Entity Frame ...

  6. EF Core 1.0中使用Include的小技巧

    (此文章同时发表在本人微信公众号"dotNET每日精华文章",欢迎右边二维码来关注.) 题记:由于EF Core暂时不支持Lazy Loading,所以利用Include来加载额外 ...

  7. .NET Core 1.0、ASP.NET Core 1.0和EF Core 1.0简介

    .NET Core 1.0.ASP.NET Core 1.0和EF Core 1.0简介 英文原文:Reintroducing .NET Core 1.0, ASP.NET Core 1.0, and ...

  8. EF Core 2.0 新特性

    前言 目前 EF Core 的最新版本为 2.0.0-priview1-final,所以本篇文章主要是针对此版本的一些说明. 注意:如果你要在Visual Studio 中使用 .NET Core 2 ...

  9. EF Core 2.0使用MsSql/Mysql实现DB First和Code First

    参考地址 EF官网 ASP.NET Core MVC 和 EF Core - 教程系列 环境 Visual Studio 2017 最新版本的.NET Core 2.0 SDK 最新版本的 Windo ...

随机推荐

  1. mysql(什么是关系型数据库?)

    关系型数据库:建立在关系模型上的数据库. 关系模型:当将日常生活中的实体(学生),和实体的属性(学生的学号,姓名)保存到数据库中,应该如何处理实体结构. 1.数据结构可以规定,同类数据,结构一致.就是 ...

  2. JavaEE之servlet相关技术

    相关技术:为了灵活实现的不同路径(/hello)执行不同的资源( HeIIoMyServlet)我们需要使用XML进行配置;为了限定XML内容,我们需要使用xml约束(DTD或schema);为了获得 ...

  3. NodeJS require路径

    项目需要用nodejs,感觉nodejs是前端装逼神器了,是通向全栈工程师的必经之路哇,接下来开始踏上学习nodejs的征程.下面是第一个hello,world的程序. 1.server.js文件,这 ...

  4. C语言——顺序表插入、删除、定位运算算法

    说明:将元素x插入到顺序表L的第i个数据元素之前,这个i是从1开始的,但是程序中数组都是从0算起的,不要混淆了. 头文件: header.h // 顺序表的结构定义 #define Maxsize 1 ...

  5. DUANG~ 万网轻云服务器,大促狂欢,不仅仅免单!

    DUANG~ 万网轻云服务器,大促狂欢,不仅仅免单!   当老板第一次知道我们要做活动的时候,其实是拒绝的.DUANG~ 打折.降价.挥泪甩卖…太俗套.客户看到一定骂我们,根本没有诚意. 所以轻云大促 ...

  6. 从尾到头打印链表(C++和Python 实现)

    (说明:本博客中的题目.题目详细说明及参考代码均摘自 “何海涛<剑指Offer:名企面试官精讲典型编程题>2012年”) 题目 输入一个链表的头结点, 从尾到头反过来打印出每个结点的值. ...

  7. Windows ->> Windows下一代文件系统 -- Resilient file system(ReFS)

    Comming soon!!! 参考文献: Building the next generation file system for Windows: ReFS ReFS: What you need ...

  8. Threading in C# 5

    Part 5: Parallel Programming In this section, we cover the multithreading APIs new to Framework 4.0 ...

  9. Python学习---模版/包的概念

    1.1. 模块/包的概念 在Python中,一个.py文件就称之为一个模块(Module) 模块一共三种: python标准库 第三方模块 应用程序自定义模块 模块的使用:模块是用来组织函数的 解释器 ...

  10. WAKE-WIN10-SOFT-环境

    操作系统名称 Microsoft Windows 10 专业版版本 10.0.14393 版本 14393其他操作系统描述 没有资料操作系统制造商 Microsoft Corporation系统名称 ...