寫在前面
.NET 5.0已經發布,C# 9.0也為我們帶來了許多新特性,其中最讓我印象深刻的就是init和record type,很多文章已經把這兩個新特性討論的差不多了,本文不再詳細討論,而是通過使用角度來思考這兩個特性。
init
init是C# 9.0中引入的新的訪問器,它允許被修飾的屬性在物件初始化的時候被賦值,其他場景作為只讀屬性的存在。直接使用的話,可能感受不到init的意義,所以我們先看看之前是如何設定屬性為只讀的。
private set設定屬性為只讀
設定只讀屬性有很多種方式,本文基於private set來討論。
首先宣告一個產品類,如下程式碼所示,我們把Id設定成了只讀,這個時候也就只能通過建構函式來賦值了。在通常情況下,實體的唯一標識是不可更改的,同時也要防止Id被意外更改。
1: public class Product
2: {
3: public Product(int id)
4: {
5: this.Id = id;
6: }
7:
8: public int Id { get; private set; }
9: //public int Id { get; }
10:
11: public string ProductName { get; set; }
12:
13: public string Description { get; set; }
14: }
15:
16: class Program
17: {
18: static void Main(string[] args)
19: {
20: Product product = new Product(1)
21: {
22: ProductName = "test001",
23: Description = "Just a description"
24: };
25:
26: Console.WriteLine($"Current Product Id: {product.Id},\n\rProduct Name: {product.ProductName}, \n\rProduct Description: {product.Description}");
27:
28: //執行結果
29: //Current Product Id: 1,
30: //Product Name: test001,
31: //Product Description: Just a description
32:
33: Console.ReadKey();
34: }
35: }
record方式設定只讀
使用init方式,是非常簡單的,只需要把private set改成init就行了:
1: public int Id { get; init; }
兩者比較
為了方便比較,我們可以將ProductName設定成了private set,然後通過ILSpy來檢視一下編譯後的程式碼,看看編譯後的Id和ProductName有何不同
咋一看,貌似沒啥區別,都使用到了initonly來修飾。但是如果僅僅只是替換宣告方式,那麼這個新特性似乎就沒有什麼意義了。
接下來我們看第二張圖:
如圖示記的那樣,區別還是很明顯的,通過init修飾的屬性並沒有完全替換掉set,由此看來微軟在設計init的時候,還是挺用心思的,也為後面的賦值留下了入口。
1: instance void modreq([System.Runtime]System.Runtime.CompilerServices.IsExternalInit) set_Id (
2: int32 'value'
3: )
另外在賦值的時候,使用private set修飾的屬性,需要定義建構函式,通過建構函式賦值。而使用了init修飾的屬性,則不需要定義建構函式,直接在物件初始化器中賦值即可。
1: Product product = new Product
2: {
3: Id = 1,
4: ProductName = "test001",
5: Description = "Just a description"
6: };
7:
8: product.Id = 2;//Error CS8852 Init-only property or indexer 'Product.Id' can only be assigned in an object initializer, or on 'this' or 'base' in an instance constructor or an 'init' accessor.
如上程式碼所示,只讀屬性Id的賦值並沒有在建構函式中賦值,畢竟當一個類的只讀欄位十分多的時候,建構函式也變得複雜。而且在賦值好之後,無法修改,這和我們對只讀屬性在通常情況下的理解是一致的。另外通過init修飾的好處便是省卻了一部分只讀屬性在操作上的複雜性,使得物件的宣告與賦值更加直觀。
在合適的場景下選擇最好的程式設計方式,是程式設計師的一貫追求,千萬不要為了炫技而把init當成了茴字的第N種寫法到處去問。
record
record是一個非常有用的特性,它是不可變型別,其相等性是通過內部的幾個屬性來確定的,同時它支援我們以更加方便的方式、像定義值型別那樣來定義不可變引用型別。
我們把之前的Product類改成record型別,如下所示:
1: public record Product
2: {
3: public Product(int id, string productName, string description) => (Id, ProductName, Description) = (id, productName, description);
4:
5: public int Id { get; }
6:
7: public string ProductName { get; }
8:
9: public string Description { get; }
10: }
然後檢視一下IL,可以看到record會被編譯成類,同時繼承了System.Object,並實現了IEquatable泛型介面。
編譯器為我們提供的幾個重要方法如下:
- Equals
- GetHashCode()
- Clone
- PrintMembers和ToString()
比較重要的三個方法
Equals:
通過圖片中的程式碼,我們知道比較兩個record物件,首先需要比較型別是否相同,然後再依次比較內部屬性。
GetHashCode():
record型別通過基型別以及所有的屬性及欄位的方式來計算HashCode,這在整個繼承層次結構中增強了基於值的相等性,也就意味著兩個同名同姓的人不會被認為是同一個人
Clone:
這個方法貌似非常簡單,實在看不出有什麼特別的地方,那麼我們通過後面的內容再來解釋這個方法。
record在DDD值物件中的應用
record之前的定義方式:
瞭解DDD值物件的小夥伴應該想到了,record型別的特性非常像DDD中關於值物件的描述,比如不可變性、其相等於是基於其內部的屬性的等等,我們先來看下值型別的定義方式。
1: public abstract class ValueObject
2: {
3: public static bool operator ==(ValueObject left, ValueObject right)
4: {
5: if (ReferenceEquals(left, null) ^ ReferenceEquals(right, null))
6: {
7: return false;
8: }
9: return ReferenceEquals(left, null) || left.Equals(right);
10: }
11:
12: public static bool operator !=(ValueObject left, ValueObject right)
13: {
14: return !(left == right);
15: }
16:
17: protected abstract IEnumerable<object> GetEqualityComponents();
18:
19:
20: public override bool Equals(object obj)
21: {
22: if (obj == null || obj.GetType() != GetType())
23: {
24: return false;
25: }
26:
27: var other = (ValueObject)obj;
28:
29: return this.GetEqualityComponents().SequenceEqual(other.GetEqualityComponents());
30: }
31:
32: public override int GetHashCode()
33: {
34: return GetEqualityComponents()
35: .Select(x => x != null ? x.GetHashCode() : 0)
36: .Aggregate((x, y) => x ^ y);
37: }
38: // Other utility methods
39: }
40: public class Address : ValueObject
41: {
42: public string Street { get; private set; }
43: public string City { get; private set; }
44: public string State { get; private set; }
45: public string Country { get; private set; }
46: public string ZipCode { get; private set; }
47:
48: public Address(string street, string city, string state, string country, string zipcode)
49: {
50: Street = street;
51: City = city;
52: State = state;
53: Country = country;
54: ZipCode = zipcode;
55: }
56:
57: protected override IEnumerable<object> GetEqualityComponents()
58: {
59: // Using a yield return statement to return each element one at a time
60: yield return Street;
61: yield return City;
62: yield return State;
63: yield return Country;
64: yield return ZipCode;
65: }
66:
67: public override string ToString()
68: {
69: return $"Street: {Street}, City: {City}, State: {State}, Country: {Country}, ZipCode: {ZipCode}";
70: }
71: }
main方法如下:
1: static void Main(string[] args)
2: {
3: Address address1 = new Address("aaa", "bbb", "ccc", "ddd", "fff");
4: Console.WriteLine($"address1: {address1}");
5:
6: Address address2 = new Address("aaa", "bbb", "ccc", "ddd", "fff");
7: Console.WriteLine($"address2: {address2}");
8:
9: Console.WriteLine($"address1 == address2: {address1 == address2}");
10:
11: string jsonAddress1 = address1.ToJson();
12: Address jsonAddress1Deserialize = jsonAddress1.FromJson<Address>();
13: Console.WriteLine($"jsonAddress1Deserialize == address1: {jsonAddress1Deserialize == address1}");
14:
15: Console.ReadKey();
16: }
執行結果如下:
1: 基於class:
2: address1: Street: aaa, City: bbb, State: ccc, Country: ddd, ZipCode: fff
3: address2: Street: aaa, City: bbb, State: ccc, Country: ddd, ZipCode: fff
4: address1 == address2: True
5: jsonAddress1Deserialize == address1: True
採用record方式定義:
如果有大量的值物件需要我們編寫,這無疑是加重我們的開發量的,這個時候record就派上用場了,最簡潔的record風格的程式碼如下所示,只有一行:
1: public record Address(string Street, string City, string State, string Country, string ZipCode);
IL程式碼如下圖所示,從圖中我們也可以看到record型別的物件,預設情況下用到了init來限制屬性的只讀特性。
main方法程式碼不變,執行結果也沒有因為Address從class變成record而發生改變
1: 基於record:
2: address1: Street: aaa, City: bbb, State: ccc, Country: ddd, ZipCode: fff
3: address2: Street: aaa, City: bbb, State: ccc, Country: ddd, ZipCode: fff
4: address1 == address2: True
5: jsonAddress1Deserialize == address1: True
如此看來我們的程式碼節省的不止一點點,而是太多太多了,是不是很爽啊。
record物件屬性值的更改
使用方式如下:
1: class Program
2: {
3: static void Main(string[] args)
4: {
5: Address address1 = new Address("aaa", "bbb", "ccc", "ddd", "fff");
6: Console.WriteLine($"1. address1: {address1}");
7:
8: Address addressWith = address1 with { Street = "############" };
9:
10: Console.ReadKey();
11: }
12: }
13:
14: public record Address(string Street, string City, string State, string Country, string ZipCode);
通過ILSpy檢視如下所示:
1: private static void Main(string[] args)
2: {
3: Address address1 = new Address("aaa", "bbb", "ccc", "ddd", "fff");
4: Console.WriteLine($"1. address1: {address1}");
5: Address address2 = address1.<Clone>$();
6: address2.Street = "############";
7: Address addressWith = address2;
8: Console.ReadKey();
9: }
由此可以看到record在更改的時候,實際上是通過呼叫Clone而產生了淺拷貝的物件,這也非常符合DDD ValueObject的設計理念。
參考:
- https://docs.microsoft.com/en-us/dotnet/architecture/microservices/microservice-ddd-cqrs-patterns/implement-value-objects
- https://deviq.com/value-object/