zoukankan      html  css  js  c++  java
  • 开源实体映射框架EmitMapper介绍

    • 综述
          EmitMapper是一个开源实体映射框架,地址:http://emitmapper.codeplex.com/
          EmitMapper映射效率比较高,接近硬编码。EmitMapper采用emit方式在运行时动态生成IL,而其他映射框架多是采用反射机制。此外EmitMapper最大限度地减少了拆箱装箱操作和映射过程中的额外的调用。
          EmitMapper支持.net的所有平台:Framework 3.5、Microsoft Silverlight 3、Mono。
     
          EmitMapper的使用非常简单,不需要指定任何的映射策略。系统会采用默认的映射配置器DefaultMapConfig完成映射操作。
     
    映射代码
    复制代码
     1     public class Sourse
    2 {
    3 public int A;
    4 public decimal? B;
    5 public string C;
    6 public Inner D;
    7 public string E;
    8 }
    9
    10 public class Dest
    11 {
    12 public int? A;
    13 public decimal B;
    14 public DateTime C;
    15 public Inner D;
    16 public string F;
    17 }
    18
    19 public class Inner
    20 {
    21 public long D1;
    22 public Guid D2;
    23 }
    24
    25 Sourse src = new Sourse
    26 {
    27 A = 1,
    28 B = 0M,
    29 C = "2011/9/21 0:00:00",
    30 D = new Inner
    31 {
    32 D2 = Guid.NewGuid()
    33 },
    34 E = "test"
    35 };
    36
    37 ObjectsMapper<Sourse, Dest> mapper =
    38 ObjectMapperManager.DefaultInstance.GetMapper<Sourse, Dest>();
    39 Dest dst = mapper.Map(src);
    40
    41 Assert.AreEqual<string>(dst.B.ToString(), src.B);
    42 Assert.AreEqual<long>(dst.C.C1, 0L);
    43 Assert.AreEqual<Guid>(dst.C.C2, src.C.C2);
    44 Assert.IsNull(dst.E);
    复制代码
     
     
    • 默认映射配置器
          默认的映射配置器能自动转换以下几种类型:
    任何类型到string类型使用ToString()方法;
    可以使用System.Convert类转换的原始类型;
    可空类型、枚举类型、各种集合类型、结构与类;
    复杂的嵌套类型采用递归方式转换;
     
          如果默认的转换满足不了需求,默认的映射配置器还允许指定命名约定,自定义构造函数,自定义转换器,忽略成员等。
         

    支持的方法

    描述

    ConvertUsing

    为指定的成员提供自定义的转换逻辑

    ConvertGeneric

    为指定的泛型类型成员提供自定义的转换逻辑

    ConstructBy

    为目标对象使用指定的构造函数替代默认构造函数

    NullSubstitution

    当源对象中指定的成员在为null时,给目标对象的成员赋值

    IgnoreMembers

    忽略指定成员的映射

    PostProcess

    在映射完成后执行指定的方法

    ShallowMap

    指定的成员采用浅拷贝方式映射

    DeepMap

    指定的成员采用深拷贝方式映射

    MatchMembers

    如果成员名称的映射不采用精确匹配,可以指定具体的映射逻辑

     
          选择几个方法简单示例如下:
     
    Default映射配置器
     1     public class Sourse
    2 {
    3 public int A;
    4 public decimal? B;
    5 public string C;
    6 public Inner D;
    7 public string E;
    8 }
    9
    10 public class Dest
    11 {
    12 public int? A;
    13 public decimal B;
    14 public DateTime C;
    15 public Inner2 D;
    16 public string F;
    17 }
    18
    19 public class Inner
    20 {
    21 public long D1;
    22 public Guid D2;
    23 }
    24
    25 public class Inner2
    26 {
    27 public long D12;
    28 public Guid D22;
    29 }
    30
    31 ObjectsMapper<Sourse, Dest> mapper1 =
    32 new ObjectMapperManager().GetMapper<Sourse, Dest>(
    33 new DefaultMapConfig()
    34 .IgnoreMembers<Sourse, Dest>(new string[] { "A" })
    35 .NullSubstitution<decimal?, decimal>((value) => -1M)
    36 .ConvertUsing<Inner, Inner2>(value => new Inner2 { D12 = value.D1, D22 = value.D2 })
    37 .PostProcess<Dest>((value, state) => { value.F = "nothing"; return value; })
    38 );
    39 Dest dst = mapper1.Map(src);
    40
    41 Assert.IsNull(dst.A);
    42 Assert.AreEqual<decimal>(dst.B, -1M);
    43 Assert.AreEqual<Guid>(dst.D.D22, src.D.D2);
    44 Assert.AreEqual<string>(dst.F, "nothing");
      
     
    • 自定义映射配置器
          当然EmitMapper是个非常灵活的框架,也可以自定义映射配置器,实现定制的映射操作。
          自定义的映射配置器可以继承自DefaultMapConfig或CustomMapConfig,利用基类的一些功能实现定制的映射,也可以继承自接口ImappingConfigurator,完全从头实现。
     
          比如可以实现从HTTP中通过Post方式提交的Form数据到具体业务实体类的映射,下面通过继承ImappingConfigurator来实现。
     
    自定义映射配置器
     1     public class FormCollectionMapConfig : IMappingConfigurator
    2 {
    3 public static TPostData GetPostData<TPostData>() where TPostData : class , new()
    4 {
    5 ObjectsMapper<NameValueCollection, TPostData> mapper
    6 = new ObjectMapperManager().GetMapper<NameValueCollection, TPostData>(new FormCollectionMapConfig());
    7
    8 return mapper.Map(HttpContext.Current.Request.Form);
    9 }
    10
    11 public IMappingOperation[] GetMappingOperations(Type from, Type to)
    12 {
    13 var members = ReflectionUtils.GetPublicFieldsAndProperties(to);
    14 return members
    15 .Select(m => new DestWriteOperation()
    16 {
    17 Destination = new MemberDescriptor(m),
    18 Getter =
    19 (ValueGetter<object>)
    20 (
    21 (form, status) =>
    22 {
    23 FormCollection forms = new FormCollection((NameValueCollection)form);
    24 IValueProvider valueProvider = forms.ToValueProvider();
    25 ValueProviderResult res = valueProvider.GetValue(m.Name);
    26 if (res != null)
    27 {
    28 return ValueToWrite<object>.ReturnValue(res.ConvertTo(ReflectionUtils.GetMemberType(m)));
    29 }
    30 else
    31 {
    32 return ValueToWrite<object>.Skip();
    33 }
    34 }
    35 )
    36 }
    37 )
    38 .ToArray();
    39 }
    40
    41 public string GetConfigurationName()
    42 {
    43 return null;
    44 }
    45
    46 public IRootMappingOperation GetRootMappingOperation(Type from, Type to)
    47 {
    48 return null;
    49 }
    50
    51 public StaticConvertersManager GetStaticConvertersManager()
    52 {
    53 return null;
    54 }
    55 }
     
     

    Customization using default configurator

    Customization using default configurator

    Customization overview
    Custom converters
    Custom converters_for_generics
    Null substitution
    Ignoring members
    Custom constructors
    Shallow and_deep_mapping
    Names matching
    Post processing

    Customization overview.

    Default configurator is the easiest way to customize mapping. Using is rather simple: create instance of type DefaultMapConfig, setup it and pass it to the method GetMapper of ObjectMapperManager. For example:

    var mapper = ObjectMapperManager
    	.DefaultInstance
    	.GetMapper<Source, Destination>(
    		new DefaultMapConfig().NullSubstitution<int?, int>(() => -1)
    	);
    ...
    Destination d = mapper.Map(new Source());
    

    Class "DefaultMapConfig" contains number of configuration methods some of them are generic methods and are parameterized by argument types. This types are the types of fields and properties to which this configuration rule should be applied. For example the configuration method:
    new DefaultMapConfig().PostProcess<Foo>( 
        (value, state) => Console.WriteLine("Post processing a Foo or it's descendant: " + value.ToString()) 
    )
    
    will be applied to all fields and properties that are instances of class Foo or any it's descendants.

    If you want the configuration method to apply to any type then use "object" as an argument type. For example:
    new DefaultMapConfig().PostProcess<object>( (value, state) => Console.WriteLine("Post processing: " + value.ToString()) )
    
    will be applied to any field or property and will write to the console all mapped values.

    Custom converters.

    You can setup custom converters for every type of source member and destination member. It is possible using the following method:

    /// <summary>
    /// Define custom type converter
    /// </summary>
    /// <typeparam name="From">Source type</typeparam>
    /// <typeparam name="To">Destination type</typeparam>
    /// <param name="converter">Function which converts an inctance of the source type to an instance of the destination type</param>
    /// <returns></returns>
    DefaultMapConfig ConvertUsing<From, To>(Func<From, To> converter)
    

    Example:

    public class A
    {
    	public string fld1;
    	public string fld2;
    }
    
    public class B
    {
    	public string fld1 = "B::fld1";
    	public string fld2 = "B::fld2";
    }
    
    [TestMethod]
    public void Test_CustomConverter2()
    {
    	A a = Context.objMan.GetMapper<B, A>(
    		new DefaultMapConfig().ConvertUsing<object, string>(v => "converted " + v.ToString())
    	).Map(new B());
    	Assert.AreEqual("converted B::fld1", a.fld1);
    	Assert.AreEqual("converted B::fld2", a.fld2);
    }
    

    Custom converters for generics.

    You may want to specify a custom conversion method for one gneric class to another, for example for "HashSet<T>" to "List<T>" where "T" can be anything. Because you don't know concrete argument type you cannot using method "ConvertUsing". Instead of this method "ConvertGeneric" should be used.

    /// <summary>
    /// Define conversion for a generic. It is able to convert not one particular class but all generic family
    /// providing a generic converter.
    /// </summary>
    /// <param name="from">Type of source. Can be also generic class or abstract array.</param>
    /// <param name="to">Type of destination. Can be also generic class or abstract array.</param>
    /// <param name="converterProvider">Provider for getting detailed information about generic conversion</param>
    /// <returns></returns>
    public IMappingConfigurator ConvertGeneric(Type from, Type to, ICustomConverterProvider converterProvider);
    ...
    /// <summary>
    /// Provider for getting detailed information about generic conversion
    /// </summary>
    public interface ICustomConverterProvider
    {
    	/// <summary>
    	/// Getting detailed information about generic conversion
    	/// </summary>
    	/// <param name="from">Type of source. Can be also generic class or abstract array.</param>
    	/// <param name="to">Type of destination. Can be also generic class or abstract array.</param>
    	/// <param name="mappingConfig">Current mapping configuration</param>
    	/// <returns>Detailed description of a generic converter.</returns>
        CustomConverterDescriptor GetCustomConverterDescr(Type from, Type to, MapConfigBaseImpl mappingConfig);
    }
    ...
    /// <summary>
    /// Detailed description of a generic converter. 
    /// </summary>
    public class CustomConverterDescriptor
    {
    	/// <summary>
    	/// Type of class which performs conversion. This class can be generic which will be parameterized with types 
    	/// returned from "ConverterClassTypeArguments" property.
    	/// </summary>
        public Type ConverterImplementation { get; set; }
    
    	/// <summary>
    	/// Name of conversion method of class which is specified by "ConverterImplementation" property.
    	/// </summary>
        public string ConversionMethodName { get; set; }
    	
    	/// <summary>
    	/// Type arguments for parameterizing generic converter specified by "ConverterImplementation" property.
    	/// </summary>
        public Type[] ConverterClassTypeArguments { get; set; }
    }
    

    In order to setup custom generic conversion you should create a generic class with conversion method:

    class HashSetToListConverter<T>
    {
    	public List<T> Convert(HashSet<T> from, object state)
    	{
    		if (from == null)
    		{
    			return null;
    		}
    		return from.ToList();
    	}
    }
    

    The next step is to specify a custom converter provider for the "HashSetToListConverter<T>". In our case we can directly use class "DefaultCustomConverterProvider" which returns string "Convert" as method name for conversion method and type arguments of the source generic object. So, if we are converting HashSet<int> to List<int> then DefaultCustomConverterProvider returns array of one item: "new Type[]{ typeof(int) }" and this type item from the array will be used as type argument for class "HashSetToListConverter<T>"

    Using custom generic converter:

    var mapper = ObjectMapperManager
    	.DefaultInstance
    	.GetMapper<Source, Destination>(
    		new DefaultMapConfig()
    		.ConvertGeneric(
    			typeof(HashSet<>), 
    			typeof(List<>), 
    			new DefaultCustomConverterProvider(typeof(HashSetToListConverter<>))
    		)
    	);
    

    Null substitution.

    There can be situations when if a source object has null in a field (or property) then mapper should write some value (not null!) to the destination field. For example we have the following classes:

    class Source
    {
    	public field int? i;
    }
    class Destination
    {
    	public field int i;
    }
    

    If we map instance of class "Source" to instance of class "Destination", Emit Mapper library should decide what to write to Destination::i when Source::i is null. It can be specified using the following code:

    var mapper = ObjectMapperManager
    	.DefaultInstance
    	.GetMapper<Source, Destination>(
    		new DefaultMapConfig().NullSubstitution<int?, int>(() => -1)
    	);
    var d = mapper.Map(new Source());
    Assert.AreEqual(-1, d.i);
    

    Ignoring members

    If you don't want to map some fields or properties use method "IgnoreMembers":

    [TestMethod]
    public class B
    {
    	public string str1 = "B::str1";
    }
    
    public class A
    {
    	public string str1 = "A::str1";
    }
    
    public void GeneralTests_Ignore()
    {
    	var a = ObjectMapperManager.DefaultInstance.GetMapper<B, A>(
    		new DefaultMapConfig().IgnoreMembers<B, A>(new[]{"str1"})
    	).Map(new B());
    	Assert.AreEqual("A::str1", a.str1);
    }
    

    Custom constructors

    In some cases you don't want to use default constructor for a destination member but instead some custom construction method or non-default constructor.

    public class ConstructBy_Source
    {
    	public class NestedClass
    	{
    		public string str = "ConstructBy_Source::str";
    	}
    	public NestedClass field = new NestedClass();
    }
    
    public class ConstructBy_Destination
    {
    	public class NestedClass
    	{
    		public string str;
    		public int i;
    		public NestedClass(int i)
    		{
    			str = "ConstructBy_Destination::str";
    			this.i = i;
    		}
    	}
    	public NestedClass field;
    }
    [TestMethod]
    public void ConstructByTest()
    {
    	var mapper = ObjectMapperManager
    		.DefaultInstance
    		.GetMapper<ConstructBy_Source, ConstructBy_Destination>(
    			new DefaultMapConfig().ConstructBy <ConstructBy_Destination.NestedClass>(
    				() => new ConstructBy_Destination.NestedClass(3)
    			)
    		);
    	var d = mapper.Map(new ConstructBy_Source());
    	Assert.AreEqual("ConstructBy_Source::str", d.field.str);
    	Assert.AreEqual(3, d.field.i);
    }
    

    Shallow and deep mapping

    If a field of destination object is the same type as field of source object and this type is complex then the field can be copied by reference (shallow copying) or for the destination field can be constructed separate object and then all nested fields and properties will be copied recursively (deep copying). This behavior can be defined using methods "ShallowMap" or "DeepMap"

    public class A1
    {
    	public int fld1;
    	public string fld2;
    }
    
    public class A2
    {
    	public string[] fld3;
    }
    
    public class A3
    {
    	public A1 a1 = new A1();
    	public A2 a2 = new A2();
    }
    
    public class B3
    {
    	public A1 a1 = new A1();
    	public A2 a2 = new A2();
    }
    
    
    b = new B3();
    Context.objMan.GetMapper<B3, A3>(new DefaultMapConfig().ShallowMap<A1>().DeepMap<A2>()).Map(b, a);
    b.a1.fld1 = 333;
    b.a2.fld3 = new string[1];
    
    Assert.AreEqual(333, a.a1.fld1); // shallow map
    Assert.IsNull(a.a2.fld3); // deep map
    

    Names matching

    If member names of source object don't exactly match to member names of destination object (for example, destination names can have prefix "m_") you can define a function which matches names using method "MatchMembers"

    public class Source
    {
    	public string field1 = "Source::field1";
    	public string field2 = "Source::field2";
    	public string field3 = "Source::field3";
    }
    
    public class Destination
    {
    	public string m_field1;
    	public string m_field2;
    	public string m_field3;
    }
    
    [TestMethod]
    public void GeneralTests_Example2()
    {
    	var mapper = ObjectMapperManager.DefaultInstance.GetMapper<Source, Destination>(
    		new DefaultMapConfig().MatchMembers((m1, m2) => "m_" + m1 == m2)
    	);
    
    	Source src = new Source();
    	Destination dst = mapper.Map(src);
    	Assert.AreEqual(src.field1, dst.m_field1);
    	Assert.AreEqual(src.field2, dst.m_field2);
    	Assert.AreEqual(src.field3, dst.m_field3);
    }
    

    The first argument of the matching function is neme of source member, the second argument is name of destination member.

    Post processing

    If special actions should be executed after mapping you can use the following method:

    /// <summary>
    /// Define postprocessor for specified type
    /// </summary>
    /// <typeparam name="T">Objects of this type and all it's descendants will be postprocessed</typeparam>
    /// <param name="postProcessor"></param>
    /// <returns></returns>
    public IMappingConfigurator PostProcess<T>(ValuesPostProcessor<T> postProcessor)
    

    Example:

    public class Destination
    {
    	public List<int> list;
    }
    
    
    public class Source
    {
    	public int[] list;
    }
    
    var mapper = ObjectMapperManager
    	.DefaultInstance
    	.GetMapper<Source, Destination>(
    		new DefaultMapConfig()
    		.PostProcess<Destination>( (destination, state) => { destination.list.Sort(); return destination;} )
    	);
    

    Last edited Jan 9 2010 at 2:47 AM by romankovs, version 17

     
     
     
     
  • 相关阅读:
    软件开发 [CJOJ 1101] [NOIP 模拟]
    OI中卡常数技巧
    疫情控制 [NOIP2012]
    开车旅行 [NOIP 2012]
    观光公交 [NOIP 2011] [思维推导]
    选择客栈 [NOIP 2011]
    2016级算法期末上机-F.中等·AlvinZH's Fight with DDLs II
    2016级算法期末上机-E.中等·ModricWang's Fight with DDLs II
    2016级算法期末上机-D.简单·AlvinZH's Fight with DDLs I
    2016级算法期末上机-C.简单·Bamboo's Fight with DDLs III
  • 原文地址:https://www.cnblogs.com/aaa6818162/p/2557879.html
Copyright © 2011-2022 走看看