RC
3.1 介绍
Reference Classes(RC) 是在 R 2.12 版本开始引入的新一代的面向对象系统，也被称为 R5（这并不是官方的名称，只是为了和 S3、S4 保持队形（￣。。￣））。
这个面向对象系统不同于 S3 和 S4 使用泛型函数模型实现类和方法，RC 的方法被封装在类的定义中。
RC 面向对象系统在行为、风格上更像其他面向对象编程语言，如 Java 和 C++ 等。
RC 使用 $ 符号来调用方法，获取和修改对象的属性，调用方法或设置属性的值会修改对象，这种方式不同于常用的函数式编程模型。
3.2 创建 RC 类
首先，我们可以使用 setRefClass 来定义类，并返回一个生成器对象。
setRefClass(Class, fields = , contains = , methods =,
where =, inheritPackage =, ...)
参数列表：
-
Class: 字符串类名 -
fields: 定义属性名称与类型，可以是命名字符串向量或命名列表。 -
contains: 定义父类，多重继承传递父类向量。如果父类也是RC，会继承父类的属性和方法 -
methods: 一个命名列表，定义对象可调用的方法。也可以使用$methods方法定义函数 -
where: 类定义的存储空间 -
inheritPackage: 是否继承父类的环境 -
...: 其他参数类似setClass
定义一个 RC 类
> Person <- setRefClass("Person", fields = c(name='character',age='numeric',gender='factor'))
> Person
Generator for class "Person":
Class fields:
Name: name age gender
Class: character numeric factor
Class Methods:
"sing", "say", "field", "trace", "getRefClass", "initFields", "copy", "callSuper",
".objectPackage", "export", "untrace", "getClass", "show", "usingMethods", ".objectParent",
"import"
Reference Superclasses:
"envRefClass"
使用 $new 实例化对象，也可以直接使用类名实例化
> genders <- factor(c("F", "M"))
> tom <- Person$new(name="tom", age=19, gender=genders[1])
# tom <- Person(name="tom", age=19, gender=genders[1])
> tom
Reference class object of class "Person"
Field "name":
[1] "tom"
Field "age":
[1] 19
Field "gender":
[1] F
Levels: F M
查看类型
> otype(tom)
[1] "RC"
> class(tom)
[1] "Person"
attr(,"package")
[1] ".GlobalEnv"
3.3 访问属性
我们可以使用 $ 符号，访问及修改属性值
> tom$name
[1] "tom"
> tom$age
[1] 19
> tom$age <- tom$age + 1
> tom$age
[1] 20
赋值给另一个对象
> sam <- tom
> sam
Reference class object of class "Person"
Field "name":
[1] "tom"
Field "age":
[1] 20
Field "gender":
[1] F
Levels: F M
> sam$name <- "sam"
> sam
Reference class object of class "Person"
Field "name":
[1] "sam"
Field "age":
[1] 20
Field "gender":
[1] F
Levels: F M
> tom
Reference class object of class "Person"
Field "name":
[1] "sam"
Field "age":
[1] 20
Field "gender":
[1] F
Levels: F M
以直接赋值的方式，只是传递了对象的引用，而不是重新构建了一个实例。
可以调用内置的 copy() 方式，创建一份拷贝
> sam <- tom$copy()
> sam$age <- 28
> sam
Reference class object of class "Person"
Field "name":
[1] "sam"
Field "age":
[1] 28
Field "gender":
[1] F
Levels: F M
> tom
Reference class object of class "Person"
Field "name":
[1] "sam"
Field "age":
[1] 20
Field "gender":
[1] F
Levels: F M
3.4 定义方法
在 RC 对象系统中，我们可以在创建类的同时指定对应的方法，而不需要将类和函数的定义分离。
例如，我们定义 Person 类
Person <- setRefClass("Person", fields = c(name='character',age='numeric',gender='factor'),
methods = list(
setName = function(x) {
name <<- x
},
setAge = function(x) {
age <<- x
}
))
然后使用方法，改变属性值
> jay <- Person(name="jay", age=21, gender=genders[2])
> jay$setAge(28)
> jay
Reference class object of class "Person"
Field "name":
[1] "jay"
Field "age":
[1] 28
Field "gender":
[1] M
Levels: F M
注意：我们在函数内部使用了 <<- 赋值方式。
该赋值方式通常在函数中使用，会在其上层环境中搜索该变量，如果找到了，则重新为该变量赋值；否则会创建为一个全局变量
如果我们将上面的代码改为
Person <- setRefClass("Person", fields = c(name='character',age='numeric',gender='factor'),
methods = list(
setName = function(x) {
name <- x
},
setAge = function(x) {
age <- x
}
))
执行相同的代码
> jay <- Person(name="jay", age=21, gender=genders[2])
> jay$setAge(28)
> jay
Reference class object of class "Person"
Field "name":
[1] "jay"
Field "age":
[1] 21
Field "gender":
[1] M
Levels: F M
发现 age 的值并没有被修改
为了不让类定义看起来很臃肿，可以将函数的定义剥离出来。
我们可以使用 $methods() 的方式为类定义相应的方法。
Person <- setRefClass("Person", fields = c(name='character',age='numeric',gender='factor'))
Person$methods(
setName = function(x) {
name <<- x
},
setAge = function(x) {
age <<- x
}
)
调用方法
> jay <- Person(name="jay", age=21, gender=genders[2])
> jay$setAge(28)
> jay
Reference class object of class "Person"
Field "name":
[1] "jay"
Field "age":
[1] 28
Field "gender":
[1] M
Levels: F M
3.5 定义构造函数
RC 类在实例化对象时，会自动调用构造器函数 $initialize()，我们可以使用该函数来初始化属性值
Person <- setRefClass("Person", fields = c(name='character',age='numeric',gender='factor'),
methods = list(
initialize = function(name="Unknown", age=18, gender=genders[1]) {
name <<- name
age <<- age
gender <<- gender
}
))
实例化
> Person()
Reference class object of class "Person"
Field "name":
[1] "Unknown"
Field "age":
[1] 18
Field "gender":
[1] F
Levels: F M
> Person(name="tom")
Reference class object of class "Person"
Field "name":
[1] "tom"
Field "age":
[1] 18
Field "gender":
[1] F
Levels: F M
3.6 继承
RC 也是通过 contains 参数来指定父类
User <- setRefClass("User", fields = c(username="character", password="character"))
User$methods(
getName = function() {
return(username)
}
)
VIP <- setRefClass("VIP", contains = "User", fields = c(level="numeric"))
使用
tom <- VIP(username="tom", password="123456", level=3)
> tom
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "123456"
Field "level":
[1] 3
> tom$getName()
[1] "tom"
3.7 RC 的内置方法
在我们定义完类之后
> Person
Generator for class "Person":
Class fields:
Name: name age gender
Class: character numeric factor
Class Methods:
"initialize", "field", "trace", "getRefClass", "initFields", "copy", "callSuper",
".objectPackage", "export", "untrace", "getClass", "show", "usingMethods", ".objectParent",
"import"
Reference Superclasses:
"envRefClass"
从 Class Methods 输出的方法中我们可以看到，很多方法都不是我们定义的，这些都是内置的方法。像我们前面用到的 copy 和 initialize 也在其中
我们为上面的继承的例子，添加一些功能
User <- setRefClass("User", fields = c(username="character", password="character"))
User$methods(
getName = function() {
return(username)
}
)
VIP <- setRefClass("VIP", contains = "User", fields = c(level="numeric"))
VIP$methods(
getName = function() {
cat("VIP：", callSuper())
},
add = function(x, y) {
return(x+y)
},
multiple = function(x, y) {
return(x*y)
}
)
我们在子类中重写了 getName 方法，通过 callSuper() 调用父类的 getName 方法获取 name 属性，并在前面添加 VIP 标记
> tom <- VIP(username="tom", password="123456", level=3)
> tom$getName()
VIP： tom
我们重新为 tom 的属性赋值
> tom$initFields(username="sam", password="1234")
Reference class object of class "VIP"
Field "username":
[1] "sam"
Field "password":
[1] "1234"
Field "level":
[1] 3
> tom
Reference class object of class "VIP"
Field "username":
[1] "sam"
Field "password":
[1] "1234"
Field "level":
[1] 3
获取或设置某一属性
> tom$field("username")
[1] "sam"
> tom$field("username", "tom")
> tom
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
[1] 3
获取对象的类的定义
> tom$getClass()
Reference Class "VIP":
Class fields:
Name: username password level
Class: character character numeric
Class Methods:
"getName#User", "multiple", "add", "import", ".objectParent", "usingMethods", "show", "getClass",
"untrace", "export", ".objectPackage", "callSuper", "copy", "initFields", "getRefClass", "trace",
"field", "getName"
Reference Superclasses:
"User", "envRefClass"
> tom$getRefClass()
Generator for class "VIP":
Class fields:
Name: username password level
Class: character character numeric
Class Methods:
"getName#User", "multiple", "add", "import", ".objectParent", "usingMethods", "show", "getClass",
"untrace", "export", ".objectPackage", "callSuper", "copy", "initFields", "getRefClass", "trace",
"field", "getName"
Reference Superclasses:
"User", "envRefClass"
# 查看类型
> otype(tom$getClass())
[1] "S4"
> otype(tom$getRefClass())
[1] "RC"
tom$show() 和 show(tom) 以及直接输入 tom 都可以输出对象的信息
> tom$show()
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
[1] 3
> show(tom)
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
[1] 3
追踪方法
> tom$trace("add")
Tracing reference method "add" for object from class "VIP"
[1] "add"
> tom$add(1, 3)
Tracing tom$add(1, 3) on entry
[1] 4
> tom$add(4, 5)
Tracing tom$add(4, 5) on entry
[1] 9
> tom$untrace("add")
Untracing reference method "add" for object from class "VIP"
[1] "add"
> add(1, 1)
[1] 2
将子类转换为父类类型
> tom$export('User')
Reference class object of class "User"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
> tom
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
[1] 3
从上面的结果可以看到，转换为父类型之后，level 属性被删除了，但是原始对象并没有被修改
使用一个对象给另一个对象赋值
> sam <- VIP()
> sam
Reference class object of class "VIP"
Field "username":
character(0)
Field "password":
character(0)
Field "level":
numeric(0)
> sam$import(tom$export("User"))
> sam
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
numeric(0)
> sam$import(tom)
> sam
Reference class object of class "VIP"
Field "username":
[1] "tom"
Field "password":
[1] "1234"
Field "level":
[1] 3
3.8 RC 类方法
我们使用 setRefClass 函数定义的类，会自动包含一些方法帮助我们查看类的属性和方法。
比如，我们上面使用到的 $new() 函数，还有一些其他函数
查看 User 类中的属性及其类型
> User$fields()
username password
"character" "character"
查看 User 中定义的方法
> User$methods()
[1] ".objectPackage" ".objectParent" "callSuper" "copy" "export"
[6] "field" "getClass" "getName" "getRefClass" "import"
[11] "initFields" "show" "trace" "untrace" "usingMethods"
查看函数的调用方式
> VIP$help("add")
Call:
$add(x, y)
为 User 的属性增加 get 和 set 方法
> User$accessors("password")
> User$methods()
[1] ".objectPackage" ".objectParent" "callSuper" "copy" "export"
[6] "field" "getClass" "getName" "getPassword" "getRefClass"
[11] "import" "initFields" "setPassword" "show" "trace"
[16] "untrace" "usingMethods"
将属性固定，一旦赋值后便不可更改
> User$lock("username")
# 查看被锁定的属性
> User$lock()
[1] "username"
> a <- User(username="fly")
> a
Reference class object of class "User"
Field "username":
[1] "fly"
Field "password":
character(0)
> a$username <- "a"
错误: invalid replacement: reference class field ‘username’ is read-only
3.9 示例
Person <- setRefClass(
"Person",
fields = list(
name="character",
age="numeric"
),
methods = list(
initialize = function(name="Unknown", age=18) {
name <<- name
age <<- age
},
talk = function() {
return("talking...")
}
)
)
Chinese <- setRefClass(
"Chinese",
contains = "Person",
fields = list(
language="character"
),
methods = list(
initialize = function(name, age, language="chinese") {
callSuper(name, age)
language <<- language
},
talk = function() {
return(paste0(callSuper(), language))
}
)
)
Chinese$lock("language")
American <- setRefClass(
"American",
contains = "Person",
fields = list(
language="character"
),
methods = list(
initialize = function(name, age, language="english") {
callSuper(name, age)
language <<- language
},
talk = function() {
return(paste0(callSuper(), language))
}
)
)
American$lock("language")
使用
> a <- American("lusy", 20)
> a
Reference class object of class "American"
Field "name":
[1] "lusy"
Field "age":
[1] 20
Field "language":
[1] "english"
> a$talk()
[1] "talking...english"
> b <- Chinese("lisin", 18)
> b
Reference class object of class "Chinese"
Field "name":
[1] "lisin"
Field "age":
[1] 18
Field "language":
[1] "chinese"
> b$talk()
[1] "talking...chinese"
网友评论