我正在构建一个FormConditions
接口,其中将有一个带有任意键的对象,每个键都是实现Condition
接口的类的实例。我想将此对象字面量分配给变量,并让结果对象的类型A)仅响应对象字面量中的键,B)尊重每个这些键可能具有的任何子类或其他扩展。>
如果您检查下面的代码,则所有实际类型都可以正常工作。我遇到的问题是,在不显式声明每个键的子类型的情况下,我不知道如何直接将对象分配给变量。相反,我可以通过标识函数makeFormConditions
来传递它,该函数使用泛型正确地推断所得对象的类型。这是执行此操作的唯一方法,还是可以直接分配它?可以根据需要随意更改FormCondition
的定义。
interface Condition {
name: string
id: number
}
type FormConditions<T extends Record<string, Condition>> = {
[P in keyof T]: T[P]
}
class SimpleCondition implements Condition {
constructor(public name: string, public id: number) {}
}
class ListCondition<T> implements Condition {
constructor(public name: string, public id: number, public entries: T[]) {}
}
// This is a passthrough function just to make the types work
function makeFormConditions<T extends Record<string, Condition>>(obj: T): FormConditions<T> {
return obj;
}
// Would prefer to avoid the function call to make types work
const conditions = makeFormConditions({
simpleOne: new SimpleCondition('simpleOne', 1),
simpleTwo: new SimpleCondition('simpleTwo', 2),
list: new ListCondition('list', 3, ['foo', 'bar'])
})
// This works but is redundantly verbose
// const conditions : FormConditions<{
// simpleOne: SimpleCondition;
// simpleTwo: SimpleCondition;
// list: ListCondition<string>;
// }> = {
// simpleOne: new SimpleCondition('simpleOne', 1),
// simpleTwo: new SimpleCondition('simpleTwo', 2),
// list: new ListCondition('list', 3, ['foo', 'bar'])
// }
//
// would instead prefer to not use the function or be
// really specific about the type declaration:
// const conditions : FormConditions = {
// simpleOne: new SimpleCondition('simpleOne', 1),
// simpleTwo: new SimpleCondition('simpleTwo', 2),
// list: new ListCondition('list', 3, ['foo', 'bar'])
// }
conditions.list.name
conditions.list.entries
conditions.simpleOne.name
conditions.simpleOne.entries // error, as expected
答案 0 :(得分:1)
简短的回答::不能,您不能分配包含异构类型的对象文字, 并且 保持通用类型约束。需要受约束的辅助函数(当前已实现)。
Condition
以包括所有子类型属性 Condition
的定义可以扩展为接受可选的entries
数组,这样FormConditions<E, T extends Record<string, Condition<E>>>
可以同时容纳SimpleConditions
和ListConditions
。这具有不希望的副作用,即SimpleCondition
的实例可以正确地引用缺少的entries
属性。
interface Condition<E> {
name: string
id: number
entries?: E[]
}
type FormConditions<E, T extends Record<string, Condition<E>>> = {
[P in keyof T]: T[P]
}
class SimpleCondition<E = never> implements Condition<E> {
constructor(public name: string, public id: number) {}
}
class ListCondition<E> implements Condition<E> {
constructor(public name: string, public id: number, public entries: E[]) {}
}
const conditions: FormConditions<string, Record<string, Condition<string>>> = {
simpleOne: new SimpleCondition('simpleOne', 1),
simpleTwo: new SimpleCondition('simpleTwo', 2),
list: new ListCondition('list', 3, ['foo', 'bar'])
}
conditions.list.name;
conditions.list.entries;
conditions.simpleOne.name;
conditions.simpleOne.entries; // Expected error; however, no error, since `entries` is optional parameter.
Condition
仅包含name
和id
由于Condition
受限制,因此尝试访问entries
实例上的SimpleCondition
时发生错误(如预期的那样)。但是,在type FormConditions<E, T extends Record<string, Condition>>
的上下文中,ListCondition
的实例在引用entries
时会导致错误,因为类型已缩小为Condition
。
interface Condition {
name: string
id: number
}
type FormConditions<E, T extends Record<string, Condition>> = {
[P in keyof T]: T[P]
}
class SimpleCondition<E = never> implements Condition {
constructor(public name: string, public id: number) {}
}
class ListCondition<E> implements Condition {
constructor(public name: string, public id: number, public entries: E[]) {}
}
const conditions: FormConditions<string, Record<string, Condition>> = {
simpleOne: new SimpleCondition('simpleOne', 1),
simpleTwo: new SimpleCondition('simpleTwo', 2),
list: new ListCondition('list', 3, ['foo', 'bar'])
}
conditions.list.name;
conditions.list.entries; // Error: Property 'entries' does not exist on type 'Condition'.
conditions.simpleOne.name;
conditions.simpleOne.entries; // error (as expected - Good)