Kotlin Basics: Standard Extension Functions kotlin

https://lmller.github.io/kotlin-standard-extensions

Kotlin’s most powerful feature (compared to Java) are probably Extension Functions.
They allow you to add new functionality to an existing class, without using inheritance or reflection. You can define your own extension functions and Kotlin ships with a lot of pre-defined ones.
There are e.g. a lot extensions for collections, but also those that extend every class in your program. That’s right - every single class is extended by the functions in Standard.kt
When I started with Kotlin, I found them quite confusing. With this post, I hope to un-confuse them a little bit.

Before we get to the details, some basics.

An Extension Function is a function that adds new functionality to an existing class. It’s functionally equivalent to a static utility function in Java, but we can call it from the object itself.

In Java, you might have something like this:

class Util {
  public static final boolean isNumeric(String receiver) {
     return reveiver.matches("\\d+");
  }
}
...

String myString = ...;

if(Util.isNumeric(myString)) ...

By using an extension function, Kotlin allows us to call the isNumeric method directly on the receiving object:

fun String.isNumeric(): Boolean {
  return this.matches("\\d+".toRegex())
}
...

val myString = ...

if(myString.isNumeric()) ...

Pretty cool, right? You might agree that using extension functions increases the code’s readability.

The example shows how an extension function is defined. We use the fun keyword, followed by the type we want to extend (in this case String). Then a . and the name of the function. That’s it!
The extension function will always receive an implicit parameter this, which is the object we called the extension function on (in this case, myString).

Let’s take a look at Kotlin’s standard extension functions.

run

public inline fun <T, R> T.run(block: T.() -> R): R = block()

This looks a bit messy in the beginning. But it’s actually quite simple. run is a generic extension function on any type T, that executes another extension function on this type (extension functions as parameters are symbolized as T.()) and returns the result of that function.

Consider this example:

val generator = PasswordGenerator()
generator.seed = "someString"
generator.hash = {s -> someHash(s)}
generator.hashRepititions = 1000

val password: Password = generator.generate()

Someone didn’t quite think through the design of this password generator class. Its constructor does nothing, but it needs a lot of initialization. To use this class, I need to introduce a variable generator, set all necessary parameters and use generate to generate the actual password. This is cool if I want to generate multiple passwords with the same generator, but if I throwaway the generator variable anyway, I can save some typing by using run:

val password: Password = PasswordGenerator().run {
       seed = "someString"
       hash = {s -> someHash(s)}
       hashRepetitions = 1000

       generate()
   }

Lambdas in Kotlin implicitly return the result of the last line. That’s why I can omit the temporary variable and store the password directly. Because an extension function is passed to run I can also access the password generator’s properties like seed or hash directly.
This way “the things that belong together” stay together. It’s one line more, I give you that. But it’s still less code with less redundancy I had to write.

apply

public inline fun <T> T.apply(block: T.() -> Unit): T { block(); return this }

This function looks almost like run, but there is a slight difference. apply will always return this, which is the receiver of the extension function.
That means that we can use apply for builder-style initialization, even if the designer of the class didn’t implement it with a builder in mind.
If we stick to the password generator example:

val generator = PasswordGenerator().apply {
       seed = "someString"
       hash = {s -> someHash(s)}
       hashRepetitions = 1000
   }

val password = generator.generate()

This is particularly useful if you need an object with the same settings more than once.
It can also be used to avoid init{} blocks during the initialization of a class.
Instead of:

class Message(message: String, signature: String) {
  val body = MessageBody()
  
  init {
    body.text = message + "\n" + signature
  }
}

You can write:

class Message(message: String, signature: String) {
  val body = MessageBody().apply {
    text = message + "\n" + signature
  }
}

let

public inline fun <T, R> T.let(block: (T) -> R): R = block(this)

let is particularly useful to use it instead of a null check:

val fruitBasket = ...

val result = apple?.let {
  fruitBasket.add(it)
}

The apple (it) will only be added to the basket if it’s not null
Notice that let will return the result of the last line in the lambda (which is the result of add).
Just like runlet also helps to keep the scope of your variables small. Instead of this you have to use it or a custom variable name to reference it:

val fruitBasket = ...

appleTree.pick()?.let {
  fruitBasket.add(it)
}

also

public inline fun <T> T.also(block: (T) -> Unit): T { block(this); return this }

Kotlin 1.1 introduced this new extension function called also. It fills the gap between let and apply.
Just like apply, it will always return its receiver. But instead of an extension function it takes a normal function as an argument.

class FruitBasket {
    private var weight = 0

    fun addFrom(appleTree: AppleTree) {
        val apple = appleTree.pick().also { apple ->
            this.weight += apple.weight
            add(apple)
        }
        ...
    }
    ...
    fun add(fruit: Fruit) = ...
}

I renamed the implicit it to an explicit apple this time. Assuming the function appleTree.pick() returns an apple, the weight of the whole basket increases.
Notice that both the apple and the basket have a weight property. If I had used apply, it would not be possible* to access the basket’s weight. Since apply takes an extension function, thiswould refer to the apple and not the basket. With also this is possible.

takeIf and takeUnless

public inline fun <T> T.takeIf(predicate: (T) -> Boolean): T? = if (predicate(this)) this else null

public inline fun <T> T.takeUnless(predicate: (T) -> Boolean): T? = if (!predicate(this)) this else null

The last two functions I want to mention have also been added to Kotlin with 1.1.

takeIf does exactly that - it takes the receiver if it satisfies a condition.
takeUnless inverts the condition and takes this only if the condition is not satisfied.

For example:

val redApple = apple.takeIf { it.color == RED }
val otherApple = apple.takeUnless { it.color == RED }

Those two methods are the functional equivalent to the filter function to collections, but they operate on a single variable.

Summary

All extension functions can be used to avoid temporary variables/Re-scope a variable (or the result of a function). Many of the extension functions in Standard.kt can be used interchangeable. Pick the ones that match your style and that offer the best readability in your context. The table below shows a common usage for each of the presented functions.

FunctionArgumentReturnsUsage examples
runthisany resultmake an object a builder for another object
applythisthismake an object its own builder
letitany resultnull checks
alsoitthisYou need apply but don’t want to shadow this
takeIfitit or nullshorthand for if(predicate(it)) it else null
takeUnlessitit or nullshorthand for if(!predicate(it)) it else null

* You still can by using this@FruitBasket, but do you want to do this?

深度学习是机器学习的一个子领域,它基于人工神经网络的研究,特别是利用多层次的神经网络来进行学习和模式识别。深度学习模型能够学习数据的高层次特征,这些特征对于图像和语音识别、自然语言处理、医学图像分析等应用至关重要。以下是深度学习的一些关键概念和组成部分: 1. **神经网络(Neural Networks)**:深度学习的基础是人工神经网络,它是由多个层组成的网络结构,包括输入层、隐藏层和输出层。每个层由多个神经元组成,神经元之间通过权重连接。 2. **前馈神经网络(Feedforward Neural Networks)**:这是最常见的神经网络类型,信息从输入层流向隐藏层,最终到达输出层。 3. **卷积神经网络(Convolutional Neural Networks, CNNs)**:这种网络特别适合处理具有网格结构的数据,如图像。它们使用卷积层来提取图像的特征。 4. **循环神经网络(Recurrent Neural Networks, RNNs)**:这种网络能够处理序列数据,如时间序列或自然语言,因为它们具有记忆功能,能够捕捉数据中的时间依赖性。 5. **长短期记忆网络(Long Short-Term Memory, LSTM)**:LSTM 是一种特殊的 RNN,它能够学习长期依赖关系,非常适合复杂的序列预测任务。 6. **生成对抗网络(Generative Adversarial Networks, GANs)**:由两个网络组成,一个生成器和一个判别器,它们相互竞争,生成器生成数据,判别器评估数据的真实性。 7. **深度学习框架**:如 TensorFlow、Keras、PyTorch 等,这些框架提供了构建、训练和部署深度学习模型的工具和库。 8. **激活函数(Activation Functions)**:如 ReLU、Sigmoid、Tanh 等,它们在神经网络中用于添加非线性,使得网络能够学习复杂的函数。 9. **损失函数(Loss Functions)**:用于评估模型的预测与真实值之间的差异,常见的损失函数包括均方误差(MSE)、交叉熵(Cross-Entropy)等。 10. **优化算法(Optimization Algorithms)**:如梯度下降(Gradient Descent)、随机梯度下降(SGD)、Adam 等,用于更新网络权重,以最小化损失函数。 11. **正则化(Regularization)**:技术如 Dropout、L1/L2 正则化等,用于防止模型过拟合。 12. **迁移学习(Transfer Learning)**:利用在一个任务上训练好的模型来提高另一个相关任务的性能。 深度学习在许多领域都取得了显著的成就,但它也面临着一些挑战,如对大量数据的依赖、模型的解释性差、计算资源消耗大等。研究人员正在不断探索新的方法来解决这些问题。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值