跳至正文

🎯 協程基礎

📖 協程入門​

要使用協程，首先必須了解協程建構器。讓我們來了解最基本的三個建構器！

💡 runBlocking​

基本用法​

import kotlinx.coroutines.*

fun main() = runBlocking {  // 啟動協程！
    println("Hello")
    delay(1000)  // 等待 1 秒
    println("World!")
}

特點​

fun main() {
    println("Before")

    runBlocking {
        delay(1000)
        println("Inside coroutine")
    }

    println("After")  // 等待協程結束
}
// Before
// (等待 1 秒)
// Inside coroutine
// After

注意：runBlocking 會阻塞當前執行緒！主要只在測試或 main 函式中使用。

🚀 launch​

啟動非同步任務​

fun main() = runBlocking {
    launch {
        delay(1000)
        println("Task 1")
    }

    launch {
        delay(500)
        println("Task 2")
    }

    println("Main")
}
// Main
// (500ms 後) Task 2
// (1000ms 後) Task 1

返回 Job​

fun main() = runBlocking {
    val job = launch {
        repeat(5) { i ->
            println("Working... $i")
            delay(500)
        }
    }

    delay(1300)
    println("取消任務！")
    job.cancel()  // 取消任務
    job.join()    // 等待取消完成
}

管理多個任務​

fun main() = runBlocking {
    val jobs = List(5) { i ->
        launch {
            delay(1000L * i)
            println("任務 $i 完成")
        }
    }

    jobs.forEach { it.join() }  // 等待所有任務
    println("所有任務結束！")
}

⚡ async​

接收結果​

fun main() = runBlocking {
    val deferred = async {
        delay(1000)
        return@async "結果值"
    }

    println("進行中...")
    val result = deferred.await()  // 等待結果
    println("收到的值: $result")
}

launch vs async​

fun main() = runBlocking {
    // launch - 無結果
    launch {
        delay(1000)
        println("launch 完成")
    }

    // async - 返回結果
    val result = async {
        delay(1000)
        "async 完成"
    }

    println(result.await())
}

🎯 實戰範例​

平行處理​

import kotlin.system.measureTimeMillis

fun main() = runBlocking {
    val time = measureTimeMillis {
        val one = async { fetchData1() }
        val two = async { fetchData2() }

        println("結果: ${one.await()}, ${two.await()}")
    }

    println("所需時間: ${time}ms")  // ~1000ms (平行執行)
}

suspend fun fetchData1(): String {
    delay(1000)
    return "資料1"
}

suspend fun fetchData2(): String {
    delay(1000)
    return "資料2"
}

循序 vs 平行​

fun main() = runBlocking {
    // ❌ 循序執行 (慢)
    val time1 = measureTimeMillis {
        val one = fetchData1()
        val two = fetchData2()
        println("$one, $two")
    }
    println("循序: ${time1}ms")  // ~2000ms

    // ✅ 平行執行 (快)
    val time2 = measureTimeMillis {
        val one = async { fetchData1() }
        val two = async { fetchData2() }
        println("${one.await()}, ${two.await()}")
    }
    println("平行: ${time2}ms")  // ~1000ms
}

例外處理​

fun main() = runBlocking {
    val deferred = async {
        delay(1000)
        throw Exception("發生錯誤！")
    }

    try {
        deferred.await()
    } catch (e: Exception) {
        println("捕獲的錯誤: ${e.message}")
    }
}

🔥 實用模式​

逾時​

fun main() = runBlocking {
    try {
        withTimeout(1300) {
            repeat(3) { i ->
                println("任務 $i")
                delay(500)
            }
        }
    } catch (e: TimeoutCancellationException) {
        println("逾時！")
    }
}

多個 API 呼叫​

data class UserData(
    val profile: String,
    val posts: List<String>,
    val friends: List<String>
)

suspend fun loadUserData(userId: String): UserData = coroutineScope {
    val profile = async { fetchProfile(userId) }
    val posts = async { fetchPosts(userId) }
    val friends = async { fetchFriends(userId) }

    UserData(
        profile.await(),
        posts.await(),
        friends.await()
    )
}

suspend fun fetchProfile(id: String): String {
    delay(500)
    return "Profile($id)"
}

suspend fun fetchPosts(id: String): List<String> {
    delay(800)
    return listOf("Post1", "Post2")
}

suspend fun fetchFriends(id: String): List<String> {
    delay(600)
    return listOf("Friend1", "Friend2")
}

重試邏輯​

suspend fun <T> retryIO(
    times: Int = 3,
    delay: Long = 1000,
    block: suspend () -> T
): T {
    repeat(times - 1) { attempt ->
        try {
            return block()
        } catch (e: Exception) {
            println("嘗試 ${attempt + 1} 失敗")
            delay(delay)
        }
    }
    return block()  // 最後嘗試
}

fun main() = runBlocking {
    val result = retryIO(times = 3) {
        fetchDataFromAPI()
    }
    println(result)
}

var attempt = 0
suspend fun fetchDataFromAPI(): String {
    attempt++
    if (attempt < 3) {
        throw Exception("網路錯誤")
    }
    return "成功！"
}

🛠️ coroutineScope​

結構化並行​

suspend fun doWork() = coroutineScope {
    launch {
        delay(1000)
        println("任務 1")
    }

    launch {
        delay(2000)
        println("任務 2")
    }

    println("所有任務啟動")
    // 等待所有子任務結束
}

fun main() = runBlocking {
    doWork()
    println("完成！")
}

例外傳播​

suspend fun riskyWork() = coroutineScope {
    launch {
        delay(500)
        throw Exception("錯誤！")
    }

    launch {
        delay(1000)
        println("這段程式碼不會執行")
    }
}

fun main() = runBlocking {
    try {
        riskyWork()
    } catch (e: Exception) {
        println("捕獲錯誤: ${e.message}")
    }
}

🤔 常見問題​

Q1. launch 和 async 的差異是？​

A: 是否返回結果！

fun main() = runBlocking {
    // launch - 不需要結果
    launch {
        println("簡單任務")
    }

    // async - 需要結果
    val result = async {
        "要返回的值"
    }
    println(result.await())
}

Q2. join() 和 await() 的差異是？​

A: join 是等待完成，await 是等待結果！

fun main() = runBlocking {
    val job = launch {
        delay(1000)
    }
    job.join()  // 只等待完成

    val deferred = async {
        delay(1000)
        "結果"
    }
    val result = deferred.await()  // 等待結果
}

Q3. 協程會自動取消嗎？​

A: 父協程取消時，子協程也會取消！

fun main() = runBlocking {
    val parent = launch {
        val child = launch {
            repeat(10) {
                println("子任務 $it")
                delay(500)
            }
        }

        delay(1300)
        println("父任務完成")
    }

    delay(2000)
    parent.cancel()  // 子任務也會一起取消
}

🎬 結語​

您已經掌握了協程的基本建構器！

核心總結:
✅ runBlocking - 用於測試/main 函式
✅ launch - 不需要結果的任務
✅ async - 接收返回結果
✅ coroutineScope - 結構化並行
✅ 透過平行處理提升效能

下一步：在 suspend 函式中深入了解暫停函式！