foreach Java“for each”循环是如何工作的？

Question

考虑：

List<String> someList = new ArrayList<String>();
// add "monkey", "donkey", "skeleton key" to someList

for (String item : someList) {
    System.out.println(item);
}

如果不对每个语法使用，那么等效的for循环会是什么样子

Answer 1

Java中的for-each循环使用底层迭代器机制。因此，它与以下内容相同：

Iterator<String> iterator = someList.iterator();

while (iterator.hasNext()) {
  String item = iterator.next();
  System.out.println(item);
}

Answer 2

这里有一个答案，它并不假定您了解Java迭代器。它不太精确，但对教育很有用

编程时，我们通常编写如下代码：

char[] grades = ....
for(int i = 0; i < grades.length; i++) {   // for i goes from 0 to grades.length
    System.out.print(grades[i]);           // Print grades[i]
}

foreach语法允许以更自然、更少语法噪音的方式编写此公共模式

for(char grade : grades) {   // foreach grade in grades
    System.out.print(grade); // print that grade
}

此外，此语法对于不支持数组索引但实现Java Iterable接口的列表或集合等对象有效

Answer 3

在Java 5中添加的for-each loop（也称为“增强的for循环”），相当于使用^{}——它是相同事物的语法糖。因此，在逐项按顺序读取每个元素时，应始终在迭代器上选择一个for-each，因为它更方便、更简洁

对于每个

for (int i : intList) {
   System.out.println("An element in the list: " + i);
}

迭代器

Iterator<Integer> intItr = intList.iterator();
while (intItr.hasNext()) {
   System.out.println("An element in the list: " + intItr.next());
}

在某些情况下，您必须直接使用Iterator。例如，在为每个can（will？）使用时尝试删除一个元素结果是ConcurrentModificationException

For each与For循环：基本区别

for循环和for each之间唯一实际的区别是，对于可索引对象，您无权访问索引。需要基本for循环时的示例：

for (int i = 0; i < array.length; i++) {
   if(i < 5) {
      // Do something special
   }  else {
      // Do other stuff
   }
}

尽管您可以手动创建一个单独的索引int变量

int idx = -1;
for (int i : intArray) {
   idx++;
   ...
}

。。。不建议这样做，因为variable-scope并不理想，基本for循环只是这个用例的标准和预期格式

For each与For循环：性能

访问集合时，for each比基本for循环的数组访问权限大significantly faster。然而，当访问数组时——至少在使用原语和包装器数组时——通过索引进行访问的速度要快得多

对基本int数组的迭代器和索引访问之间的差异进行计时

在访问int或Integer数组时，索引比迭代器快23-40。这是本文底部测试类的输出，它对100个元素的原始int数组中的数字求和（a是迭代器，B是索引）：

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 358,597,622 nanoseconds
Test B: 269,167,681 nanoseconds
B faster by 89,429,941 nanoseconds (24.438799231635727% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 377,461,823 nanoseconds
Test B: 278,694,271 nanoseconds
B faster by 98,767,552 nanoseconds (25.666236154695838% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 288,953,495 nanoseconds
Test B: 207,050,523 nanoseconds
B faster by 81,902,972 nanoseconds (27.844689860906513% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,373,765 nanoseconds
Test B: 283,813,875 nanoseconds
B faster by 91,559,890 nanoseconds (23.891659337194227% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 375,790,818 nanoseconds
Test B: 220,770,915 nanoseconds
B faster by 155,019,903 nanoseconds (40.75164734599769% faster)

[C:\java_code\]java TimeIteratorVsIndexIntArray 1000000
Test A: 326,373,762 nanoseconds
Test B: 202,555,566 nanoseconds
B faster by 123,818,196 nanoseconds (37.437545972215744% faster)

我还为一个Integer数组运行了这个程序，索引仍然是明显的赢家，但速度仅在18%到25%之间

对于集合，迭代器比索引快

然而，对于List中的Integers，迭代器显然是赢家。只需将测试类中的int数组更改为：

List<Integer> intList = Arrays.asList(new Integer[] {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100});

并对测试函数进行必要的更改（int[]到List<Integer>，length到size()，等等）：

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,429,929,976 nanoseconds
Test B: 5,262,782,488 nanoseconds
A faster by 1,832,852,512 nanoseconds (34.326681820485675% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,907,391,427 nanoseconds
Test B: 3,957,718,459 nanoseconds
A faster by 1,050,327,032 nanoseconds (26.038700083921256% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,566,004,688 nanoseconds
Test B: 4,221,746,521 nanoseconds
A faster by 1,655,741,833 nanoseconds (38.71935684115413% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 2,770,945,276 nanoseconds
Test B: 3,829,077,158 nanoseconds
A faster by 1,058,131,882 nanoseconds (27.134122749113843% faster)

[C:\java_code\]java TimeIteratorVsIndexIntegerList 1000000
Test A: 3,467,474,055 nanoseconds
Test B: 5,183,149,104 nanoseconds
A faster by 1,715,675,049 nanoseconds (32.60101667104192% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,439,983,933 nanoseconds
Test B: 3,509,530,312 nanoseconds
A faster by 69,546,379 nanoseconds (1.4816434912159906% faster)

[C:\java_code\]java TimeIteratorVsIndexIntList 1000000
Test A: 3,451,101,466 nanoseconds
Test B: 5,057,979,210 nanoseconds
A faster by 1,606,877,744 nanoseconds (31.269164666060377% faster)

在一个测试中，它们几乎是等价的，但对于集合，迭代器获胜

*本文基于我对堆栈溢出的两个回答：

• Uses and syntax for for-each loop in Java

• Should I use an Iterator or a forloop to iterate?

更多信息：Which is more efficient, a for-each loop, or an iterator?

完整的测试等级

我在阅读了堆栈溢出上的this question之后，创建了这个比较类来做任何两件事所需的时间：

import  java.text.NumberFormat;
import  java.util.Locale;

/**
   <P>{@code java TimeIteratorVsIndexIntArray 1000000}</P>

   @see  <CODE><A HREF="https://stackoverflow.com/questions/180158/how-do-i-time-a-methods-execution-in-java">https://stackoverflow.com/questions/180158/how-do-i-time-a-methods-execution-in-java</A></CODE>
 **/
public class TimeIteratorVsIndexIntArray {

    public static final NumberFormat nf = NumberFormat.getNumberInstance(Locale.US);

    public static final void main(String[] tryCount_inParamIdx0) {
        int testCount;

        // Get try-count from a command-line parameter
        try {
           testCount = Integer.parseInt(tryCount_inParamIdx0[0]);
        }
        catch(ArrayIndexOutOfBoundsException | NumberFormatException x) {
           throw  new IllegalArgumentException("Missing or invalid command line parameter: The number of testCount for each test. " + x);
        }

        //Test proper...START
        int[] intArray = new int[] {1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100};

        long lStart = System.nanoTime();
        for(int i = 0; i < testCount; i++) {
           testIterator(intArray);
        }

        long lADuration = outputGetNanoDuration("A", lStart);

        lStart = System.nanoTime();
        for(int i = 0; i < testCount; i++) {
           testFor(intArray);
        }

        long lBDuration = outputGetNanoDuration("B", lStart);

        outputGetABTestNanoDifference(lADuration, lBDuration, "A", "B");
    }

    private static final void testIterator(int[] int_array) {
       int total = 0;
       for(int i = 0; i < int_array.length; i++) {
          total += int_array[i];
       }
    }

    private static final void testFor(int[] int_array) {
       int total = 0;
       for(int i : int_array) {
          total += i;
       }
    }
    //Test proper...END

    //Timer testing utilities...START
    public static final long outputGetNanoDuration(String s_testName, long l_nanoStart) {
        long lDuration = System.nanoTime() - l_nanoStart;
        System.out.println("Test " + s_testName + ": " + nf.format(lDuration) + " nanoseconds");
        return  lDuration;
    }

    public static final long outputGetABTestNanoDifference(long l_aDuration, long l_bDuration, String s_aTestName, String s_bTestName) {
        long lDiff = -1;
        double dPct = -1.0;
        String sFaster = null;
        if(l_aDuration > l_bDuration) {
            lDiff = l_aDuration - l_bDuration;
            dPct = 100.00 - (l_bDuration * 100.0 / l_aDuration + 0.5);
            sFaster = "B";
        }
        else {
            lDiff = l_bDuration - l_aDuration;
            dPct = 100.00 - (l_aDuration * 100.0 / l_bDuration + 0.5);
            sFaster = "A";
        }
        System.out.println(sFaster + " faster by " + nf.format(lDiff) + " nanoseconds (" + dPct + "% faster)");
        return  lDiff;
   }

   //Timer testing utilities...END

}

Answer 4

for (Iterator<String> i = someIterable.iterator(); i.hasNext();) {
    String item = i.next();
    System.out.println(item);
}

请注意，如果需要在循环中使用i.remove();，或者以某种方式访问实际的迭代器，则不能使用for ( : )习惯用法，因为实际的迭代器只是推断出来的

正如Denis Bueno所指出的，此代码适用于实现^{} interface的任何对象

此外，如果for (:)习惯用法的右侧是array而不是Iterable对象，则内部代码使用一个int索引计数器并检查array.length。见Java Language Specification

Answer 5

如JLS中所定义，每个循环可以有两种形式：

1. 如果表达式的类型是Iterable的子类型，则翻译如下：

   List<String> someList = new ArrayList<String>();
   someList.add("Apple");
   someList.add("Ball");
   for (String item : someList) {
       System.out.println(item);
   }
   
   // IS TRANSLATED TO:
   
   for(Iterator<String> stringIterator = someList.iterator(); stringIterator.hasNext(); ) {
       String item = stringIterator.next();
       System.out.println(item);
   }
   

2. 如果表达式必须具有数组类型T[]，则：

   String[] someArray = new String[2];
   someArray[0] = "Apple";
   someArray[1] = "Ball";
   
   for(String item2 : someArray) {
       System.out.println(item2);
   }
   
   // IS TRANSLATED TO:
   for (int i = 0; i < someArray.length; i++) {
       String item2 = someArray[i];
       System.out.println(item2);
   }
   

Java8引入了流，这些流在使用下降大小数据集时通常表现得更好。我们可以将其用作：

someList.stream().forEach(System.out::println);
Arrays.stream(someArray).forEach(System.out::println);

Answer 6

每个的构造也适用于数组。e、 g

String[] fruits = new String[] { "Orange", "Apple", "Pear", "Strawberry" };

for (String fruit : fruits) {
    // fruit is an element of the `fruits` array.
}

这基本上相当于

for (int i = 0; i < fruits.length; i++) {
    String fruit = fruits[i];
    // fruit is an element of the `fruits` array.
}

因此，总体总结：
[nsayer]下面是正在发生的事情的较长形式：

for(Iterator<String> i = someList.iterator(); i.hasNext(); ) {
  String item = i.next();
  System.out.println(item);
}

Note that if you need to use i.remove(); in your loop, or access the actual iterator in some way, you cannot use the for( : ) idiom, since the actual Iterator is merely inferred.

[Denis Bueno]

It's implied by nsayer's answer, but it's worth noting that the OP's for(..) syntax will work when "someList" is anything that implements java.lang.Iterable -- it doesn't have to be a list, or some collection from java.util. Even your own types, therefore, can be used with this syntax.