NIO之ByteBuffer的使用
NIO中有三大组件：Channel、Selector、Buffer，今天就来讲讲Buffer的使用。
Buffer的继承结构：
Buffer为java中的每个数据类型都提供了对应的实现类，但是实际使用中用得最多的还是ByteBuffer。
ByteBuffer的创建
在堆上创建：
log.info("freeMemory:{}MB", Runtime.getRuntime().freeMemory() / _1_MB);
ByteBuffer allocate = ByteBuffer.allocate(20 * _1_MB); // 堆上内存分配
log.info("freeMemory:{}MB", Runtime.getRuntime().freeMemory() / _1_MB);运行结果如下：
2020-11-04 15:49:01,421 INFO [main] (ByteBufferNewDemo.java:20) - freeMemory:230MB
2020-11-04 15:49:01,436 INFO [main] (ByteBufferNewDemo.java:23) - freeMemory:209MB在直接内存上创建：
OperatingSystemMXBean osmxb = (OperatingSystemMXBean) ManagementFactory.getOperatingSystemMXBean();
log.info("freeMemory:{}MB", osmxb.getFreePhysicalMemorySize() / _1_MB);
final ByteBuffer byteBuffer = ByteBuffer.allocateDirect(20 * _1_MB);
log.info("freeMemory:{}MB", osmxb.getFreePhysicalMemorySize() / _1_MB);运行结果如下：
2020-11-04 15:53:38,808 INFO [main] (ByteBufferNewDemo.java:35) - freeMemory:4524MB
2020-11-04 15:53:38,816 INFO [main] (ByteBufferNewDemo.java:37) - freeMemory:4503MBByteBuffer的使用
ByteBuffer有三个重要的属性：
- capacity：ByteBuffer底层其实是一个字节数组，capacity代表数组的长度。
- position：表示下一个要操作（读或写）的元素的位置。
- limit：在写模式下，limit=capacity，在读模式下，limit代表数组中实际元素的个数。
get和put值
ByteBuffer byteBuffer = ByteBuffer.allocate(8);
byteBuffer.put((byte) 'a');
byteBuffer.put((byte) 'b');
byteBuffer.put((byte) 'c');
byteBuffer.put((byte) 'd');
log.info("put a b c d, {}", byteBuffer);
byteBuffer.flip(); // 切换为读模式
log.info("after flip, {}", byteBuffer);
log.info("get byte: {}", (char) byteBuffer.get());
log.info("after get, {}", byteBuffer);运行结果如下：
2020-11-04 16:04:08,130 INFO [main] (ByteBufferNewDemo.java:47) - put a b c d, java.nio.HeapByteBuffer[pos=4 lim=8 cap=8]
2020-11-04 16:04:08,133 INFO [main] (ByteBufferNewDemo.java:49) - after flip, java.nio.HeapByteBuffer[pos=0 lim=4 cap=8]
2020-11-04 16:04:08,133 INFO [main] (ByteBufferNewDemo.java:50) - get byte: a
2020-11-04 16:04:08,133 INFO [main] (ByteBufferNewDemo.java:51) - after get, java.nio.HeapByteBuffer[pos=1 lim=4 cap=8]
一次性get
ByteBuffer byteBuffer = ByteBuffer.allocate(8);
byteBuffer.put((byte) 'a');
byteBuffer.put((byte) 'b');
byteBuffer.put((byte) 'c');
byteBuffer.put((byte) 'd');
byteBuffer.flip(); // 切换为读模式
byte[] bytes = new byte[byteBuffer.remaining()];
log.info("before get, {}", byteBuffer);
byteBuffer.get(bytes, 0, bytes.length);
log.info("bytes:{}", new String(bytes));
log.info("after get, {}", byteBuffer);运行结果如下：
2020-11-04 16:32:36,032 INFO [main] (ByteBufferNewDemo.java:64) - before get, java.nio.HeapByteBuffer[pos=0 lim=4 cap=8]
2020-11-04 16:32:36,034 INFO [main] (ByteBufferNewDemo.java:66) - bytes:abcd
2020-11-04 16:32:36,034 INFO [main] (ByteBufferNewDemo.java:67) - after get, java.nio.HeapByteBuffer[pos=4 lim=4 cap=8]指定位置读和写
使用put和get方法时，指定index进行操作，不会改变position的位置。
ByteBuffer byteBuffer = ByteBuffer.allocate(8);
byteBuffer.put(0, (byte) 'a');
byteBuffer.put(1, (byte) 'b');
byteBuffer.put(2, (byte) 'c');
byteBuffer.put(3, (byte) 'd');
log.info("put a b c d, {}", byteBuffer);
log.info("get byte: {}", (char) byteBuffer.get(0));
log.info("after get, {}", byteBuffer);运行结果如下：
2020-11-04 16:35:23,152 INFO [main] (ByteBufferNewDemo.java:77) - put a b c d, java.nio.HeapByteBuffer[pos=0 lim=8 cap=8]
2020-11-04 16:35:23,154 INFO [main] (ByteBufferNewDemo.java:78) - get byte: a
2020-11-04 16:35:23,154 INFO [main] (ByteBufferNewDemo.java:79) - after get, java.nio.HeapByteBuffer[pos=0 lim=8 cap=8]clear
clear()方法会加将position设置为0，limit设置成capacity，也就是说ByteBuffer被清空了，其实Buffer中的数据并未被清除，只是这些标记告诉我们可以从哪里开始往Buffer里写数据。如果Buffer中有一些未读的数据，调用clear()方法，数据将“被遗忘”，意味着不再有任何标记会告诉你哪些数据被读过，哪些还没有。
ByteBuffer byteBuffer = ByteBuffer.allocate(32);
byteBuffer.put((byte) 'x');
log.info("before clear: {}" + byteBuffer);
byteBuffer.clear();
log.info("after clear: {}" + byteBuffer);运行结果如下：
2020-11-04 16:44:36,956 INFO [main] (ByteBufferNewDemo.java:86) - before clear: {}java.nio.HeapByteBuffer[pos=1 lim=32 cap=32]
2020-11-04 16:44:36,958 INFO [main] (ByteBufferNewDemo.java:88) - after clear: {}java.nio.HeapByteBuffer[pos=0 lim=32 cap=32]compact
compact()方法将所有未读的数据拷贝到Buffer起始处（压缩），然后将position设到最后一个未读元素正后面，limit属性依然像clear()方法一样，设置成capacity。
ByteBuffer byteBuffer = ByteBuffer.allocate(8);
byteBuffer.put((byte) 'a');
byteBuffer.put((byte) 'b');
byteBuffer.put((byte) 'c');
byteBuffer.put((byte) 'd');
byteBuffer.flip(); // 切换为读模式
byteBuffer.get();
log.info("before compact: {}", byteBuffer);
byteBuffer.compact();
log.info("after compact: {}", byteBuffer);
byteBuffer.flip(); // 切换为读模式
log.info("{}",byteBuffer.remaining());运行结果如下：
2020-11-04 16:51:19,010 INFO [main] (ByteBufferNewDemo.java:100) - before compact: java.nio.HeapByteBuffer[pos=1 lim=4 cap=8]
2020-11-04 16:51:19,013 INFO [main] (ByteBufferNewDemo.java:102) - after compact: java.nio.HeapByteBuffer[pos=3 lim=8 cap=8]
2020-11-04 16:51:19,013 INFO [main] (ByteBufferNewDemo.java:104) - 3rewind
rewind()方法会将position设置为0，limit的值不变，类似磁带的倒带功能。
ByteBuffer byteBuffer = ByteBuffer.allocate(16);
byteBuffer.position(10);
byteBuffer.limit(15);
log.info("before rewind: {}", byteBuffer);
byteBuffer.rewind();
log.info("after rewind: {}", byteBuffer);运行结果如下：
2020-11-04 16:56:53,935 INFO [main] (ByteBufferNewDemo.java:112) - before rewind: java.nio.HeapByteBuffer[pos=10 lim=15 cap=16]
2020-11-04 16:56:53,939 INFO [main] (ByteBufferNewDemo.java:114) - after rewind: java.nio.HeapByteBuffer[pos=0 lim=15 cap=16]mark和reset
ByteBuffer中有一个属性用来存放临时的位置下标，调用mark()会将mark设为当前的position的值，以后调用reset()会将position属性设置为mark的值。
ByteBuffer中的几个属性总是满足以下关系：0 <= mark <= position <= limit <= capacity。
ByteBuffer byteBuffer = ByteBuffer.allocate(16);
byteBuffer.position(10);
byteBuffer.mark();
byteBuffer.position(15);
log.info("before reset: {}", byteBuffer);
byteBuffer.reset();
log.info("after reset: {}", byteBuffer);运行结果如下：
2020-11-04 17:00:13,572 INFO [main] (ByteBufferNewDemo.java:123) - before reset: java.nio.HeapByteBuffer[pos=15 lim=16 cap=16]
2020-11-04 17:00:13,577 INFO [main] (ByteBufferNewDemo.java:125) - after reset: java.nio.HeapByteBuffer[pos=10 lim=16 cap=16]
