NIO review of java

Following the java IO, review about NIO! !!

• IO : https://qiita.com/liguofeng29/items/e08dc21b16c0057f601e

What is NIO

Simply put, it's an improvement over existing IO.

Bad points of existing IO

1. https://docs.oracle.com/javase/jp/8/docs/api/java/io/InputStream.html#read-- "This __method will be blocked __ until the input data can be read, the end of the stream is detected, or an exception is raised."

2. Inefficient Since the processing unit is byte, char, it is not efficient. Using hige level stream (such as PrintStream) does not change the underlying processing unit.

NIO's core API

• Channel
• It's like a pipe that connects a data source and an app.
• Buffer
• Something like a container for Channle to IO.
• The essence is an array.

Buffer Buffer is an abstract class, and there are subclasses corresponding to basic types other than boolean.

Important elements in Buffer

The above four factors satisfy the following 0 <= mark <= position <= limit <= capacity

Regular subclass of Buffer

• ByteBuffer
  • MappedByteBuffer
• CharBuffer
• ShortBuffer
• IntBuffer
• LongBuffer
• FolatBuffer
• DoubleBuffer

1. ByteBuffer is set / get
2. ByteBuffer is put / get

Buffer's regular method

CharBuffer sample

• CharBuffer toString () is return toString (position (), limit ());

Output method

private static void showProperty(CharBuffer buffer) {
    System.out.println("capacity : " + buffer.capacity());
    System.out.println("limit : " + buffer.limit());
    System.out.println("position : " + buffer.position());
    System.out.println("toString() : " + buffer.toString());
}

Generate

CharBuffer buffer = CharBuffer.allocate(8);
System.out.println("-----After generation-----");
showProperty(buffer);

output ----- After generation ----- capacity : 8 limit : 8 position : 0 toString() :

![Buffer-allocate.png](https://qiita-image-store.s3.amazonaws.com/0/104977/2497625d-f231-645a-d64d-19076d0ed910.png)
1.The end of buffer is 8
2.IO start position(position)is 0
3. toString()Is the position~Since it is a limit, 8 nulls


#### **`put()`**
```java

buffer.put('a');
buffer.put('b');
buffer.put('c');
 System.out.println ("----- After put -----");
showProperty(buffer);

 ----- After put -----
capacity : 8
limit : 8
position : 3
toString() :      

1. put()I did it three times, so the posiont+3
2. toString()Is the position~Since it is a limit, 5 nulls

flip()

buffer.flip();
 System.out.println ("----- after flip -----");
showProperty(buffer);

 ----- After flip -----
capacity : 8
limit : 3
position : 0
toString() : abc

1.First INDEX that cannot be IO(limit)Is 3 2.The first INDEX that can be IO(position)Start position is 0 3. toString()Is the position~Because it is a limitabc

get()

buffer.get();
buffer.get();
System.out.println("-----get-----");
showProperty(buffer);

 ----- After get -----
capacity : 8
limit : 3
position : 2
toString() : c

1. get()Since it is twice, the position is+2、cPoint to
2. 3. toString()Is the position~Because it is a limitc

clear()

buffer.clear();
 System.out.println ("----- after clear -----");
showProperty(buffer);
System.out.println("get(2) : " + buffer.get(2));

 ----- After clear -----
capacity : 8
limit : 8
position : 0
toString() : abc     
get(2) : c

1.Since it returns to the initial position, position=0, limit,capacity=8 2.Not shown here, but mark=-1 3. clear()Just return the position__Because it does not clear the buffer__、 clear()Get even after(2)socが取得soきる

###bonus ByterBuffer has allocateDirect()There is a method to get an instance with. The IO efficiency is high, but the production cost is high, so it is not suitable for disposable use.

Channel

Direct contact(HDD, net, etc.)It is a pipe that connects to, and all IO__Through Buffer__To do.

###Partial implementation of Channel

• FileChannel
• Pipe.SinkChannel, Pipe.SourceChannle
• ServerSocketChannel
• SelectableChannel
• DatagramChannel

FileChannel

map(), write()

channel-in.txt(UTF-8)

 ab12
 Ai

ab12 = 4byte Ai= 9byte \r\n = 2byte

map(),write()

try (
    FileChannel channelIn = new FileInputStream(f).getChannel();
    FileChannel channelOut = new FileOutputStream("./NIO/channel-out.txt").getChannel()) {

    // map()
    MappedByteBuffer mappedByteBuffer = channelIn.map(
        FileChannel.MapMode.READ_ONLY, 0, f.length()
    );
    System.out.println("----channel#map----");
    System.out.println("buffer capacity : " + mappedByteBuffer.capacity());
    System.out.println("buffer limit : " + mappedByteBuffer.limit());
    System.out.println("buffer position : " + mappedByteBuffer.position());
    System.out.println("channel position : " + channelIn.position());

    // write
    channelOut.write(mappedByteBuffer);
    System.out.println("----channel#write----");
    System.out.println("buffer capacity : " + mappedByteBuffer.capacity());
    System.out.println("buffer limit : " + mappedByteBuffer.limit());
    System.out.println("buffer position : " + mappedByteBuffer.position());
    System.out.println("channel position : " + channelOut.position());
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

----channel#map----
buffer capacity : 15
buffer limit : 15
buffer position : 0
channel position : 0
----channel#write----
buffer capacity : 15
buffer limit : 15
buffer position : 15
channel position : 15

*map processing * getChannel() ->FileChannel instantiation * 0, f.length() -> 0~Get 15 bytes * buffer position : 0 -> flip()Done * channel position -?Do not keep the read position in the file

• write
  • out.write -> position ~Write limit
  • buffer position : 15 ->The position was moved by the amount written out
  • channel position : 15 ->Next time file index= 15(16th byte)Write from

read()

File f = new File("./NIO/channel-in.txt");
try (
    FileChannel inChannel = new FileInputStream(f).getChannel()) {
    
    ByteBuffer byteBuffer = ByteBuffer.allocate(8);
    int hasRead = 0;

 System.out.println ("---- channel # read 1st ----");
    hasRead = inChannel.read(byteBuffer);
    System.out.println("buffer capacity : " + byteBuffer.capacity());
    System.out.println("buffer limit : " + byteBuffer.limit());
    System.out.println("buffer position : " + byteBuffer.position());
    System.out.println("channel position : " + inChannel.position());
    System.out.println("hasRead : " + hasRead);

    byteBuffer.clear();

 System.out.println ("---- channel # read 2nd ----");
    hasRead = inChannel.read(byteBuffer);
    System.out.println("buffer capacity : " + byteBuffer.capacity());
    System.out.println("buffer limit : " + byteBuffer.limit());
    System.out.println("buffer position : " + byteBuffer.position());
    System.out.println("channel position : " + inChannel.position());
    System.out.println("hasRead : " + hasRead);
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

 ---- channel # read 1st time ----
buffer capacity : 8
buffer limit : 8
buffer position : 8
channel position : 8
hasRead : 8
 ---- channel # read 2nd time ----
buffer capacity : 8
buffer limit : 8
buffer position : 7
channel position : 15
hasRead : 7

*read 1st time * buffer position : 8 ->8 bytes read * channel position : 8 ->Next is file index= 8(9th byte)Read from *read 2nd time * buffer position : 7 ->7 bytes read * channel position : 15 ->Next is file index= 15(9th byte)Read from

• flip()Otherwise get()I can't use it

Charset

Java uses Unicode by default, but garbled characters may occur when reading other character codes. Charset is provided to convert between byte and char.

###Get charset supported by java

Charset.availableCharsets()

 // Charset list
Charset.availableCharsets().entrySet().forEach(System.out::println);

// defalut
System.out.println("default charset : " + Charset.defaultCharset());　// UTF-8

part

 ・ ・ ・ Abbreviation
Shift_JIS=Shift_JIS
UTF-16=UTF-16
UTF-32=UTF-32
UTF-8=UTF-8
 ・ ・ ・ Abbreviation

ByteBuffer to CharBuffer

Read SJIS file

File f = new File("./NIO/file-sjis-in.txt");
try (
    FileChannel inChannel = new FileInputStream(f).getChannel()) {

    // FileChannel to ByteBuffer
    ByteBuffer byteBuffer = ByteBuffer.allocate(6);
    inChannel.read(byteBuffer);
    byteBuffer.flip();

    // Shift_JIS
    Charset sjis = Charset.forName("Shift_JIS");

    // decode buff with SJIS
    CharBuffer charBuffer = sjis.decode(byteBuffer);
    System.out.println("str : " + charBuffer.toString());
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

String to ByteBuffer

String(unicode)Is written in SJIS

try (
    FileChannel outChannel = new FileOutputStream("./NIO/file-sjis-out.txt").getChannel()) {

    // unicode
 String str = "123 Ai" + System.lineSeparator () + "SHIFT-JIS";
    // Shift_JIS
    Charset sjis = Charset.forName("Shift_JIS");
    // encode buff with SJIS
    ByteBuffer byteBuffer = sjis.encode(str);
    // write to file
    outChannel.write(byteBuffer);
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

Or

It's also in String! !!

new String(byte[] xx, "Shift_JIS");

###One unclear point utf-8(1byte ~ 6byte)When the number of bytes differs depending on the character, such as What should I do with the Buffer size? ?? ??

UTF-8 files

 ab ai

ab = 2byte Ai= 3byte + 3byte

ByteBuffer byteBuffer = ByteBuffer.allocate(6)If set to ab ah + 1st byteGoes into the buffer and decodesab ah�become.

What should I do in this case? ?? ??

1. ~~Use the common multiple of the character set.~~ ~~「1,2,3,4,5,Since it is "6", use a multiple of 60 bytes as the Buffer size.~~

2.Determine the garbled characters, put them at the beginning of the Buffer, adjust the position, and read. ⇒ map()Does not keep block reading & file position! ~~Do I have to specify a size that does not garble? ??~~

The solution is below.

CodingErrorAction.IGNORE

IGNORE sample

File f = new File("./a/file-utf8-in.txt");
try (
    FileChannel inChannel = new FileInputStream(f).getChannel()) {
    ByteBuffer byteBuffer = ByteBuffer.allocate(6);
    while (inChannel.read(byteBuffer) > -1) {
 // IO preparation
        byteBuffer.flip();

        Charset utf8 = Charset.forName("UTF-8");
        CharsetDecoder decoder = utf8.newDecoder();

        CharBuffer charBuffer2 = decoder
            .onMalformedInput(CodingErrorAction.IGNORE)
            .decode(byteBuffer);
        System.out.print(charBuffer2.toString());
        byteBuffer.clear();
    }
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

output

 ab ah

IWas ignored

CodingErrorAction.IGNORE

IGNORE

File f = new File("./a/file-utf8-in.txt");
try (
    FileChannel inChannel = new FileInputStream(f).getChannel()) {
    ByteBuffer byteBuffer = ByteBuffer.allocate(6);
    while (inChannel.read(byteBuffer) > -1) {
 // IO preparation
        byteBuffer.flip();
        Charset utf8 = Charset.forName("UTF-8");
        CharsetDecoder decoder = utf8.newDecoder();
        CharBuffer charBuffer2 = decoder
 .onMalformedInput (CodingErrorAction.REPLACE) .replaceWith ("O")
            .decode(byteBuffer);
        System.out.print(charBuffer2.toString());
        byteBuffer.clear();
    }
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

output

 ab Aoo

I1st byte ⇒ I2nd byte ⇒ I3rd byte ⇒

CodingErrorAction.REPORT

IGNORE,REPLACE destroys the original data !!

REPORT

File f = new File("./a/file-utf8-in.txt");
try (
    FileChannel inChannel = new FileInputStream(f).getChannel()) {
    ByteBuffer byteBuffer = ByteBuffer.allocate(6);
    while (inChannel.read(byteBuffer) > -1) {
        byteBuffer.flip();
        Charset utf8 = Charset.forName("UTF-8");
        CharsetDecoder decoder = utf8.newDecoder();
        try {
            CharBuffer charBuffer = decoder
                .onMalformedInput(CodingErrorAction.REPORT)
                .decode(byteBuffer);

 // Output
            System.out.print(charBuffer.toString());
            byteBuffer.clear();
        } catch (MalformedInputException ex) {
 // Error occurrence position and end position
            int errorIndexStart = byteBuffer.position();
            int errorIndexEnd = byteBuffer.limit();

 // Output to normal position
            byteBuffer.flip();
            CharBuffer charBuffer = decoder
                .decode(byteBuffer);
            System.out.print(charBuffer.toString());

 // Move error position start-end to buffer top
            byteBuffer.clear();
            for (int i = errorIndexStart; i < errorIndexEnd; i++) {
                byteBuffer.put(byteBuffer.get(i));
            }
        }
    }
} catch (IOException ex) {
    System.out.println(ex.getMessage());
}

 ab ai

1.When an error occurs, first output to the normal position. 2.Move from the error occurrence position to the end of the buffer 3.The next read is from after the error length, not from 0 4. utf-8 is 1 to 6 bytes, so allocate at least(6)Do you need?

Is this really the best? ??

FileLock

Represents a lock on a file

lock()And tryLock()The difference of

lock()Blocks threads until the lock is acquired. tryLock returns null if the lock cannot be acquired.

lock()

lock sample

try (FileChannel fileChannel = new FileOutputStream("file-lock-out.txt").getChannel();) {
    FileLock lock = fileChannel.lock();
 System.out.println ("lock was obtained");
 System.out.println ("Keep lock for 60 seconds");
    Thread.sleep(60 * 1000);
    lock.release();
} catch (IOException | InterruptedException ex) {
    System.out.println("1" + ex.getMessage());
}

When the above is executed twice with different jvm

output of jvm1

 I was able to get the lock
 Keep lock for 60 seconds

There is no output because the output of jvm2 waits.

tryLock()

tryLock()sample

try (FileChannel fileChannel = new FileOutputStream("file-lock-out.txt").getChannel();) {
    FileLock lock = fileChannel.tryLock();
    if (lock != null) {
 System.out.println ("tryLock acquired");
        Thread.sleep(60 * 1000);
        lock.release();
    } else {
        System.out.println("lock is null");
    }
} catch (IOException | InterruptedException ex) {
    System.out.println(ex.getMessage());
}

jvm1 output

 I was able to get tryLock

jvm2 output

lock is null

###Other 1.Since the lock is owned by the jvm, the same file cannot be locked with the same jvm. 2.On some platforms, the lock is released when the FileChannel is closed. So it's not good to open multiple File Channels for locked files. 3.On some platforms, file locking is recommended, not compulsory. ⇒ You can read and write without locking.

#Next is NIO.2 https://qiita.com/liguofeng29/items/3d0ba350a1547630727b