利用hadoop命令rcc生成Record 一种简单的方式实现自定义的writable对象

 

hadoop 的bin目录下面一个rcc命令，网上介绍很少。看了一下源码。发现可以用来生成 java和c++的hadoop Record 类代码，比起自己直接实现writable 接口，简单很多而且不容易出错，不妨试一下。具体方法如下：

>rcc

Usage: rcc --language [java|c++] ddl-files

 

1 新建 ddl文件

>vim stud.ddl

module com.xschao.bean {

class Stud{

ustring name;

long id;

int age;

ustring time;

}

}

其他的类型补充，具体是啥自行脑补：

"\"byte\"",

"\"boolean\"",

"\"int\"",

"\"long\"",

"\"float\"",

"\"double\"",

"\"ustring\"",

"\"buffer\"",

"\"vector\"",

"\"map\"",

2 执行编译命令

> rcc --language java stud.ddl

查看生成的文件：

>tree

.

|-- com

| `-- xschao

| `-- bean

| `-- Stud.java

|-- command_example.sh

`-- stud.ddl

3 directories, 3 files

Stud.java即为生成的java代码

// File generated by hadoop record compiler. Do not edit.

package com.xschao.bean;

public class Stud extends org.apache.hadoop.record.Record {

private static final org.apache.hadoop.record.meta.RecordTypeInfo _rio_recTypeInfo;

private static org.apache.hadoop.record.meta.RecordTypeInfo _rio_rtiFilter;

private static int[] _rio_rtiFilterFields;

static {

_rio_recTypeInfo = new org.apache.hadoop.record.meta.RecordTypeInfo("Stud");

_rio_recTypeInfo.addField("name", org.apache.hadoop.record.meta.TypeID.StringTypeID);

_rio_recTypeInfo.addField("id", org.apache.hadoop.record.meta.TypeID.LongTypeID);

_rio_recTypeInfo.addField("age", org.apache.hadoop.record.meta.TypeID.IntTypeID);

_rio_recTypeInfo.addField("time", org.apache.hadoop.record.meta.TypeID.StringTypeID);

}

private String name;

private long id;

private int age;

private String time;

public Stud() { }

public Stud(

final String name,

final long id,

final int age,

final String time) {

this.name = name;

this.id = id;

this.age = age;

this.time = time;

}

。。。。。。。。。。。。。。。。。。。。。

ok

格格巫--蓝精灵-