hadoop自定义类型

hadoop为用户提供了自定义类型的接口，对于key，需要重写实现WritableComparable，而对于value，只需重写实现Writable即可。但是在作为reduce的输出的时，输出格式为TextOutputFormat时，不能正常显示。还需要自定义OutputFormat。

常用的几个函数：

private double []values = null;
	
	public VectorWritable(){}
	
	public VectorWritable(double []a){
		this.values =a;
	}
	public double get(int index){
		return this.values[index];
	}
	public int getLen(){
		return this.values.length;
	}
	public void set(double []a){
		this.values = a;
	}

 需要重写的几个函数：

@Override
	public void readFields(DataInput arg0) throws IOException {
		// TODO Auto-generated method stub
		this.values = new double[arg0.readInt()];
		for(int i=0;i<this.values.length;i++){
			double tmp = arg0.readDouble();
			this.values[i] = tmp;
		}
	}

@Override
	public void write(DataOutput arg0) throws IOException {
		// TODO Auto-generated method stub
		arg0.writeInt(this.values.length);
		for(int i=0;i<this.values.length;i++){
			arg0.writeDouble(this.values[i]);
		}
	}

 其实就是序列化和反序列化的需要重写

Map：

public static class TestMap 
	extends Mapper<Object,Text,Text,VectorWritable>{
		private static Text outputkey = new Text("1");
		private static VectorWritable Vector = new VectorWritable();
		public void map(Object key, Text rawLine, Context context)
		throws IOException, InterruptedException{
			String []parts = rawLine.toString().trim().split(",");
			int len = parts.length;
			double []tmp = new double[len];
			for(int i=0;i<len;i++){
				 tmp[i] = Double.valueOf(parts[i]);	
			}
			Vector.set(tmp);
			context.write(outputkey, Vector);
		}
	}

 

 Reduce：

public static class TestReduce
	extends Reducer<Object,VectorWritable,Text,DoubleWritable>{
		private static Text outputkey = new Text("1");
		private static DoubleWritable output = new DoubleWritable();
		public void reduce(Object key,Iterable<VectorWritable>value,Context context)
		throws IOException, InterruptedException{
			for(VectorWritable vm : value){
				int len = vm.getLen();
				for(int i=0;i<len;i++){
					double x = vm.get(i);
					//System.out.println(x);
					output.set(x);
					context.write(outputkey, output);
				}
			}
		}
	}