#!/bin/sh

hadoop dfs -put input input

HADOOP_HOME=/usr/lib/hadoop
HADOOP_VERSION=0.20.2-cdh3u1
JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"

OUT=mapper_test_out

$HSTREAMING \
    -input input \
    -output $OUT \
    -mapper "python mapper.py" \
    -file mapper_test.py \
    -reducer org.apache.hadoop.mapred.lib.IdentityReducer \
    -jobconf mapred.reduce.tasks=3

hadoop dfs -get $OUT .
hadoop dfs -rmr $OUT
hadoop dfs -rm input