本文介绍了如何使用Hadoop的MultipleInputs功能处理不同格式的多个文件，并通过具体案例展示了配置和实现过程，包括自定义数据类型及Mapper和Reducer的设计。
转载自：http://blog.youkuaiyun.com/fansy1990/article/details/26267637
hadoop多文件格式输入，一般可以使用MultipleInputs类指定不同的输入文件路径以及输入文件格式。
比如现在有如下的需求：
现有两份数据：
phone：
在CODE上查看代码片派生到我的代码片
123,good number
124,common number
125,bad number user：
在CODE上查看代码片派生到我的代码片
zhangsan,123
lisi,124
wangwu,125 现在需要把user和phone按照phone number连接起来，得到下面的结果：
在CODE上查看代码片派生到我的代码片
zhangsan,123,good number
lisi,124,common number
wangwu,125,bad number 那么就可以使用MultipleInputs来操作，这里把user和phone上传到hdfs目录中，分别是/multiple/user/user , /multiple/phone/phone。
设计的MultipleDriver如下：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.MultipleInputs;
import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
//import org.slf4j.Logger;
//import org.slf4j.LoggerFactory;
/**
* input1(/multiple/user/user):
* username,user_phone
*
* input2(/multiple/phone/phone):
* user_phone,description
*
* output: username,user_phone,description
*
* @author fansy
*
*/
public class MultipleDriver extends Configured implements Tool{
// private Logger log = LoggerFactory.getLogger(MultipleDriver.class);
private String input1=null;
private String input2=null;
private String output=null;
private String delimiter=null;
public static void main(String[] args) throws Exception {
Configuration conf=new Configuration();
// conf.set("fs.defaultFS", "hdfs://node33:8020");
// conf.set("mapreduce.framework.name", "yarn");
// conf.set("yarn.resourcemanager.address", "node33:8032");
ToolRunner.run(conf, new MultipleDriver(), args);
}
@Override
public int run(String[] arg0) throws Exception {
configureArgs(arg0);
checkArgs();
Configuration conf= getConf();
conf.set("delimiter", delimiter);
@SuppressWarnings("deprecation")
Job job = new Job(conf, "merge user and phone information ");
job.setJarByClass(MultipleDriver.class);
job.setReducerClass(MultipleReducer.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(FlagStringDataType.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(NullWritable.class);
job.setNumReduceTasks(1);
MultipleInputs.addInputPath(job, new Path(input1), TextInputFormat.class, Multiple1Mapper.class);
MultipleInputs.addInputPath(job, new Path(input2), TextInputFormat.class, Multiple2Mapper.class);
FileOutputFormat.setOutputPath(job, new Path(output));
int res = job.waitForCompletion(true) ? 0 : 1;
return res;
}
/**
* check the args
*/
private void checkArgs() {
if(input1==null||"".equals(input1)){
System.out.println("no user input...");
printUsage();
System.exit(-1);
}
if(input2==null||"".equals(input2)){
System.out.println("no phone input...");
printUsage();
System.exit(-1);
}
if(output==null||"".equals(output)){
System.out.println("no output...");
printUsage();
System.exit(-1);
}
if(delimiter==null||"".equals(delimiter)){
System.out.println("no delimiter...");
printUsage();
System.exit(-1);
}
}
/**
* configuration the args
* @param args
*/
private void configureArgs(String[] args) {
for(int i=0;i<args.length;i++){
if("-i1".equals(args[i])){
input1=args[++i];
}
if("-i2".equals(args[i])){
input2=args[++i];
}
if("-o".equals(args[i])){
output=args[++i];
}
if("-delimiter".equals(args[i])){
delimiter=args[++i];
}
}
}
public static void printUsage(){
System.err.println("Usage:");
System.err.println("-i1 input \t user data path.");
System.err.println("-i2 input \t phone data path.");
System.err.println("-o output \t output data path.");
System.err.println("-delimiter data delimiter , default is comma .");
}
} 这里指定两个mapper和一个reducer，两个mapper分别对应处理user和phone的数据，分别如下：
mapper1（处理user数据）：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import java.io.IOException;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
/**
* input :
* username,phone
*
* output:
* <key,value> --> <[phone],[0,username]>
* @author fansy
*
*/
public class Multiple1Mapper extends Mapper<LongWritable,Text,Text,FlagStringDataType>{
private Logger log = LoggerFactory.getLogger(Multiple1Mapper.class);
private String delimiter=null; // default is comma
@Override
public void setup(Context cxt){
delimiter= cxt.getConfiguration().get("delimiter", ",");
log.info("This is the begin of Multiple1Mapper");
}
@Override
public void map(LongWritable key,Text value,Context cxt) throws IOException,InterruptedException{
String info= new String(value.getBytes(),"UTF-8");
String[] values = info.split(delimiter);
if(values.length!=2){
return;
}
log.info("key-->"+values[1]+"=========value-->"+"[0,"+values[0]+"]");
cxt.write(new Text(values[1]), new FlagStringDataType(0,values[0]));
}
} mapper2（处理phone数据）：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import java.io.IOException;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Mapper;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
/**
* input :
* phone,description
*
* output:
* <key,value> --> <[phone],[1,description]>
* @author fansy
*
*/
public class Multiple2Mapper extends Mapper<LongWritable,Text,Text,FlagStringDataType>{
private Logger log = LoggerFactory.getLogger(Multiple2Mapper.class);
private String delimiter=null; // default is comma
@Override
public void setup(Context cxt){
delimiter= cxt.getConfiguration().get("delimiter", ",");
log.info("This is the begin of Multiple2Mapper");
}
@Override
public void map(LongWritable key,Text value,Context cxt) throws IOException,InterruptedException{
String[] values= value.toString().split(delimiter);
if(values.length!=2){
return;
}
log.info("key-->"+values[0]+"=========value-->"+"[1,"+values[1]+"]");
cxt.write(new Text(values[0]), new FlagStringDataType(1,values[1]));
}
} 这里的FlagStringDataType是自定义的：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import org.apache.hadoop.io.WritableComparable;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import com.google.common.primitives.Ints;
public class FlagStringDataType implements WritableComparable<FlagStringDataType> {
private Logger log = LoggerFactory.getLogger(FlagStringDataType.class);
private String value;
private int flag;
public FlagStringDataType() {
}
public FlagStringDataType(int flag,String value) {
this.value = value;
this.flag=flag;
}
public String get() {
return value;
}
public void set(String value) {
this.value = value;
}
@Override
public boolean equals(Object other) {
return other != null && getClass().equals(other.getClass())
&& ((FlagStringDataType) other).get() == value
&&((FlagStringDataType) other).getFlag()==flag;
}
@Override
public int hashCode() {
return Ints.hashCode(flag)+value.hashCode();
}
@Override
public int compareTo(FlagStringDataType other) {
if (flag >= other.flag) {
if (flag > other.flag) {
return 1;
}
} else {
return -1;
}
return value.compareTo(other.value);
}
@Override
public void write(DataOutput out) throws IOException {
log.info("in write()::"+"flag:"+flag+",vlaue:"+value);
out.writeInt(flag);
out.writeUTF(value);
}
@Override
public void readFields(DataInput in) throws IOException {
log.info("in read()::"+"flag:"+flag+",vlaue:"+value);
flag=in.readInt();
value = in.readUTF();
log.info("in read()::"+"flag:"+flag+",vlaue:"+value);
}
public int getFlag() {
return flag;
}
public void setFlag(int flag) {
this.flag = flag;
}
public String toString(){
return flag+":"+value;
}
} 这个自定义类，使用一个flag来指定是哪个数据，而value则对应是其值。这样做的好处是在reduce端可以根据flag的值来判断其输出位置，这种设计方式可以对多种输入的整合有很大帮助，在mahout中也可以看到这样的设计。
reducer（汇总输出数据）：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import java.io.IOException;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Reducer;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
public class MultipleReducer extends Reducer<Text,FlagStringDataType,Text,NullWritable>{
private Logger log = LoggerFactory.getLogger(MultipleReducer.class);
private String delimiter=null; // default is comma
@Override
public void setup(Context cxt){
delimiter= cxt.getConfiguration().get("delimiter", ",");
}
@Override
public void reduce(Text key, Iterable<FlagStringDataType> values,Context cxt) throws IOException,InterruptedException{
log.info("================");
log.info(" =======");
log.info(" ==");
String[] value= new String[3];
value[2]=key.toString();
for(FlagStringDataType v:values){
int index= v.getFlag();
log.info("index:"+index+"-->value:"+v.get());
value[index]= v.get();
}
log.info(" ==");
log.info(" =======");
log.info("================");
cxt.write(new Text(value[2]+delimiter+value[0]+delimiter+value[1]),NullWritable.get());
}
} 这样设计的好处是，可以针对不同的输入数据采取不同的逻辑处理，而且不同的输入数据可以是序列文件的格式。
下面介绍一种方式和上面的比，略有不足，但是可以借鉴。
首先是Driver：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
//import org.slf4j.Logger;
//import org.slf4j.LoggerFactory;
/**
* input1(/multiple/user/user):
* username,user_phone
*
* input2(/multiple/phone/phone):
* user_phone,description
*
* output: username,user_phone,description
*
* @author fansy
*
*/
public class MultipleDriver2 extends Configured implements Tool{
// private Logger log = LoggerFactory.getLogger(MultipleDriver.class);
private String input1=null;
private String input2=null;
private String output=null;
private String delimiter=null;
public static void main(String[] args) throws Exception {
Configuration conf=new Configuration();
// conf.set("fs.defaultFS", "hdfs://node33:8020");
// conf.set("mapreduce.framework.name", "yarn");
// conf.set("yarn.resourcemanager.address", "node33:8032");
ToolRunner.run(conf, new MultipleDriver2(), args);
}
@Override
public int run(String[] arg0) throws Exception {
configureArgs(arg0);
checkArgs();
Configuration conf= getConf();
conf.set("delimiter", delimiter);
@SuppressWarnings("deprecation")
Job job = new Job(conf, "merge user and phone information ");
job.setJarByClass(MultipleDriver2.class);
job.setMapperClass(MultipleMapper.class);
job.setReducerClass(MultipleReducer.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(FlagStringDataType.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(NullWritable.class);
job.setNumReduceTasks(1);
FileInputFormat.addInputPath(job, new Path(input1));
FileInputFormat.addInputPath(job, new Path(input2));
FileOutputFormat.setOutputPath(job, new Path(output));
int res = job.waitForCompletion(true) ? 0 : 1;
return res;
}
/**
* check the args
*/
private void checkArgs() {
if(input1==null||"".equals(input1)){
System.out.println("no user input...");
printUsage();
System.exit(-1);
}
if(input2==null||"".equals(input2)){
System.out.println("no phone input...");
printUsage();
System.exit(-1);
}
if(output==null||"".equals(output)){
System.out.println("no output...");
printUsage();
System.exit(-1);
}
if(delimiter==null||"".equals(delimiter)){
System.out.println("no delimiter...");
printUsage();
System.exit(-1);
}
}
/**
* configuration the args
* @param args
*/
private void configureArgs(String[] args) {
for(int i=0;i<args.length;i++){
if("-i1".equals(args[i])){
input1=args[++i];
}
if("-i2".equals(args[i])){
input2=args[++i];
}
if("-o".equals(args[i])){
output=args[++i];
}
if("-delimiter".equals(args[i])){
delimiter=args[++i];
}
}
}
public static void printUsage(){
System.err.println("Usage:");
System.err.println("-i1 input \t user data path.");
System.err.println("-i2 input \t phone data path.");
System.err.println("-o output \t output data path.");
System.err.println("-delimiter data delimiter , default is comma .");
}
} 这里添加路径直接使用FileInputFormat添加输入路径，这样的话，针对不同的输入数据的不同业务逻辑可以在mapper中先判断目前正在处理的是那个数据，然后根据其路径来进行相应的业务逻辑处理：
在CODE上查看代码片派生到我的代码片
package multiple.input;
import java.io.IOException;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.InputSplit;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.lib.input.FileSplit;
/**
* input1 :
* username,phone
*
* input2
* phone,description
*
* output:
* <key,value> --> <[phone],[0,username]>
* <key,value> --> <[phone],[1,description]>
* @author fansy
*
*/
public class MultipleMapper extends Mapper<LongWritable,Text,Text,FlagStringDataType>{
private String delimiter=null; // default is comma
private boolean flag=false;
@Override
public void setup(Context cxt){
delimiter= cxt.getConfiguration().get("delimiter", ",");
InputSplit input=cxt.getInputSplit();
String filename=((FileSplit) input).getPath().getParent().getName();
if("user".equals(filename)){
flag=true;
}
}
@Override
public void map(LongWritable key,Text value,Context cxt) throws IOException,InterruptedException{
String[] values= value.toString().split(delimiter);
if(values.length!=2){
return;
}
if(flag){
cxt.write(new Text(values[1]), new FlagStringDataType(0,values[0]));
}else{
cxt.write(new Text(values[0]), new FlagStringDataType(1,values[1]));
}
}
} 总体来说，这种处理方式其实是不如第一种的，在每个map函数中都需要进行判断，比第一种多了很多操作；同时，针对不同的序列文件，这种方式处理不了（Key、value的类型不一样的情况下）。所以针对多文件格式的输入，最好还是使用第一种方式。
扫一扫
361
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
