标签:leo img 返回 声明 jar enum ddl imp 集合
不多说,直接上代码。
生成的结果,作为输入源。
代码
package zhouls.bigdata.myMapReduce.ParseTVDataCompressAndCounter;
import java.net.URI;
import java.util.List;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.io.LongWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.io.compress.GzipCodec;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
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;
/**
*
* @function 统计无效数据和对输出结果进行压缩
* @author 小讲
*
*/
public class CompressAndCounter extends Configured implements Tool
{
// 定义枚举对象
public static enum LOG_PROCESSOR_COUNTER
{
BAD_RECORDS
};
/**
*
* @function Mapper 解析数据,统计无效数据,并输出有效数据
*
*/
public static class CompressAndCounterMap extends Mapper<LongWritable, Text, Text, Text>
{
protected void map(LongWritable key, Text value, Context context) throws java.io.IOException, InterruptedException
{
// 解析每条机顶盒记录,返回list集合
List<String> list = ParseTVData.transData(value.toString()); //调用ParseTVData.java下的transData方法
int length = list.size();
// 无效记录
if (length == 0)
{
// 动态自定义计数器
context.getCounter("ErrorRecordCounter", "ERROR_Record_TVData").increment(1);
// 枚举声明计数器
context.getCounter(LOG_PROCESSOR_COUNTER.BAD_RECORDS).increment(1);
} else
{
for (String validateRecord : list)
{
//输出解析数据
context.write(new Text(validateRecord), new Text(""));
}
}
}
}
/**
* @function 任务驱动方法
*
*/
@Override
public int run(String[] args) throws Exception
{
// TODO Auto-generated method stub
//读取配置文件
Configuration conf = new Configuration();
//文件系统接口
URI uri = new URI("hdfs://HadoopMaster:9000");
//输出路径
Path mypath = new Path(args[1]);
// 创建FileSystem对象
FileSystem hdfs = FileSystem.get(uri, conf);
if (hdfs.isDirectory(mypath))
{
//删除已经存在的文件路径
hdfs.delete(mypath, true);
}
Job job = new Job(conf, "CompressAndCounter");//新建一个任务
job.setJarByClass(CompressAndCounter.class);//设置主类
job.setMapperClass(CompressAndCounterMap.class);//只有 Mapper
job.setOutputKeyClass(Text.class);//输出 key 类型
job.setOutputValueClass(Text.class);//输出 value 类型
FileInputFormat.addInputPath(job, new Path(args[0]));//输入路径
FileOutputFormat.setOutputPath(job, new Path(args[1]));//输出路径
FileOutputFormat.setCompressOutput(job, true);//对输出结果设置压缩
FileOutputFormat.setOutputCompressorClass(job, GzipCodec.class);//设置压缩类型
job.waitForCompletion(true);//提交任务
return 0;
}
/**
* @function main 方法
* @param args 输入 输出路径
* @throws Exception
*/
public static void main(String[] args) throws Exception
{
String[] date = {"20120917","20120918","20120919","20120920","20120921","20120922","20120923"};
int ec = 1;
for(String dt:date)
{
String[] args0 = { "hdfs://HadoopMaster:9000/middle/tv/"+dt+".txt",
"hdfs://HadoopMaster:9000/junior/tvCompressResult/"+dt };
// String[] args0 = { "./data/compressAndCounter/"+dt+".txt",
// "hdfs://HadoopMaster:9000/junior/tvCompressResult/"+dt };
ec = ToolRunner.run(new Configuration(), new CompressAndCounter(), args0);
}
System.exit(ec);
}
}
package zhouls.bigdata.myMapReduce.ParseTVDataCompressAndCounter;
import java.util.ArrayList;
import java.util.List;
import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;
/**
*
* @function 解析数据
*
*
*/
public class ParseTVData
{
/**
* @function 使用 Jsoup 工具,解析输入数据,
* @param text
* @return list
*/
public static List<String> transData(String text)
{
List<String> list = new ArrayList<String>();
Document doc;
String rec = "";
try
{
doc = Jsoup.parse(text);// jsoup解析数据
Elements content = doc.getElementsByTag("WIC");
String num = content.get(0).attr("cardNum");// 记录编号
if (num == null || num.equals(""))
{
num = " ";
}
String stbNum = content.get(0).attr("stbNum");// 机顶盒号
if (stbNum.equals(""))
{
return list;
}
String date = content.get(0).attr("date");// 日期
Elements els = doc.getElementsByTag("A");
if (els.isEmpty())
{
return list;
}
for (Element el : els)
{
String e = el.attr("e");// 结束时间
String s = el.attr("s");// 开始时间
String sn = el.attr("sn");// 频道名称
rec = stbNum + "@" + date + "@" + sn + "@" + s + "@" + e;
list.add(rec);
}
} catch (Exception e)
{
System.out.println(e.getMessage());
return list;
}
return list;
}
}
Hadoop MapReduce编程 API入门系列之压缩和计数器(三十)
标签:leo img 返回 声明 jar enum ddl imp 集合
原文地址:http://www.cnblogs.com/zlslch/p/6171823.html