温馨提示×

温馨提示×

您好,登录后才能下订单哦!

密码登录×
登录注册×
其他方式登录
点击 登录注册 即表示同意《亿速云用户服务条款》

Mapreduce RCFile如何写入和读取API

发布时间:2021-12-16 16:20:31 来源:亿速云 阅读:147 作者:小新 栏目:云计算

这篇文章主要介绍Mapreduce RCFile如何写入和读取API,文中介绍的非常详细,具有一定的参考价值,感兴趣的小伙伴们一定要看完!

RCFile是FaceBook开发的高压缩比、高效读的行列存储结构。通常在Hive中可以直接对一张Text表使用insert-select转换,但有时希望使用Mapreduce进行RCFile的读写。

        <dependency>
            <groupId>org.apache.hadoop</groupId>
            <artifactId>hadoop-client</artifactId>
            <version>2.5.0-cdh6.2.1</version>
        </dependency>

        <dependency>
            <groupId>org.apache.hive</groupId>
            <artifactId>hive-serde</artifactId>
            <version>0.13.1-cdh6.2.1</version>
        </dependency>

        <dependency>
            <groupId>org.apache.hive.hcatalog</groupId>
            <artifactId>hive-hcatalog-core</artifactId>
           <version>0.13.1-cdh6.2.1</version>
       </dependency>

读取文本文件,使用mapreduce生成RCFile格式文件

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hive.serde2.columnar.BytesRefArrayWritable;
import org.apache.hadoop.hive.serde2.columnar.BytesRefWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hive.hcatalog.rcfile.RCFileMapReduceInputFormat;

import java.io.IOException;

public class RcFileReaderJob {
    static class RcFileMapper extends Mapper<Object, BytesRefArrayWritable, Text, NullWritable> {
        @Override
        protected void map(Object key, BytesRefArrayWritable value,
                           Context context)
                throws IOException, InterruptedException {
            Text txt = new Text();
            StringBuffer sb = new StringBuffer();
            for (int i = 0; i < value.size(); i++) {
                BytesRefWritable v = value.get(i);
                txt.set(v.getData(), v.getStart(), v.getLength());
                if (i == value.size() - 1) {
                    sb.append(txt.toString());
                } else {
                    sb.append(txt.toString() + "\t");
                }
            }
            context.write(new Text(sb.toString()), NullWritable.get());
        }

        @Override
        protected void cleanup(Context context) throws IOException,
                InterruptedException {
            super.cleanup(context);
        }

        @Override
        protected void setup(Context context) throws IOException,
                InterruptedException {
            super.setup(context);

        }
    }

    static class RcFileReduce extends Reducer<Text, NullWritable, Text, NullWritable> {
        @Override
        protected void reduce(Text key, Iterable<NullWritable> values,
                              Context context) throws IOException, InterruptedException {
            context.write(key, NullWritable.get());
        }
    }

    public static boolean runLoadMapReducue(Configuration conf, Path input, Path output) throws IOException,
            ClassNotFoundException, InterruptedException {
        Job job = Job.getInstance(conf);
        job.setJarByClass(RcFileReaderJob.class);
        job.setJobName("RcFileReaderJob");
        job.setNumReduceTasks(1);
        job.setMapperClass(RcFileMapper.class);
        job.setReducerClass(RcFileReduce.class);
        job.setInputFormatClass(RCFileMapReduceInputFormat.class);
//        MultipleInputs.addInputPath(job, input, RCFileInputFormat.class);
        RCFileMapReduceInputFormat.addInputPath(job, input);
        job.setOutputKeyClass(Text.class);
        job.setOutputValueClass(NullWritable.class);
        FileOutputFormat.setOutputPath(job, output);
        return job.waitForCompletion(true);
    }

    public static void main(String[] args) throws Exception {
        Configuration conf = new Configuration();
        if (args.length != 2) {
            System.err.println("Usage: rcfile <in> <out>");
            System.exit(2);
        }
        RcFileReaderJob.runLoadMapReducue(conf, new Path(args[0]), new Path(args[1]));
    }
}  

读取RCFile格式文件,使用mapreduce生成Text格式文件

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.conf.Configured;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hive.serde2.columnar.BytesRefArrayWritable;
import org.apache.hadoop.hive.serde2.columnar.BytesRefWritable;
import org.apache.hadoop.io.NullWritable;
import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.util.GenericOptionsParser;
import org.apache.hadoop.util.Tool;
import org.apache.hadoop.util.ToolRunner;
import org.apache.hive.hcatalog.rcfile.RCFileMapReduceOutputFormat;

import java.io.IOException;

public class RcFileWriterJob extends Configured implements Tool{
    public static class Map extends Mapper<Object, Text, NullWritable, BytesRefArrayWritable>{
        private byte[] fieldData;
        private int numCols;
        private BytesRefArrayWritable bytes;
        
        @Override
        protected void setup(Context context) throws IOException, InterruptedException {
            numCols = context.getConfiguration().getInt("hive.io.rcfile.column.number.conf", 0);
            bytes = new BytesRefArrayWritable(numCols);
        }
        
        public void map(Object key, Text line, Context context
                ) throws IOException, InterruptedException {
            bytes.clear();
            String[] cols = line.toString().split("\t", -1);
            System.out.println("SIZE : "+cols.length);
            for (int i=0; i<numCols; i++){
                fieldData = cols[i].getBytes("UTF-8");
                BytesRefWritable cu = new BytesRefWritable(fieldData, 0, fieldData.length);
                bytes.set(i, cu);
            }
            context.write(NullWritable.get(), bytes);
        }
    }
    
    public int run(String[] args) throws Exception {
        Configuration conf = new Configuration();
        String[] otherArgs = new GenericOptionsParser(conf, args).getRemainingArgs();
        if(otherArgs.length < 2){
            System.out.println("Usage: " +
                    "hadoop jar RCFileLoader.jar <main class> " +
                    "-tableName <tableName> -numCols <numberOfColumns> -input <input path> " +
                    "-output <output path> -rowGroupSize <rowGroupSize> -ioBufferSize <ioBufferSize>");
            System.out.println("For test");
            System.out.println("$HADOOP jar RCFileLoader.jar edu.osu.cse.rsam.rcfile.mapreduce.LoadTable " +
                    "-tableName test1 -numCols 10 -input RCFileLoaderTest/test1 " +
                    "-output RCFileLoaderTest/RCFile_test1");
            System.out.println("$HADOOP jar RCFileLoader.jar edu.osu.cse.rsam.rcfile.mapreduce.LoadTable " +
                    "-tableName test2 -numCols 5 -input RCFileLoaderTest/test2 " +
                    "-output RCFileLoaderTest/RCFile_test2");
            return 2;
        }

        String tableName = "";
        int numCols = 0;
        String inputPath = "";
        String outputPath = "";
        int rowGroupSize = 16 *1024*1024;
        int ioBufferSize = 128*1024;
        for (int i=0; i<otherArgs.length - 1; i++){
            if("-tableName".equals(otherArgs[i])){
                tableName = otherArgs[i+1];
            }else if ("-numCols".equals(otherArgs[i])){
                numCols = Integer.parseInt(otherArgs[i+1]);
            }else if ("-input".equals(otherArgs[i])){
                inputPath = otherArgs[i+1];
            }else if("-output".equals(otherArgs[i])){
                outputPath = otherArgs[i+1];
            }else if("-rowGroupSize".equals(otherArgs[i])){
                rowGroupSize = Integer.parseInt(otherArgs[i+1]);
            }else if("-ioBufferSize".equals(otherArgs[i])){
                ioBufferSize = Integer.parseInt(otherArgs[i+1]);
            }
            
        }
        
        conf.setInt("hive.io.rcfile.record.buffer.size", rowGroupSize);
        conf.setInt("io.file.buffer.size", ioBufferSize);

        Job job = Job.getInstance(conf);
        job.setJobName("RcFileWriterJob");
        job.setJarByClass(RcFileWriterJob.class);
        job.setMapperClass(Map.class);
        job.setMapOutputKeyClass(NullWritable.class);
        job.setMapOutputValueClass(BytesRefArrayWritable.class);
//        job.setNumReduceTasks(0);
        
        FileInputFormat.addInputPath(job, new Path(inputPath));
        
        job.setOutputFormatClass(RCFileMapReduceOutputFormat.class);
        RCFileMapReduceOutputFormat.setColumnNumber(job.getConfiguration(), numCols);
        RCFileMapReduceOutputFormat.setOutputPath(job, new Path(outputPath));
        RCFileMapReduceOutputFormat.setCompressOutput(job, false);

        System.out.println("Loading table " + tableName + " from " + inputPath + " to RCFile located at " + outputPath);
        System.out.println("number of columns:" + job.getConfiguration().get("hive.io.rcfile.column.number.conf"));
        System.out.println("RCFile row group size:" + job.getConfiguration().get("hive.io.rcfile.record.buffer.size"));
        System.out.println("io bufer size:" + job.getConfiguration().get("io.file.buffer.size"));
        
        return (job.waitForCompletion(true) ? 0 : 1);
    }
    
    public static void main(String[] args) throws Exception {
        int res = ToolRunner.run(new Configuration(), new RcFileWriterJob(), args);
        System.exit(res);
    }

}

以上是“Mapreduce RCFile如何写入和读取API”这篇文章的所有内容,感谢各位的阅读!希望分享的内容对大家有帮助,更多相关知识,欢迎关注亿速云行业资讯频道!

向AI问一下细节

免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。

AI