开发者社区> 问答> 正文

MaxCompute用户指南:MapReduce:示例程序:Pipeline示例



测试准备


  1. 准备好测试程序的 Jar包,假设名字为 mapreduce-examples.jar,本地存放路径为 data\resources。

  2. 准备好 Pipeline 的测试表和资源。

    • 创建测试表。create table wc_in (key string, value string);
    • create table wc_out(key string, cnt bigint);

  • 添加测试资源。
    1. add jar data\resources\mapreduce-examples.jar -f;

    使用 tunnel 导入数据。
    1. tunnel upload data wc_in;

    导入 wc_in 表的数据文件 data 的内容,如下所示:
    1. hello,odps


    测试步骤


    在 odpscmd 中执行 WordCountPipeline,如下所示:
    1. jar -resources mapreduce-examples.jar -classpath data\resources\mapreduce-examples.jar
    2. com.aliyun.odps.mapred.open.example.WordCountPipeline wc_in wc_out;


    预期结果


    作业成功结束后,输出表 wc_out 中的内容,如下所示:
    1. +------------+------------+
    2. | key        | cnt        |
    3. +------------+------------+
    4. | hello      | 1          |
    5. | odps       | 1          |
    6. +------------+------------+


    代码示例

    1.     package com.aliyun.odps.mapred.open.example;
    2.     import java.io.IOException;
    3.     import java.util.Iterator;
    4.     import com.aliyun.odps.Column;
    5.     import com.aliyun.odps.OdpsException;
    6.     import com.aliyun.odps.OdpsType;
    7.     import com.aliyun.odps.data.Record;
    8.     import com.aliyun.odps.data.TableInfo;
    9.     import com.aliyun.odps.mapred.Job;
    10.     import com.aliyun.odps.mapred.MapperBase;
    11.     import com.aliyun.odps.mapred.ReducerBase;
    12.     import com.aliyun.odps.pipeline.Pipeline;
    13.     public class WordCountPipelineTest {
    14.       public static class TokenizerMapper extends MapperBase {
    15.         Record word;
    16.         Record one;
    17.         @Override
    18.         public void setup(TaskContext context) throws IOException {
    19.           word = context.createMapOutputKeyRecord();
    20.           one = context.createMapOutputValueRecord();
    21.           one.setBigint(0, 1L);
    22.         }
    23.         @Override
    24.         public void map(long recordNum, Record record, TaskContext context)
    25.             throws IOException {
    26.           for (int i = 0; i < record.getColumnCount(); i++) {
    27.             String[] words = record.get(i).toString().split("\\s+");
    28.             for (String w : words) {
    29.               word.setString(0, w);
    30.               context.write(word, one);
    31.             }
    32.           }
    33.         }
    34.       }
    35.       public static class SumReducer extends ReducerBase {
    36.         private Record value;
    37.         @Override
    38.         public void setup(TaskContext context) throws IOException {
    39.           value = context.createOutputValueRecord();
    40.         }
    41.         @Override
    42.         public void reduce(Record key, Iterator<Record> values, TaskContext context)
    43.             throws IOException {
    44.           long count = 0;
    45.           while (values.hasNext()) {
    46.             Record val = values.next();
    47.             count += (Long) val.get(0);
    48.           }
    49.           value.set(0, count);
    50.           context.write(key, value);
    51.         }
    52.       }
    53.       public static class IdentityReducer extends ReducerBase {
    54.         private Record result;
    55.         @Override
    56.         public void setup(TaskContext context) throws IOException {
    57.           result = context.createOutputRecord();
    58.         }
    59.         @Override
    60.         public void reduce(Record key, Iterator<Record> values, TaskContext context)
    61.             throws IOException {
    62.           while (values.hasNext()) {
    63.             result.set(0, key.get(0));
    64.             result.set(1, values.next().get(0));
    65.             context.write(result);
    66.           }
    67.         }
    68.       }
    69.       public static void main(String[] args) throws OdpsException {
    70.         if (args.length != 2) {
    71.           System.err.println("Usage: WordCountPipeline <in_table> <out_table>");
    72.           System.exit(2);
    73.         }
    74.         Job job = new Job();
    75.         /***
    76.          * 构造Pipeline的过程中,如果不指定Mapper的OutputKeySortColumns,PartitionColumns,OutputGroupingColumns,
    77.          * 框架会默认使用其OutputKey作为此三者的默认配置
    78.          ***/
    79.         Pipeline pipeline = Pipeline.builder()
    80.             .addMapper(TokenizerMapper.class)
    81.             .setOutputKeySchema(
    82.                     new Column[] { new Column("word", OdpsType.STRING) })
    83.             .setOutputValueSchema(
    84.                     new Column[] { new Column("count", OdpsType.BIGINT) })
    85.             .setOutputKeySortColumns(new String[] { "word" })
    86.             .setPartitionColumns(new String[] { "word" })
    87.             .setOutputGroupingColumns(new String[] { "word" })
    88.             .addReducer(SumReducer.class)
    89.             .setOutputKeySchema(
    90.                     new Column[] { new Column("word", OdpsType.STRING) })
    91.             .setOutputValueSchema(
    92.                     new Column[] { new Column("count", OdpsType.BIGINT)})
    93.             .addReducer(IdentityReducer.class).createPipeline();
    94.         job.setPipeline(pipeline);
    95.         job.addInput(TableInfo.builder().tableName(args[0]).build());
    96.         job.addOutput(TableInfo.builder().tableName(args[1]).build());
    97.         job.submit();
    98.         job.waitForCompletion();
    99.         System.exit(job.isSuccessful() == true ? 0 : 1);
    100.       }
    101.     }
  • 展开
    收起
    行者武松 2017-10-23 17:49:09 2166 0
    0 条回答
    写回答
    取消 提交回答
    问答排行榜
    最热
    最新

    相关电子书

    更多
    Data+AI时代大数据平台应该如何建设 立即下载
    大数据AI一体化的解读 立即下载
    极氪大数据 Serverless 应用实践 立即下载