我正在使用Pig Latin/MapReduce中的简单UDF。 Pig Latin 查询为:
REGISTER PigStringOperations.jar
sensitive = LOAD '/mdsba/sample2.csv' using PigStorage(',') as (AGE:int,EDU:chararray,SEX:chararray,SALARY:chararray);
BV= group sensitive by (EDU,SEX) ;
BVA= foreach BV generate sensitive.AGE as AGE;
anon = FOREACH BVA GENERATE PigStringOperations.StringSplit(sensitive.AGE);
DUMP anon;
UDF是一个简单的Java程序如下图所示
public String exec(Tuple input) throws IOException
String data = (String)input.get(0);
if (data.contains(" "))
{
this.data2 = data.split(" ");
return this.data2[0].toString();
}
return data;}}
这取自成人数据库成人数据库示例分组的AGE输出(EDU,)从一个元组到另一个元组不同,如下所示
AGE(12,10,35,20)
AGE(4,56,10)
AGE(70)
每次运行程序时,我都会收到以下错误:
ERROR 1066: Unable to open iterator for alias anon. Backend error : org.apache.pig.backend.executionengine.ExecException: ERROR 0: Scalar has more than one row in the output. 1st : (,EDU,SEX,SALARY), 2nd :(39,Bachelors,Male,<=50K)
对
敏感数据进行分组后。年龄将是一个袋子!在规划 UDF 时,请考虑这一点。如果你对你的投影进行描述,比如:
DESCRIBE BVA;
这将帮助您了解数据结构并相应地计划处理。