计算Apache Pig中的不同项目

Question

我有一个带模式的表用户表

|Column 1 | USER ID |int|

|Column 2 |EMAIL|chararray|

|Column 3 |LANGUAGE |chararray|

|Column 4 |LOCATION |chararray|

和具有模式的事务表

|Column 1 | ID |int|

|Column 2 |PRODUCT|int|

|Column 3 |USER ID |int|

|Column 4 |PURCHASE AMOUNT |double|

|Coulmn 5 |DESCRIPTION |chararray|

是...在不同的位置查找每种产品的数量。

我写了如下的猪脚本：-

user = LOAD '/tmp/users.txt' USING PigStorage ('    ')
AS (USER_ID:int, EMAIL:chararray, LANGUAGE:chararray, LOCATION:chararray);

transaction = LOAD '/tmp/transaction.txt' USING PigStorage ('   ')
AS (ID:int, PRODUCT:int,USER_ID:int, PURCHASE_AMOUNT:double,DESCRIPTION:chararray);

u1 = JOIN user by USER_ID, transaction by USER_ID;

u2 = GROUP u1 by LOCATION;

Result = FOREACH u2 GENERATE COUNT(u2.PRODUCT);

DUMP Result;

错误---错误org.apache.pig.tools.grunt.Grunt-错误1200：Pig脚本无法解析：标量投影无效：u2

这就是我要得到的。

Answer 1

在您的脚本中Result = FOREACH u2 GENERATE COUNT(u2.PRODUCT);是错误的。分组操作后，数据结构发生变化。您可以用describe u2找到它。请尝试以下操作（我假设字段在txt文件中用逗号分隔）：

user = LOAD 'user.txt' USING PigStorage (',') AS (USER_ID:int, EMAIL:chararray, LANGUAGE:chararray, LOCATION:chararray);

transaction = LOAD 'transaction.txt' USING PigStorage (',') AS (ID:int, PRODUCT:int,USER_ID:int, PURCHASE_AMOUNT:double,DESCRIPTION:chararray);

u1 = JOIN user by USER_ID, transaction by USER_ID;

u2 = GROUP u1 by (LOCATION,PRODUCT);

Result = FOREACH u2 GENERATE FLATTEN(group) as (LOCATION,PRODUCT), COUNT($1);

DUMP Result;