Question

我有一个返回扩展名的UDF（GetUrlExt）。（例如：/ abc/models/xyz/images/top.jpg中的jpg）。数据如下所示：

Date Time TimeTaken uristem  
9/5/2011 0:00:10 234 /abc/models/xyz/images/top.jpg  
9/5/2011 0:00:11 456 /abc/models/xyz/images/bottom.jpg  
9/5/2011 0:00:14 789 /abc/models/xyz/images/left.gif  
9/5/2011 0:00:16 234 /abc/models/xyz/images/top.pdf  
9/5/2011 0:00:18 734 /abc/models/xyz/images/top.pdf  
9/5/2011 0:00:19 654 /abc/models/xyz/images/right.gif  
9/5/2011 0:00:21 346 /abc/models/xyz/images/top.pdf  
9/5/2011 0:00:24 556 /abc/models/xyz/images/front.pdf  
9/5/2011 0:00:26 134 /abc/models/xyz/images/back.jpg

没有'GROUP BY'的查询工作正常：

SELECT GetUrlExt(uristem) AS extn FROM LogTable;

结果： JPG JPG GIF PDF格式 PDF格式 GIF PDF格式 PDF格式 JPG

现在我需要'GROUP BY'来获取GetUrlExt UDF的结果预期结果：
jpg 3 274.6
gif 2 721.5
pdf 4 467.5

但以下查询无效：

SELECT GetUrlExt(uristem) AS extn, Count(*) AS PerCount, Avg(TimeTaken) AS AvgTime FROM LogTable GROUP BY extn;

感谢任何形式的帮助！

Answer 1

请使用子查询进行分组。

Hive不直接按计算值支持分组。

SELECT a.extn, Count(*) AS PerCount, Avg(TimeTaken) AS AvgTime 
FROM
(
    SELECT GetUrlExt(uristem) AS extn, TimeTaken
    FROM LogTable 
) a
GROUP BY a.extn;

Answer 2

您可以按别名启用分组，也可以按

分组启用整个语句

SELECT GetUrlExt(uristem) AS extn, Count(*) AS PerCount, Avg(TimeTaken) AS AvgTime 
FROM LogTable 
GROUP BY GetUrlExt(uristem);

Hive：UDF和GROUP BY

2 个答案: