Row number functionality in Hive

问题:

How can I generate row numbers for an existing table while running a select query?
For example:

select row_number(), * from emp;

I am using hive 0.13. I can't access external jars or udfs in my environment. The underlying files are in parquet format.

Thanks in advance!

回答1:

ROW_NUMBER() is a windowing function so it needs to be used in conjunction with an OVER clause. Just don't specify any PARTITION.

SELECT *, ROW_NUMBER() OVER () AS row_num
FROM emp
--- other stuff

回答2:

row_number() can be used to find for example, recent visit of a user on your site.

SELECT user_id,user_name,timestamp
FROM (
SELECT user_id,user_name,timestamp,row_number() over (partition by userid order by timestamp desc) as visit_number 
from user) user_table
    WHERE visit_number = 1