（子）组内的 Oracle Rank()

Question

我试图在子组中获得正确的排名，但结果有问题。我需要根据时间戳获得组每次更改的排名。

例如，使用这个 table:

create table syntrans (
    transid     number,
    launchtime  timestamp,
    status      varchar2(10)
);

insert into syntrans values ( 1, '19-APR-17 07.34.05.824875 PM','FAIL');
insert into syntrans values (  1, '19-APR-17 07.34.06.828753 PM','FAIL');
insert into syntrans values (  1, '19-APR-17 07.34.08.567579 PM','SUCCESS');
insert into syntrans values (  1, '19-APR-17 08.07.31.731745 PM','SUCCESS');
insert into syntrans values (  1, '19-APR-17 08.07.32.735582 PM','SUCCESS');
insert into syntrans values (  2, '19-APR-17 08.17.51.332804 PM','FAIL');
insert into syntrans values (  2, '19-APR-17 08.17.52.336530 PM','FAIL');
insert into syntrans values (  2, '19-APR-17 08.19.27.993327 PM','SUCCESS');
insert into syntrans values (  2, '19-APR-17 08.25.54.860077 PM','FAIL');
insert into syntrans values (  2, '19-APR-17 08.25.55.862830 PM','SUCCESS');

我目前得到的是

SELECT transid,
       launchtime,
       status,
       rank() over (partition by status order by launchtime) rnk
  FROM syntrans 
 order by transid, launchtime, status;

   TRANSID LAUNCHTIME                     STATUS            RNK
---------- ------------------------------ ---------- ----------
         1 19-APR-17 07.34.05.824875 PM   FAIL                1
         1 19-APR-17 07.34.06.828753 PM   FAIL                2
         1 19-APR-17 07.34.08.567579 PM   SUCCESS             1
         1 19-APR-17 08.07.31.731745 PM   SUCCESS             2
         1 19-APR-17 08.07.32.735582 PM   SUCCESS             3
         2 19-APR-17 08.17.51.332804 PM   FAIL                3
         2 19-APR-17 08.17.52.336530 PM   FAIL                4
         2 19-APR-17 08.19.27.993327 PM   SUCCESS             4
         2 19-APR-17 08.25.54.860077 PM   FAIL                5
         2 19-APR-17 08.25.55.862830 PM   SUCCESS             5

需要的输出是这样的：

   TRANSID LAUNCHTIME                     STATUS            RNK
---------- ------------------------------ ---------- ----------
         1 19-APR-17 07.34.05.824875 PM   FAIL                1
         1 19-APR-17 07.34.06.828753 PM   FAIL                2
         1 19-APR-17 07.34.08.567579 PM   SUCCESS             1
         1 19-APR-17 08.07.31.731745 PM   SUCCESS             2
         1 19-APR-17 08.07.32.735582 PM   SUCCESS             3
         2 19-APR-17 08.17.51.332804 PM   FAIL                1
         2 19-APR-17 08.17.52.336530 PM   FAIL                2
         2 19-APR-17 08.19.27.993327 PM   SUCCESS             1
         2 19-APR-17 08.25.54.860077 PM   FAIL                1
         2 19-APR-17 08.25.55.862830 PM   SUCCESS             1

...其中 "rank" 每次状态值更改时重新开始（按日期排序）。我知道我得到的输出是按整体状态组排名的，但我无法找到任何函数组合来获得所需的输出。

这几乎明白了，但不完全是：

SELECT transid, launchtime, status, rnk
  FROM (SELECT transid,
              status,
              launchtime,
              RANK() OVER (PARTITION BY transid, status ORDER BY launchtime) rnk
         FROM syntrans )
ORDER BY transid, launchtime;

Answer 1

您可以使用不同的行号方法将连续的相同状态行归为一组。（运行内部查询以查看组的分配方式。）然后对这些组使用 row_number。

SELECT transid, launchtime, status
,ROW_NUMBER() over(PARTITION BY transid,grp ORDER BY launchtime) as rnk
FROM (SELECT  transid,
              status,
              launchtime,
              ROW_NUMBER() OVER (PARTITION BY transid ORDER BY launchtime)
              -ROW_NUMBER() OVER (PARTITION BY transid, status ORDER BY launchtime) as grp
         FROM syntrans) t

（子）组内的 Oracle Rank()

Oracle Rank() within (sub)group

sql

oracle

rank

oracle12c