Update each row of the table with data from table itself - sql-server

Question

I have a problem with updating work table - values from that table come from table itself. Here is my table:
+----------+----------+-----+---------+
| EVEN_KEY | INVE_KEY | QUA | QUA_MAX |
+----------+----------+-----+---------+
| 1 | 2 | 1 | NULL |
| 2 | 2 | 2 | NULL |
| 3 | 2 | 3 | NULL |
| 1 | 1 | 2 | NULL |
| 4 | 2 | 3 | NULL |
+----------+----------+-----+---------+
What I would like to do is update column qua_max - summarize column qua for given inve_key for each row. So, results in above table should look like after update:
+----------+----------+-----+---------+
| EVEN_KEY | INVE_KEY | QUA | QUA_MAX |
+----------+----------+-----+---------+
| 1 | 2 | 1 | 9 |
| 2 | 2 | 2 | 9 |
| 3 | 2 | 3 | 9 |
| 1 | 1 | 2 | 1 |
| 4 | 2 | 3 | 9 |
+----------+----------+-----+---------+
And here is my problem - query from this example is giving me error, I can not even run it. What is wrong?
Error:
Query:
UPDATE #TEMP_FINAL
SET QUA_MAX = (SELECT SUM(QUA)
FROM #TEMP_FINAL t2
WHERE #TEMP_FINAL.INVE_KEY = t2.INVE_KEY
GROUP BY INVE_KEY
)
EXAMPLE TABLE:
DECLARE #TEMP_FINAL TABLE
(
EVEN_KEY INT,
INVE_KEY INT,
QUA INT,
QUA_MAX INT
)
insert into #TEMP_FINAL (even_key, inve_key, qua)
values(1, 2, 1),
(2,2,2),
(3,2,3),
(1,1,2),
(4,2,3)

Answer 1

You can try this..
UPDATE t1
SET t1.QUA_MAX = a.sum_qua
from #temp_final t1,
(SELECT SUM(QUA) as sum_qua,inve_key
FROM #TEMP_FINAL t2
GROUP BY INVE_KEY
) a
where t1.INVE_KEY = a.INVE_KEY

Answer 2

You are looking for window function
sum(QUA) over (partition by INVE_KEY)
Note that your sample data for #TEMP_FINAL has QUA = 2 for INVE_KEY = 1 which isn't what the original sample data was.
select
*,
QUA_MAX = sum(QUA) over (partition by INVE_KEY)
from #TEMP_FINAL
And one way to update it would be with a correlated subquery
update t1
set QUA_MAX = (select top 1 sum(QUA) over (partition by INVE_KEY) from #TEMP_FINAL t2 where t2.INVE_KEY = t1.INVE_KEY)
from #TEMP_FINAL t1
select * from #TEMP_FINAL
Or a CTE
;with cte as(
select
EVEN_KEY,
INVE_KEY,
QUA,
QUA_MAX = sum(QUA) over (partition by INVE_KEY)
from #TEMP_FINAL)
update #TEMP_FINAL
set QUA_MAX = c.QUA_MAX
from cte c
where c.INVE_KEY = [#TEMP_FINAL].INVE_KEY
select * from #TEMP_FINAL

Answer 3

i think that you should add temp_final in the from clause, because you're doing an implicit selfjoin, but you declare just a table as alias t2

Answer 4

So here is your complete code:
DECLARE #TEMP_FINAL TABLE
(
EVEN_KEY INT,
INVE_KEY INT,
QUA INT,
QUA_MAX INT
)
insert into #TEMP_FINAL (even_key, inve_key, qua)
values(1, 2, 1),
(2,2,2),
(3,2,3),
(1,1,2),
(4,2,3)
UPDATE #TEMP_FINAL
SET QUA_MAX = (SELECT SUM(QUA)
FROM #TEMP_FINAL t2
WHERE #TEMP_FINAL.INVE_KEY = t2.INVE_KEY
GROUP BY INVE_KEY
)
I don't know if I am looking at this wrong but to my it looks like you are doing a where clause comparing that same thing to itself? But instead of doing #TEMP_FINAL.INVE_KEY = #TEMP_FINAL.INVE_KEY you're doing #TEMP_FINAL.INVE_KEY = t2.INVE_KEY. Sorry for bringing this up I just wanted to verify that this was intentional or not?