New to SQL so looking for help
I'm trying to write a query which would print values starting from column A till the column B excluding the value present in column 'ANS' of second table.
Like here are the two tables X and Y
Table1
A FROM TO
a 6 9
b 3 6
c 0 3
d 2 3
Table2
A ANS
a 7
b 5
c 1
And I want the output as
A ANS
a 6
a 8
a 9
b 3
b 4
b 6
c 0
c 2
c 3
d 2
d 3
I've tried to write something like this but it doesn't work
WITH y(n) AS
(SELECT 1 AS n
FROM dual
UNION ALL
SELECT n + 1 AS n
FROM y, table1 T
WHERE n <= T.TO AND n>= T.FROM )
SELECT * FROM y;
Which prints 5000+ rows (that's why I am not attaching output)
Thanks in advance
After you get all the numbers between from and to with a recursive cte, left join on the generated table and get only those numbers which don't exist in table2 using not exists.
--Get the maximum value of `to` column and generate all numbers between 0 and that value
WITH maxto(maxt) as (SELECT MAX(TO) FROM TABLE1)
,y(n) AS
(SELECT 0 AS n FROM dual
UNION ALL
SELECT n + 1 AS n FROM y WHERE n < (SELECT maxt FROM maxto))
SELECT * FROM
(SELECT t1.a, y.n
FROM y
LEFT JOIN table1 t1 on y.n between t1.from and t1.to
WHERE t1.a IS NOT NULL) x
WHERE NOT EXISTS (SELECT 1 FROM table2 WHERE x.a = a and x.n = ans)
ORDER BY 1,2
Sample demo
WITH y(n) AS
(SELECT level - 1 FROM dual connect by level <= select max(TO- FROM) +2 from table1)
SELECT t1.a, t1.from + y.n FROM table1 t1
JOIN y on 1 = 1
left JOIN table2 on y.n + t1.FROM = t2.ANS and t2.a = t1.a
where y.n < t1.TO-t1.FROM
and t2.ANS is null;
You can use a "hierarchical query" and a MINUS operation and avoid joins altogether. MINUS is easy to understand if you are somewhat familiar with set theory. Generating numbers using hierarchical queries is somewhat unnatural (and may only be available in Oracle, I don't know any other db products), but it is used very often and it works very fast.
I changed the column names to from_n and to_n; I don't remember if "from" and/or "to" are reserved words in Oracle, but why take the risk.
with
table1 ( a, from_n, to_n ) as (
select 'a', 6, 9 from dual union all
select 'b', 3, 6 from dual union all
select 'c', 0, 3 from dual union all
select 'd', 2, 3 from dual
),
table2 ( a, ans ) as (
select 'a', 7 from dual union all
select 'b', 5 from dual union all
select 'c', 1 from dual
)
-- everything above this point is for testing only and can be removed
-- solution (SQL query) begins below
select a, from_n + level - 1 as ans
from table1
connect by level <= 1 + to_n - from_n
and prior a = a
and prior sys_guid() is not null
minus
select a, ans
from table2
;
Output:
A ANS
- ----------
a 6
a 8
a 9
b 3
b 4
b 6
c 0
c 2
c 3
d 2
d 3
11 rows selected
Related
My actual problem involves a larger row source and more involved math, but this is a small example that still exhibits the challenge faced. Using Oracle 19c.
Suppose we have a table X with four rows of data as follows.
x
-
1
2
3
4
Further, suppose we want to derive two columns, a and b, from X such that
a = x + sum(previous rows of b ordered by x)
b = a - 1.
Where if there are no previous rows, then the sum is 0.
Therefore, the new table would have rows like the following.
x a b
- - -
1 1 0
2 2 1
3 4 3
4 8 7
The following is invalid SQL, but provides an example of what is being attempted.
with
X AS
(
select 1 x from dual
union all select 2 from dual
union all select 3 from dual
union all select 4 from dual
)
, A AS
(
select
x
, x + sum(b) over (order by x range between unbounded preceding and 1 preceding) AS a
, a - 1 AS b
from x
)
select * from A
;
Perhaps a hierarchical query may help, but not sure on what it is to connect by.
Any ideas would be appreciated. Thanks in advance.
You can do this using a recursive CTE:
with X AS (
select 1 x from dual
union all select 2 from dual
union all select 3 from dual
union all select 4 from dual
),
cte(x, a, b, b_sum) as (
select x, x as a, x - 1 as b, x - 1 as b_sum
from x
where x = 1
union all
select x.x, x.x + cte.b_sum, x.x + cte.b_sum - 1, cte.b_sum + (x.x + cte.b_sum - 1)
from cte join
x
on x.x = cte.x + 1
)
select *
from cte;
Here is a db<>fiddle.
If I have two columns:
col1 col2 amount
1 2 15
2 3 12
1 3 10
3 1 4
3 2 3
And I perform a group by col1,col2 then I get a row for each combination (present) in the data.
My problem though is, that I dont always have all combinations, but I would want to return a row of each combination still. So if there isn't a combination. for example 2 -> 1 then I would want its value to be 0.
Can I somehow specify the "levels" of the group by?
I'm using SQL Oracle.
and the outcome I would want is:
1 -> 2 15
1 -> 3 10
2 -> 1 0
2 -> 3 12
3 -> 1 4
3 -> 2 3
With their respective amount, and 0 if they dont exist, or null works. ( I have a filter to exclude where col1 and col2 are same)
Generate all the rows using cross join and then filter for the ones you want:
select c1.col1, c2.col2, coalesce(t.amount, 0)
from (select 1 as co1l from dual union all
select 2 as co1l from dual union all
select 3 as co1l from dual
) c1 cross join
(select 1 as co12 from dual union all
select 2 as co12 from dual union all
select 3 as co12 from dual
) c2 left join
t
on t.col1 = c1.col1 and t.col2 = c2.col2
where c1.col1 <> c2.col2;
I have something like
id | name
---|-----
1 | Sarah
3 | Pat
4 | Lea
I'm looking for missing rows. I've tried to use generate_series and a left join but this is something you can't do in Redshift because generate_series is not supported.
Is it possible to do it without temporary table?
EDIT
Finally did with a temporary table (0 to 1_000_000) see answer.
That's probably not optimal. But this is how I did
-- create temporary table
CREATE TABLE series (id INT) SORTKEY(id);
-- insert 0 to 1_000_000
INSERT INTO series WITH seq_0_9 AS
(SELECT 0 AS num
UNION ALL SELECT 1 AS num
UNION ALL SELECT 2 AS num
UNION ALL SELECT 3 AS num
UNION ALL SELECT 4 AS num
UNION ALL SELECT 5 AS num
UNION ALL SELECT 6 AS num
UNION ALL SELECT 7 AS num
UNION ALL SELECT 8 AS num
UNION ALL SELECT 9 AS num),
seq_0_999 AS
(SELECT a.num + b.num * 10 + c.num * 100 AS num
FROM seq_0_9 a,
seq_0_9 b,
seq_0_9 c)
SELECT a.num + b.num * 1000 AS num
FROM seq_0_999 a,
seq_0_999 b
ORDER BY num;
-- Why not
VACUUM series;
-- LEFT OUTER JOIN with table inverted and with the interval
SELECT *
FROM series
LEFT OUTER JOIN other_table ON series.id = other_table.id
WHERE series.id BETWEEN 0 AND 4
ORDER BY series.id;
I have below table.
TABLE: ABCD
B column have value 1 whenever there is a change in A column. Now I have to update the table like below. How can I do that?
You can do this using a correlated subquery:
update t
set b = (select sum(t2.b) from t t2 where t2.A <= t.A);
This is standard SQL and should work in either Oracle or Teradata.
Lets have a slightly more complicated example (where the changes in B are not correlated to the changes in A):
Oracle Setup:
CREATE TABLE ABCD( A, B ) AS
SELECT 1, 0 FROM DUAL UNION ALL
SELECT 1, 0 FROM DUAL UNION ALL
SELECT 1, 1 FROM DUAL UNION ALL
SELECT 2, 1 FROM DUAL UNION ALL
SELECT 2, 0 FROM DUAL UNION ALL
SELECT 3, 0 FROM DUAL UNION ALL
SELECT 3, 1 FROM DUAL UNION ALL
SELECT 3, 0 FROM DUAL;
Update:
UPDATE ABCD t1
SET b = (
SELECT sm
FROM (
SELECT ROWID AS id,
SUM( b ) OVER ( ORDER BY a, ROWNUM ) AS sm
FROM ABCD
) t2
WHERE t1.ROWID = t2.ID
);
Output:
SELECT * FROM ABCD;
A B
- -
1 0
1 0
1 1
2 2
2 2
3 2
3 3
3 3
(Note: This is an Oracle solution; I have no idea if it will or won't work in Teradata.)
I have a query on 2 views which when it unions
View X
A B C
2 3
View Y
A B C
3 4
this is my query from the 2 views
select * from view X
UNION
select * from view Y;
the result I had:
A B C
2 3
3 4
the result I wanted (the 2 overrides the null value in view Y # attribute A):
A B C
2 3
2 3 4
how can i obtained that ?
Try this
select nvl(A, lag(A) over (order by rownum)), B, C from (
select A, B, C from X
union
select A, B, C from Y
)
sqlfiddle
here is one answer in oracle
select * from
(select x from tbl1
union
select x from tbl2) t1
,
(select y from tbl1
union
select y from tbl2) t2
,
(select z from tbl1
union
select z from tbl2) t3
where t1.x is not null
order by t1.x desc nulls first;