How can I split the column data into rows with basic SQL.
COL1 COL2
1 A-B
2 C-D
3 AAA-BB
Result
COL1 Col2
1 A
1 B
2 C
2 D
3 AAA
3 BB
From Oracle 12, if it is always two delimited values then you can use:
SELECT t.col1,
l.col2
FROM table_name t
CROSS JOIN LATERAL (
SELECT SUBSTR(col2, 1, INSTR(col2, '-') - 1) AS col2 FROM DUAL
UNION ALL
SELECT SUBSTR(col2, INSTR(col2, '-') + 1) FROM DUAL
) l
Which, for the sample data:
CREATE TABLE table_name (COL1, COL2) AS
SELECT 1, 'A-B' FROM DUAL UNION ALL
SELECT 2, 'C-D' FROM DUAL UNION ALL
SELECT 3, 'AAA-BB' FROM DUAL;
Outputs:
COL1
COL2
1
A
1
B
2
C
2
D
3
AAA
3
BB
db<>fiddle here
Snowflake is tagged, so here's the snowflake way of doing this:
WITH TEST (col1, col2) as
(select 1, 'A-B' from dual union all
select 2, 'C-D' from dual union all
select 3, 'AAA-BB' from dual
)
SELECT test.col1, table1.value
FROM test, LATERAL strtok_split_to_table(test.col2, '-') as table1
ORDER BY test.col1, table1.value;
As of Oracle:
SQL> with test (col1, col2) as
2 (select 1, 'A-B' from dual union all
3 select 2, 'C-D' from dual union all
4 select 3, 'AAA-BB' from dual
5 )
6 select col1,
7 regexp_substr(col2, '[^-]+', 1, column_value) col2
8 from test cross join
9 table(cast(multiset(select level from dual
10 connect by level <= regexp_count(col2, '-') + 1
11 ) as sys.odcinumberlist))
12 order by col1, col2;
COL1 COL2
---------- ------------------------
1 A
1 B
2 C
2 D
3 AAA
3 BB
6 rows selected.
SQL>
For MS-SQL 2016 and higher you can use:
SELECT Col1, x.value
FROM t CROSS APPLY STRING_SPLIT(t.Col2, '-') as x;
BTW: If Col2 contains null, it does not appear in the result.
Related
I am new to oracle thus this question.
There is a table already existed and I have added a new column to it.
There are 5 rows and I do not want to use update table with where clause to insert the values one by one in the new column. Is there a statement like INSERT ALL to insert the values into the new column in one shot ?
Thanks
You can also use something like below which in-effect I would say is multiple update only, wrapped in single statement.
SQL> select * from test_upd;
ID1 ID2
---------- ----------
1
2
3
4
SQL> update test_upd a set a.id2 =
2 (select
3 case
4 when id1=1 then 100
5 when id1=2 then 200
6 when id1=3 then 300
7 else 5000 end
8 from test_upd b
9 where a.id1=b.id1);
4 rows updated.
SQL> select * from test_upd;
ID1 ID2
---------- ----------
1 100
2 200
3 300
4 5000
Use a MERGE statement:
MERGE INTO your_table dst
USING (
SELECT 1 AS id, 'aaa' AS newvalue FROM DUAL UNION ALL
SELECT 2, 'bbb' FROM DUAL UNION ALL
SELECT 3, 'ccc' FROM DUAL UNION ALL
SELECT 4, 'ddd' FROM DUAL UNION ALL
SELECT 5, 'eee' FROM DUAL
) src
ON (dst.id = src.id)
WHEN MATCHED THEN
UPDATE SET value = src.newvalue;
Which, if you have the table:
CREATE TABLE your_table (id, value) AS
SELECT 1, CAST(NULL AS VARCHAR2(3)) FROM DUAL UNION ALL
SELECT 2, NULL FROM DUAL UNION ALL
SELECT 3, NULL FROM DUAL UNION ALL
SELECT 4, NULL FROM DUAL UNION ALL
SELECT 5, NULL FROM DUAL;
Then, after the MERGE, the table contains:
ID
VALUE
1
aaa
2
bbb
3
ccc
4
ddd
5
eee
db<>fiddle here
I am trying to get some default value in my resultset if query does not return anything. I am trying nvl for the same but it is not returning the expected default value. To simulate, Consider following query,
select nvl(null, '10') from dual where 1=0;
I want to get 10 in case of given condition is not true and query does not return any value. However above query not returning any row.
Your query returns zero rows. NVL() isn't going to change that (*).
The correct solution is for the program which executes the query to handle NO_DATA_FOUND exception rather than fiddling the query.
However, you need a workaround so here is one using two sub-queries, one for your actual query, one to for the default.
When your_query returns an empty set you get this:
SQL> with your_qry as
2 ( select col1 from t42 where 1=0 )
3 , dflt as
4 ( select 10 as col1 from dual )
5 select col1
6 from your_qry
7 union all
8 select col1
9 from dflt
10 where not exists (select * from your_qry );
COL1
----------
10
SQL>
And when it returns a row you get this:
SQL> with your_qry as
2 ( select col1 from t42 )
3 , dflt as
4 ( select 10 as col1 from dual )
5 select col1
6 from your_qry
7 union all
8 select col1
9 from dflt
10 where not exists (select * from your_qry );
COL1
----------
12
13
SQL>
The WITH clause is optional here, it just makes it easier to write the query without duplication. This would have the same outcome:
select col1
from t42
where col0 is null
union all
select 10
from dual
where not exists (select col1
from t42
where col0 is null)
;
(*) Okay, there are solutions which use NVL() or COALESCE() with aggregations to do this. They work with single column projections in a single row as this question poses, but break down when the real query has more than one row and/or more than one column. Aggregations change the results.
So this looks alright ...
SQL> with cte as (
2 select 'Z' as col0, 12 as col1 from dual where 1=0 union all
3 select 'X' as col0, 13 as col1 from dual where 1=0 )
4 select
5 nvl(max(col0), 'Y') as col0, nvl(max( col1), 10) as col1
6 from cte;
COL0 COL1
---------- ----------
Y 10
SQL>
... but this not so much:
SQL> with cte as (
2 select 'Z' as col0, 12 as col1 from dual union all
3 select 'X' as col0, 13 as col1 from dual )
4 select
5 nvl(max(col0), 'Y') as col0, nvl(max( col1), 10) as col1
6 from cte;
COL0 COL1
---------- ----------
Z 13
SQL>
May be something like this is what you need
You could change WHERE clause (in this case WHERE COL > 1) similarly in both places.
WITH T(COL) AS(
SELECT 1 FROM DUAL UNION ALL
SELECT 2 FROM DUAL UNION ALL
SELECT 3 FROM DUAL
)
SELECT COL FROM T WHERE COL > 1
UNION ALL
SELECT 10 AS COL FROM DUAL WHERE NOT EXISTS( SELECT 1 FROM T WHERE COL > 1)
You can use aggregation. An aggregation query always returns one row:
select coalesce(max(null), '10')
from dual
where 1 = 0;
I prefer coalesce() to nvl() because coalesce() is the ANSI standard function. But, nvl() would work here just as well.
I have written this query
update cmart_m_branch
set code=concat('000',cmart_m_branch.code)
where length(code)=1.
But it is not working..it was updating with all zeros..column code should have only 4 characters. But it is working like this.... just I am changing the query.....
Update cmart_m_branch
set code=concat('00',cmart_m_branch.code)
where length(code)=2.
Why it is not working with the length 1 ..please suggest me..
Sounds like you're after LPAD:
WITH sample_data AS (SELECT '1' col1 FROM dual UNION ALL
SELECT '12' col1 FROM dual UNION ALL
SELECT '123' col1 FROM dual UNION ALL
SELECT '1234' col1 FROM dual)
SELECT col1,
LPAD(col1, 4, '0') new_col1
FROM sample_data;
COL1 NEW_COL1
---- --------
1 0001
12 0012
123 0123
1234 1234
Of course, this is assuming that your cmart_m_branch.code column is of string datatype. If it's of number datatype then there is no use in doing the update - you should instead format the data on-the-fly when you're selecting it (or you could use a virtual column to do the formatting for you). This can be done using to_char():
WITH sample_data AS (SELECT 1 col1 FROM dual UNION ALL
SELECT 12 col1 FROM dual UNION ALL
SELECT 123 col1 FROM dual UNION ALL
SELECT 1234 col1 FROM dual)
SELECT col1,
to_char(col1, '0999') new_col1
FROM sample_data;
COL1 NEW_COL1
---------- --------
1 0001
12 0012
123 0123
1234 1234
If you're absolutely dead-set on using concatenation (why?) then you can do it using a case expression:
WITH sample_data AS (SELECT '1' col1 FROM dual UNION ALL
SELECT '12' col1 FROM dual UNION ALL
SELECT '123' col1 FROM dual UNION ALL
SELECT '1234' col1 FROM dual)
SELECT col1,
CASE WHEN LENGTH(col1) = 1 THEN '000'||col1
WHEN LENGTH(col1) = 2 THEN '00'||col1
WHEN LENGTH(col1) = 3 THEN '0'||col1
ELSE col1
END new_col1
FROM sample_data;
COL1 NEW_COL1
---- --------
1 0001
12 0012
123 0123
1234 1234
I have the following working query:
WITH pivot_data AS (
select PSGROUP,
PSCOLUMN as PSCOLUMN
FROM LOG_PS_STATUS
)
SELECT *
FROM pivot_data
PIVOT (
MAX(NULL) --<-- pivot_clause
FOR PSCOLUMN--<-- pivot_for_clause
IN (&PS_COLUMNS.) --<-- pivot_in_clause
);
It shows results as expected:
Values:
PSGroup PSColumn
A 1
A 2
A 3
B 1
B 2
B 3
C 3
Result is giving like:
PSGroup(Column vertically) PSColoumn(Horizontally)
1 2 3
A
B
C
Now I want to make PSGroup column as group of PSColumn and output should be like:
A
1 2 3
B
1 2 3
C
3
You can use listagg:
WITH pivot_data(PSGroup,PSColumn) AS (
select 'A', 1 FROM dual UNION all
select 'A', 2 FROM dual UNION all
select 'A', 3 FROM dual UNION all
select 'B', 1 FROM dual UNION all
select 'B', 2 FROM dual UNION all
select 'B', 3 FROM dual UNION all
select 'C', 3 FROM DUAL),
res(PSGroup,PSColumns) as(
SELECT PSGroup, LISTAGG(PSColumn, ' ') WITHIN GROUP (order by PSColumn)
FROM pivot_data
GROUP BY PSGroup)
select DECODE(PSColumns,NULL,PSGroup,NULL) AS PSGroup, PSColumns from(
select PSGroup, NULL as PSColumns from res
union all
select PSGroup, PSColumns from res)t
ORDER BY t.PSGroup, t.PSColumns NULLS first
Also note that LISTAGG(PSColumn, ' ') WITHIN GROUP (order by PSColumn) is limited to 4000 characters
The query does not give the same results in oracle 10g and oracle 11g:
with table1 as
(select 1 as id, 'A' as col1, 'X' as col2
from dual
union all
select 2 as id, 'Y' as col1, 'A' as col2
from dual
union all
select 3 as id, 'Z' as col1, 'Z' as col2
from dual)
select *
from table1
where (1 = 1 and (ID=2 or ID=3))
or EXISTS (SELECT 0 FROM dual WHERE col1 = 'A')
or EXISTS (SELECT 0 FROM dual WHERE col2 = 'A')
result in oracle 11g:
ID COL1 COL2
1 A X
2 Y A
result in oracle 10g:
ID COL1 COL2
1 A X
2 Y A
3 Z Z
why???