Oracle SQL - How to Cut out characters from a string with SUBSTR? - sql

Question

I have values like "ABC1234", "ABC", "DEF456", "GHI" etc. in a specific column which I need.
Now I need to split this string but only if the character (e.g. "ABC") are followed by digits.
So if the value is "ABC1234" then I need to cut out ABC and 1234 seperated. But if there is only "ABC" as a value, I just need the "ABC". I can't find any solution with SUBSTR. Do you have any idea?
Note: The length of the characters can differ from 1 to 10 and also the length from the digits (sometimes there isn't any like I showed you).

Answer 1

So if the value is "ABC1234" then I need to cut out ABC and 1234
seperated. But if there is only "ABC" as a value, I just need the
"ABC".
Amidst of other solutions, I propose one solution as shown below:
Logic:
1) Replace all the digits to 1. Check the position of the digit occurring in the string. If
there is no digit in the string then use the String.
2) Extract the alphabets from 1st position to the position where
digit starts.
3) Extract the digit from the position it starts till end. If digit doesnot exists the set it NULL
--Dataset Preparation
with test (col) as
(select 'ABC1234' from dual union all
select 'ABC' from dual union all
select 'dEfH456' from dual union all
select '123GHI' from dual union all
select '456' from dual
)
--Query
select col Original_Column,
CASE
WHEN (instr(regexp_replace(col,'[0-9]','1'),'1',1)) = 0
then col
else
substr( col,1,instr(regexp_replace(col,'[0-9]','1'),'1',1)-1)
end Col_Alp,
CASE
WHEN (instr(regexp_replace(col,'[0-9]','1'),'1',1)) = 0
then NULL
Else
substr( col,instr(regexp_replace(col,'[0-9]','1'),'1',1))
END col_digit
from test
where regexp_like(col, '^[a-zA-Z0-9]+$');
Result:
SQL> /
Original_Column Col_Alp col_digit
---------- ----- -----
ABC1234 ABC 1234
ABC ABC NULL
dEfH456 dEfH 456
123GHI NULL 123GHI
456 NULL 456

Answer 2

Using SUBSTR (and INSTR and TRANSLATE):
SQL Fiddle
Oracle 11g R2 Schema Setup:
CREATE TABLE data ( value ) AS
SELECT 'ABC1234' FROM DUAL UNION ALL
SELECT 'ABC123D' FROM DUAL UNION ALL
SELECT 'ABC ' FROM DUAL UNION ALL
SELECT 'ABC' FROM DUAL UNION ALL
SELECT 'DEFG456' FROM DUAL UNION ALL
SELECT 'GHI' FROM DUAL UNION ALL
SELECT 'JKLMNOPQRS9' FROM DUAL;
Query 1:
SELECT value,
SUBSTR( value, 1, first_digit - 1 ) AS prefix,
TO_NUMBER( SUBSTR( value, first_digit ) ) AS suffix
FROM (
SELECT value,
INSTR(
TRANSLATE( value, '-1234567890', ' ----------' ),
'-',
1
) AS first_digit
FROM data
)
WHERE SUBSTR( value, first_digit ) IS NOT NULL
AND TRANSLATE( SUBSTR( value, first_digit ), '-1234567890', ' ' ) IS NULL
Results:
| VALUE | PREFIX | SUFFIX |
|-------------|------------|--------|
| ABC1234 | ABC | 1234 |
| DEFG456 | DEFG | 456 |
| JKLMNOPQRS9 | JKLMNOPQRS | 9 |

Answer 3

Try this below query for scenarios mentioned , I didn't split if characters followed by numbers:
with test (col) as
(select 'ABC1234' from dual union all
select 'ABC' from dual union all
select 'dEfH456' from dual union all
select '123GHI' from dual union all
select '456' from dual
)
select col,reverse(trim(regexp_replace(reverse(col),'^[0-9]+',' '))) string ,trim(regexp_replace(col,'^[a-zA-Z]+',' ')) numbers from test
if like to move that characters&string to any place my case statement
with test (col) as
(select 'ABC1234' from dual union all
select 'ABC' from dual union all
select 'dEfH456' from dual union all
select '123GHI' from dual union all
select '456' from dual
)
select v.col,case when v.string=v.numbers THEN NULL ELSE string end string , v.numbers
from (select col,reverse(trim(regexp_replace(reverse(col),'^[0-9]+',' '))) string ,trim(regexp_replace(col,'^[a-zA-Z]+',' ')) numbers from test) v

Answer 4

Would something like this do?
SQL> with test (col) as
2 (select '"ABC1234", "ABC", "dEf456", "123GHI", "456"' from dual),
3 inter as
4 (select trim(regexp_substr(replace(col, '"', ''), '[^,]+', 1, level)) token
5 from test
6 connect by level <= regexp_count(col, ',') + 1
7 )
8 select regexp_substr(token, '^[a-zA-Z]+') letters,
9 regexp_substr(token, '[0-9]+$') digits
10 from inter
11 where regexp_like(token, '^[a-zA-Z]+[0-9]+$');
LETTERS DIGITS
---------- ----------
ABC 1234
dEf 456
SQL>