I’m looking to currently split a string in Oracle by more than 1 character.

Currently I’m splitting a set of numbers by a space like so:

SELECT regexp_substr('123 456', '(^| )([^ ]*)') from dual
SELECT regexp_substr('123 456', '(^| )([^ ]*)', 1, 2, null, 2) from dual

etc. which returns back 123 and then 456.

However I’m looking to have a scenario where I could input "123 456-789" and be able to split the numbers into "123" "456" "789".

I’ve tried

(^| )([^ ]*)|^[^-]*[^ -]

as my new regex expression but no joy.

>Solution :

You can use:

SELECT value,
       REGEXP_SUBSTR(value, '(^|[ -])([^ -]*)', 1, 1, null, 2) AS first,
       REGEXP_SUBSTR(value, '(^|[ -])([^ -]*)', 1, 2, null, 2) AS second,
       REGEXP_SUBSTR(value, '(^|[ -])([^ -]*)', 1, 3, null, 2) AS third
FROM   table_name

Or, more simply, don’t look for delimiters, just look for numbers:

SELECT value,
       REGEXP_SUBSTR(value, '\d+', 1, 1) AS first,
       REGEXP_SUBSTR(value, '\d+', 1, 2) AS second,
       REGEXP_SUBSTR(value, '\d+', 1, 3) AS third
FROM   table_name

Which, for the sample data:

CREATE TABLE table_name (value) AS
SELECT '123 456'     FROM DUAL UNION ALL
SELECT '123-456 789' FROM DUAL;

Both output:

---

However, if you only ever have single character delimiters, you can use simple string functions (which is more to type but may be more efficient):

SELECT value,
       CASE
       WHEN sep1 > 0 THEN SUBSTR(value, 1, sep1 - 1)
       ELSE value
       END AS first,
       CASE
       WHEN sep2 > 0 THEN SUBSTR(value, sep1 + 1, sep2 - sep1 - 1)
       WHEN sep1 > 0 THEN SUBSTR(value, sep1 + 1)
       END AS second,
       CASE
       WHEN sep2 > 0 THEN SUBSTR(value, sep2 + 1)
       END AS third
FROM   (
  SELECT value,
         normalised_value,
         INSTR(normalised_value, ' ', 1, 1) AS sep1,
         INSTR(normalised_value, ' ', 1, 2) AS sep2
  FROM   (
    SELECT value,
           TRANSLATE(value, ' -', '  ') AS normalised_value
    FROM   table_name
  )
)

Which has the same output.

fiddle