REGEXP_REPLACE to split a string

Question

HI everyone,
I have a little weird problem... I have the following scenario:
WITH temp_1 AS
  SELECT 'AA.1.0a01.25.1' as t FROM dual
  UNION ALL
  SELECT 'AA.2.Aax1.2p5.1' FROM dual
  UNION ALL
  SELECT 'ABC.123.DB' FROM DUAL
  UNION ALL
  SELECT 'ABC.AAX.01' FROM DUAL
SELECT t,
       regexp_replace(t, '(^AA\.)([0-9]*)\..*$','\1\2')
  FROM temp_1;I have a string like 'AA.11.xce' and I need to retrieve untill the second . (dot) because the string starts with AA. But, if the string starts with something else, I only need to retrieve untill the first . (dot). Is this possible with regexp_replace? [I'm using Oracle 10g]
Thank you

Answer 1

Hi,
So you want to do one thing if t starts with 'AA.', and you want to do something if t does not start with 'AA.'. That sounds like a job for CASE:
SELECT     t
,     REGEXP_REPLACE ( t
                 , CASE
                           WHEN  SUBSTR (t, 1, 3)     = 'AA.'
                    THEN  '([^.]*\.[^.]).*'
                    ELSE  '([^.]*).*'
                       END
                 , '\1'
                 )     AS new_t
FROM    temp_1;The 2nd argument to REGEXP_REPLACE is a string. You can use any kind of string expression, including a literal, a column, or (as above) the results of a CASE expression.
For the 2nd argument, you could also use the results of a multi-part || operation, like this:
SELECT     t
,     REGEXP_REPLACE ( t
                 , '('               || -- \1 is ...
                      '[^.]*'          || --   the first .-delimited group, plus
                   CASE
                       WHEN  SUBSTR (t, 1, 3) = 'AA.'
                    THEN  '\.[^.]'        --   another ,-delimited group
                   END               || --   (sometimes)
                ')'               || -- End \1
                '.*'                  -- the rest of the string
                 , '\1'
                 )     AS new_t
FROM    temp_1;Edited by: Frank Kulash on Nov 14, 2011 12:30 PM
Added 2nd example