Oracle RTRIM trims more than asked for

Question

Does anyone know why these:

SELECT RTRIM('123R_CLUSTER', '_CLUSTER') -- should give '123R' 
FROM DUAL;
SELECT RTRIM('123S_CLUSTER', '_CLUSTER') -- should give '123S' 
FROM DUAL;
SELECT RTRIM('123T_CLUSTER', '_CLUSTER') -- should give '123T'
FROM DUAL;
SELECT RTRIM('123U_CLUSTER', '_CLUSTER') -- should give '123U'
FROM DUAL;

return '123' instead of the expected?

I'm on Oracle Database 12c Enterprise Edition Release 12.1.0.2.0 - 64bit Production.

The fun begins when you try these:

• replace 123 with anything else (no change still wrong results, ie trims one character more),
• replace "R" / "S" / "T" / "U" with anything else, (works OK)
• replace "_CLUSTER" with anything else, (works OK)
• add anything after "_CLUSTER" (no change).

Answer 1

The documentation is quite clear:

The Oracle/PLSQL RTRIM function removes all specified characters from the right-hand side of a string.

So it doesn't remove the string _CLUSTER at the end of your string - it removes characters until there is one that isn't _, C, L, U, S, T, E or R. Since your postfixes are R/S/T/U, they also match the rtrim condition, and are removed. As would 123S_SLURTE , for example.

As an easier to understand example,

rtrim('LK_123aababaabbbababbaa', 'ab') // returns LK_123

rtrim simply isn't the tool for the job at hand :)

Answer 2

And in case you want to know how to easily do this with, try this:

SELECT regexp_substr('123R_CLUSTER', '^(.*)_CLUSTER$', 1, 1, null, 1) from dual
union all
SELECT regexp_substr('123S_CLUSTER', '^(.*)_CLUSTER$', 1, 1, null, 1) from dual
union all
SELECT regexp_substr('123T_CLUSTER', '^(.*)_CLUSTER$', 1, 1, null, 1) from dual
union all
SELECT regexp_substr('123U_CLUSTER', '^(.*)_CLUSTER$', 1, 1, null, 1) from dual;

yields

REGEXP_SUBST
------------
123R
123S
123T
123U

And to decipher the function

regexp_substr(
     '123T_CLUSTER',    -- source_field
     '^(.*)_CLUSTER$',  -- regular expression 
                        -- to capture all characters from start of
                        -- data up to "_CLUSTER"
     1,                 -- start looking at position 1 of string
     1,                 -- which occurance to return
     null,              -- used for match behaviour
     1)                 -- return what is in first set of parentheses