Oracle 'printf' equivalent

Question

Is there an equivalent or alternative to the following?

SELECT mix_type || ' (' || mix_num || ')' as description
  FROM acid_batch
 WHERE mix_num < 10

Does Oracle have something like printf style formatting?

SELECT printf("%s (%s)", mix_type, mix_num) as description,
  FROM acid_batch
 WHERE mix_num < 10

Answer 1

Just another idea for you: I've found REPLACE to be useful for this kind of thing, especially when the template is complex:

SELECT REPLACE(REPLACE(
        '%mix_type% (%mix_num%)' /*template*/
       ,'%mix_type%', mix_type)
       ,'%mix_num%' , mix_num ) as description,
FROM   acid_batch
WHERE  mix_num < 10

The only downside is you need to add as many REPLACE('s as there are variables to replace - but at least you only need to have one per variable, regardless of how many times it appears in the template.

(NOTE: There is no particular significance to using "%" as a delimiter, it's just a personal convention of mine - you might choose a different pattern, e.g. <mix_type> or [mix_type])

For this particular instance it looks like overkill, but in some cases it can make things much easier, e.g.:

template := 'bla bla %a% %b% %a%';
output := REPLACE(REPLACE(template
    ,'%a%', some_complex_expression)
    ,'%b%', b);

Compare the above with:

output := 'bla bla ' || some_complex_expression || ' ' || b || ' ' || some_complex_expression;

Answer 2

You can resolve it in the select.

SELECT mix_type || '(' ||  mix_num || ')' as description,
FROM acid_batch
WHERE mix_num < 10

you should also take a look at the functions

to_char

to_date

to_number

as they give your a finer granularity on how you want the things represented.

Answer 3

No there are no built-in Oracle functions that apply a formatting string in this fashion. Although it would be easy to write a custom function for this specific example, writing a PL/SQL-based implementation of printf would be challenging.

If you have a frequent need for this, perhaps you could write an Oracle function that wraps a Java call for a richer string handling environment.

Answer 4

The closest standard approximation to printf for Oracle I can think of is utl_lms.format_message. However, it won't work in SQL statements, that is, this is ok:

begin
  dbms_output.put_line(
    utl_lms.format_message('hello %s, the number is %d', 'world', 42)
  );
end;
/

but this gives a ORA-00902: invalid datatype error:

select utl_lms.format_message('hello %s, the number is %d', 'world', 42)
  from dual

Answer 5

I've made a simple template engine named ora_te (on GitHub) for Oracle SQL / PLSQL. With the help of it your goal can be achieved in the following ways:

Noneffective implementation with multiple parsings of template string:

with acid_batch as (
  select rownum as mix_type, rownum + 2 as mix_num 
  from all_objects
  where rownum < 10
)
--
SELECT pk_te.substitute('$1 ($2)', ty_p( mix_type, mix_num ) ) as description
FROM acid_batch
WHERE mix_num < 10;

An effective implementation with one time compilation (parsing):

with acid_batch as (
  select rownum as mix_type, rownum + 2 as mix_num 
  from all_objects
  where rownum < 10
),
--
o as ( 
  select ty_te.compile_numbered( '$1 ($2)' ) te from dual
)
SELECT pk_te.substitute( o.te, ty_p( mix_type, mix_num ) ) as description
FROM acid_batch, o
WHERE mix_num < 10;

BTW it also supports named placeholders.