Question

I have a text field that is a date, although sometimes it also has the time component. Ex:

"23/1/1999"
"25/2/2003 05:18:00"

Now, i simply want to query them and order by the date component only. My attempt

TO_DATE(
       TO_CHAR(
           TO_DATE(mytextfield  , 'DD-MM-YYYY HH:MI:SS')
       ,'DD-MM-YYYY')
,'MON-YYYY')

The above fails on the outer TO_DATE (ORA-01843 invalid month.)

TO_CHAR(
    TO_DATE(mytextfield   , 'DD-MM-YYYY HH:MI:SS')
,'DD-MM-YYYY')

The above works, but i’m left with a string, not a date.

——– Solution using accepted answer

Just use

EXTRACT(MONTH FROM  TO_DATE(timetable.string_time)  , 'DD-MM-YYYY HH24:MI:SS'))

EXTRACT(YEAR FROM   TO_DATE(timetable.string_time)  , 'DD-MM-YYYY HH24:MI:SS'))

As stated on the answer, you can swap the order you use it in the ORDER BY and in the SELECT.

Will need a little number formatting on the month, and some concats, but this works very well.

Answer 1

NLS_DATE_FORMAT specifies the default date format to use with the TO_CHAR and TO_DATE functions. Let’s change it to get nice output from SQL*Plus:

ALTER SESSION SET NLS_DATE_FORMAT = 'YYYY-MM-DD HH24:MI:SS';

First we use TO_DATE to convert the textual representation of date into real date type (note that the date format below accepts the shorter text):

SELECT TO_DATE('25/2/2003 05:18:00', 'DD-MM-YYYY HH24:MI:SS') FROM DUAL;
2003-02-25 05:18:00

SELECT TO_DATE('25/2/2003', 'DD-MM-YYYY HH24:MI:SS') FROM DUAL;
2003-02-25 00:00:00

Second – we want to truncate the date to get rid of unnecessary fields (TRUNC function):

SELECT TRUNC(TO_DATE('25/2/2003 05:18:00', 'DD-MM-YYYY HH24:MI:SS')) FROM DUAL;
2003-02-25 00:00:00

SELECT TRUNC(TO_DATE('25/2/2003 05:18:00', 'DD-MM-YYYY HH24:MI:SS'),'MONTH') FROM DUAL;
2003-02-01 00:00:00

Now you can use this in ORDER BY or GROUP BY provided that you want the ordering based on year/month/day (and so on).

If you want to do some exotic sorting (for grouping you don’t care: {year,month} isn’t any different from {month, year})- e.g. month/year, then I think you need to compose the individual fields and use the following:

SELECT EXTRACT(MONTH FROM DATE '1998-03-07') FROM DUAL;
3

SELECT EXTRACT(YEAR FROM DATE '1998-03-07') FROM DUAL;
1998