SimpleDateFormat can parse 2009102x

Question

Hello,
If you look to the piece of code below, it seems my users can enter 2009102w.
However, when one tries to enter 20091w20, it throws an error.
I don't understand why an incorrect character in my date can get my date parsed?!
Is this a bug or is there a reason?
          try {
                  DateFormat formatter = new SimpleDateFormat("yyyyMMdd");
                  //this is ok
                  //Date date = (Date)formatter.parse("20091020");
                  //this is also ok !!!!!
                  //Date date = (Date)formatter.parse("2009102w");
                  //this is not ok
                  //Date date = (Date)formatter.parse("20091w20");
                  System.out.println(date);
              } catch (ParseException e) {
                   System.out.println("exception");
              }Thanks

Answer 1

The date formatter doesn't feel the need to parse all the characters you give it. If it reads an acceptable date it will ignore any garbage following. The Formatter objects in java.text seem to have been designed to parse a string with multiple elements, dates, numbers, etc., each taking it's own chunk. You'll find the same with the NumberFormat objects.
Also, by default, the parsing mode on the data formatter is set to "lenient". If you want to tighten it up call setLenient(false) on the formatter. This is mostly about things like if you put 31/02/2009 then with lenient on you'll get 03/03/2009, but may affect the acceptance of a single digit day.
P.S. if you really want to be strict about validating the format then check with a regexp first.
Edited by: malcolmmc on Sep 8, 2009 8:59 AM