dev@javaserverfaces.java.net

Re: NumberConverter using buggy NumberFormat ?

From: Ryan Lubke <Ryan.Lubke_at_Sun.COM>
Date: Fri, 30 Nov 2007 10:53:56 -0800

Matthias Wessendorf wrote:
> Hi Ryan,
>
> I noticed some strange things, when using the NumberConverter for:
> -currency
> -fr_FR locale
>
> I guess it is a bug in the underlying JDK (1.5.0_11 I am using)........
>
> I did a quick JUnit test-case against the MyFaces API (which contains
> the *base* number-converter)
>
> protected void setUp() throws Exception
> {
> super.setUp();
>
> mock = new NumberConverter();
> mock.setLocale(Locale.FRANCE);
> FacesContext.getCurrentInstance().getViewRoot().setLocale(Locale.GERMANY);
> }
>
> public void testFranceLocale()
> {
> UIInput input = new UIInput();
> mock.setType("currency");
> Number number = (Number)
> mock.getAsObject(FacesContext.getCurrentInstance(), input, "12 345,68
> ");
> assertNotNull(number);
> }
>
> And............ it fails :-)
>
> So... what is the work-around?
> I assume it is not to not use fr_FR :-))
>
> Also, a simple Java-test fails and shows why:
>
> Doing this:
>
> String va = "12 345,68 ";
> NumberFormat nf = NumberFormat.getCurrencyInstance(Locale.FRANCE);
> Number n = (Number) nf.parseObject(va);
>
> and you'll see that n is NULL.
>
> Why?
> So, here it is:
> the String va contains two blanks (" "), which are between 2 and 3, and
> between 8 and as well.
>
> In fr_FR, however, the *grouping separator * is not " ", but it is a
> special char for blank (\u00a0).
> So, my little test will pass, when the first BLANK is replaced by the
> special char...
>
> I thought, that the NumberFormat actually does parse the object for me!!
> Looks like for the "fr_FR" locale, I have to create a *custom parser*...
> Which is odd, IMO
>
> Now, do this:
>
> String va1 = "12 345,68 ";
> NumberFormat nf = NumberFormat.getCurrencyInstance(Locale.FRANCE);
> String va2 = nf.format(12345.68));
> System.out.println(va1.equals(va2));
>
> and you see, what the issue from another side :-)
>
> Any ideas?
> (Or perhaps known bugs?)
>
> Thx,
> Matthias
>
>
Hi Mathias,

This looks pretty similar to http://bugs.sun.com/view_bug.do?bug_id=6318800