Character data

Hi,

data that looks like this in a XML file:

<foo>
1234 4567 891011
</foo>

is sometimes passed through a character data callback as this:
1234 45

and then in a second call:
67 891011

As this integer data has no fixed size, there seems to be no obvious way to
know whether the XML file contained these 3 numbers:
1234 4567 891011

or these these 4 numbers:
1234 45 67 891011

The only way I see to go around this issue is to add some special 
character (a i in this example), to mark where each number actually 
ends, for example:

1234i 4567i 891011i

could be passed in the callback as:
1234i 45

and then in another call:
67i 891011i

so I could see that 45 and 67i belong to the same number. However, this 
seems a bit ugly... what about XML formats that may include large blocks 
of numerical data (for example X3D...)?? is there a better solution? 
what am I missing here?

Thanks a lot for your help!
Carlos Pereira