stripping leading and trailing blanks of a value?

Assume I have a XML doc where occasionally values appear with leading and trailing blanks similar to

<aaa>
<bbb>
'3 blanks'somevalue'4 blanks'
</bbb>
</aaa>

Because blanks are probably removed by the mailing list system I put 'n blanks' in the code above.
In reality real blanks are meant.

So how can I strip all such leading and trailing blanks (nad line breaks) with XSLT?

The stripping should applied only to "atomic" values. That means that blanks between e.g.
<aaa> and <bbb> should be kept. So the result should look like

<aaa>
<bbb>somevalue</bbb>
</aaa>

Moreover how can I strip all "atomic" blanks in a whole XML doc and not only from a particular tag value?

Ben