I haven't used the Apache SAX parser, but here is a general scheme that should work for any SAX parser.
The ElementStart handler in your parser looks for the field2 element. When it's found, it installs new start element, end element, and characters handlers. The new start element handler spits out the name and attribute/value pairs within angle brackets. The new characters handler spits out the text value. The new end element handler spits out the closing tag, unless it is for field2, in which case it swaps back in all the old handlers.