I haven't used dom4j so someone else can probably give you
better advice than I can, but in case nobody else jumps in, here
are some ideas.

SAX is a parser.  The concept behind it is that it makes a pass through
an XML document looking for standard types of objects - the start of
an element, end of an element, a text node, an attribute, etc.  You use it
by registering event handlers (references to callback routines) to be called
when the objects of interest to you are recognized by the parser.

SAX is a reasonable interface to use if you are going to read an XML
document and do something with the data - possibly create a new XML
document, or possibly just do things with the data you've read.

However if you want to read in an XML document, update it, and
write it out again, SAX is probably NOT the interface you want to
use.  The DOM ("document object model") interface is designed to
allow that.

SAX makes a single pass through a document, telling you what it finds.

DOM makes a pass through the document, builds an in-memory tree
of nodes, and gives you a reference to the top of the tree.  You can
then walk through it, reading nodes, text and attributes, and adding
deleting, or replacing nodes, text and attributes.  When you're done,
you "serialize" the tree, converting it back into a stream of characters
with "<" and ">" delimiters, etc.  Then you can write that to whatever
output sink you're interested in.

The DOM interface, a W3C standard, is complex and not always very
intuitive.  There is a learning curve to be climbed.  But it's useful and
ubiquitous.  You can find DOM implementations for Java, C++, Perl,
Python, and who knows what else.

Looking at the API docs for dom4j, they're mighty skimpy.  But
since DOM is a standard, you may be able to find documentation
at W3C, xml4c, or other places that will shed additional light on what
you see in dom4j.

Hope that helps.

   Alan