ldap question

Hi,

I have an LDAP directory with a very large number of records (some possibly
duplicated in their entirety or partially as superset/subset) which I would like to 
condense and repair and correct insofar as the individual subrecords/fields 
within each record are concerned.

The format of this LDAP directory, when dumped, is a set of millions of rows
of data which when sorted and uniqued on the cname results in a small fraction of
the original total (.00746% to be exact), though whether the duplicates themselves
have same fields is another matter entirely.

The records are in XML format and consist of key/value pairs.

My suspicion is that this directory has never been properly maintained so I have some questions:

  + what are the accepted ways via automation to maintain this directory

  + what methods or code exist to condense and verify a hitherto-unmaintained LDAP directory

  + simplifying the data to the bare bones number of records, discarding the others after
     making a general full backup

So what I am asking for is a general set of already written Perl tools using Net::LDAP which deal
with LDAP directories intelligently.

--Stuart