SQL & PL/SQL

Our Problem: Customers that have problems we can not recreate are asked to export their data as a dmp and we then download the dmp to investigate to import locally and establish were the problem is.
Some customers datasets are too big to download (3-4gb - i know this shouldnt be an issue but some customers have very slow connection speeds and this would take a lifetime) .
To get round this, they put onto a DVD and send us (which takes a couple of days).
The thing is, that these datasets are probably around 400mb with one table taking up the rest with blobs (and this table is unlikely to be the cause of the problem). So in effect we can exclude this from the export.

What we want to achieve:
There are potentially a few of these tables that could be large so we would like to produce a script to determine which tables are the largest on disk space. We will then (if the problem the customer experiences is nothing to do with these tables) not include these in the export on the DMP. Meaning we can get their data quicker and subsequently solve their issue in a shorter time frame.

Any help on where to start will be much appreciated. We use 10g db