I too have this problem. Some of my indexed fields are actually the URL of a Word/PDF/HTML document. My solution is to add info the the key so that when you retrieve the results you can parse out the real key plus the extra info that reveals which fields.

E.g I index the rows of a database table, so I have one document with the key being the Primary key of that table (OID), and the various fields of the document representing the columns of the table.

Some of my columns correspond to URLs that lead to a Word/PDF or HTML external document. So in order to be able to identify whether the search found a match in the database row or in an external document, I can encode the Lucene document key to include the OID plus, say, #fieldName.

There's no reason why you can't have multiple Lucene documents representing a single searchable entity.

NOTE - if someone knows how to extract this info from Lucene's results in the first place, I'd love to hear about it as well.

Ian White