Parsing reports.

Hello!

I have many kinds of text reports that are output of some legacy programs.
These are tabular reports: titles, headers, footers, pages, details, 
totals, subtotals, an so on.

My task is to transform these reports into a structured data, and then feed 
these data to
a contemporary report engine.

I'm trying to understand now, if it's worth to approach to the task from 
the xslt side.
Technically the task is solvable, however this would mean creation and 
support of many such 
report parsers within xslt.

If anybody have already dealt with similar task?
Any advices?
Any xslt support libraries?
Is it worth to deal with xslt here?
--
Vladimir Nesterovsky
http://www.nesterovsky-bros.com/