Ok, from your clariification it sounds like you have a text database whose format was never really decided upon. I guess that a number of different ad hoc formats got appended.

Your best approach will probably be to try to characterize the different formats with regular expressions. Fixed length formats can be decoded with unpack, and delimited fields with split. It will take a certain amount of eyeball inspection to know you have detected all the formats. The trouble with ad hoc formats is that they usually are inadequate.

Once you can detect and decode the whole file, use one of the CSV modules to write to a single, capable, standard.

After Compline,
Zaxo