Hi Lisa!
This is precisely what we have done (and are continuing to do). There are a few different ways to get data from one to the other, and we started out with a slightly more complex method (perhaps an understatement) way back before the days of csv importing. The attached diagram shows all the sources we are moving into AtoM, where each of the parenthetical statements identifies the source - those identified as DB are InMagic textbases. Our main method was to export from InMagic into XML, conduct a number of cleanup procedures, then run a PERL script to connect previously unconnected data from the Fonds level (turquoise in the diagram) down through the xls inventories to the File/Box DB (item-level data - brown in the diagram) - the script links three separate sources and produces EAD records. For further details on this exact procedure, you can take a look at the second document I have attached. Depending on the structure of your original data, it may be possible to adjust the scripts we wrote to use in your case (and I'm happy to share)...but I suspect you might have better success with the second option.
This second option would be to export your InMagic data to Delimited ASCII Format, import this data into Excel, and adjust each field (column) in preparation for csv import. You could probably automate these adjustments in some fashion using Google Refine, or even with some of the concatenation or search/replace features in Excel (depending on how you need to massage the data for import). Using the csv templates as a guide (
https://wiki.accesstomemory.org/Resources/CSV_templates), it should be fairly easy to map out the crosswalks from your InMagic field names to the AtoM fields.
I hope this helps!
Jeremy
---------------------------------
Jeremy Heil
Digital and Private Records Archivist, Queen's University