So I wrote a small tool to process the data from Discussions that you can get from the Canvas Data Portal. It does two things extract and save records from a date range and download the canvas stored files linked in those records.
You download the data files and add a header (documented in readme) and then run the tool against them. It is written in Ruby and uses the Pandarus gem to extract the display name of the file and the current down url. You will need to provide an admin level api token to run it.