We collect DMARC aggregate (rua) reports for our common mail domains, but the reports are XML and not human-readable. Typically organizations parse and stick them in a database, and query said db.
I took a look at open source scripts for doing this and did not find anything great, so I wrote something in python (see https://gerrit.wikimedia.org/r/#/c/163881/ ), which can process an mbox or accept input on a pipe.
Whether or not we use this ^^^ script, the architecture would be similar--the parser should run on a server which is not otherwise sensitive since (at least with python . . . https://docs.python.org/2/library/xml.html) there's some risk in processing XML, The database itself would be fairly small--within a few GB.