Docovery: toward generic automatic document recovery

Abstract

Application crashes and errors that occur while loading a document are one of the most visible defects of consumer software. While documents become corrupted in various ways---from storage media failures to incompatibility across applications to malicious modifications---the underlying reason they fail to load in a certain application is that their contents cause the application logic to exercise an uncommon execution path which the software was not designed to handle, or which was not properly tested.