Digital preservation and active software stewardship are both cycli- cal processes. While digital preservation strategies have to be reevaluated regularly to ensure that they still meet technological and organizational requirements, software needs to be tested with every new release to ensure that it functions correctly. JHOVE is an open source format validation tool which plays a central role in many digital preservation workflows and the PDF module is one of its most important features. Unlike tools such as Adobe PreFlight or veraPDF which check against requirements at profile level, JHOVE’s PDF-module is the only tool that can validate the syntax and structure of PDF files. Despite JHOVE’s widespread and long-standing adoption, the underlying validation rules are not formally or thoroughly tested, leading to bugs going undetected for a long time. Furthermore, there is no ground-truth data set which can be used to understand and test PDF validation at the structural level. The authors present a corpus of light-weight files designed to test the validation criteria of JHOVE’s PDF module against “well-formedness”. We conclude by measuring the code coverage of the test corpus within JHOVE PDF validation and by feeding detected inconsistencies of the PDF-module back into the open source development process.

application/pdf