Monthly Archives: April 2015

New open-source file validation project

The VeraPDF Consortium has announced that it has begun the prototyping phase for a new open-source validator of PDF/A. This is a piece of the PREFORMA (PREservation FORMAts) project; other branches will cover TIFF and audio-visual formats. Participants in VeraPDF are the Open Preservation Foundation, the PDF Association, the Digital Preservation Coalition, Dual Lab, and Keep Solutions.

Documents are available, including a functional and technical specification. It aims at being the “definitive” tool for determining if a PDF document conforms to the ISO 19005 requirements. It will separate the PDF parser from the higher-level validation, so a different parser can be plugged in.

Validating PDF is tough In JHOVE, I designed PDF/A validation as an afterthought to the PDF module. PDF/A requirements affect every level of the implementation, so that approach led to problems that never entirely went away. Making PDF/A validation a primary goal should help greatly, but having it sit on top of and independent from the PDF parser may introduce another form of the same problem.

PDF files can include components which are outside the spec, and PDF/A-3 permits their inclusion. This means that really validating PDF/A-3 is an open-ended task. Even in the earlier version of PDF/A, not everything that can be put into a file is covered by the PDF specification per se. The specification addresses this by providing for extensibility; add-ons can address these aspects as desired. In particular, the core validator won’t attempt thorough validation of fonts.

A Metadata Fixer will not just check documents for conformance, but in some cases will perform the necessary fixes to make a file PDF/A compliant.

JHOVE ignores the content streams, focusing only on the structure, so it could report a thoroughly broken file as well-formed and valid. JHOVE2 doesn’t list PDF in its modules. Analyzing the content stream data is a big task. In general, the project looks hugely ambitious, and not every ambitious digital preservation project has reached a successful end. If this one does, it will be a wonderful accomplishment.

Update on the JHOVE handover

There’s a brief piece by Becky McGuinness in D-Lib Magazine on the handover of JHOVE to the Open Preservation Foundation. It describes upcoming plans:

During March the OPF will be working with Portico and other members to complete the transfer of JHOVE to its new home. The latest code base will move to the OPF GitHub organisation page. All documentation, source code files, and full change history will be publicly available, alongside other OPF supported software projects, including JHOVE2, Fido, jpylyzer, and the SCAPE project tools.

Once the initial transfer is complete the next step will be to set up a continuous integration (CI) build on Travis, an online CI service that’s integrated with GitHub. This will ensure that all new code submissions are built and tested publicly and automatically, including all external pull requests. This will establish a firm foundation for future changes based on agile software development best practises.

With this foundation in place OPF will test and incorporate JHOVE fixes from the community into the new project. Several OPF members have already developed fixes based on their own automated processes, which they will be releasing to the community. Working as a group these fixes will be examined and tested methodically. At the same time the OPF’s priority will be to produce a Debian package that can be downloaded and installed from its apt repository.

Following the transfer OPF will gather requirements from its members and the wider digital preservation community. The OPF aims to establish and oversee a self-sustaining community around JHOVE that will take these requirements forward, carrying out roadmapping exercises for future development and maintenance. The OPF will also assess the need for specific training and support material for JHOVE such as documentation and online or virtual machine demonstrators.

It’s great to know that JHOVE still has a future a decade after its birth, but what boggles my mind is the next sentence:

The transfer of JHOVE is supported by its creators and developers: Harvard Library, Portico, the California Digital Library, and Gary McGath.

I never expected to see my name in a list like that!