- Rewrite MarkBuilder as something more reusable.
- Add file globbing.
- Complete more unit tests.
- Remove use of iText in favour of PDFBox (it does everything we need it to, while iText does a subset).
- Expand MarkBuilder or whatever replaces it to support query response for non-article DOIs.
- Expand unixref model to support non-article types.
- Implement scanning of PDF documents.