The CURSUS project was as much an investigation into the technologies appropriate to the digitisation and encoding of medieval liturgical manuscripts as about creating the resource itself. The project quickliy realised the benefits of XML encoding for the structuring and recording of manuscript editions.
Almost of our working files are encoded in XML. The Text Encoding Initiative (TEI) Guidelines and DTD were used. However, a number of extensions to the TEI DTD were created using the extension methods provided. This allowed us to use a more specific taxonomy and included division elements for liturgical days, and services, but also textual elements to structure antiphons, responds, prayers, and rubrics, amongst many others.
The majority of encoding has been completed with the EMACS editor with psgml and nxml modes. This allowed a set of project-specific tools, written in elisp, to be created that facillitated data entry and encoding. These include an extensive set of short-cuts for inserting templates of frequently-used component elements, routines to automate the insertion of cross-references between the project files, and various specific validity-checking devices beyond the generalised ones included with psgml. For more information see our EMACS page.
The CURSUS files that form the core of the content delivered on the website are the manuscript editions, and our repository of antiphons, responds and prayers (both from CAO and our manuscripts). Every full antiphon, respond, or prayer in our manuscript files is extracted from this repository when the manuscript editions are generated. All of the biblical readings for these editions are also plucked out of a standardised XML version of the Vulgate we created for this purpose. As with almost all of the transformations of CURSUS data this is done through the use of extensible stylesheet language transformations (XSLT).
The processing transformations that take place include: bursting our quite large repository into individual files for each antiphon, respond or prayer; pre-generation of html versions of our manuscript files incorporating individual repository items and Vulgate readings; and the creation of our alphabeticised incipit lists of those antiphons, responds and prayers in our repository.
The website is mainly run through Pycoon, a Python-based XML web publishing framework which allows (amongst many other things) the dynamic pipelining of XSLT transformations. So as a user clicks on a particular URL, an xml source source may be transformed a number of times (and/or aggregated from a number of sources) before final XSLT stylesheet adds on stylistic and navigational aids such as our header, sidebar and footer. All of this is invisible and (given network connectivity and not too much of a strain on the old machine runing it) fairly swiftly completed. This kind of on-the-fly transformation is used for the critical editions and individual readings of our repository items, as well as for pages such as this one.
If you have any other questions about the technology behind the CURSUS project feel free to contact James Cummings via twitter.