Speaker
Description
CIDEHUS Digital seeks to share information generated by ongoing research projects, already completed, historical cartography and other datasets, while providing the user the ability to carry out advanced search, faceting, extract information through an API, among other features. The heterogeneity of data implied a technical solution that allowed full and dynamic data visualization of text, raster and vector formats, but also spatial queries and advanced searches.
Among the datasets available, stands out “Ophir'', edited by the extinct “National Commission for the Commemorations of the Portuguese Discoveries”, from the end of the 1990s onward. This collection, with the subtitle “Virtual Library of the Portuguese Discoveries”, was created aiming Windows 95, and included some datasets with major importance for research and some are now difficult or impossible to use for the majority of the users. Part of our challenge was to maintain the possibility of viewing information that specialized users would like to continue consulting, and, therefore, the recovery process implied that the software ran virtualized (Oracle VirtualBox) for observation and annotation of the tasks to be performed. Interestingly, some software kept the original protections not allowing copying and only works on alternative OS, namely through Linux (Wine).
On a second step, all the extracted data was processed through Python scripts to transform the text into HTML, enriching it with the original footnotes and to recreate in web format the links present in the original versions of the software. But also the conversion of proprietary formats and aiming to recreate the relations present in the databases.
The ambition to provide users with a dynamic application that would allow efficient loading with different views and the move to a Progressive Web APP (PWA), led us on a first stage to a MEVN stack implementation (MongoDB, Express.js, Vue.js, Node.js). The major disadvantage of the MEVN option was the rapid evolution of the libraries associated with Vue.js, this meant that the release of several versions with breaking changes implied rewriting working code in each release iteration.
In name of long-term sustainability of the code and ensure that the information will be available and accessible, we are now exploring a new solution using a headless CMS (Directus) associated with a static site generator (Eleventy), to create HTML5 enriched versions with Microdata and JSON-LD.
