We use the Feeds module to import the documents. We create a custom fetcher that checks a folder on a cron run and a custom parser to be able to use it for importing the transformed DITAMap XHTML. We also create a special GUID (or UUID) for books and topics to be able to update in place (just new or updated nodes, this should be composed of a map and a topic element ID). Each DITAMap contains metadata for the software version that also gets imported into Drupal, this can then be used in a faceted search interface to differentiate between software versions. Currently the XHTML maps import is controlled by a 2 step import workflow, where the first step is to recognize and recreate the map structure in the Drupal Book node framework, then the second step is to populate the created nodes with the DITA specific content, including rich formatted text, URLs, internal link references, embedded images and tags.

Mass publication is a new feature that allows content editors to bulk upload content by uploading the XHTML output in a zip file. Single maps or multiple maps can be uploaded at a time. This semi-automatized solution will unzip the files on the web server to a ‘watched’ folder, and upon the next cron runs the importer will fetch and process the files by creating and/or updating the content with the latest changes. The use of the UUIDs will make sure that only the respective nodes will be updated.
The file upload interface is available for authenticated users with a special permission.

To make sure that XHTML outputs will work fine with the importer, we set up a validation process to check the structure of the XHTML, the cross-references between the index.html file and the rest of the HTML files, as well as the referenced images. In case of any inconsistencies, the validation will stop the import process and send an email report to the editorial team with the discovered issues.

To preserve cross references between DITA topic maps, a script scans through all the existing x-refs and translates DITA links to Drupal links, so we can make sure that all of the linking structure implemented in DITA gets translated over to Drupal correctly. Basically, we ensure links work within a book and across multiple books.

We automatically process images found in your DITA content, and we store them as actual image assets in the Drupal files directory. The filter makes sure that all image references are mapped to the content and all image files are embedded properly. This allows us, if necessary, to further customise the image experience (e.g. image processing to improve site performance, responsive images for mobile users, etc.).