jjmccollum
commented
4 years ago Alternatively, much of the current structure can be maintained in a TEI-compliant manner if the <fs/> and <graph/> elements are placed under a <note/> element under the <app/>. This may be the simplest and least error-prone approach moving forward.
Presently, the populate_db.cpp script and the variation_unit class operate under the assumption that the
<graph/>element containing the local stemma for a variation unit is a child element of the<app/>element representing the corresponding variation unit. According to the TEI guidelines for the<app/>element (https://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-app.html), an<app/>can only have children in a very limited set of elements (e.g., notes, lemma or variant readings, or witnesses), so this hierarchy is invalid. The same is true of the<fs/>subelement containing a variation unit's connectivity feature.This can be resolved easily by requiring that the
<graph/>and<fs/>elements be moved to a position parallel to the corresponding<app/>elements and that they haveidornattributes matching that of the<app/>. As long as these elements can be matched this way, where the<graph/>and<fs/>elements are placed in the XML tree is immaterial; they could appear directly after the<app/>, if this is convenient, or they could be relegated to a position past the end of the text.This change in implementation would require code changes involving the XPath searches used in populate_db.cpp, modifications to the variation_unit and local_stemma constructors, and updates to the example XML files to reflect the new expected structure. Since this will make the library incompatible with earlier inputs, it should be handled on a dev branch and incorporated into the next major release; the code with the present input structure, while not strictly TEI-compliant, functions as expected.