linglit's data extraction capabilities depend on typographic conventions used in the source
publications. Thus, parsing may break when these conventions change. Since you may be the first
person running linglit on a brand new Glossa paper or Language Science Press book, you may also
be the first one to encounter such breakage.
Let us know.
If you know of a linguistic publication (a journal or an archive for example) which publishes
- IGT examples and/or bibliographies or reference lists
- under an open license
- in a format that is structured enough to allow for automated extraction
Due to the diverity of typographic conventions to indicate an IGT example's language
(for both, glossa and langsci), detecting this language is a sketchy and imprecise process.
To guide this process, linglit can make use of extra information, though. For langsci books
on individual languages (typically grammars or dictionaries), this language (if indicated in the
langsci catalog) is taken as default. But sometimes, books have individual chapters devoted to a
single language. Such cases can be indicated in
the linglit configuration, mapping filenames of the
LaTeX files containing the chapter to a language name.
A similar mechanism exists for glossa, mapping glossa article
numbers to example languages specified by Glottocode.
If you find Glossa papers or Language Science Press chapters missing in these lists, let us know.
- Fork
cldf/linglit - Clone your fork
- Install
linglitfor development (preferably in a separate virtual environment) runningpip install -r requirements.txt
- Run the tests
pytest