A set of tables encoding USFM 3 markers from https://ubsicap.github.io/usfm (forked here as https://github.com/Freely-Given-org/usfm). See also https://docs.usfm.bible.
Any basic Bible software needs to be able to iterate through all the books of the Bible. For Western Protestants, this is typically 66 books consisting of what's called in English, the Old Testament and the New Testament.
Thus many starting programmers write code to iterate through the integers 1..66. But this fails if you go to add the original manuscripts where, of course, the Hebrew only has books 1..39 and the Greek has books 40..66. It can also fail if you want to view a translation that's incomplete (not all books are there yet so you need to skip that book number), and/or is from a different culture (e.g., includes the Laodiceans letter or more which you don't have a number for, and that and many of the other books occur inside the numbers 1..66, e.g., using say 39a, 39b, etc. for between-Testament books).
Then if you use USFM files, they often have a three-character book code (e.g., "2CO") in the filename, plus they usually have a two-character number where Matthew is "41". Note that that's different from your 40 above. Note also that that number can also contain letters (e.g., "A0" for front matter, or "C3" for Letter to the Laodiceans)!
Then you decide to include some OSIS XML files. They can contain two to seven characters like "Ps", "2Cor", and "AddEsth" and are a variant of the SBL abbreviations with any spaces removed.
So now you need to map between two different numbering systems as well as two different book abbreviation systems. The data table in this repository provides our own much wider book code and numbering system, and then provides conversions to other systems, including SBL, OSIS, Sword, CCEL, USFM, USX, Unbound Bible, Bibledit, NETBible, DrupalBible, BibleWorks, and the Robinson-Pierpont Byzantine Greek text books codes.
The source file for the data table is XML. Although not our favourite format, it does allow commenting unlike JSON, and it has good schema support.
JSON, however, is more popular these days, and so the XML data table is converted into multiple JSON Objects (all within a single JSON file). So too with pickled Python dictionaries and more formats to come.
Python, Rust, and Javascript utility packages are/will-be supplied and available on package managers like PyPI, Crates.io and NpmJs.