A modern BibTex implementation

This repository contains (or will contain, eventually, hopefully) a reimplementation of the BibTex, the venerable citation manager.

Eventually, hopefully?

Yes. So far, we just have a mostly-complete BibTex parser. There's a lot more left to do (mainly, a .bst-file interpreter).

This repository does not currently contain any binaries: it consists only of rust-language libraries ("crates", in rust-speak). Eventually, our goal is to have a drop-in replacement for the bibtex binary, along with some other tools.

Why reimplement BibTex?

If you use TeX or LaTeX on a regular basis, BibTex is everywhere. You use it because your co-authors do, or because Google Scholar or mathscinet export references in BibTex format. But BibTex is also extremely, ahem, venerable. Its only interface is a command-line batch program, and so integration into other workflows is achieved by shelling out and parsing the error messages. Wouldn't it be great if there were a modern BibTex engine that was modular enough to embed it into an editor (with support for accurate syntax highlighting and incremental execution), while retaining 100% compatibility with old databases? This repository doesn't contain that modern engine yet, but maybe one day it will.

Why reimplement BibTex again?

There are plenty of other BibTex parsers around. Thirty seconds of searching found two (here and here) implementations just in rust. There are two things about this implementation that are a little bit special, though. Firstly, we aim for compatibility with BibTex: we will accept exactly the same files and report errors in exactly the same locations. This is a little more tricky than it sounds, because BibTex doesn't have a formally specified grammar. Secondly, we (eventually, hopefully) aim to provide a full implementation, including a parser for .aux files and an interpreter for .bst files. Most reimplementations of BibTex just parse .bib files.

Known incompatiblities

This section lists situations where we our behavior intentionally differs from that of BibTex. All of these concern invalid input.

• When we encounter an I/O error reading the input, we report an error. BibTex treats an I/O error the same as it does an EOF.
• When we, as part of an error message, display part of the input file, our display will match the input file more closely than BibTex's. This is because BibTex sometimes changes parts of the input to lowercase before printing the error message.
• BibTex has (in my opinion) some bugs while parsing names containing unbalanced braces. For example, in the string "{First} } {Last", it declares the first name to be "{First} }" instead of "{First}". We don't replicate this behavior (which is quite hard to hit, anyway, because the .bib parser won't let unbalanced braces through).