mathbib 0.7.10

A mathematics BibLaTeX bibliography manager.

MathBib

MathBib is a mathematics BibLaTeX bibliography manager.

WARNING: MathBib is currently in alpha state. The API and other implementation details may change substantially!

Installation and basic usage

First, ensure that mbib is installed with

pip install mathbib

This installs the mbib command-line interface. The first useful subcommand is mbib get. For instance, running

mbib get arxiv:1212.1873

will retrieve the arxiv record at https://arxiv.org/abs/1212.1873. If you check the output, you will notice that it actually returns

@article{arxiv:1212.1873,
  author = {Hochman, Michael},
  eprint = {1337.28015},
  eprinttype = {zbl},
  journal = {Ann. Math.},
  language = {English},
  number = {2},
  pages = {773--822},
  publisher = {Annals of Mathematics},
  title = {On self-similar sets with overlaps and inverse theorems for entropy},
  volume = {180},
  year = {2014}
}

This contains more information than what can be found on arxiv! In general mbib will search multiple locations for records to build the most updated citation possible.

Generating records from command-line arguments is nice, but a much more common use case is to generate records corresponding to a file. Suppose you have a file doc.tex containing the contents

\documentclass{article}
\usepackage{biblatex}
\addbibresource{doc.bib}
\begin{document}
This document references an article \cite{arxiv:1212.1873}.
It also contains more references \cite{zbl:1251.28008, doi:10.4007/annals.2019.189.2.1}.
\printbibliography
\end{document}

Simply run

mbib generate doc.tex --out doc.bib

to build the .bib file, and compile.

Of course, it can be somewhat inconvenient to remember a large number of citation keys, especially ones that are designed to be machine-readable. One solution is to create aliases. Let's define a few:

mbib alias add Hoc2014 arxiv:1212.1873
mbib alias add HocShm2012 zbl:1251.28008
mbib alias add Shm2019 doi:10.4007/annals.2019.189.2.1

We could now instead write our file as

\documentclass{article}
\usepackage{biblatex}
\addbibresource{doc.bib}
\begin{document}
This document references an article \cite{Hoc2014}.
It also contains more references \cite{HocShm2012, Shm2019}.
\printbibliography
\end{document}

Remember to regenerate the doc.bib file with mbib generate doc.tex --out doc.bib (the citation keys need to be updated), and the file will compile! Aliases also work as command line arguments anywhere a key:id is expected.

Suppose you have been working on the above TEX document for a while, and now you want to cite a specific theorem from the paper Hoc2014. Simply run

mbib file open Hoc2014

to open the PDF file associated with the resource. This will automatically download a PDF resource for the file (if mbib can find one online) and then open it. The PDF file is also saved locally, so the next time you run this command, you do not need an internet connection.

Of course, if mbib cannot find a resource for your file, you can also add one yourself.

mbib file add Hoc2014 path/to/file.pdf

Finally, if you have loaded a record but there are mistakes (for instance, improperly formatted LaTeX), you can apply local modifications which will always be applied when requesting the record. Edit the local record with, for example,

mbib file open zbl:1251.28008

There are also other features implemented: either read more below, or run mbib --help and mbib <subcommand> --help for more information.

Remote and local records

Internally, whenever you request a new record, MathBib searches a few online data repositories for the information associated with the record. Currently, the following KEY: types are supported

• local: (WARNING: not yet supported) internal local records
• zbl: zbMATH Identifiers
• doi: Digital Object Identifier
• mr: (WARNING: not yet supported) MathSciNet
• zbmath: zbMath Internal Identifier
• arxiv: arXiv Identifier
• isbn: International Standard Book Number.
• ol: Open Library resource identifiers.

The requests to the various sources are cached in $XDG_CACHE_HOME/mathbib. In particular, the first search will be slow, but afterwards the request will resolve quickly.

You can see all the information collected by MathBib by running, for instance,

mbib get -r json arxiv:1212.1873

Note that MathBib also has an internal notion of record priority. Local modifications of records have the highest priority, and otherwise the priority is precisely as specified in the list above. For instance, if you run mbib show arxiv:1212.1873, your web browser will open a record on zbMATH. If you view the corresponding JSON record, you will see the zbl: entry has the highest priority.

Useful scripts

The output of mbib list is designed to be machine-readable. For example, to make it easier to find keys associated with documents, I find the following script (written in Fish) useful:

function pa --description 'Search MathBib record keys using fzf'
    if count $argv > /dev/null
        set --function fzf_opts --query "$argv"
    end
    set --local captured (mbib list | fzf $fzf_opts | cut -d " " -f 1)

    if test -n "$captured"
        echo $captured
    end
end

For instance, you can use this command whenever a key is required:

mbib file open (pa)

will prompt you to choose a file name, and then open the file.

Contributing and future improvements

MathBib is still under active development! Some planned features include:

1. Rework the code for remote record searching and parsing.
2. Asynchronous requests for faster record downloading.
3. Nicer printing and viewing using rich.
4. Proper implementation of local records.