bibfish 0.2.0

Extract entries from a .bib file that are cited in a .tex file

bibfish

bibfish generates a local BibTeX file from a central BibTeX database based on the citations found in a LaTeX file. This is useful if you want to maintain a single master BibTeX file (or several) and automatically generate a separate, independent BibTeX file for each manuscript you're working on. It is similar to bibexport and makebib, except bibfish is not dependent on any particular LaTeX tool and is therefore agnostic about your choice of bibliographic software (BibTeX vs. BibLaTeX, etc) or general typesetting pipeline.

Installation

bibfish is written in Python and can be installed using pip:

pip install bibfish

Usage

Once installed, bibfish may be used from the command line like this:

bibfish manuscript.tex ~/master.bib references.bib

By default, bibfish will not overwrite a local .bib file if it already exists. To override this behavior, use the -f option:

bibfish -f manuscript.tex ~/master.bib references.bib

By default, bibfish searches your manuscript for \cite{}, \citet{}, and \citep{}. If you are using a different set of cite commands, you can specify them with the --cc option:

bibfish --cc "textcite,parencite,possessivecite" manuscript.tex ~/master.bib references.bib

bibfish can also be used as an integral part of your typesetting procedure, with the following intended usage pattern.

1. Start a new LaTeX document and, when setting up the bibliography, point it to e.g. references.bib (no need to create this file; it will be generated automatically). For example, depending on how you set things up, you might have a line like \bibliography{references.bib} or \addbibresource{references.bib}.
2. Add any BibTeX entries you want to cite to your master.bib (stored e.g. in your home directory).
3. Cite some BibTeX entries in your LaTeX document using their citekeys as normal.
4. Before typesetting, run bibfish first; this will fish out the relevant entries from master.bib and place them in references.bib.
5. Continue with the rest of your typesetting procedure, e.g. run pdflatex, latex, xelatex, bibtex, biber, dvipdf, or whatever else you normally do in your pipeline.

For example, you might create a typesetting script like this:

#!/bin/bash

bibfish -f manuscript.tex ~/master.bib references.bib
latex manuscript.tex
bibtex manuscript.aux
latex manuscript.tex
dvipdfm manuscript.dvi

Each time you run this script, bibfish will search manuscript.tex for citekeys, extract the relevant entries from ~/master.bib, and write them out to references.bib, allowing the rest of the typesetting process to proceed as normal.

The benefit of this is that your LaTeX document does not need to have any dependence on or reference to ~/master.bib. This means you can maintain a single master.bib, while also maintaining each manuscript as its own independent self-contained package. You could, for example, send manuscript.tex and references.bib to a coauthor or publisher without needing to supply your entire master.bib, and manuscript.tex and references.bib can be kept under version control without any connection to master.bib.

Caveats

bibfish uses the BibtexParser library to parse the .bib files and we think it is configured in a relatively permissive fashion. If you have trouble with bibfish parsing your .bib files, please post an Issue so we can investigate it.

License

bibfish is licensed under the terms of the MIT License.