Check me out on github.com/dgerosa/filltex.
filltex
is a simple tool to fill LaTex reference lists with records from the ADS and INSPIRE databases. ADS and INSPIRE are the most common databases used among the astronomy and theoretical physics scientific communities, respectively. filltex
automatically looks for all citation labels present in a tex document and, by means of web-scraping, downloads all the required citation records from either of the two databases. filltex
significantly speeds up the LaTex scientific writing workflow, as all required actions (compile the tex file, fill the bibliography, compile the bibliography, compile the tex file again) are automated in a single command. We also provide an integration of filltex
for the macos LaTex editor TexShop.
If you use filltex
for your research, please drop a citation to this paper:
- filltex: Automatic queries to ADS and INSPIRE databases to fill LaTex bibliography, Davide Gerosa, Michele Vallisneri, The Journal of Open Source Software 2 (2017) 13.
Of course, you can use filltex
to cite filltex
! Just put \cite{2017JOSS....2..222G}
in your tex file!
Installation
filltex
can be installed from the python package index Pypi:
pip install filltex
If you're a TexShop user and want to use this feature, run
filltex install-texshop
If you want to give it a try, you can run it on the example.tex
file provided in this repository:
git clone https://github.com/dgerosa/filltex.git
cd filltex/example
filltex example
and you should get a filled .bib
file and a finished .pdf
.
What's about?
What happens when you compile a LaTex file? How's bibliography handled?
- Run
pdflatex
and all requested citation keys are dumped into a.aux
file. - You should have the required entries in you
.bib
file. - Run
bibtex
, which looks for citations inside the.bib
file and writes the results into a.bbl
. - Run
pdflatex
again, which processes the.bbl
into the compiled paper, and dumps the citation records into.aux
. - Finally run
pdflatex
again, which puts the correct citation records into the paper.
The commands you need to run are: pdflatex
, bibtex
, pdflatex
, pdflatex
. These, of course can be put into a script or a makefile and done in one goal.
filltex
is meant to automatically solve the second point as well: look for citations on ADS, INSPIRE or both.
So, here is the deal:
- The
fillbib
python script queries both databases and creates/updates a.bib
file without getting each record manually. - The
filltex
bash script puts everything together to go from a.tex
(and no.bib
) into a.pdf
. - I also provide TexShop engines for mac users
Of course, all of this works if your citations are specified in the ADS or INSPIRE format, e.g. \cite{2016PhRvL.116f1102A}
, \cite{Abbott:2016blz}
. If you use your personal keys \cite{amazing_paper}
there's no way to get them from a database.
Usage
fillbib (script)
fillbib
has two working modes. It can either look for citations into a .aux
file and create/update a bibtex file with the records found on ADS and INSPIRE, or it can fetch a list of bibtex entries specified from the command line from ADS or INSPIRE.
usage: fillbib.py [-h] [--generate] [--journal_arXiv_fallback]
[--max-num-authors MAX_NUM_AUTHORS]
[--num-authors-short NUM_AUTHORS_SHORT]
{tex,list} ...
positional arguments:
{tex,list} Subcommands
tex Create a bibliography for a tex document
list Create a bibliography given a list of ADS/iNSPIRE keys
optional arguments:
-h, --help show this help message and exit
--generate Generate the BibTeX entries from the metadata (this is
useful to customize the generated BiBTeX file)
--journal_arXiv_fallback
Set the journal entry to be arXiv for unpublished
preprints (iNSPIRE entries only, requres --generate)
--max-num-authors MAX_NUM_AUTHORS
Include at most this many authors for each bibtex
entry(iNSPIRE entries only, requres --generate)
--num-authors-short NUM_AUTHORS_SHORT
Number of authors to list if the number of authors is
larger than max_num_authors(iNSPIRE entries only,
defaults to max-num-authors, requres --generate)
The first argument specifies the subcommand to run.
tex
will produce a bibtex file given an.aux
filelist
will print a bibtex file given a list of keys from CLI
The help for the two subcommands can be obtained with
fillbib.py {tex,list} --help
When working in tex
mode it is possible to specify the name of the bibtex file using the option --bibtex
. Otherwise, the code will scan the .aux
file to guess the name of your bibliography file. Arguments can be typed with or without extension, and the script is smart enough to figure it out. You need to have .aux
file already, not just the .tex
. If you don't have it, run pdflatex
once.
fillbib.py
contains two short unit tests, to make sure the web-scarping part is done correctly. You can run them from the filltex
directiory using
python
> import fillbib
> fillbib.test_ads()
> fillbib.test_inspire()
or simply using pytest
pytest fillbib
fillbib
supports both python 2 (2.6 or higher) and python 3.
filltex (script)
filltex
does the whole thing: compiles LaTex, fills the bibliography and gives you the final .pdf
. Usage:
filltex <tex file>
Argument can be with or without extension, and the script is smart enough to figure it out.
Since ADS bibliography items contains journal abbreviations, you need to use aas_macros.sty
(available here). Don't worry, you just put \include{aas_macros}
in your .tex
file, and filltex
will download the file for you if you need it.
At the end, filltex
also runs TexCount which counts the words in your document.
TexShop
I use the TexShop editor, so I wrote an implementation of filltex
for it. If you copied the filltex.engine
file as specified above, just open your paper with TexShop and select filltex from the drop menu on the left. Now automagically compile your paper with Typeset
or cmd-T.
Example
A short example.tex
file is provided, where you can try this new way of writing papers!
Known limitations
Treating arXiv e-prints with ADS is tricky. When an e-print gets published they change the database key, but make the old key point to the new version! For instance, the key switches from
2016arXiv160203837T
to2016PhRvL.116f1102A
. If you're citing an e-print which is not yet published, everything is fine: only the arXiv key (e.g.2016arXiv160203837T
) is available and your reference list will show the arXiv version. If you're citing a paper that is published, both the e-print key (e.g.2016arXiv160203837T
) and the published-version key (e.g.2016PhRvL.116f1102A
) are available. When used, they will both point to the same published version! If you write a document with citations to both, this will cause the same record to appear twice in your reference list (see the example file). To avoid the issue, always use the published-paper key if a published version is out. INSPIRE doesn't have this problem, because they don't change the citation key when a paper gets published.Multiple bibliographies are not allowed, only one
.bib
file per paper. I don't plan to implement multiple bibliographies in here, because you're not going to need them with this script: one paper, one bibliography, that's all.
Manual installation from repository
If you don't like pip (but why wouldn't you?), you can install the code manually:
git clone https://github.com/dgerosa/filltex.git # Clone repo
cd filltex
chmod +x bin/* # Make bin content executable
PATH=$PATH:$(pwd)/bin # Add bin directory to path
echo "PATH=$PATH:$(pwd)/bin" >> ${HOME}/.bashrc # To add the new path to your .bashrc
cp filltex.engine ~/Library/TeXshop/Engines/filltex.engine # To install the Texshop engine
filltex
uses TexCount, which is included in most Tex distribution. In case it's not in yours, here you can find installation instruction.
References to filltex
filltex
is included in the suggested tools from the INSPIRE team.filltex
is included the official v3.80 release of Texshop.
Credits
The code is developed and maintained by Davide Gerosa. If you find bugs, want to contribute to this project (any help is welcome!) or need help with it, just open an issue here on GitHub.
The idea started from this python
course taught by Michele Vallisneri at Caltech (and in particular from this example) and was later developed with key contributions from David Radice. We also thank Lars Holm Nielsen, reviewer for The Journal of Open Software, for several suggestions which improved filltex
. TexCount is developed by Einar Andreas Rodland. Useful info on the INSPIRE and ADS APIs are available here and here.
Changes
v1.0: Initial release, main functionalities.
v1.1: Version accepted in JOSS.
v1.2: Uploaded on pip.
v1.3: Compatible with new ADS "Bumblebee".
v1.4: Compatible with new INSPIRE API.
v1.5: New tex
and list
subcommands.