dginev / ar5iv-bindings

Extended LaTeXML bindings for converting the sources of arXiv.org

ar5iv-bindings

Extended LaTeXML bindings for converting the older sources of arXiv.org

Sourced from: https://arxiv.org/help/macro_list

Excluded files that are:

• already supported by the mainline LaTeXML distribution
• available under texlive 2021

The initial work in this repository went as far as to tease away the Fatal status of a report over 100 arXiv documents. Just dipping our toes in...

🚧 High Velocity

For all intents and purposes, the bindings in this repository are highly experimental.

As support stabilizes and matures, bindings of sufficient quality will swim up to the mainline LaTeXML "Package" support, and will be removed from this repository.

Note on Licensing

All files under bindings/ are released in the public domain and you are invited to reuse them.

All other assets currently in the repository come from the wider LaTeX publishing ecosystem, and are distributable under their own licenses.

Ideally, avoid using this repository in any way that relies on raw .sty or .cls assets. Only the .ltxml files are ours to fully distribute.

ar5iv-bindings's Issues

\parencite[][]{} is not compiled correctly

See arXiv/zzzArchived_arxiv-readability#10

TeXlive trees and arXiv

Since we don't have a dedicated arXMLiv discussion forum nowadays, opening an issue in the latexml tracker for now. This posting is meant to seek out some discussion w.r.t managing the ever drifting tex toolchain versions in arXiv, and fish for a pragmatic approach to tackle that in latexml.

The motivating example for that is article 1305.3026 in particular, and the error:unexpected:_ error report in general. The running example relies on a package called osajnl2 from the Optical Society, which used to be in TeXlive 2009, but is no longer present in any of the more modern texlives out there. The way it connects to the error class is that this package (among other macros) loads graphicx.sty, which allows \includegraphics on a filename containing underscores. Without it the underscores are interpreted as wrongly used outside of math mode, hence the error.

Back to the general problem, one wonders how to approach all missing files in arXiv, including such that are neither to be found in a modern texlive installation, nor in the local repository of the article. It may be worthwhile to compile a list of those files and do one of the following:

1. Ignore/drop all arXiv articles that require them from the official corpus, marking them invalid
2. Obtain the full range of texlive distributions and extract the now obsolete files, to address them as a separate dependency for the conversion (texlive-obsolete?)
3. Something even more creative?

arXiv has some notes on its texlive toolchains here.

A separate set of package files are the few packages that are specially distributed by arXiv here.