LoginSign Up
← Back to all sparks
T

texor

INFRA · APIS
Velocity0.0

Converting 'LaTeX' 'R Journal' Articles into 'RJ-web-articles'

github.com/abhi-1u/texor ↗

The LaTeX-to-R-Markdown converter that stopped transcribing articles and started making them executable.

latexr-markdowndocument-conversionsweavepandocr-journal
◆Current state
texor converts legacy R Journal articles from LaTeX into R Markdown, wrestling with the parts that never translate cleanly: numbering, cross-references, algorithm environments, and macros nobody outside the R Journal uses. Since 1.5.0 it can also emit figures as code chunks and tables as CSV read back through kable(), and target bookdown, litedown or biocstyle. The 2025 releases are repair work — pandoc keeps changing underneath it.
◆Where it's heading
Two expansions define the arc. 1.4.0 took on Sweave as a source format, opening the pre-Rmd back catalogue of literate R documents. 1.5.0 changed what comes out the other end, offering executable code chunks and data files where earlier versions produced flat markup. Both moves point the same way: toward converted articles that still compute rather than merely resemble the original. Everything since has been keeping that machinery working against pandoc's release cadence.
◆Prediction
The last two releases were both triggered by upstream pandoc changes breaking numbering or Lua filters, so the next release is most likely more of the same; new output styles remain the plausible feature direction given how recently that list was extended.

◆Recent moves

  1. 1y ago

    Broken figure and table numbering repaired after pandoc updates

    Repairs cross-referencing that pandoc updates had broken: table, figure and algorithm numbering, and table references rendering as ?? instead of a number. A soft fuse now skips further image conversion when a non-standard image is hit rather than failing the run. Visible output correctness, not new capability.

    View source ↗
  2. 1y ago

    DESCRIPTION path, cleanup ordering, and a pandoc filter patch

    Internal fixes: the DESCRIPTION lookup path, cleanup moved to on.exit() so it runs even on failure, and a Lua filter patched for a newer pandoc. Nothing changes in a document that already converted successfully.

    View source ↗
  3. 1y ago

    Executable figure chunks, CSV-backed tables, and new vignette styles

    A dense batch of options rather than a change of default: equations number automatically without depending on number_sections, figures can be emitted as R Markdown code chunks, most tables can become a CSV read back with kable(), and vignettes can target bookdown, litedown or biocstyle. Algorithm environments now render via pseudocodejs.

    View source ↗
  4. 2y ago

    Sweave becomes a supported source format

    ⚡ SPARK

    The release that widened what texor accepts. Until now the tool read R Journal LaTeX; from here it also reads Sweave, which is a different class of source document — one that carries live R code rather than typeset output. The rnw_to_rmd() path introduced here is what 1.5.0 then extended with DESCRIPTION handling.

    View source ↗
  5. 2y ago

    Bookdown-style equation references and LaTeX macro workarounds

    Equation handling is reworked to match bookdown's reference format, and workaround macros arrive for \nameref, \autoref, \scalebox, \fbox and friends — the long tail of LaTeX that has no markdown equivalent. Every processed image now gets default alt text, the package's first explicit accessibility work.

    View source ↗
  6. 3y ago

    Numbering for code blocks and wide tables, plus seven more macros

    Code blocks and wide tables get their own numbering, seven more R-documentation macros are understood, and interactive_mode starts asking the user for input rather than guessing. Incremental coverage of the R Journal's LaTeX conventions.

    View source ↗