Using lwarp on an existing latex document without modifying it

[tvercaut]

I am wondering if it's possible to generate an html document from existing latex sources using lwarp but without modifying the source files. That is, can lwarp be used solely using command line arguments?

My best attempt so far looks like this but the resulting html is not complete so I guess I am missing something:

#!/usr/bin/env sh

# Choose bibtex of biber - TODO automate
bibexe=bibtex
#bibexe=biber

# Get the file path without the file extension
F=$1
F=${F%.*}
Fpath=$(dirname "${F}.tex")

# Convert image files
find ${Fpath} -name "*.eps" -exec lwarpmk epstopdf {} \;
find ${Fpath} -name "*.pdf" -exec lwarpmk pdftosvg {} \;

# Compile document while injecting lwarp
pdflatex -jobname=${F} '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
${bibexe} ${F}
pdflatex -jobname=${F} '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
pdflatex -jobname=${F} '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'

# Compile _html document while injecting lwarp
pdflatex -jobname=${F}_html '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
${bibexe} ${F}_html
pdflatex -jobname=${F}_html '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
pdflatex -jobname=${F}_html '\AtBeginDocument{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'

# Converrt _html pdf to actual html
pdftotext -enc UTF-8 -nopgbrk -layout ${F}_html.pdf ${F}_html.html

And the example file I test it with:

\documentclass{article}

% lwarp loading commented out to test external injection
%\usepackage[mathjax]{lwarp}

\usepackage{blindtext}

\author{Jane Doe}
\title{Test}

\begin{document}

\maketitle

\blindtext

Test citation \cite{article-full}.

Test equation:
\begin{equation}
\frac{1}{x} = \sqrt{a^2+b^2}  
\end{equation}
    
\bibliographystyle{plain}
\bibliography{xampl}

\end{document}

Am I missing something relatively simple here or is what I am trying to achieve much more complex than I would hope for?

[bdtc]

Lwarp wants to be loaded early, just after the fonts are set. You could try:
\AddToHook{package/lmodern/after}{\RequirePackage[mathjax]{lwarp}}
or to load immediately after the class:
\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}

And to convert the PDF to HTML, use
lwarpmk html
and maybe
lwarpmk limages

I'm guessing you have a large number of files to process in bulk, thus the desire to avoid editing each file. If you can get this to work then I'll add something about it in the Lwarp docs.

[tvercaut]

Many thanks! The second suggestion worked for my MWE but this is restricted to documents that use the article class. The first suggestion seemed more flexible but doesn't work with my MWE. Maybe lmodern isn't loaded by default?

Anyway, for now, here is the updated version of my script:

#!/usr/bin/env sh

# Choose bibtex of biber - TODO automate
bibexe=bibtex
#bibexe=biber

# Get the file path without the file extension
F=$1
F=${F%.*}
Fpath=$(dirname "${F}.tex")

# Convert image files
find $Fpath -name "*.eps" -exec lwarpmk epstopdf {} \;
find $Fpath -name "*.pdf" -exec lwarpmk pdftosvg {} \;

# Compile document while injecting lwarp
pdflatex -jobname=${F} '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
${bibexe} ${F}
pdflatex -jobname=${F} '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
pdflatex -jobname=${F} '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'

# Compile _html document while injecting lwarp
pdflatex -jobname=${F}_html '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
${bibexe} ${F}_html
pdflatex -jobname=${F}_html '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
pdflatex -jobname=${F}_html '\AddToHook{class/article/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'

# Convert _html pdf to actual html
lwarpmk pdftohtml

[tvercaut]

I couldn't find a better latex hook than class/<document_class>/after so used a bit of sed to find out the document class. Putting this together, my working albeit certainly brittle script looks like this:

#!/usr/bin/env sh

# Get the file path without the file extension
F=$1
F=${F%.*}
Fpath=$(dirname "${F}.tex")

# Compile the document to parse the log file
pdflatex ${F}

# Find out the document class
docclass=$(sed -n 's/.*Document Class: \([^[:space:]]*\).*/\1/p' ${F}.log)
echo "The latex document uses documentclass=${docclass}"

# Find out if bibtex or biber should be used
bibexe=bibtex
if grep -q 'Package biblatex' ${F}.log; then
    bibexe=biber
fi
echo "Bibliography will be processed using ${bibexe}"

# Convert image files
find $Fpath -name "*.eps" -exec lwarpmk epstopdf {} \;
find $Fpath -name "*.pdf" -exec lwarpmk pdftosvg {} \;

# Compile document while injecting lwarp
pdflatex -jobname=${F} '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
${bibexe} ${F}
pdflatex -jobname=${F} '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
pdflatex -jobname=${F} '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'

# Compile _html document while injecting lwarp
pdflatex -jobname=${F}_html '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
${bibexe} ${F}_html
pdflatex -jobname=${F}_html '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
pdflatex -jobname=${F}_html '\AddToHook{class/'${docclass}'/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'

# Convert _html pdf to actual html
lwarpmk pdftohtml

[bdtc]

For docs on the hooks,
texdoc lthooks
and
texdoc ltcmdhooks

These are new features as of a few years ago.

[bdtc]

Found the doc:
texdoc ltfilehook-doc
where it mentions you can use
class/after
for any class, which seems like it would work well.

[tvercaut]

That's nicer! Updated script for reference.

#!/usr/bin/env sh

# Get the file path without the file extension
F=$1
F=${F%.*}
Fpath=$(dirname "${F}.tex")

# Compile the document to parse the log file
pdflatex ${F}

# Find out if bibtex or biber should be used
bibexe=bibtex
if grep -q 'Package biblatex' ${F}.log; then
    bibexe=biber
fi
echo "Bibliography will be processed using ${bibexe}"

# Convert image files
find $Fpath -name "*.eps" -exec lwarpmk epstopdf {} \;
find $Fpath -name "*.pdf" -exec lwarpmk pdftosvg {} \;

# Compile document while injecting lwarp
pdflatex -jobname=${F} '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
${bibexe} ${F}
pdflatex -jobname=${F} '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'
pdflatex -jobname=${F} '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'.tex}'

# Compile _html document while injecting lwarp
pdflatex -jobname=${F}_html '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
${bibexe} ${F}_html
pdflatex -jobname=${F}_html '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'
pdflatex -jobname=${F}_html '\AddToHook{class/after}{\RequirePackage[mathjax]{lwarp}}\input{'${F}'_html.tex}'

# Convert _html pdf to actual html
lwarpmk pdftohtml

[bdtc]

Remember that not everyone uses pdflatex, they may use lualatex or xelatex.

[tvercaut]

Indeed, bibtex/biber may also not be needed in some documents. The above script is mostly serving my own needs at this stage.