RawBlock not found by pandoc depending on how file is passed (tex to html)

[Pecnut]

Hello Pandoc team!

I am following the example of Building images with TikZ.

Why does this command work (exactly as given in the example):

pandoc --lua-filter tikz.lua -s -o cycle.html <<EOF
Here is a diagram of the cycle:

\begin{tikzpicture}

\def \n {5}
\def \radius {3cm}
\def \margin {8} % margin in angles, depends on the radius

\foreach \s in {1,...,\n}
{
  \node[draw, circle] at ({360/\n * (\s - 1)}:\radius) {$\s$};
  \draw[->, >=latex] ({360/\n * (\s - 1)+\margin}:\radius)
    arc ({360/\n * (\s - 1)+\margin}:{360/\n * (\s)-\margin}:\radius);
}
\end{tikzpicture}
EOF

(and in so-doing, converts the tikz into an SVG file and correctly generates the HTML page)

but if I place the LaTeX code into a separate file, say cycle.tex, and run

pandoc --lua-filter tikz.lua -s -o cycle.html cycle.tex

then the output file is still created, but the image isn't converted at all?

Some digging suggests that Pandoc isn't interpreting the \begin{tikzpicture} \end{tikzpicture} environment at all: as far as I can tell from looking at what I think is the document tree, this is being ignored rather than being parsed as a RawBlock, whereas when I give the tex file using <<EOF, pandoc does parse it as a RawBlock within the document tree.

I am so confused. Please help!

(I am running Pandoc 3.5 on a Mac)

[tarleb]

Pandoc tries to guess the input format based of the filename extension, so it uses the latex reader in your second case above. If there is no input filename, as in the first case, then pandoc defaults to markdown.

The LaTeX reader skips TeX commands that it cannot handle, while the Markdown reader puts all LaTeX blocks in the RawBlock elements that are then processed by the filter. You could try to set --from=markdown or --from=latex+raw_tex when using an input file with a .tex extension. Enabling the +raw_tex extension on the LaTeX reader ensures that unknown TeX blocks are kept around instead of just being skipped.

[Pecnut]

This is super helpful, thank you @tarleb ! The example above now works and I feel enlightened. 👍

This somewhat annoyingly runs into a problem with the larger project I'm working on, where --from=latex correctly parses \subfile (from the subfiles LaTeX package), but --from=latex+raw_tex appears to ignore the \subfile command (and so doesn't include the contents of the named subfile).

Any idea how to get it to work? Obviously --from=latex+raw_tex is what I need to get the image conversion working, but it doesn't appear to interpet \subfile in the way that --from=latex does. At the moment it seems I get one or the other! 🫤

(If you think this is better asked as a separate question, I am happy to do that!)

[tarleb]

I'm afraid you'll need some kind of filter to get it to work.

[Pecnut]

OK that's great, thank you @tarleb !