generate HTML version for online documentation using pdf2htmlEX

[boegel]

https://github.com/coolwanglu/pdf2htmlEX

[redfast00]

I had some issues installing pdfhtmlEX, but I managed to install it in Docker. I ran the default on the intro-HPC pdf, and it seems to work (in Firefox at least): there are no issues I saw. It's basically the pdf webview:
intro.zip

[redfast00]

I now tried to use latex2html: it's old and no longer supported, and breaks as soon as it encounters something it doesn't know. It seems that imports don't work, conditionals don't work, ... I don't think latex2html does a latex build, I think it just converts it directly.

Edit: it seems to have a styles directory with a whole bunch of perl files in it. I think it just takes those perl files an use them instead of tex libraries. It would take a lot of work to port over all the libraries we use, so looking for an alternative.

[redfast00]

Issues with pdf2htmllEx

• It generates a huge single html file => can be solved, there's a split option
• Linking isn't nice: works by page or by location, but haven't been able to get it to go to a specific label

Advantages:

• It looks just like the PDF
• It's pretty fast (~30 seconds)
• It's opensource
• It works

[redfast00]

Now trying the htlatex command from tex4ht. It doesn't seem to work, due to a bug in tex4ht that hasn't been patched (solution at https://tex.stackexchange.com/a/232739/169881).

[redfast00]

I've used the patch, then tried to compile it. It got stuck at the beginning: it couldn't find logo_vsc. There is an \graphicspath that htlatex seems to ignore, so I copied all image files to the current directory. This got me further, but not much: it gets stuck at the same line,

LaTeX Error: Cannot determine size of graphic in logo_vsc.png (no BoundingBox
).

[redfast00]

Basically, what we want is to render to HTML instead of converting our LaTeX to HTML: all programs that do the latter don't support arbitrary LaTeX package, and instead have a sort of "emulator" for the most common packages (see for example this directory). That doesn't work for us, since we have a large document with lots of packages and a complicated structure.