I've used `pdftocairo -svg` (it's part of the Poppler[0] utilities) to do this in the past, but Cairo insists on converting all of the text to shapes. (They are/were planning to fix that, so maybe it's already fixed.) You've got all of the text there, so you're a step closer to what I wanted, but I noticed that the spaces are all missing. I'm actually pretty impressed that SVG can render properly without the spaces for the text.

[0] http://poppler.freedesktop.org/

The fonts are not kept. Any real use case?

I really wonder about the utility of something like this tool. I enjoy and encourage building new things "just because" but I can help but think the point to PDFs is to have complete visual control of layout and formatting while providing the textual and tagging information layers is critical to transform and adapt the information for artificial intelligent agents and people with disabilities that rely upon assistive technologies such as screen readers.

I would be interested in knowing about the Sizes of SVG produced by this tool. Some of the SVGs converted by poppler utility tool 'pdftocairo' are just too large i.e few MB's which is absolutely unacceptable as these large SVG'S tends to crash the mobile browsers.

When would one need a SVG version of a PDF instead of just using pdf.js by itself?

What about something that goes the other way?