@jcsteh, am I misremembering things, or is Firefox supposed to render any tagged content from PDFs to screen readers? I know that saving as PDF doesn't keep the tags like Chromium browsers, but I thought Firefox was the only browser to render the tags when opening. See my previous post in the thread for some example documents.

@jcsteh Never mind. PDF tagging is supported as of v.89. Will create a Bugzilla account and file something after work.

@jcsteh FYI, the bug is now filed, though I'm not the best at this kind of thing:
bugzilla.mozilla.org/show_bug.…

Thanks. This seems to have regressed in Firefox 126, which doesn't quite match your experience because I think you tested with Firefox 125? Is that correct?

@jcsteh Yeah, 125.0.3.

@jcsteh Just grabbed 124 portable, and PDF tags are exposed there.

Thanks so much for reporting this and I'm really sorry. I've given it pretty high severity, diagnosed the problem and asked for it to be tracked for 126, so hopefully we can get it fixed there. I'm not very familiar with this code, so I can't push a fix myself, but hopefully it will get attention fairly quickly.

@jcsteh No need for apologies. I hate PDFs on good days, and do use Adobe Acrobat for the rare times a tagged PDF exists in the wild. I just happened to test this in Firefox because of the Zoom Essential documentation tagging their supplemental manuals. I didn't realize that even when working, image descriptions are still not exposed to screen readers (bug 1708040). But this is the only Windows browser to support PDF tags, so figure I'd file a bug report. Thanks so much for looking into it.

IMO, it'd be nice if we didn't have to use Acrobat Reader, since although it has the best support for tagged PDF, I feel it's also super clunky/bloated. So yeah, I'd really love to see the remaining issues get fixed too.

I'd also love it if we could infer tags for untagged PDFs like Acrobat does, but that's a much harder problem unfortunately.

@jcsteh Acrobat has, as far as I know, the only complete support for PDF tags in Windows, but even it breaks word boundaries sometimes, even when tagged, and I end up quoting something poorly because words were unintentionally combined and I didn't notice. If my work didn't require interfacing with PDFs on the regular, I wouldn't touch it with a ten-foot pole. And as for its auto-tagging, it bites something fierce and leaves much to be desired.

@jcsteh I do wonder if something like this would offer far better output as compared to Acrobat though...
github.com/VikParuchuri/marker

Yeah, I saw that a while ago. Haven't had a chance to check it out. I think it might need some pretty decent models/hardware to work though, so probably not something we could derive from in Firefox.