PDF Associated File Type Data

[lexicism]

Are there currently plans to add support for Associated Files with AFRelationship "Data"/"Source" in PDFs? (Meaning source data for something like a table or line graph, rather than an AF with AFRelationship /Supplement for MathML.) Or is there an ongoing project for this that I overlooked in my searches?/Has this feature already been implemented?

I even made a Figure-only PDF with an Associated File (tried both CSV and XML), but didn't hear any results when testing in Foxit. (Reads fine for MathML.) I think the functionality that I'm envisioning would be NVDA reads out Alt Text when in browse mode but navigates through the Associated File when in Focus Mode. If work starts (or has started) on this, would that be more of a MathCAT project?

[SaschaCowley]

@NSoiffer do you have any thoughts on this?

[NSoiffer]

I like your idea for supporting data behind a graph. AFs are the right way to do it. The unfortunate thing is that the accessible interface for PDFs was designed before AFs and doesn't have a way to get access to them.

The clever idea (or horrible hack) that Roman Toda (Foxit) and I had was to map the AF that contains MathML to the existing mechanism used if MathML is in the structure tree. That way, no change to the accessibility interface is needed and almost no code changes need to be made in NVDA and elsewhere to use it. I think this approach is reasonable because the two means of representing MathML should result in the same experience for AT users. Adobe also agrees this is reasonable and one of these days (they won't say when), they will do the mapping also in Adobe Reader/Acrobat.

Your suggestion is an enhancement to the existing experience, so there really isn't an equivalence like there is with MathML. Hence, I don't think there is a similar hack that makes sense. If (hopefully when) a new accessibility interface for PDF is created, then I think it would be possible to move forward with an idea like yours.

FYI: I can think of some workarounds/hacks that could be done, perhaps as an experiment. However, experiments tend to lead to "it works here and so others should do this too", and the hacks I that I can imagine lead away from standards and towards fragmentation. That's why I don't even want to mention them.