Hi Holger, I don't agree that we should change default behaviour of FOP to suit docbook users. Instead I propose that we implement support for role attribute on fo:block, such that it would be possible to specify role="artifact" in docbook xsl so that the nested structure is not represented as nested p tags in accessibility structure.
Thanks, Chris On 19/04/2017 13:48, Holger Bast wrote: > Hi there, > we're using docbook5 to write our technical documents which needs to be > published as accessible PDF files and we would like to use FOP as > fo-processor. > FOP already supports generating tagged PDF content but I'm not happy with the > results and would like to discuss this topic further. > > Docbook5 provides XSL sheets to convert docbook to fo which then can be > processed by fo-processors. The XSL sheets often generate deep nested > fo:block structures like the following example: > > <fo:block> > <fo:block> > <fo:block ...> > <fo:block keep-with-next.within-column="always"> > <fo:block ...> > <fo:marker marker-class-name="section.head.marker">Level 1</fo:marker> > <fo:block font-size="20.735999999999997pt">1.1. Level 1</fo:block> > </fo:block> > </fo:block> > </fo:block> > </fo:block> > </fo:block> > </fo:block> > > This code also generates a deep nested p(aragraph) structure in the pdf file, > because every fo:block automatically is > tagged as paragraph. I would like to get rid of this to get a flat document > structure. > > I propose that fo:blocks are not automatically recognized as paragraphs > because they can contain different kinds of content, not only paragraph-like > content. So in my opinion they should not be affected by the tagging > mechanism automatically, so they are not included in the structural > information. The user should decide (opt-in) how to treat fo:blocks (like p, > h1 or something else). > > What do you think about this approach? > Is this something that can be (easily) achieved in FOP? > > If you need further information, I'll can provide sample documents and files. > > Any information relating this topic is appreciated. > thx & bye, Holger > . >