Aw: Re: [docbook-apps] tagged and accessible PDF document with DocBook

["Holger Bast" <HolgerBast@gmx.de>]

Hello Tony,
thank you for your answer and the feedback you provided.

I re-worked the document based on your it. Below you can find
my answers. 

Bye, Holger

> 1. In Section 4.1.4.1, Strong vs. weak block-level structures in PDF
> files, you say 'DocBook already provides a very strong structure' [1],
> but then your examples use 'H1' to 'H5' PDF tags. The definition of
> 'Strongly structured' that you copied from the PDF 1.5 reference says to
> use 'H' in strongly structured PDF files. Section 7.4.4, Unnumbered
> headings', of ISO 14289 includes both "'H' ... should be used in
> strongly structured documents" and "Documents that are strongly
> structured may use numbered headings.", so ISO 14289 would also rather
> that you use 'H' in strongly structured PDF.

Answer: fixed.
Using H tags is not a must according to the ISO 14289 but we should follow
the recommendation.


> 2. It really shouldn't be necessary to specify that 'fo:static-content'
> is tagged as 'Artifact'. It should just happen, as specified in Section
> 7.8, Page headers and footers, of ISO 14289.

Answer: If I understood the ISO 14290 and 32000 right, then these parts
get an Artifact tag as well as a subtype (Pagination, Layout, Page and
Background). Apache FOP doesn't allow to tag a subtype. I also couldn't
find any information about Artifact subtypes in AH.

If I do not tag header and footer as Artifact Apache FOP will tag them
as usual. If we can't find an agreement I can do this during my
post-processing run.


> 3. Similarly, it shouldn't be necessary to supply PDF tags for
> 'fo:list-block', 'fo:list-item', 'fo:list-item-label', and
> 'fo:list-item-body': the XSL-FO Formatter should be providing the right
> tags for those FOs. AH Formatter will do it, and it seems from the code
> in Section 3.1.2, Automatic tagging by Apache FOP, that FOP will do it.

Answer: That's right. But I understood Bobs wish for a specification in
such a way that the generated XSL-FO code can be processed with (mostly)
all FO processors. I added a table for these structures where we can
annotate if the processors supports auto-tagging.


> 4. AFAICT, PDF tags are case-sensitive, so you probably should use the
> specified forms in your examples, e.g., 'Document' instead of 'DOCUMENT'.

Answer: fixed.


> 5. Within your 'TOCI' for a ToC entry, you should use 'Lbl' for the
> entry's title, 'NonStruct' for the leader, and 'Reference' for the page
> number citation.
> 
> Sorry, I should have said to also use 'Reference' for the entry's title.
> If you wanted to be as correct as possible, you'd probably use 'Lbl'
> for the chapter or section number preceding the title text in the ToC entry.

Answer: I tested your proposal but I'm not really happy with the rendered
result in Apache FOP. Now I have two examples inside the document:
Example one only includes the must-have tags TOC and TOCI; the other
one includes much more information.


> 6. My understanding of the 'NonStruct' tag changes on alternate days,
> but you might be able to use it on some of the 'fo:block' (that you
> can't magic away with your post-processing) to indicate that the
> 'fo:block' has 'no inherent structural significance'.

Answer: We could use this inside the TOC. The TOCI elements are also
nested. Therefore it would be useful.


> 7. Putting PDF tag names in @role won't do anything in AH Formatter. If
> you need to override the default PDF tag [2] for a particular FO, you
> should use @axf:pdftag [3].

Answer: Ok, so we should implement some switch to decide whether the
fo output is rendered for processors which understand @role and 
for processors which do not?


> 8. You have a comment about preferring a flat rather than nested ToC
> structure. You can represent a hierarchical ToC structure in Tagged
> PDF, but I don't know whether flat or hierarchical is better for people
> using assistive technology. Dave Pawson may know.
> 
> Dave:
> I think 'visual' presentation isn't an issue?
> When text is read by a machine, one line with indent x is the same as
> one with indent y.

I'd suggest a flat structure would be more readily accessible, unless
each entry is numbered etc?

Answer: Ok. See answers to 5.

> Gesendet: Donnerstag, 27. April 2017 um 19:27 Uhr
> Von: "Dave Pawson" <dave.pawson@gmail.com>
> An: "Tony Graham" <tgraham@antenna.co.jp>
> Cc: docbook-apps <docbook-apps@lists.oasis-open.org>
> Betreff: Re: [docbook-apps] tagged and accessible PDF document with DocBook
>
> On 27 April 2017 at 17:49, Tony Graham <tgraham@antenna.co.jp> wrote:
> 
> > Which brings me to...
> >
> > 8. You have a comment about preferring a flat rather than nested ToC
> > structure.  You can represent a hierarchical ToC structure in Tagged
> > PDF, but I don't know whether flat or hierarchical is better for people
> > using assistive technology.  Dave Pawson may know.
> 
> I think 'visual' presentation isn't an issue?
> When text is read by a machine, one line with indent x is the same as
> one with indent y.
> 
> I'd suggest a flat structure would be more readily accessible, unless
> each entry is numbered etc?
> 
> HTH
> 
> 
> 
> >
> > Regards,
> >
> >
> > Tony Graham.
> > --
> > Senior Architect
> > XML Division
> > Antenna House, Inc.
> > ----
> > Skerries, Ireland
> > tgraham@antenna.co.jp
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: docbook-apps-unsubscribe@lists.oasis-open.org
> > For additional commands, e-mail: docbook-apps-help@lists.oasis-open.org
> >
> 
> 
> 
> -- 
> Dave Pawson
> XSLT XSL-FO FAQ.
> Docbook FAQ.
> http://www.dpawson.co.uk
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: docbook-apps-unsubscribe@lists.oasis-open.org
> For additional commands, e-mail: docbook-apps-help@lists.oasis-open.org
> 
>