[tex-live] PDF/A-1b compliancy with (pdf,lua)TeX

Reinhard Kotucha reinhard.kotucha at web.de
Fri Dec 20 01:07:21 CET 2013

On 2013-12-19 at 09:13:20 +0100, Florian Breitwieser wrote:

 > Hi Reinhard,
 > > good to know that my assumption was wrong.  I must admit that I
 > > only compiled the testfiles shipped with pdfx.sty with pdfTeX and
 > > sent the PDF files to two online PDFA validators.
 > I also struggled get a PDF/A with pdfLaTeX and pdfx, and it did not
 > work out in the end.
 > I changed to a different route - and I am nearly there now - where
 > I use hyperxmp, which is nice to use and which also supports
 > LuaTeX.  Color profile inclusion (there's a small bug in the
 > original pdfx code) and glyph management are taken from pdfx. I
 > think these should work in LuaTeX, too. The code, in the end, is
 > very compact, as hyperxmp provides higher-level access to metadata
 > writing. It requires modifications in the package source to write
 > the PDF/A compliancy tag and other things, but I am in contact with
 > the author to include it.

Hi Florian,
thanks for the hints.  As far as XMP data is concerned I only have to
know how a valid XMP record looks like.  In my case it makes more
sense to create them with Lua.  The reason is that everything to be
put into the PDF file exists in Lua already and I use TeX only for

 > > Unfortunately everything is time-consuming.  One PDFA validator has a
 > > nice web interface but the other one accepts files by email and sends
 > > its results as email attachments.  The latter is more picky, hence I
 > > can't avoid it though it's quite inconvenient.
 > True, I also spent a week. In the end, I found a combination I am
 > quite happy with: pdf-tools online validator [1] and Apache pdfbox
 > preflight [2]. I did not find the later very much used by other
 > people, but it is very powerful and sensitive to issues.

Looks promising.  The reason that very few people are using it is
probably the lack of end-user documentation.  Java programmers
certainly know how to install and invoke it.  Installation of the
whole pdfbox package failed.  Then I downloaded preflight-1.8.3.jar
but I didn't find out how to use it.

 > To be really useful, also there I needed to modify the code to
 > include the byte offset, etc, to error messages [3].

This is definitely a great improvement.  On the other hand it's
probably sufficient if the object number is reported.  What I really
miss is an Emacs pdf-mode which reads the xref table whenever a PDF
file is opened and allows to jump to particular objects conveniently.

 > There is maybe one more iteration for the LaTeX code, then I can
 > supply it if you are interested.

Yes, I'm interested.  Maybe we better discuss it off-list if nobody
else intends to contribute.


Reinhard Kotucha                                      Phone: +49-511-3373112
Marschnerstr. 25
D-30167 Hannover                              mailto:reinhard.kotucha at web.de
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.

More information about the tex-live mailing list