Out of sequence entries break TOC

[orebmur]

Just noted an unexpected behaviour for out of sequence TOC entries in koreader for Linux and Android.

Manually created a TOC (see attached toc.txt) for the new koreader user guide . In addition to the TOC entries made from the section headers, an additional custom tree named 'IMAGES' directly pointing to the pictures of the document was inserted.

Following command was used to merge both the pdf and the TOC text file (in Linux):

pdftk koreader-user-guide.pdf update_info bookmarks.txt output koreader-ug-with-toc.pdf

The resulting pdf file 'koreader-ug-with-toc.pdf' is attached to this post.

As can be seen already on the attached screenshots, while the additional entries show the referenced pages within brackets, the actual redirection always points to the same page 1.

This is reproducible for any koreader version on both amd64/arm64 Linux and Android/arm using a Boyue Note/Mimas.

Using either xpdf or acroread-9.5.5 (via Wine) in Linux, the TOC display and jumping to the expected page works as expected.

Expected behaviour for koreader TOC would be to function the same as the forementioned pdf readers.

[Frenzie]

More directly relevant to us in the present, MuPDF 1.13 deals with it correctly.

Edit:

This seems to be by design; let's see why…

07/10/22-22:37:07 DEBUG BOGUS TOC: 32 3 < 31 25 - 28 29
07/10/22-22:37:07 DEBUG   fix prev 31 25 => 1
07/10/22-22:37:07 DEBUG   fix prev 30 22 => 1
07/10/22-22:37:07 DEBUG   fix prev 29 21 => 1
07/10/22-22:37:07 DEBUG   fix prev 28 21 => 1
07/10/22-22:37:07 DEBUG   fix prev 27 21 => 1
07/10/22-22:37:07 DEBUG   fix prev 26 20 => 1
07/10/22-22:37:07 DEBUG   fix prev 25 19 => 1
07/10/22-22:37:07 DEBUG   fix prev 24 18 => 1
07/10/22-22:37:07 DEBUG   fix prev 23 17 => 1
07/10/22-22:37:07 DEBUG   fix prev 22 15 => 1
07/10/22-22:37:07 DEBUG   fix prev 21 15 => 1
07/10/22-22:37:07 DEBUG   fix prev 20 15 => 1
07/10/22-22:37:07 DEBUG   fix prev 19 14 => 1
07/10/22-22:37:07 DEBUG   fix prev 18 14 => 1
07/10/22-22:37:07 DEBUG   fix prev 17 13 => 1
07/10/22-22:37:07 DEBUG   fix prev 16 12 => 1
07/10/22-22:37:07 DEBUG   fix prev 15 12 => 1
07/10/22-22:37:07 DEBUG   fix prev 14 11 => 1
07/10/22-22:37:07 DEBUG   fix prev 13 9 => 1
07/10/22-22:37:07 DEBUG   fix prev 12 9 => 1
07/10/22-22:37:07 DEBUG   fix prev 11 8 => 1
07/10/22-22:37:07 DEBUG   fix prev 10 8 => 1
07/10/22-22:37:07 DEBUG   fix prev 9 7 => 1
07/10/22-22:37:07 DEBUG   fix prev 8 7 => 1
07/10/22-22:37:07 DEBUG   fix prev 7 6 => 1
07/10/22-22:37:07 DEBUG   fix prev 6 5 => 1
07/10/22-22:37:07 DEBUG   fix prev 5 5 => 1
07/10/22-22:37:07 DEBUG   fix prev 4 5 => 1

Edit 2: https://github.com/koreader/koreader/pull/5907

Edit 3: https://github.com/koreader/koreader/issues/9319

[orebmur]

Wow, this was taken care of really fast!

Guess i'll then have to work around this koreader quirk regarding nested TOC's by fixing them manually whenever (rarely!) i happen to stumble over any book PDF with this "feature"...

For the curious:

Using pdftk, the table of contents (aka bookmarks) can easily be extracted with following command:

pdftk sample.pdf dump_data output datadump.txt

Use any suitable text editor of choice to modify the bookmark data contained in the resulting file (admittedly, not for the faint of heart). Rejoin the results with PDF file as follows:

pdftk sample.pdf update_info output datadump.txt output corrected.pdf

PDFtk is availabel for free at www.pdflabs.com/tools/pdftk-the-pdf-toolkit

[Frenzie]

I also like https://github.com/hellerbarde/stapler btw. Particularly since pdftk isn't really updated anymore afaik, though of course it's still useful for the moment.

[orebmur]

Originally Posted by Frenzie

I also like https://github.com/hellerbarde/stapler btw. Particularly since pdftk isn't really updated anymore afaik, though of course it's still useful for the moment.

Thanks for this pointer!
In Debian Linux, the upstream pdftk has been replaced by gitlab.com/pdftk-java/pdftk, an enhanced(?) reimplementation of the original pdftk, and which is in fact what i use.
Good to know there are even more options.

[Frenzie]

Originally Posted by orebmur

Thanks for this pointer!
In Debian Linux, the upstream pdftk has been replaced by gitlab.com/pdftk-java/pdftk, an enhanced(?) reimplementation of the original pdftk, and which is in fact what i use.
Good to know there are even more options.

Ahh, pdftk is now a transitional package for pdftk-java. I hadn't realized that. Good to know.

In my defense, it was quite dead between 2013 and 2017 afaik.

[Jellby]

I've been using qpdf as a replacement to pdftk, but I'm not sure it has this particular feature (and the syntax is hideous, I don't think I'll ever understand it).