epub to PDF - specific letters are transparent and MISSING

[Jam]

Recently when exporting from epub to pdf my pdf exports have been crippled by some of the letters, specific alphabet letters, eg d and n being trasparent and not visible.

They are there, because when I select various paragraphs of pdf text and paste it into a text editor it is all there.

I have tried different pdf viewers and none show the transparent letters.

I've also tried viewing the pdf with "night mode" and the letters are still not there.

Anyone know what the fix is ?

[Jam]

I get the same problem when using the ebook-convert utility

Here is my one liner script

ebook-convert \
book_name.epub \
book_name.pdf \
--embed-all-fonts \
--paper-size=a4 \
--pdf-add-toc \
--pdf-page-numbers \
--pdf-sans-family \
--pretty-print

It used to work. I converted lots of epub books to pdf with this line in the past, but not any more.

Has the same issue as described in OP above.

Random, but specific letters are transparent, making the pdf export unusable.

For about 5 months (?) this problem has existed.

It's not the epub files, because pdf2go can convert them all successfully.

But pdf2go deos not produce a TOC, and does not number the pdf pages, so it is pretty useless.

[Quoth]

Use Calibre to export the epub to docx. Edit in LO Writer using odt format. Have styles, TOC/Index etc as desired. Export a PDF.

[Jam]

Originally Posted by Quoth

Use Calibre to export the epub to docx. Edit in LO Writer using odt format. Have styles, TOC/Index etc as desired. Export a PDF.

ty. That's a solution, I appreciate the suggestion, but not at all ideal compared to point-and-click direct export.

As someone who has done quite a lot of bash scripting, I know that errors are easily introduced into software. I wonder if the software was re-factored, in some upgrade, and things are still being ironed out.

epub files that used to convert no longer convert, and also suffer the transparent letters problem, when being converted to pdf.

I wonder why it could be that some letters are transparent but not all. This seems very odd.

When I buy a epub book, this was a very hand feature for printing. Probably one of the most important features of Calibre.

Does anyone know which is the last version that does not have this issue. I will then downgrade?

[kovidgoyal]

https://www.mobileread.com/forums/sh...d.php?t=186697

[Jam]

Originally Posted by kovidgoyal

https://www.mobileread.com/forums/sh...d.php?t=186697

thank you. I appreciate Calibre, as do so many others.
I understand that coding is complex and does many many tasks.

When I get a moment I will provide more detail and information on the problem I am seeing.
I am running up to date Arch Linux. Perhaps I have a dependency missing.

I get the feeling it is some kind of PDF layering issue.

My temporary fix is to take the ebook-convert output pdf file using the above command, see previous comment, and then post process this as follows

inkscape in_file.pdf --pages=all --export-filename=out_file.pdf

Inkscape get's rid of the text transparency problem but removes the useful pdf document index and in document clickable links from the output pdf.

It lets me print the books for now.

[Jam]

Before post processing this is how te doc looks.

https://drive.google.com/file/d/1AdH...oCPssIpYa/edit

Note letters v and h and other letters are missing. Makes doc un-readable. It is all there which is confirmed by selecting and copying sections of text from the pdf and pasting them into the text editor such as Vim.

It's just a transparency / visibility issue.

[Quoth]

Elephant in the room.
I stopped printing on paper about 2013. I "print" to epub and proof / annotate on eink.

I sometimes have to check PDFs or read PDFs, so have choice of TCL Nxtpaper 11 (matt and almost like colour print paper or eink, laptop with matt add-on 2560x1440 16″ screen or matt 23″ 4K screen on workstation. Saved about 250,000 A4 sheets of paper and associated ink/toner environment and time in the last 10 years. Sometimes the 8″ Sage eink is big enough for a PDF.

For an entire book or something that needs to be on paper for someone else I'd use Print24 or Lulu Print On Demand. Or loan them an ereader. It's cheaper to buy and give someone an ereader than print 15 novels.

Normally I only generate PDF for POD / Publishing, hence the preferred route of LO Writer. I have used print from Calibre viewer occasionally to create a PDF, though not to have it on paper.

If it's just a paper copy you need you can try printing from the Calibre viewer.

A also sometimes use The GIMP to process poor PDFs (only available as PDF) or ImageMagick or K2PDFOPT https://www.mobileread.com/forums/sh...d.php?t=144711
I do use Inkscape, but not for PDFs.

[kovidgoyal]

I suggest you uninstall the arch calibre package and install the official binaries from https://calibre-ebook.com/download_linux

[Quoth]

Originally Posted by kovidgoyal

I suggest you uninstall the arch calibre package and install the official binaries from https://calibre-ebook.com/download_linux

Which has worked for me for 5 Linux distros, x86 (now only 5.44 max), x86-64 and ARM64 (3 Linux distros and several kinds of Desktop).

[Jam]

Originally Posted by kovidgoyal

I suggest you uninstall the arch calibre package and install the official binaries from https://calibre-ebook.com/download_linux

Thank you. I will do so and let you know what happens.

[Jam]

Originally Posted by kovidgoyal

I suggest you uninstall the arch calibre package and install the official binaries from https://calibre-ebook.com/download_linux

That fixed everything, Thank you.

[Jam]

I had a detailed discussion on the Arch Linux bug reporting site about this issue.

One of the packagers and myself did some pdf font export testing.
He did it, I replicated his results.
It seems the issue with text transparencey in epub to pdf exports may be related to font setting not being specific in the pdf export settings.

The Arch packager suggested this is an upstream issue in the original Calibre package which should be looked into.

While I am now using a Calibre install, direct from the Calibre web site, which does not seem to contain this issue, it might still be worth looking into this.

The discussion on the Arch bug reporting web is shown here
https://gitlab.archlinux.org/archlin...bre/-/issues/4

Calibre is a great package, thank you for it.

[kovidgoyal]

This has nothing to do with fonts. The PDF is generated by Qt WebEngine. And its subset by fonttools and post processed using PoDoFo. Most likely the arch package of one of those is broken in some way or the other. I am afraid I lost my patience with debugging distro packages many many years ago. Just use the calibre official binaries if you want a bug free experience.

[Jam]

Originally Posted by kovidgoyal

This has nothing to do with fonts. The PDF is generated by Qt WebEngine. And its subset by fonttools and post processed using PoDoFo. Most likely the arch package of one of those is broken in some way or the other. I am afraid I lost my patience with debugging distro packages many many years ago. Just use the calibre official binaries if you want a bug free experience.

That is very clear. I understand your insights. Thank you for your insightful response.

I will stick with your binary going forward.