LibreOffice Writer messes up encoding when exporting to PDF

JonnyRobbie · 2015-09-21 20:14:21

I'm not sure if it's an Arch problem or LO problem, but simce I'm on arch, I've decided I might be able to post it here.

I'm tyring to export to pdf from LOWriter. But the pdf has messed up encoding. See linked sceenshot. On the left is Writer, on the right is Okular, but the PDF is messed up everywhere, even on iPad.

http://i.imgur.com/elQmYpU.png

    $ uname -roms; pacman -Qi libreoffice-fresh | grep Version
    Linux 4.1.6-1-ARCH x86_64 GNU/Linux
    Version        : 5.0.1-1

I've tried searching through Settings and export settings, but did not find anything related to what might help me.

My locale should be set properly:

    $ locale -a                
    C
    cs_CZ.utf8
    en_GB.utf8
    en_US.utf8
    ja_JP.utf8
    POSIX
    $ locale charmap
    UTF-8

I may have messed up the topic. I was trying to post it in Applications & Desktop Environments and I had MM&Games opened by accident.

Last edited by JonnyRobbie (2015-09-21 20:27:13)

ewaller · 2015-09-21 20:31:12

Moved by request

olive · 2015-09-21 20:58:49

You should give more information if you want help. First and more importantly, how can we reproduce the problem? What was the original odt file? Also, what are you locale (output of locale and not locale -a that gives all available locale, not the used one). In what language have you set LibreOffice? What version of Libreoffice do you use? The text in the image of Libreoffice is actually the good text? What language is it?

Last edited by olive (2015-09-21 20:59:03)

Roken · 2015-09-21 21:01:25

It would help to have a copy of the file

JonnyRobbie · 2015-09-21 21:13:18

Yeah, sorry for not providing that. Here's the original odt: https://dl.dropboxusercontent.com/u/286 … angram.odt and here's the messed up exported pdf: https://dl.dropboxusercontent.com/u/286 … angram.pdf

dice · 2015-09-21 21:25:13

I just tried it and it seems to be a problem with the font 'Nimbus Roman No9 L'. Using this font it gets messed up. With e.g. 'Bitstream Vera Sans' or 'DejaVu Sans' the pdf has the correct characters in it. Though I am not familiar with any of these. What language is it?

My findings may be influenced by having Infinality installed though.

JonnyRobbie · 2015-09-21 21:42:27

I have infinality too.

Yes, you may be right it might be a font issue. I've just tried Linux Libertine Display and it seems to work fine with that one. I haven't thought of font issue immediately because I've had the same problem with another document that someone created on Win and was using Times New Roman, but looking at fc-match "Times New Roman", it falls back to my Nimbus Roman.

But why does it happen with Nimbus? It shows up fine in LOWriter, so why does it mess up when exporting?

It's czech (slightly nonsencial) pangram simmilar to the english quick brown fox designed to print all letters with diacritics.

Last edited by JonnyRobbie (2015-09-21 21:43:01)

jsoy9pQbYVNu5nfU · 2015-09-22 01:09:45

I have encountered the same issue with the Nimbus fonts before (actually, also with LibreOffice).

When inspecting the font information of your PDF file, you'll find that the Nimbus fonts are Type1 fonts while probably all other major fonts you have thought of testing against the Nimbus case are TrueType fonts (I suppose not Type3 .

name                                 type              encoding         emb sub uni object ID
------------------------------------ ----------------- ---------------- --- --- --- ---------
NimbusRomNo9L-Reg                    Type 1            Builtin          yes no  yes     11  0
NimbusRomNo9L-Reg                    Type 1            Custom           yes no  yes     14  0

Here is the sample provided typeset cleanly by pdflatex: text.pdf, also using Type1 fonts:

name                                 type              encoding         emb sub uni object ID
------------------------------------ ----------------- ---------------- --- --- --- ---------
BPVWSD+NimbusRomNo9L-Regu            Type 1            Custom           yes yes no       4  0

Here is another forum thread where an user also had problems with Type1 fonts. The broken text is probably a result of a bug in the LO PDF exporter. You'll find more results when googling using keywords going into that direction ("libreoffice type1 pdf export") . Inkscape manages to produce clean text using the very same Type1 fonts (pdflatex uses fonts packaged by TeXLive) just fine: sample. Unsurprisingly, its PDF export handles the matter in yet a different way:

name                                 type              encoding         emb sub uni object ID
------------------------------------ ----------------- ---------------- --- --- --- ---------
YHMHCI+NimbusRomNo9L-Reg             Type 1            WinAnsi          yes yes yes      5  0
TFZLUN+NimbusRomNo9L-Reg             Type 1            Builtin          yes yes yes      6  0

(note the different encodings).

I imagine a solution would be either to obtain or create a TTF version of Nimbus and embed that, to use a different Times font (personally, I recommend TeX Gyre Termes – noticably nicer to look at than Nimbus IMHO) or an alternative creation method (pdflatex, xelatex, lualatex, Inkscape, …).

I suppose it's a LO problem, but that'd need testing against more Type1 fonts.

Last edited by jsoy9pQbYVNu5nfU (2015-09-22 01:17:28)

olive · 2015-09-22 06:36:37

I can reproduce the problem. This seems to appear only with Nimbus Roman No9L and libreoffice. Abiword renders the file correctly. It is also correctly rendered if you print the file to a ps in Libreoffice file and then convert the ps to pdf with ps2pdf. Seems there is a bug in Libreoffice, but it is strange that it only seem to appear with this particular font. I have not seen the problem submitted, probably because we ussually don't use this font (it is a ghostscript font), but it is still a bug. You can install Times New Roman legally with no Microsoft license with https://aur.archlinux.org/packages/ttf-ms-fonts/ (these fonts were once freely redistribuable)

dkremer · 2016-04-20 10:39:12

Hi, it seems this issue has not been solved with the latest issues.

I filled a bug report : https://bugs.documentfoundation.org/sho … i?id=99412

Please, if you have any additional information, feel free to make some comments about it.

Arch Linux

#1 2015-09-21 20:14:21

LibreOffice Writer messes up encoding when exporting to PDF

#2 2015-09-21 20:31:12

Re: LibreOffice Writer messes up encoding when exporting to PDF

#3 2015-09-21 20:58:49

Re: LibreOffice Writer messes up encoding when exporting to PDF

#4 2015-09-21 21:01:25

Re: LibreOffice Writer messes up encoding when exporting to PDF

#5 2015-09-21 21:13:18

Re: LibreOffice Writer messes up encoding when exporting to PDF

#6 2015-09-21 21:25:13

Re: LibreOffice Writer messes up encoding when exporting to PDF

#7 2015-09-21 21:42:27

Re: LibreOffice Writer messes up encoding when exporting to PDF

#8 2015-09-22 01:09:45

Re: LibreOffice Writer messes up encoding when exporting to PDF

#9 2015-09-22 06:36:37

Re: LibreOffice Writer messes up encoding when exporting to PDF

#10 2016-04-20 10:39:12

Re: LibreOffice Writer messes up encoding when exporting to PDF

Board footer