[SOLVED] Can't use Century Schoolbook L font on Ubuntu 20.04 even though its .pfb file is shown by fc-list

gonthcarovd · 06-25-2020, 05:17 AM

Hello,

I'm running Ubuntu 20.04 as a VM on a MacOS host.

I require the Century Schoolbook L font (and likely others in the future). I checked that this font is present under the form of a .pfb file on my system:

Code:

$ fc-list | grep Century                            
/usr/share/fonts/type1/gsfonts/c059013l.pfb: Century Schoolbook L:style=Roman
/usr/share/fonts/type1/gsfonts/c059033l.pfb: Century Schoolbook L:style=Italic
/usr/share/fonts/type1/gsfonts/c059016l.pfb: Century Schoolbook L:style=Bold
/usr/share/fonts/type1/gsfonts/c059036l.pfb: Century Schoolbook L:style=Bold Italic

In addition the /usr/share/fonts/type1/gsfonts/ folder contains a .afm and a .pfm file for each .pfb file.
However, there is no match for this font in the font cache:

Code:

$ fc-match Century
DejaVuSans.ttf: "DejaVu Sans" "Book"

Here are the permissions of one of these files:

Code:

$ ls -l /usr/share/fonts/type1/gsfonts/c059013l.pfb
-rw-r--r-- 1 root root 106309 Okt 31  2017 /usr/share/fonts/type1/gsfonts/c059013l.pfb

I have tried rebuilding the cache multiple times to no avail with:

Code:

% fc-cache -v -r

and

Code:

% sudo fc-cache -v -r

I have read that this might be due to a permission issue. Therefore I tried copying the /usr/share/fonts folder to $HOME/.local/share/fonts and rebuilding the cache, with no success.

I did have success with installing Microsoft fonts by copying the relevant .ttf and .TTF files into /usr/share/fonts and rebuilding the cache with the above command.

Code:

$ ls msfonts                                                 
AndaleMo.TTF  AriBlk.TTF   couri.ttf     Georgiaz.TTF  Times.TTF     Verdanab.TTF
Arialbd.TTF   Comicbd.TTF  cour.ttf      Impact.TTF    Trebucbd.TTF  Verdanai.TTF
Arialbi.TTF   Comic.TTF    Georgiab.TTF  Timesbd.TTF   Trebucbi.TTF  Verdana.TTF
Ariali.TTF    courbd.ttf   Georgiai.TTF  Timesbi.TTF   Trebucit.TTF  Verdanaz.TTF
Arial.TTF     courbi.ttf   Georgia.TTF   Timesi.TTF    Trebuc.TTF    Webdings.TTF

For Microsoft fonts the cache can match the correct font for Arial:

Code:

fc-match Arial                                                
Arial.TTF: "Arial" "Regular"

This makes me wonder:

Can I only install fonts using .ttf or .TTF files?

If .pfb files (postscript fonts?) serve no purpose, why are they shown by fc-list?

How do I proceed?

shruggy · 06-25-2020, 07:59 AM

Quote:

Originally Posted by gonthcarovd

Code:

$ fc-match Century
DejaVuSans.ttf: "DejaVu Sans" "Book"

Because Century doesn't match the font family:

Code:

$ fc-match "Century Schoolbook L"
c059013l.pfb: "Century Schoolbook L" "Roman"
$ fc-match :postscriptname=CenturySchL-Roma
c059013l.pfb: "Century Schoolbook L" "Roman"
$ fc-match "Century Schoolbook L":style=bold
c059016l.pfb: "Century Schoolbook L" "Bold"

gonthcarovd · 06-25-2020, 08:53 AM

You are right, I mistakenly thought that the font name didn't have to be exact.
I get the same output as you for "Century Schoolbook L".

Nevertheless, the font is still unavailable in LibreOffice so I fear it's not quite available on my system yet.
(My real application involves fine tuning a Tesseract 4.0 OCR model whose training data requires this font.)

shruggy · 06-25-2020, 10:33 AM

Ah yes, see the LO bug #104701, particularly comments 4 and 6 there:

Quote:

Originally Posted by Adolfo Jayme

Support for old Type-1-format fonts has been dropped. Ensure the URW fonts you have installed are SFNT (TrueType or OpenType).

Quote:

Originally Posted by V Stuart Foote

the non Unicode Type 1 PS fonts have been completely dropped with ESC approval

It seems AbiWord, Calligra Words and GIMP still support them.

Alternatively, you may use the newer Unicode variants of these fonts from the package fonts-urw-base35. They're supported by LO. E.g.

Code:

$ fc-list :foundry=urw:style=roman
/usr/share/fonts/type1/gsfonts/c059013l.pfb: Century Schoolbook L:style=Roman
/usr/share/fonts/type1/urw-base35/C059-Roman.t1: C059:style=Roman
/usr/share/fonts/type1/urw-base35/P052-Roman.t1: P052:style=Roman
/usr/share/fonts/type1/gsfonts/p052003l.pfb: URW Palladio L:style=Roman

C059 is basically the same as Century Schoolbook L, and P052 is basically the same as URW Palladio L (compare the file names for Ghostscript fonts).

gonthcarovd · 06-26-2020, 01:49 AM

I downloaded the Century fonts from from this GitHub repo.

Now the Century Schoolbook is available in LO.

However, for my real application I still need the font "Century Schoolbook L Medium".

I tried downloading another set of Century fonts from wfonts. The font still isn't available.

Would you know where I could find this particular font?

shruggy · 06-26-2020, 07:53 AM

Sorry, but you've lost me here. Why are you trying to download the font with this exact name? As I said, you have a bug-for-bug clone of Century Schoolbook L in C059 from fonts-urw-base35. Besides, you have a reworked and enhanced version of it in TeX Gyre Schola from fonts-texgyre.

Just specify either of them in the --fontlist option to tesstrain.sh (and the appropriate location in --fonts_dir):

Code:

./tesstrain.sh --fonts_dir /usr/share/fonts --fontlist $(fc-list -f '%{?fullname{"%{fullname[1]}" }{}}' C059) ...

or

Code:

./tesstrain.sh --fonts_dir /usr/share/texmf/fonts --fontlist $(fc-list -f '"%{fullname[1]}" ' 'TeX Gyre Schola') ...

Alternatively, you can adjust the font list in src/training/language-specific.sh to your liking. E.g., to reflect the actual name of the font face you have installed:

Code:

sed -i '/Century Schoolbook L/s/Medium/Roman/' src/training/language-specific.sh

gonthcarovd · 06-29-2020, 01:05 AM

I guess I wanted to download the font with the exact name because I didn't know how to configure Tesseract to use the clone font. Your comment explained it very clearly. I managed to create the desired training data. Thank you for your help! It really got me forward