'State of Hebrew vowels in Linux'

[prev in list] [next in list] [prev in thread] [next in thread] 

List:       koffice-devel
Subject:    State of Hebrew vowels in Linux
From:       Ken Bloom <kbloom () gmail ! com>
Date:       2006-08-20 18:02:14
Message-ID: 200608201302.15640.kbloom () gmail ! com
[Download RAW message or body]

[Note to recipients: I had the great idea to write this up for LUGOD's 
vox-tech mailing list so that interested users could find it in the 
list archives. Thereafter, I decided it would be to CC: to the relevant 
developers as well.]

I've been looking for ways to typeset Hebrew with vowels (nikudot) in 
Linux with acceptable quality. The executive summary about doing this is 
that it's tricky to find appropriate software, considering that the 
cumlus[1] fonts treat PostScript as a real programming language[2], and 
not all software supports that. For software that doesn't support it, 
this means that vowel points will usually be placed about .5em away from 
where they should be, thus appearing underneath space between letters 
rather than directly underneath letters. Some software can do worse, by 
treating each vowel as a separate letter, thus adding extra spaces.

The text I am trying to typeset is a highly corrected version of a text 
from Hebrew WikiSource[3]. The most prevalant correction to the text was 
the addition of the meteg in the many words where it is appropriate.

I imagine some of the language here (specifically, the names of various 
nikudot) will go over the head of normal vox-tech readers, but this is 
mostly for Google, and for those interested.

OpenOffice
==========
Using the culmus fonts, nikudot are placed incorrectly (generally 
centered underneath two letters) when displayed on screen, and when 
printed to a PostScript file. (OpenOffice's letter placement algorithms 
must interfere with creating well-behaved postscript. Note that in 
Debian, OpenOffice doesn't depend on pango.)

When using Times New Roman (from the Microsoft Core Fonts), many nikudot 
are given their own full letter spaces.

KWord
=====

Many nikudot are placed correctly. Others are given their own letter 
spaces (a circle of dots indicates that a letter should be placed 
there). Still other nikudot are replaced with "missing character" boxes.  
This occurs regardless of the font.

AbiWord
=======
AbiWord has a dependancy on Pango. Using Times New Roman (a Microsoft 
core font), it places most nikudot correctly, but when a letter has two 
nikdot, for example when one of those nikudot is a meteg, they both end 
up centered under the letter, both in the same place. The cholam appears 
centered over letters, rather than at the left top corner of a letter.  
The text appers the same when printed to a PostScript file.

When using the Culmus fonts, the text has the same problems as 
OpenOffice+culmus fonts did.

Konqueror
=========
Technically konqueoror doesn't want to display my file, because it's a 
text file which it wants to hand off to a text editor. Konqueror display 
of the original data appears mostly correct, with a couple of nikudot 
that don't display (and display in one of the two ways noted as 
incorrect for KWord).

Printed in PostScript, the typesetting appears in print as it does on 
screen. Mostly correct.

It's possible that I corrected the few screwy nikudot in the text file 
that I was using to test.

Firefox
=======
Firefox also wants to hand the text file off to an editor. The original 
WikiSource text appears correctly on screen, but when printed to 
PostScript, the font changes and the results appear like OpenOffice.

LaTeX: Makor2+Lambda
====================
The makor2[4] package doesn't accept native hebrew text, rather using a 
transliteration-like code to enter hebrew. I wrote a sed-based 
transliterator to convert from UTF-8 hebrew to this transliteraton code, 
and the results are spot-on and beautiful.

[Implementation note: when installing on TeTeX 3.0, instead of following 
the directions given and putting psfonts.mkr2 into your psfonts.map, the 
correct procedure is to run:
updmap --enable MixedMap /path/to/psfonts.mkr2]

LaTeX: Nikud+eLaTeX
===================
Nikud[5] is a package for eLaTeX to use the Culmus fonts to typeset 
hebrew with nikudot. The results in xdvi look like the OpenOffice 
results I have described above, but when compiled to PostScript, the 
fonts' native nikud rules come into play, and the result is a beautiful  
PostScript file. Running ps2pdf on the PostScript file segfaults.  
Compiling using pdflatex gives results which look like OpenOffice's 
results, probably because pdflatex can't run the PostScript code that 
does the nice nikud placement.

[Implementation note: when installing on TeTeX 3.0, install the fonts 
map using:
updmap --enable Map /path/to/culmus.map]

Executive summary:
the "Nikud" package is the way to go if you want native hebrew input, 
and beautiful output in PostScript only. The makor2 package gives 
beautiful output for any output format, but doesn requires a somewhat 
wierd syntax for hebrew.

Screenshots are attached.

--Ken

[1] http://culmus.sourceforge.net/
[2] http://groups.google.com/group/comp.text.tex/msg/1de2968ac005719e
    http://www.cs.tau.ac.il/%7Estoledo/Pubs/vowels.ps
[3] http://tinyurl.com/mr5ye (provided as TinyURL because the URL
    because the URL includes hebrew)
[4] http://www.ctan.org/tex-archive/help/Catalogue/entries/makor2.html
[5] http://nikud.berlios.de/

-- 
Ken Bloom. PhD candidate. Linguistic Cognition Laboratory.
Department of Computer Science. Illinois Institute of Technology.
http://www.iit.edu/~kbloom1/

["abiword-frankruehlCLM.png" (image/png)]
["konqueror-defaultfont.png" (image/png)]
["kword-frankruehlCLM.png" (image/png)]
["makor2.png" (image/png)]
["nikud-dvips.png" (image/png)]
["nikud-pdflatex.png" (image/png)]
["nikud-xdvi.png" (image/png)]
["openoffice-frankruehlCLM.png" (image/png)]
["openoffice-timesnewroman.png" (image/png)]
["abiword-timesnewroman.png" (image/png)]
["firefox-postscript.png" (image/png)]
["firefox-screen.png" (image/png)]

_______________________________________________
koffice-devel mailing list
koffice-devel@kde.org
https://mail.kde.org/mailman/listinfo/koffice-devel

[prev in list] [next in list] [prev in thread] [next in thread] 