Red Hat Bugzilla – Bug 128137
RFE: please add aspell-bg
Last modified: 2014-03-16 22:46:48 EDT
I've created a rpm package from
http://bgoffice.sourceforge.net/aspell/ for aspell - aspell-bg. It is
I used aspell-en-0.51-6.1.ker.rhel3.src.rpm's spec file as template.
The size is of the source package is 1.7Mb, binary package is 13Mb
and when installed it takes 33-36 Mb. I belive we'll translate the
required po files in time to include Bulgarian in FC3 and aspell-bg
will be good to have too. I've tested the package while translating
system-logviewer and it works with kbabel, kwrite, kmail...
Created attachment 102343 [details]
aspell-bg spec file
It only needs
As Fedora Core 3 test2 will have Bulgarian included it would be nice
to have this one too.
I've recompiled aspell-bg for FC3t1 and it is available from
http://mr700.cjb.net/pub/fc/3/i386/aspell-bg-4.0-0.i386.rpm. I did
some tests and it works fine with the following settings in kde:
and 'Consider run-together words as spelling errors' selected.
Is there something more that must/can be done?
> Is there something more that can be done?
On a related note:
gnome-spell needs some fixing (like adding bg as a known language)
that would allow spellchecking in Evolution and other gnome apps.
I've managed to hack this together, but it was just a hack and not in
line with the extensive patching thats done in the Fedora gnome-spell rpm.
Somehow adding the spelling package without being able to support it
in the default desktop seems backwards.
Thanks for aspell-bg rpm.
I use only KDE (and it works fine with this package). Can you
provide more details about gnome-spell? What did you hack?
I added Bulgarian in gnome-spell-1.0.5/gnome-spell/dictionary.c,
but this did not help.
PS: does this gnome problem belong here or a new bug against
gnome-spell should be entered?
*** This bug has been marked as a duplicate of 129520 ***
I reopen this bug since it's not fixed - there's still no
aspell-bg in the development tree. I have attached a spec file to bug
# 129520, so I don't attach it here again.
It should be in the development tree for fc3 now.
Many thanks for aspell-bg, but I reopen this bug because there are
problems with the package (If I should open separate bug please tell me).
I see your aspell-bg package is based on aspell-bg-0.50-0.tar.bz2,
which is older than aspell-bg-4.0.tar.gz. The problem with 0.50 it is
that it uses koi8-r (our 'official' encoding is cp-1251) and
'soundslike none' in bg.dat. I tought I can iconv the dictionary, but
there's another big problem with it - it includes a huge (about 1/2)
number of wrong words. You can verify this with this command (bg.cwl
is from aspell-bg-0.50-0.tar.bz2):
word-list-compress d < bg.cwl | iconv -f koi8-r -t utf8 | sort >
At the end of the file (bg.cwl.txt) you'll see a number of words
that start with two equal letters 'Ñ'(ia) like 'ÑÑÐ¼Ð°ÑÐ°' - there's no
such word in Bulgarian.
In aspell-bg-4.0 these problems are fixed and it includes about
65000 words. There are also aspell5-bg-4.0-0 and aspell6-bg-0.02, so I
contacted Radostin Radnev and am waiting for response. When I know
what package should be used (I hope in 2-3 days) I'll attach a new
spec file and add the information I've gathered.
PS: all files can be downloaded from bgoffice's sourceforge page -
http://bgoffice.sourceforge.net/ or directly -
http://prdownloads.sourceforge.net/bgoffice/. Thanks again.
I got response from Radostin Radnev. He says that all version
except the last one on bgoffice.sf.net are wrong. He says all previous
versions were converted from the ispell's author to koi8-r and thet
don't work (infact, they have huge amount of mistakes).
I assume that by 'last version' he means aspell-bg-0.50-2, because
all other are released on the same date too
I'll not attach the whole spec file, because the original one
needs just these changes:
--- cut ---
%define langrelease 2
--- cut ---
PS: I see there are aspell-bg-4.0.tar.gz, aspell6-bg-0.02.tar.bz2
and aspell5-bg-4.0-0.tar.bz2, but the files inside do not differ (only
comments and unesential files). I tested the new package and it works
fine with encoding CP-1251 as it must be.