Re: shaping arabic

[Nadim Shaikli <shaikli at yahoo dot com>]

Sorry Chris for the delay - we're in the midst of releasing the first
ever Arabic Live-CD distro and we all have lots of details to workout
which is very time consuming.  I'll answer your 4 emails here (do
please subscribe to arabeyes' "developer" list and post all future
inquires there).

--- Chris Whiting <chris at riverslime dot com> wrote:
> Actually, the ALEF MAKSURA appeared at the end of a word (i.e. on the left
> side).  So it seems that it should have used the 'final' version of the
> character.  I tested the routine with the attached file (I do not know what
> it says, but I was sent a pdf file of what the text should look like - see
> attached pdf file).  I reversed the characters before calling your routine.

I tested out what you mailed me and the results I got were consistent and
correct with what should happen, so I don't see what you see.

> I also found a few characters reversed with your routine so this past
> weekend I combined the functionality of the two routines (shape_arabic and
> arabjoin) and created a third.  The routine also does the following:
> 
> SOURCE:
> <Arabic1> <Latin1> <Arabic2> <Latin2> <Arabic3>
> 
> RESULT:
> <3cibarA> <Latin2> <2cibarA> <Latin1> <1cibarA>

I'm not a fan of arabjoin and I think it is your source of problems.
Dump it and use fribidi instead.

> > > Thanks for the module and any help that you can give me.
> >
> > BTW: you might want to look into these Arabic fonts (note their collage),
> >
> >        http://www.arabeyes.org/project.php?proj=Khotot
> >
> > Thanks for the heads-up and for any help with the CPAN inclusion.
> >
> 
> Regarding CPAN help, I can probably help out.  I never did it before
> but I have built .pm modules using the CPAN rules.  Let's just make
> sure that we get the agorithm correct.

Do tell me what's involved as I'm not looking to simply do this and forget
about it, but to learn and teach others hopefully (I'd rather go without
'arabjoin' though for now).

> I am using imagemagick (freeware) to test out this module.

Not sure how you are doing that - but there are other means to test
(use a browser - IE or mozilla).

> I am using BabelMap (freeware) to view the fonts online.  BTW I am using a
> Windows XP OS.

Never heard of BabelMap - have a URL for it ?

> I am still testing ShapeArabic and am looking for further advice on how to
> handle Arabic test that has embedded latin numbers, which I am told is
> fairly common.  You can see some latin numbers in the test file.
>
[snip]
>
> I did not realize that there was a bidi module already developed. 
> Unfortunately, the module is not on CPAN or ActiveState and there doesn't
> seem to be a binary for the C module.  For my purpose, I do not need a
> extensive bidi support.  I am told that latin numbers appear in Arabic
> script quite often.  I can remove thes 10 lines from the module that I
> sent to you easy enough.

You should check freshmeat.net as well.

> Attached is a set of tif files from a test that I made with your shape
> routine along with the test script.
> 
> I am using the same font as the ones that I sent yesterday.  I am no longer
> getting the condition that I discussed yesterday.  I am not sure why.
> 
> The 'some characters are reversed' comment still apply.  I am assuming that
> it is because the substitutions are not being done to combine or re-arrange
> certain characters.

I don't think I saw any screenshot (except the reference PDF that
you mailed) that looked right.  I'm not sure what you are doing
since you can simply pipe data into fribidi and then pipe fribidi's
output into my 'shape_arabic.pl' script and all should be well.
Creating your own Bidi implementation is problematic at best and
again I highly suggest you subscribe to the fribidi list as it
should (again should :-) get going again soon with the next
release (from what I hear a rather extensive rewrite).

BTW: to verify your results you can always view the text in your
     browser (make sure you have the proper encoding enabled - UTF-8).
     Both IE and Mozilla 1.4+ have decent Arabic support and so a
     clearer more decisive comparison should result.

PS: if you elect to mail me (or the list) screenshots or docs, do please
    place them on a website somewhere and provide me with a link (that
    would be best).  I'm rather positive if you run both fribidi and
    the perl shaping script all will work without any problems.

Regards,

 - Nadim


__________________________________
Do you Yahoo!?
The New Yahoo! Shopping - with improved product search
http://shopping.yahoo.com