 |
Journal | Friends | Calendar | User Info | Memories |
Indic LyX
Testing Arab-LyX and CJK-LyX.
The big problem is the X-setup. Cant get my KDE to render arabic or CJK in a smooth way.
since LYX-qt is qt based we need KDE rendering proper. I have KDE-farsi, devanagari etc. installed. Working for Gnome-based applications.
testing ->
this is devanagari --> मोजीला
this is arabic --> نيقنت سنشعضىيس تسعقضشح
(no idea what it means but it looks beautiful and did type right-to-left)
(15 June) A user has been created on my machine for Indic_Lyx work. the group can access it via ssh and do work and testing. The TeX tree and such things will be on my machine.
Will look at the metafont format, ability to convert betn diff formats, how metafonts are used in the ucs package, etc.,
Current mood:
energetic.
1. http://www.unruh.de/DniQ/latex/unic
Unruh claims he has solutions. Guys test it.
Looks like the text widgets used by the different versions of lyx
would be interesting to study - seems like some of the issues of
displaying indian language text will be solved by using a suitable
text widget. I will lookup the current codebase and then that of the
arab lyx group.
Another area where things really need to be clarified is the fonts section - currently unsure of which fonts are used - the formats and how to convert between the mainstream font formats and those used by latex.
Current mood:
tired.
To: Amitabh Trehaan <amitabhtrehan@softhome.net>
cc: John Smith <john.smith@oriental.cam.ac.uk>
Subject: Re: [Devnag-general] devnag problem
On Wed, 9 Jun 2004, Amitabh Trehaan wrote:
> Thanks for the interesting mail. I suspected A. Pandeys uses
> M$Windows ;-). Problem is Omega is Omega (a slow giant). Devnag does
> work with it without a preprocessor, thats true. My problem would be
> solved rather more easily if I could get something that converts
> unicode characters directly to devnag codes (not the
> transliteration). Then, I can use a Unicode font for display and
> output using devnag in LaTeX. Do you know of any such code around?
No, and I don't think that's the way to approach it. You may regard
the preprocessor as an annoyance, but it is your friend, carrying out
a very complex task behind the scenes for you. Any code that converted
from Unicode Devanagari to devnag codes would have to incorporate all
the same trickery. Much better to convert to devnag input
representation -- a relatively simple job -- and run the preprocessor
as usual.
Best wishes,
John Smith
Subject:
Re: [Devnag-general] devnag problem
From:
John Smith
Date:
Wed, 9 Jun 2004 11:40:25 +0100 (BST)
To:
Amitabh Trehaan <amitabhtrehan@softhome.net>
CC:
John Smith
On Wed, 9 Jun 2004, Amitabh Trehaan wrote:
>> Hi Prof. Smith,
>> Yes. I figured out the problem after mailing this. Sorry for not
>> updating the list. There are two asides to this:
>> i) How did this creep up into the distro. Some developer using windows?
>> I last got this same problem 3 years ago, never with any release after
>> that. Is there a generic solution or cross-platform text will remain
>> with this problem and the program will have to handle this?
I think that if you download the files over the Web, your browser will
(/may/should) adjust text files appropriately, whereas if you do a
bulk download via a zipfile or similar, this won't happen. Anshuman
Pandey, who has been in charge of the devnag distribution, uses
Windows, hence the 0x0D characters. There's no generic solution to
this very irritating problem.
>> ii) Is it better to use
>>
>> "tr -d" (Is tr -d '\r\032' correct?) option to remove \r or use "tr
>> -s '\r\n' '\n'". In other words, Is there any other use of \r in
>> text files?
No, there's no other use. (The "\032" additionally removes the ^Z
character that older Microsoft software inserted to mean "end of file"
-- I don't think they use it any more, but it does no harm to zap it
anyway.) By the way, your command "tr -s '\r\n' '\n'" doesn't do what
you think it does: "tr '\r\n' '\n'" converts every 0x0D to 0x0A (so
you now have two linefeeds in succession); the -s option "squeezes"
repeated sequences of characters, thus reducing the two linefeeds to
one.
>> Have you read my mails on the devnag LyX work that I have started (
>> http://blog.sarai.net/community/indic_ly
>> expert opinion.
I'm vaguely aware of this, but haven't seen it. I really know nothing
about LyX, so my opinion would not be very expert at all.
>> Also, I am looking fo ran answer to why the devnag
>> pre-processor has to be there. Can it not be integrated in some
>> ArabTeX sort of way (I think Arabic is an even more complex
>> script!). Excuse me if I have missed something very elementary.
I know that the Omega people are integrating the preprocessor
functionality in some way in their version, but I don't know any more
than that -- and since they haven't talked to the devnag people about
what they are doing, they have effectively caused a fork in the
development. I haven't a clue how ArabTeX works, but even if it were
possible to implement the devnag preprocessor functionality via TeX
macros, it would be an enormous undertaking -- and not one that would
come anywhere near the top of my list of priorities.
Best wishes,
John Smith
Subject:
[Tugindia] Why a Devnag Pre-processor?
From:
Amitabh Trehaan <amitabhtrehan@softhome.net>
Date:
Mon, 07 Jun 2004 19:04:43 +0530
To:
TUGIndia Mailing List <tugindia@tug.org>, devnag-general@lists.sarovar.org
Well. this may be an obvious question (but may not be too obvious also):
Why actually is a devnag pre-processor required? Is it not possible to integrate it like ArabTeX for instance. I was playing around with LyX and trying to write devnag code and because of the requirement of devnag preprocessor to have an input file with extension .dn, the only way I am able to run it is through an external script called through lyx ;-
--
cp $1 $1.dn
devnag $1.dn $1
latex $1
---
(pretty ugly, huh) $1 is filename.tex generated by lyx.
Attempting to run devnag code in LyX.
step-1: go to Layout..Document.. Preamble and add the line \usepackage{dev}
step-2: for devanagari output, use insert TeX option and insert text between {\dn and }
step-3: How to Compile? Urrgh.. 1 ugly solution.
Go to Edit...Preferences..Converters
and LaTeX->DVI option and put something which can handle devnag. since devnag is a seperate preprocessor program which looks for a file with .dn as extension, as yet I have done it by writig a small shell script. This I call by giving the Converter option /amitabh/Sarai/LyxProject/devlatex.sh .
The script goes as:
--
cp $1 $1.dn
devnag $1.dn $1
latex $1
--
even the devnag output is not stable as yet!!! (needs checking).
1) concentrate on the QT-frontend (not X-forms).
2) As we figured, there is no relation between the screen fonts and the document fonts, so we can work on both independently.
3) (simple part) - Install the package devnag into LyX and use the scheme presently used i.e. velthuis transcription scheme *** you input the text between {\dn } braces in a (very nice) phonetic latin input scheme and it is converted to devanagari in the output ***. This should work and it will be a test which proves the present state of health of the system is fine.
4) (The project actaully...) pick up a choice of input method/screen fonts and get LyX to show them on the screen. Find a way to store the Unicode text, and convert to devnag text for the TeX file. (Angus comments on CJK and how CJK handles useful here).
5) Angus comments on the CJK patch --> 'The patch basically replaces data stored in chars with
wchar_t and enables the user to use one of the normal CJK input methods.' This is probably what we are also looking for. He has also suggested understanding src/frontends/qt2/qfont_loader.C and src/frontends/gtk/xftFontLoader.C.
That's the plot in brief,.. have I missed something. Comments please.