1. It is not necessary to change your locale to an Arabic locale to enable Arabic, however it is useful to have a default locale with UTF-8 encoding. To find out what your current default locale is give the command "locale". The easiest way to change your locale is to use the command "dpkg-reconfigure locales". You must be root to do this. This command launches a program that lets you add as many locales as you wish from a list. Hold the control key down to select more than one locale. You will then be asked whether you want to make one locale the default locale. If you choose an Arabic locale as the default you will Arabicize the desktop and many menus. If you choose "none", POSIX will become your default locale. This is not a good idea because POSIX does not support unicode (UTF-8) encoding. So choose a locale for any language with UTF-8 encoding. After you have selected your default locale you will have to reboot before the change goes into effect. If you wish you can edit the /etc/locale.gen file by hand. You will find a list of all the possible locales in /usr/share/i18n/SUPPORTED. Then when you have edited the file you must run the command "locale-gen" to generate the locales you have added to /etc/locale.gen. You can then use the command "locale -a" to see a list of the locales that are now available. You will still have to run "dpkg-reconfigure locales" to set your default locale.
 
2. Your next step after setting your default locale is to install some Arabic-script font packages from your Debian dvd. To see what packages are available use the command "apt-cache search Arabic" or "apt-cache search Farsi". Then install the packages you want by using "apt-get install [.....]". For Arabic fonts you should install the ttf-kacst and ttf-arabeyes packages. For Mozilla and Firefox install the mozilla-firefox-locale-ar package. You may also want to install katoob, an Arabic/Hebrew editor that also works with Persian if you have the Persian keyboard loaded. Most of these fonts will be stored in /usr/share/fonts/truetype. Additional Arabic-script fonts can be downloaded from:
 
WAZU JAPAN's Gallery of Unicode Fonts
 
Several of these fonts contain the full range of Arabic Unicode characters. They are Ariel, Lateef and Scheherazade. If you want to use Arabic-script languages other than Arabic and Persian you will need these fonts. Also you will need Lateef or or Scheherazade if you want to use vocalization (tashkil) with Arabic. Many of the other Arabic fonts are not able to join vocalized characters correctly. The fonts to be downloaded will probably be zipped. If they are truetype fonts put the zipped file in /usr/share/fonts/truetype and unzip it. Make sure that the unzipped files are readable by users other than root.
 
3. Finally you must load the Arabic and Persian keyboards for X. To load the Arabic keyboard you use the setxkbmap command as follows:
 
setxkbmap -v -rules xfree86 -model pc104 -layout "us,ar" -option "grp:alt_shift_toggle" -option "grp_led:caps"
 
If the command is successful in loading the Arabic keyboard, you should get the following message:
 

Warning! Multiple definitions of rules file
         Using command line, ignoring X server
Warning! Multiple definitions of keyboard model
         Using command line, ignoring X server
Warning! Multiple definitions of keyboard layout
         Using command line, ignoring X server
Trying to build keymap using the following components:
keycodes:   xfree86+aliases(qwerty)
types:      complete
compat:     complete+leds(caps)
symbols:    pc/pc(pc104)+pc/us+pc/ar:2+group(alt_shift_toggle)
geometry:   pc(pc104)

You can then toggle between English and Arabic by pressing the shift key and the alt key at the same time. When you change to the Arabic keyboard the caps-lock light will go on. When you return to the English keyboard the caps-lock light will go off. You can put the command in your .bashrc file, but make sure it is all in one line. To load the Persian keyboard use the same command but change "us,ar" to "us,ir". If you wish you can even combine the two commands by including all three languages (us,ar,ir) in the same command. You can then toggle between all three languages, but the caps-lock light will stay on for both the Arabic and Persian keyboards and will only go off when you switch to the English keyboard. The language symbols that can be used in the setxkbmap command are actually the names of the keymap files used by setxkbmap. They can be found in /usr/X11R6/lib/X11/xkb/symbols and /usr/X11R6/lib/X11/xkb/symbols/pc. (/usr/X11R6/lib/X11/xkb is a link to /etc/X11/xkb/.) Some Arabic fonts, such as Nazli and Homa, also contain Persian characters but most do not, so if you want to write in Persian you will have to make sure that the font you are using has the additional characters needed for Persian. Further information on keyboards and keymapping may be found in the files /etc/X11/xkb/README, /etc/X11/xkb/README.config, and /etc/X11/xkb/README.enhancing. The same files are also in /usr/X11R6/lib/X11/xkb.
 
4. It is fairly easy to edit the existing ar and ir keyboard files. The files are found in /usr/X11R6/lib/X11/xkb/symbols and /usr/X11R6/lib/X11/xkb/symbols/pc. If you are using a generic pc keyboard edit the files in the /pc subdirectory of the /symbols directory. By editing a keyboard file you can remove characters you do not need and substitute for them characters that you do need. You can, for example, add characters needed for other Arabic-script languages. However, you will then have to install the appropriate fonts. You can also rearrange the position of the characters on the keyboard. To make a new keyboard, even one based on an existing keyboard, is much more complicated. Do not try it unless you know what you are doing. Information on editing and creating keyboards can be found on the following web sites:
 
Arabic on Linux (Oibane's website)
XKB Configuration by Ivan Pascal
XKB Setup by Ivan Pascal
Ali El Dada's Web Page
 
For Unicode encoding charts of all the characters used in Arabic-script languages go to:
 
www.unicode.org/charts/
 
5. Installing additional truetype Unicode fonts for other Arabic-script languages is easy. As mentioned above fonts with the full range of Arabic characters are available for downloading from:
 
WAZU JAPAN's Gallery of Unicode Fonts
 
The fonts will probably be zipped. If they are truetype fonts put the zipped file in /usr/share/fonts/truetype and unzip it. Make sure that the unzipped files are readable by users other than root. That's all you have to do. You should be able to use the new font in any editor or word processor.
 
5. Once you have enabled Arabic and Persian on your computer you will be able to create Arabic or Persian files and save them with Arabic or Persian names. The file names will appear in Arabic or Persian in desktop file manager programs, but if you use "ls" in a command-line terminal you will see only question marks for an Arabic file name.
 
6. The following programs work well with both Arabic and Persian. They all support the ISO-8859-6 code page for Arabic and the Windows cp-1256 and Unicode UTF-8 code pages for both Arabic, Persian and other Arabic-script languages. They do not support the Persian code page ISIRI-3342. The default encoding is UTF-8. It is a little tricky to open previously-written files written in encodings other than UTF-8 or that may have been created with different programs or with different operating systems. How such files can be opened is described in the paragraphs below dealing with the various word processors and editors.
 
Word Processors:
 
oowriter - Open Office Writer is a bidirectional word processor that works very well with Arabic fonts. You can align text to either the right or the left. This is useful of you want to align Arabic text to the left instead of to the right. You can also change the direction of the text. To open a previously-written file pull down the "file" menu and click on "Open". A window with a list of files will appear. Select the file you wish to open and then click on "File Type" and select "Text Encoded" near the top of the list. Another window will pop up asking you to select the encoding. You can also select an Arabic or Persian font in this window as well as the language. After you have selected the correct encoding click "OK" and the file will open. You can also select the font after the file has opened. Go to the "Edit" menu and click on "Select All", then change the font and the size. Oowriter supports UTF-8, Windows cp-1256 and ISO-8859-6. It does not support ISIRI-3342 for Persian. Oowriter allows you to choose between Arabic (Western) and Hindi (Persian style) numbers. You can set the numbers in the "Tools" menu under "Options" > "Language Settings" > "Complex Text Layout". If you set Hindi numbers you will also get Hindi numbers when you are writing in English. In fact, changing to Hindi may remove the ability to type any numbers at all. If this happens change back to Arabic or System numbers.
 
abiword - Abiword is another bidirectional word processor similar to oowriter. It does not however handle all Arabic fonts well. It cannot connect the letters properly in some fonts and may put periods, for example, at the beginning of a sentence instead of at the end. It does allow one to select the alignment of text to either the right or the left. To open a previously-written Arabic-script file first load Abiword. Then pull down the "File" menu and click on "Open". A list of files will appear in a window. Select the file you wish to open. At the bottom of the window you can choose the type of file to be opened. For Arabic and Persian files chose "encoded text" and then click OK. Another window will open which will allow you to indicate the code page used for the encoding. Select the code page. For Arabic you can choose cp1256, ISO-8859-6 or UTF-8. For Persian there is only cp1256 or UTF-8. Click on "OK". If you have selected the correct code page the file will open, but you will then have to choose the font before you will see any Arabic or Persian. Pull down the "Edit" menu and click on "Select All". Then change the font to an Arabic or Persian font.
 
Editors:
 
katoob - Katoob is a bidirectional Arabic/Persian and Hebrew UTF-8 editor with its own keyboard emulators for Arabic and Hebrew. You can, however, use the keyboards for Arabic and Persian that you have loaded with setxkbmap if you wish. Katoob does not have its own fonts. Make sure you have at least one font, such as Nazli, Homa or Arial, that contains Persian characters if you wish to write in Persian. Katoob aligns Arabic and Persian text automatically to the right and English text to the left. When opening previously written files Katoob tries to convert the file to UTF-8, so you must let it know what code page the file is in. It can convert from cp1256, ISO-8859-6 but not from ISIRI-3342. When you quit Katoob make sure that you are in US/Ascii mode; otherwise your Arabic keyboard will remain loaded and you will not be able to type in a terminal. There are manual pages for Katoob.
 
gedit - Gedit is bidirectional and works with all Arabic-script fonts. Like Katoob it automatically aligns text to the right or to the left depending on the language. This can be a problem if you want to align Arabic text to the left instead of the right. There is a full set of fonts. You can save files in UTF-8, ISO-8859-6, or cp1256 encodings. When opening previously-written files it is necessary to tell Gedit what encoding the file is in. To do this you must open a file by clicking on "Open" in the "File" menu. You will get a window with a list of files. At the bottom of the window is a list of encodings. The default is "Auto Detected". "Auto Detected" does not usually work so you must indicate the encoding of the file before clicking on the name of the file. The file will then open in the correct encoding. If you open a file from a shell command line the file will open without any Arabic or Persian characters. Gedit does not support ISIRI-3342 for Persian.
 
bluefish - Bluefish is an html editor that works with Arabic script languages.
 
KDE Editors:
 
If you have a problem opening a file in these programs because they can't communicate with klauncher give the command "kdeinit".
 
kedit - Kedit works with both Arabic and Persian. It automatically aligns text to the right or to the left depending on language. Therefore you cannot align English to the right and Arabic to the left. There is a full set of fonts, To set the encoding pull down the "Settings" menu. Click on "Configure Kedit" and then on "ABC Spelling". You can change the encoding on the fourth line of the menu. For Arabic and Persian the only encoding provided is UTF8. To open previously-written files pull down the "file" menu and click on "open". A window will appear with a list of files. Select the file you want to open. Then click on "ABC" in the bar at the top left corner of the window. This will open another window in which you can indicate the encoding of the file. For Arabic and Persian you can choose cp1256, ISO-8859-6 or UTF8. If you have a problem opening a program because kedit can't communicate with klauncher give the command "kdeinit".
 
kate - Kate works with both Arabic and Persian. It does not align text to the right, however. It has a full selection of encodings and fonts. To set the font pull down the "settings" menu. Click on "configure Kate". Click on "fonts and colors" under "editor". Then click on "font". To set the encoding pull down the "view" menu and click on "set encoding". To open previously written files you must tell kate, as with the other editors, what encoding the file is in. Do this by opening files by clicking on "open" in the "file" menu. You will get a window with a list of files. At the right side of the window is a list of encodings. Click on the correct encoding before clicking on the name of the file. The file will then open in the correct encoding. If you open a file from a shell command line the file will open without any Arabic or Persian characters.
 
kwrite - Kwrite works with all Arabic-script fonts and various encodings. Like Kate, however, it does not align text to the right. When opening previously-written files make sure you first set the correct encoding in the upper-right corner of the list of files to be opened.
 
Browsers:
 
Mozilla, Firefox, Epiphany, and Konqueror all work with Arabic texts. Mozilla, Firefox, and Epiphany, however, cannot correctly join vocalized Arabic letters nor can they join Arabic letters from languages other than Arabic and Persian. Konqueror does not have either of these problems. Also it seems that Konqueror is the only browser that can print Arabic files.
 
Terminals:
 
If your default locale has UTF-8 encoding you can use Arabic and Persian in any of the terminals provided by Debian but without bidirectionality and shaping of the Arabic letters.
 
If you want bidirectionality and shaping you will have to install MultiLingual Terminal (mlterm). There are Debian packages for mlterm which you can download from the Debian website. You will need the following packages:
 
mlterm-common_2.9.2-2_i386.deb
mlterm_2.9.2-2_i386.deb
mlterm-tools_2.9.2-2_i386.deb

 
You may also have to download some unicode fonts. These fonts are in the following packages:
 
unifont_1.0-1_all.deb
xfonts-efont-unicode_0.4.0-4_all.deb

 
Put the mlterm packages in /usr/local/ and then, as root, install them with dpkg -i.
 
Install mlterm-common first, then the others.
 
The mlterm files will be installed in the following directories:
 
/etc/mlterm
/usr/share/doc/mlterm
/usr/share/terminfo/m/mlterm
/usr/share/mlterm
/usr/bin/mlterm
/usr/lib/menu/mlterm
/usr/lib/mlterm

 
If you find that you need the Debian unicode fonts, download them and install them with dpkg -i.
 
Once mlterm is installed you must configure it for Arabic with UTF-8 encoding. You will find the configuration files in /etc/mlterm/. There are other important documents in /usr/share/doc/mlterm/ which you should read, and there are also man pages for mlterm.
 
The main configuration file is /etc/mlterm/main. Make sure that you have the following lines in it:
 
use_bidi=true
ENCODING=UTF-8
input_method=kbd

 
As far as I know the only Arabic font that will work in mlterm on Debian is the one mentioned in the /etc/mlterm/font file, so uncomment the font for Arabic speakers:
 
ISO10646_UCS4_1 = -gnu-unifont-medium-r-normal--*-iso10646-1;
 
In the /etc/mlterm/key file add this line:
 
Shift+space=IM_HOTKEY
 
This line means that when you hit shift and then space the word "Arabic" will appear on the screen just under the cursor and you will be able to type in Arabic from your keyboard.
 
Make sure that the line Shift+space=XIM_OPEN is commented out.
 
Note, however, that in order for mlterm to work properly your default locale must be one with UTF-8 encoding. Use the command "locale" to check to see what your default locale is. If it does not have UTF-8 encoding change your default locale to one that does have UTF-8 encoding by using the "dpkg-reconfigure locales" command as described above. To launch mlterm from another terminal give the command "mlterm".
 
Not all editors will work in mlterm. The ones I have found to work are vi, vim and Joe's editor in its different configurations. Version 4.93 of pico, the new unicode version, will also work, as will alpine, the new unicode version of the pine mail program. For some reason nano does not work.
 
Mlterm will will also allow you to see and write Arabic file names, and to manipulate files with Arabic names just as you would files with ASCII names. And you will be able to read Arabic files with the "less" and "more" commands. If you have a Persian keyboard you will be able to do the same things in Persian. I have not tried out mlterm with any other Arabic-script languages except Jawi. With Jawi mlterm displays the additional Jawi letters but it cannot shape them.
 
Further information on mlterm may be found on the following web sites:
 
Mlterm Web Site
Oibane's pages on mlterm

 
TeX and ArabTeX:
 
TeX is a marvelous typesetting program created by Donald Knuth. The Debian distribution of Linux includes TeX in the following versions: tetex, latex, pdftex and arabtex. ArabTeX is a set of macros created by Klaus Lagally that works with plain TeX (tetex), LaTeX and pdftex. Since it is a self-contained program with its own Arabic, Persian and other Arabic-script fonts, you can use ArabTeX without enabling Arabic on your computer. One of the advantages of using ArabTeX is that it allows you to format Arabic poetry just as a printer would using old-fashioned handset type. Click here for a pdf file showing some Arabic poetry formatted with ArabTeX. Another advantage of TeX is that it contains a complete set of accents and diacritical marks useful in transliterating Arabic scripts into other alphabets.
 
Additional useful information on enabling Arabic on Debian Linux can be found on these sites:
 
Arabic on Linux (Oibane's website)
Arabic Howto on www.arabeyes.org
Web Page for Sharif Linux, a Persian version of Linux
Information on Persian word processors for Microsoft Windows (includes information useful for Linux users also such as a chart of the standard Persian keyboard in pdf.format)
www.openroad.net.au/support/notes/persian.html
 
Comments and corrections to this file are welcome. I can be reached at heer@u.washington.edu
 

 
---Nicholas Heer