The Arabic Mac and the Internet

Using Arabic on the Net has become more straightforward in recent years, so many Mac users do not need any special instruction for this, if you pick a browser and mail program such as those Apple has already put on your machine (Safari, Mail), then Arabic just works without any further ado. But for many, some care must sometimes be taken (e.g. old-time Eudora users, who could read Arabic emails on their old Macs, no longer have this possibility). Therefore, while the top part of "Arabic under OS X", is fairly short: "It works / It does not work", it may be of interest to look a little into the previous situations to see where the problems lay.
For a more detailed discussion of the background to these problems, see my note on character sets and their importance for the Internet. Below, the focus is on more practical information about using some of the more common network programs on the Mac for Arabic.

• Arabic on the net in OS X
• E-mail

  ---

• Email in Classic: Eudora
• Arabic on the Web under Classic
• Browsers
• Netscape
• Mozilla
• MS Internet Explorer
• iCab
• Writing HTML text in Arabic
• News: YA Newswatcher
• Telnet

• Net Resouces for the Arabist (text, 1996)

The Macintosh and Internet Arabic today

Arabic on the Web

Arabic texts on the Web are increasingly widespread, but although the Mac has been the primary computer for typing in Arabic since the 1980s, we were for a long time starved when it came to access to the Internet sites in that script. Happily, that time seems to have passed with modern Macs using OS X; all major Web browsers on the Mac (with one exception) will display Arabic automatically, you just click and read. You do not need to make any changes, installation or anything on your Mac to read Arabic in these browsers, everything is already pre-installed on Macs, although Arabic is better supported in the most recent Macs (OS X version 10.3 or higher) than in the earlier ones. Among these browsers that display Arabic in current versions are: Safari, Firefox, Netscape, Opera, iCab and OmniWeb. The exception is - of course - Microsoft. Internet Explorer does not support Arabic, and we must conclude that it most likely never will; avoid this program.

As you can read below, there are however innumerable different ways a web author can choose to display Arabic. Your browser will try its best to display the Arabic for you correctly. But if the author has made too many mistakes, things may come out incorrectly. If so, you must tell your browser how to display it. Many (not all) browsers have an "Encoding" item in the "Display" ("View") menu, pick either "Unicode" or one of the Arabic options (in Firefox, under "More: Middle Eastern"), until the text on the web page is meaningful. What has happened is that the author has not given the information the browser needs for automatic display, so you must set that yourself. However, this is more and more rare, Arabic will mostly appear automatically.

NB: "Bug" warning, Safari and Office 2004
Notice one problem some people have met which has a strange cause: If you have installed Microsoft Office 2004, the installer will also place in your system a corrupt version of the font Arial, which breaks up the Arabic of Safari. You will see Arabic letters, but disjointed. If this happens, just go to FontBook, deactivate Arial (and Times New Roman, which causes the same error), or replace it with the Apple-provided version of Arial that came with your machine, and Safari will be back up and running in Arabic.

Arabic in E-mail in OS X

Of the probably most common email programs on the Mac OS X, Apple's Mail, installed on all Macs, will display Arabic properly and without any special settings (in order to write Arabic yourself, you must of course activate Arabic typing on your Mac, see The Arabic Mac). The former most common program, Eudora, does not support Unicode and thus not Arabic script in OS X. See below for more information on Eudora in Classic, and how to make it read and write Arabic.

Arabic internet before OS X

E-mail: How to Arabize Eudora in Classic / OS 9

Apple's Mail, the OS X choice, does not work in the older OS 9, but there you can make Eudora, the other stalwarth, display and write Arabic, with some modifications. This is a run-down, for those who are still running pre-OS X software:

Standard Eudora from version 2 supports WorldScript and Arabic almost correctly both in the (free) Light and (commercial) Pro/Paid versions (there are some minor glitches with the arrow key direction and non-standard font sizes). In order to send and receive Arabic correctly, you must install the Arabic EudoraTables file, which you will find elsewhere on this server (click here to download). This translates between the various non-Mac versions of Arabic out there on the net and our kind of Arabic. Follow the instructions in the ReadMe file.
From Eudora vs. 4.3, it will also accept non-European scripts using the operating system's "text encoding converter" (TEC, part of OS 9 and possibly also under OS 8.5), but only if the email message coming in has been properly labelled by the sender application. However, the EudoraTables still work under Eudora 4.3, so that conversion can also be set manually.

In addition, you must set up Eudora to display the message in an Arabic font. This is the manner you do this:

Eudora Light / Pro versions 1 and 2

• Open Special: Settings: Fonts (in the oldest Eudoras, the Settings menu was called Configuration). Choose an Arabic font.
• In Settings: Sending Mail, uncheck (i.e. disable) the options "Fix curly quotes" (not in Eudora older than version 1.4).
• Open the Mac's Text Control Panel and choose: "Right to left". This will change all text alignment (including of English text!) to right. English text will still appear, but not other European special characters.

Eudora Pro 3 and 4.

• Make sure the setting in Special: Settings: Styled Text (at the very bottom of the Settings panel) is set to "Send style information with mail" and "But warn me first" (crosses in both). Otherwise, the options below will be grey.
• Click on the Editing ("pencil") icon at the top of the message's window.
• Select the text you suspect is Arabic. Go to the Edit: Text: Font menu and choose an Arabic font.
• Select Alignment to be Right-adjusted from the same menu.

A common, but unfortunate problem is caused by Arabic Windows users who write in their own kind of non-standard Arabic, but let their mail program label the message as "English". This is wrong on their part, and will lead to their message being incorrectly presented on the Mac. If this happens, what you must do is in three steps:

• First choose the Translation menu Repair ISO to undo their damage.
• Then Save the message.
• Finally, choose the Translation menu MacArabic <- WinArabic.

As for alignment, there is an unfortunate but unavoidable side-effect of the fact that Macs and Windows etc. are different. The Mac has separate Arabic period and colon characters. Windows does not have this, they have a common English and Arabic period. Therefore, any period, colon or similar punctuation in an Arabic text from Windows (or Unix) will cause the line to be broken in two, before and after the period. The text before the period will be displayed to the left, and the text after to the right of the period (as if it had been English). Alignment will not change this, nor is there any other way to do so, short of manually replacing any period, colon or other punctuation with Arabic ones. (Arabic colons, semicolons and question marks do not cause this to happen, as they are separete also in Windows). Annoying, but apparently inevitable. [If it is a long text, I would probably copy the passage and paste into Nisus or another Arabic-capable word processor that does not have this problem].

To write a message in Arabic, just choose an Arabic font from the Text: font menu and type away. Make sure that you choose an Arabic table from Change: Conversion, preferably Mac -> ISO Arabic (the standard choice). When sending, you will be asked whether to include Style information or not; I would normally choose to remove style information, as few other programs can use this (if you are sending to another Eudora Mac vs 3, however, it may be useful for the receiver).

Eudora Light

Although Eudora Light shares most of Eudora Pro's resources, and can actually display multiple fonts, it doesn't have the ability to change them. Thus, it lacks the Edit tool and Text menu, which are crucial for Arabic. So, in effect, you are reduced to the choices of Eudora 1/2 above, and set the Arabic font as a default font in order to read it.

Arabic on the Web before OS X

Those of us who may still be using old Macs, will find things a bit more complex, and the rest of this page concerns you, partly historical from OS 7 onwards, but may perhaps be readable as a history lesson.

• As images. Because of the problems of reading Arabic, many of those who write Web pages in Arabic (the "providers" of text) have made them into "pictures", where the full page has been saved and is presented to us as an image. It is a bit slower to download, but it does not require anything special from the machine that is to view them. You do not need to have Arabic of any kind installed; you can use any web browser: A picture of an Arabic text page is no more demanding to display than a picture of a house. Web images are often saved in a format known as "GIF", so a Web page with such image text is known as a "GIF" option.
• As Acrobat files. Basically similar to an image, but more sophisticated is to present the text as an Acrobat (or "pdf") file. Again, it does not require the browsing machine to have any sort of Arabic fonts installed, the pages can be viewed on any English Mac or PC. The format is also technically superior to GIF images, as pages can be zoomed for clearer reading etc. The Arabic daily al-Hayat in London puts their newspaper on the Web as Acrobat images. All you need is the plug-in or utility called Acrobat Reader, which is part of new web browsers as standard, or can be downloaded separately. You do need version 3.0 or higher of Reader to read Arabic, however.
• As text. Now we are getting into more demanding stuff. If the Arabic is presented as text, like English text, the browsing machine (yours) must have the fonts and extensions/resources etc. to present Arabic, and you may have to instruct your web browser to use Arabic fonts for the particular page when you encounter it. However, unlike the two previous options, you get full Web functionality when you use text rather than image or plug-in files, with links and indexing etc., so that will still be preferred by many users. Text files are also much smaller and download faster. The problem is, as indicated above and detailed more in the Non-European on the net page, that there are several systems for writing Arabic, most importantly Mac differs from Windows Arabic, and if the provider has used one, and the browser only can read another, you can get into trouble: the fonts on your machine do not "match" those of the provider, so instead of "misr" you see "kisr" or "hisr". There are basically three different versions of displaying Arabic that we need worry about:
  • ISO Arabic, or technically "iso-8859-6". This is the offical standard that should be used on the Net, and as it happens it is also what the Macintosh uses for its own Arabic (by and large. The Mac uses a few more characters than ISO. Unix also uses ISO). Mac users can therefore with justice feel superior and say that everyone who publishes Arabic on the net should use the Mac system. But of course they don't.
  • Windows Arabic, technically "cp-1256". As we might suspect, this is probably the most commonly used Arabic text format on the net. It is slightly different from ISO (the earlier letters of the alphabet, down to dad, are the same, but the latter are different). Until recently, no Mac web browser has been able to display WinArabic, this is about to change.
  • Unicode, the "universal" character set which includes all possible scripts (you will often find it under the name "UTF-8" on the Net, "UTF-16" is another more demanding variant), is becoming increasingly common. Some browsers will support it under Mac OS 9 (and with additions, OS 8). OS X uses only Unicode, for OS X web browsers, see below.

Netscape

This is not true for previous versions of the program, a run-down for those considering such installations: Netscape 6.2 appears to display Arabic, but does not; the letters are reversed, so the text becomes unreadable.

Netscape 4.7 is "out of the box" only able to present ISO (that is, Mac) Arabic pages. It is not able to display Arabic Windows, except in one of the adaptations below, and not Unicode/UTF text at all. It also, like Internet Explorer, suffers from the "meta bug" that means that many of the ISO-produced pages are actually also unavailable. (This is also applies to earlier versions of Netscape, from 1.1, the first that was able to display ISO Arabic text). Thus Netscape seems to be the least capable browser today, at least in its standard form.

In order to make Netscape 4.7 display even those pages it understands in Arabic, you must tell it that you wish to use an Arabic font for such pages, Netscape is not capable of understanding this by itself. You set that in Netscape's "Preferences" menu (under "Edit"). Open this, and click on the Appearance: Fonts category. Where you see "For the Encoding" and a sub-menu, choose "User Defined" from that menu, almost at the bottom. Then, select an Arabic font [those with unreadable names!] for both Variable and Fixed width font. The bottom part should normally stay with "Use my fonts" options.

Then, each time you reach an Arabic page that you suspect Netscape may understand (it says "Mac Arabic" or something), go to the Character Set item on the View menu, and select "User Defined". The page should then reappear in the Arabic font, readable if Netscape can understand it.
(Historical note: If you are still using Netscape 1.1 from the early 1990s, it did not have a "User Defined" option. In that case, you have to change the settings for the item "MacRoman" to an Arabic font, and you probably have to make this visit to the Preferences each time you come across an Arabic page.)

• However, you will sometimes find that the page stubbornly remains in unreadable characters in what looks like Times, when it should have been in Geeza or al-Qahira or other Arabic. In that case, you have come across the "meta" bug that plagues all versions of Netscape up to 4 and, surprisingly, also Internet Explorer. It seems that when Netscape comes across a certain html code in a web page's header section (a particular "<meta charset>" header, which is correctly formed, but which Netscape does not understand), it will insist on displaying Times - which of course contains no Arabic - instead of the user's choice of an Arabic font. Regrettably, there is nothing to be done about this from the user's side.

As for Arabic Windows pages in Netscape 2-4, there are actually two adaptations that can be made so that these earlier versions can display such. One requires you to purchase something called the "WinArabic script" which adds fonts and resourcse adapted to Arabic Windows to the Mac Arabic system. This utility, which is produced by a Moroccan firm, will also work with Explorer and other Internet programs, but is best only to be used for Net access, it is not compatible with standard Mac Arabic and should not be used in standard word processing and other programs. It is to be found at http://www.d-ventures.com/, e-mail: dv@d-ventures.com (the full version costs about $40, I believe). User reports say it works, although with some limitations. I have not heard whether it works with Netscape 6.

The other way is free, but requires you to modify the Netscape application itself (versions 2, 3 or 4). It is actually quite safe, but you should preferably be familiar with the ResEdit application before you attempt this. If you do not know what ResEdit is, try to find someone who does. The process is as follows:

• You must download the relevant WinArabic "xlat" resource for your version of Netscape.
  • If you are using Netscape 2 or 3, download this file.
  • If you are using Netscape 4.7, download this file (I have not tested the earlier Netscape 4 variants).
• Open [a copy of] Netscape as well as the xlat file in ResEdit.
• Copy the two resources from the file into Netscape xlat resource, telling it to replace the existing with the same numbers [which are "MacCyrillic -> iso-8859-5". Check the ID numbers, in NS 3 they are 4136 and 10256. If you often read Cyrillic pages, you may choose another encoding to replace, just change the ID numbers to what you want to replace].
• Open the Menu resource in Netscape, and open the resource item for the "Character Set" menu. Manually change the text of "Cyrillic (Win 1251)" to "Arabic (Win 1256)".
• That is it. Close and save.

One minor bug that I have found in many of these Arabic Windows solutions, both in Netscape and below, is in forms you fill in (such as "search"). Some servers do not properly understand Arabic text I ask for, when I say "find 'misr'", it replies "no "kisr" found". This may be the problem of the server, not the browser. Also, in many cases, you will find that alignment is not perfect, or that a period or parenthesis in a line can cause the words before the period to appear to the left of it, and those after to the right, causing some confusion. Registering the browser as "Arabic" in the Language registry may or may not help, as the case may be.

[Another quirk with Netscape version 2-4, quite unrelated to Arabic, is that it uses its own conversion format concerning European non-English symbols ("Latin 1"). Pure Latin 1 leaves some characters, about 30 of them, undefined. Both the Mac and Windows have defined these characters to contain certain symbols that Latin 1 does not include, such as "smart quotes", bullet, TradeMark symbol, Sigma, etc. All Mac network programs have agreed on a way to include these spare symbols, so they travel correctly from Mac to Mac, although they will not translate correctly to other kinds of machines. Except Netscape, which from version 2.0 changed its translation table for these undefined characters, for reasons known only to itself. Thus, Netscape 2 and up stands in a world of its own on the Mac side. This does not concern real accents etc., which are properly defined, but can be a drag. For this reason, I have extracted the original Netscape 1 conversion table, and made into a resource file; whith the help of ResEdit you can install this into Netscape 2 to make it conform to the Mac practice. Click here to download. You must know ResEdit to perform this operation. Netscape 6 does not have internal conversion tables, it seems to take its character conversion from the Mac OS's TEC resources.]

Mozilla / Firefox

Internet Explorer

• In the category Font : "Default Character set for" (choose:) "Universal Alphabet" : select an Arabic font
• Web Content : Allow page to set font : Click to ON (default setting)

IE in older versions can accept ISO (Mac) Arabic, but not Windows Arabic, except in the same kind of adaptations as for older Netscapes above, with the WinArabic script from Morocco, or by modifying it with ResEdit. (I have not had tried this with the later versions of IE, but the same xlat as worked for Netscape, also worked with IE 2; someone with experience with ResEdit can no doubt modify the resource if needed to later IE support). As mentioned, IE seemed to suffer from the same meta bug as Netscape did; and it may even extend to Arabic Windows pages, unlike Netscape, but this is not properly confirmed.

iCab

iCab, can be downloaded from www.iCab.de. It is still in development, so not everything works perfectly. In earlier versions, some html codes on Arabic Windows (in particular &nbsp;) were displayed as code rather than interpreted, this has been corrected in the current versions. ICab plans to charge people $30 for the full version once it is fully developed (but say that there will also be a smaller free version for download). We will have to wait and see for that, the current version you can download for free.

As mentioned above, you do not need to make any settings for iCab to present Arabic pages in Arabic. That requires that the web providers have "labelled" their pages correctly, so iCab can know it is supposed to be Arabic. If not, iCab has Character Set menus that work like those of Netscape and IE, but unlike them it has Arabic as options, both Windows and Mac, in the character set menu. iCab works also on older machines, I have tested it on an old Colour Classic with System 7.1 from the early 1990s, and it worked, although it was fairly slow under these conditions (you couldn't make current Netscape or IE to even start up, however, on such a system). The Arabic worked, but without the automatic display of Arabic text, that requires OS 9 (or possibly 8.1); on older systems, you have to select an Arabic option from the menu.

Arabic pages in Unicode also require OS 9 (or 8.1 with extras) to display, because they require a system element that appeared in that OS (the "TEC" extension). -- As for other OS 9 browser, Opera does not yet support Arabic or any other non-Western character set.

-- Opera 6 (discussed below) is also reported to work with Arabic under OS 9, this remains to be tested --

There do however remain a few glitches in many Web browsers, some for logical reasons, or because Arabic has not been specifically considered. One concerns periods, colons and other punctuation. They may cause a line to become muddled up. (This is because the browser sees all periods as "English", and, thinking in English, insists to display the part of the line before the period to the left of it, and the new sentence after the period to right of it. Neither Netscape 7 or Mozilla seem to suffer from this bug). Some remedies may be, (i) to use the Language registry to "register" the browser as an "Arabic" application. This may make it organize the line in proper order. If not, the same may be achieved by (ii) using the Mac's "Text" control panel and change the overall system direction from left-right to right-left while reading these web pages. That will take care of the problem in most cases, but apparently not in all; depending on how the web pages have been coded. Such minor, but irritating glitches will probably be with us until a stable, universal character system is in force in a few years' time.

***

For browsers in early versions of OS X (10.2-up), the below may still be relevant (not every update has been checked):

The browsers that do display Arabic have access to all kinds of older character systems, including Windows Arabic, old Mac (ISO) Arabic, etc., so those differences have disappeared with OS X, and the problems with right-left justification and confusion of words on a line seem also to be gone. However, you are mostly stuck with one Arabic font, the all-Unicode Lucida Grande. Most browsers do not list other Arabic fonts at all.

Under OSX 10.2, some browsers had display bugs that made e.g. some letters of a word turn up in larger size than others. These problems seem to have gone away with 10.3 and the latest versions of the browsers mentioned. Opera (tested version) repeats the problem with punctuation messing up the line order known to OS 9 browsers, discussed above. The other OS X browsers do not.Chimera, based on Mozilla displayed the text correctly right-adjusted, others often use some sort of centre-adjustment; a minor problem but not nice-looking.

You can also run older or non-carbon versions of browsers under the Classic environment in OS X. Since Classic works as OS 9, and you can add Arabic resources to it as you can to OS 9, those browsers should work in the same way there as they do under OS 9.

Basically, however, it seems that reading Arabic pages on the net from a Mac is possible in a number of ways. In OS 9, Netscape 7, iCab and Mozilla stand out, the preferred browser today would be Netscape, but it is also a much bigger and "heavier" browser than iCab. iCab is also a fast and good solution, but with one or two annoying bugs in the Arabic text display.

Editing Arabic HTML texts for the Web (Classic / OS 9 or older)

The specific Web authoring programs are still evolving, and many people, including myself, prefer to use our standard word processors for Web authoring. Thus, I have not made an extensive survey of such Web authoring programs. I have, however, not yet come up with any that does support Arabic editing. Significantly, Netscape 3.0 Gold, while it will display any Arabic text from the net correctly, just as the earlier Netscape 1 and 2, cannot handle writing Arabic text for new files. Thus, my advice, until anyone tells me otherwise, is to use any normal text editor or word processor that can handle Arabic, such as TexEdit, muEdit, SimpleText or any other, and add the HTML formatting codes manually (or with home-made macros. Unfortunately, BBEdit, a commonly used text editor for the Web, does not handle Arabic. See the survey of programs that support Arabic.)

I personally use Nisus for all my Web work, English or non-English, because of its superior text handling features that makes supporting a large Web archive (like the present one) a much more manageable proposition. It also has various built-in Web authoring tools, which may be sufficient for you so that you do not need to actually see any formatting code (although the HTML options are so few it is not very complicated to use "raw" code) - however, I would then upgrade to version 5 or higher, where these tools have improved markedly. Thus, if you already have Nisus because of its Arabic capability, I would look closely at it before going anywhere else.

When you author Web pages in Arabic on the Mac, there are a few things to notice:

• You should use only Web standard Arabic

• Spaces: What the Mac uses for "Arabic space" is the same as the standard's "non-breaking space". Your Mac space will appear as a space to the reader, but word wrap will not work: the paragraph will disappear off beyond the right edge of the window, because there are only "no-break" spaces in the text.
• Periods, colons, exclamation marks, and numerals 1-9: All of these are specific to the Mac, and will appear as nonsense to any non-Mac reader (the Mac uses separate Arabic numbers, ISO and others do not, they have common numerals for English and Arabic)

The safe characters, which are common to the Mac and ISO Arabic are,

• All Arabic text characters: alif-ya, marbutas, hamza on all carriers.
• All vowels, sukun, shadda.
• Comma, semicolon and question mark.

In HTML theory, all Arabic (and other non-English) should actually be coded in a particular way, thus instead of the "alif" character, one should type in the code &#199;, you will see this in most manuals. However, this formal requirement is ignored by all real-world texts I have seen, they all contain the real text, which does not appear to cause any problems, so this rule must be considered obsolete.

• Use the format code <P Align=Right> instead of <P>

As discussed above, some modern Web browsers (such as iCab) can get the correct script and font if you label the web page with what character set you use. The format for this is, in the <head> section, <meta content="text/html; charset=ISO-8859-6">. However, this is the very code that causes Netscape and IE on the Mac not to display it in Arabic, so there is a choice which problem you wish to cater for.

Excursion: Diacritics and web pages (OS X)

A small OS X-related note: You can now use diacritics in Web pages, giving full transcription. The simplest, but technically messiest way to do so, is to write your text in Word 2004, using a Unicode font such as JaghbUni, Gentium or others, typing diacritics as you would any text. Then "Save as web page". This page will then display, with diacritics in place, in browsers that support Unicode (on the Mac, Safari and Firefox, but not Internet Explorer).
The messy part is partly that Word - while it correctly inserts the correct codes for the diacritics - otherwise creates very messy html code, so web designers hate it. Also, Word will insert a command for the browser to display in the same font you typed in; which may cause problems if the reader does not have this font. A Mac user will normally still see the diacritic, but displayed in the Lucida Grande font; a Windows user may not; so it may be better to insert font code in the page like <font face="Gentium, Arial Unicode">, listing the most common Unicode fonts. If no such font is listed, Mac users will often see the text in Lucida Grande (in Safari, at least). This font command may be overridden by the user's settings, if his browser is e.g. set to always display text in Times, he will typically see the long vowels, which Times contains, but not the dot under consonants, which that font does not have.
You can also code the diacritics manually, using decimal values of the Unicode characters, so that e.g. long a - ā - is typed as &#0257; etc.
So, all in all, it works, many, but not all users will see the text correctly. To reach everyone, PDF files may still be a safer route to display transliterated text, but I notice that more and more people are starting to use diacritics directly into web pages, taking their chances.
(October 2005)

News: YA NewsWatcher

To install, you simply drop the desired file into the YA-Newswatcher Tables folder that the program has created when you first used it. To display any Arabic message in the correct format, open View: Article format and choose an Arabic font as well as the correct Article Text Conversion table (such as "MacArabic <- ISO Arabic") from the relevant menus in the dialogue box. Use the same menu to write a new message in Arabic.

You can also choose default fonts and conversions for particular groups if any happen to start using Arabic script, from the Newsgroups Settings: Article options dialog box.

Telnet

Knut S. Vikør