1. The range hex 80-9F (decimal 128-159): Accented characters are converted to their Latin 1 counterparts except when there is a conflict with upper-ASCII Arabic Windows characters. For example, Mac a-grave (decimal 136) is converted to Latin 1 decimal 224, but Mac a-acute (decimal 135) is converted to the string "#0225" rather than to Latin 1 decimal 225, which happens to be the letter "laam"(!) The idea behind this is that if you want to restore any of these accented characters in Arabic Windows (in Word, for example), you can do so by using the "Euro" font feature. For example, search for "#0225" and replace with Times New Roman Euro ^0225.
2. The range hex A0-BF (decimal 160-191): This range covers Arabic Mac right-to-left numbers, punctuation, and other symbols. Many of these lose their right-to-left nature in the Arabic Windows code page, so "results will vary" (as they say). The problem is caused by ambiguous numbers and punctuation in Arabic Windows: the Arabic Mac period (full stop) is decimal 174, but in Arabic Windows it's the same as the Latin period (decimal 46). Arabic Windows does have a dedicated, unambiguous comma (decimal 161), however. The point is that some of the Arabic numbers and punctuation symbols in the Mac input file will end up being Latin when you open the output file in Arabic Word. This depends on the context of these letters.
3. Extended Arabic Mac characters: the 3-dot Arabic Mac characters (paa', jiim/giim and vaa') are converted to their one-dot counterparts in Arabic Win. So, if you had a "viiza" or "viituu" in your Mac document, they will now become "fiiza" and "fiituu." The Mac Persian character "gaaf" ("kaaf" with a line over it) is converted to plain "kaaf" in Arabic Windows.
4. Non-convertible characters: Several Urdu [?] characters in the Arabic Mac code page have no logical counterpart in Arabic Windows, and so they are converted to the string "?nnnn" where "nnnn" is their Arabic Mac decimal value. They are: 139, 244, 246, 249, 250, 254, and 255. For example, decimal 255 is converted to the string "?0255" which you can then convert to whatever seems appropriate.
The above references are in binary format. The same files in UUencode format:
Tim Buckwalter
Responsibe for these Web pages is Knut S. Vikør. Last updated 21.11.95