Urdu

[KenDown]

I'm really not sure if this is the place for my query, but here goes.

If you take a string of Arabic characters and PRINT it (having set text to right-to-left, of course) it all comes out as the equivalent of capital letters (they are actually terminal letters, but never mind) whereas what you need is running writing. Richard has provided a very clever function that adds numbers here and does something else there and the end result is beautiful running writing.

The trouble is that when I try it with Urdu, an Indian language, it doesn't work - and the function is not REMed, so I don't know what the different manipulations do or mean. I could, possibly, try and work it out myself (emphasis on "try"!) but if anyone has any insight into what Richard's function does and how it produces the desired effect, I'd be grateful.

I'd also like to know (but I haven't done any research on this point yet) a quick and easy way for my program to recognise that the string it is presented with is Arabic or Chinese or whatever.

[KenDown]

I have had a reply from the very kind gentleman whose name we may not mention, which indicates that there would need to be a separate FN written for each language that uses the Arabic-style scripts - Urdu, Persian and no doubt others.

Very frustrating.

[jgharston]

The only way a computer program can tell what human language a text string contains is if you the programmer - or the user entering a user-supplied string - tells the program what human language the text string contains. Is "were made" English? Or German? Or Japanese written in Romanji? Is 電車 the Chinese for tram or the Japanese for train? You the programmer or you the data inputter has to tell the program. If I paste 電車 into Google Translate I have to tell it what language I want it translated from, it can't read my mind and work it out. (It actually guesses based on the last language I used, which is a fairly good pragmatic method, but it can only do that because I'm logged in to Google.)