ALPHAB. ORDER= �, e, �, etc.

[whanigsberg]

RE: ALPHAB. ORDER= �, e, �, etc.

Hi Ira,

That's an interesting approach. I hadn't thought of using the AT function. I think I will do some speed comparisons between the various approachs and report back.

Regarding the asc function, I, too, was surprised. It sort of blocked me off from one approach I might have taken. On the bright side, it was probably a lousy approach.

At any rate, the interactive window gives the following:

?asc("�")
= -23.000000

?asc("e")
= 101.000000

Go figure.

As usual, your response has opened my eyes to new ways to solve a problem. For this I thank you.

You will definitely see me in the audience of any workshop on functions you offer in Boston.

Bill

[whanigsberg]

RE: ALPHAB. ORDER= �, e, �, etc.

Hello again,

For those (probably few) who have not tuned out of this arcane forum devoted to accents, here is what I wrote last night following Ira's suggestion that syntax using $ would run faster than my original which used multiple .or. condition.

function no_accents as C(input_string as C)
' Replaces accented characters with unaccented equivalents for ordering

tmpstr = ""
for i = 1 to len(input_string)
vChar = substr(input_string,i,1)
select
case between(vChar,"A","Z")
vChar = vChar
case vChar $ "��������"
vChar = "e"
case vChar $ "��"
vChar = "c"
case vChar $ "�����������"
vChar = "a"
Case vChar $ "��������"
vChar = "u"
Case vChar $ "����������"
vChar = "o"
Case vChar $ "��������"
vChar = "i"
Case vChar $ "���"
vChar = "y"
case else
vChar = vChar
end select

tmpstr = tmpstr+vChar
next

no_accents = tmpstr

end function

It will be observed that I included more accented characters than the earlier version, or than Ira did in is later contribution.

Using a list of 400 names it ran at about the same speed as Ira's. It would be interesting to know the speed effect of lengthening the list of accented characters using Ira's approach.

The fact that the function is rather slow reinforces the idea that--always depending on circumstances--it may be better to use the function in a calculated field rule (or elsewhere) to fill a field to index. That way, the function wouldn't have to evaluate each record whenever indexes were rebuilt.

Trivial as this may seem, the teachers in my language departments have been bugging me for a year or two about my application misordering their names many of which include accented characters. I wonder if they notice that they now will display correctly.

If you ever visit Montreal, treat yourself to a look at the telephone directory. Alphabetical order in French is, as they say, piquant. A favorite of mine is the place where I buy jogging shoes called La Boutique Endurance. It is listed under "L" along with half the other businesses in the city.

Bill

[csda1]

RE: ALPHAB. ORDER= �, e, �, etc.

Bill,

The asc() function is returning a negative number for ascii values above 128. I haven't checked the mapping, but you most likely either add 256 to the value, subtract it from 256 or take the absolute of the negative values and add 128 or subtract it from 256.

Also, I would think placing all the characters into newvalues and old values would still be faster than all of the case statements (particularly the ones at the end). A5V5 has profiling (timing measurements) built into parts of it, so it will be useful for evaluating methodologies.

Finally, is it typical to convert umlat etc type characters to the normal alphabetic value for sorting, or are they considered a subsort of the letter, i.e. sorts would be something like this kind of order E�����e��� ?

Regards,

Ira J. Perlow
Computer Systems Design & Associates
[email protected]

[whanigsberg]

RE: ALPHAB. ORDER= �, e, �, etc.

Hi Ira,

I also prefer your methodology if only because all those case statements I used are ugly.

You wrote, "Finally, is it typical to convert umlat etc type characters to the normal alphabetic value for sorting, or are they considered a subsort of the letter...."

I actually investigated this, easy enough for me as all I had to do is ask a member of the French department three doors down the hall. I was surprised to learn that they just want all the e's together regardless of the accents (or no accent). I was a bit surprised but that was the answer.

I have no idea what the convention may be in other languages or other French-speaking societies.

I had to add the upper case accented characters to your function, expecially "�". They aren't always used, but when they're there they really rearrange things.

Take care,
Bill

[Jim Rowlen]

RE: ALPHAB. ORDER= �, e, �, etc.

Bill,
Since you like playing around with stuff, here's one for you. Assuming that A5's random number generator will return the same result if given the same "seed" (like Quick Basic does), you can create your own encryption function
using something like this (but, of course, some stuff is going to be missing here...) :
dim array(255)
get user's seed / password
generate random number sequence (255 spaces) (call it x)
loop through 255 (call it n)
get the n position in x
newchar = add x to n
if newchar > 255 then newchar = 1
put newchar in array(1, n)
end loop

Now you can use ASC() and CHR() to get/set values in a string for encryption/decryption. You could just use 1 random number and "shift" the ASCII values up or down by that amount, but those kinds of encryption are pretty easy to break with a little effort (letter frequency, 1 and 2 character words, etc). You could even use a multiple encryption with the random number indicating which random result to use, have that result indicate which new formula to use which creates additional random seuqnces, etc. Maybe by nesting random sequences you would encrypt/decrypt a string of chars like : the first 3 chars use random sequence 8 and the 4th uses random sequence 9, the next 2 use seq 3, the next 5 use seq 9, etc. You get the idea.

Go crazy.

[Steve Pick]

RE: ALPHAB. ORDER= �, e, �, etc.

Bill,
Just looking in the spanish dictionary the accent doesn't affect the order of things, ie centrar, c�ntrico, centrifugadora.

Not being a linguistic expert but knowing that the accent is only to help pronunciation it is not surprising that your french colleagues don't care about sorting by accent order.

However, the original question may have other considerations.

Bon weekend.

Steve

[whanigsberg]

RE: ALPHAB. ORDER= �, e, �, etc.

Steve,

Merci pour l'information.

A bient�t,

Bill

[whanigsberg]

RE: ALPHAB. ORDER= �, e, �, etc.

Jim,

Thanks for that. I printed it out and began to obsess actively. My wife noticed this and commented that I was already crazy enough without extra help from you.

Coming to Boston?

Bill

[Jim Rowlen]

RE: ALPHAB. ORDER= �, e, �, etc.

With the proper level of obsessing you'll discover that you can expand the process (multi-dimensioanl arrays, etc) and even use it to do things other than encryption/decryption, translation, etc. If she thinks ou're crazy now, wait until you're driving down the road and start pulling off to the side to write down notes. Maybe it's a good thing I never got married.

Plan to get to Boston. Depends on schedule. With as responsive as we try to be with our clients I can literally get 1 or 2 days notice that a client has an emergency and have to crank out a 5 or 10 day special project before the next time I go to bed. (Guess it's a good thing I can still go about 3 days without sleep, a few weeks with a couple hours a day. They're not always the most efficient programming sessions tho.)

[EDUTECH]

RE: ALPHAB. ORDER= �, e, �, etc.

Hi Bill and Ira,
Thanks a lot, because I arrived successfully to a solution, with all your participation:
I created a function named no_acentos (I mixed French, Spanish and English, sorry)
Then, a field named �verbe_s_a�, I maked it calculated type, with the expression : no_acentos(verbe)
And that�s all, each time I write a verb in the field �verbe�, automatically the field verbe_s_a writes the same verb, but without accents, no mater which place or haw many letters with accent exists.
The function wrote was (almost yours) :

function no_acentos as C(verbe)
' Replaces accented characters with unaccented equivalents for ordering '

tmpstr = ""
for i = 1 to len(verbe)
vChar = substr(verbe,i,1)
select
case between(vChar,"A","Z")
vChar = vChar
case vChar $ "��������"
vChar = "e"
case vChar $ "��"
vChar = "c"
case vChar $ "�����������"
vChar = "a"
Case vChar $ "��������"
vChar = "u"
Case vChar $ "����������"
vChar = "o"
Case vChar $ "��������"
vChar = "i"
Case vChar $ "���"
vChar = "y"
case else
vChar = vChar
end select
tmpstr = tmpstr+vChar
next
no_acentos = tmpstr
end function
Then I index by "verbe_s_a" and I'd finish.
Thank you both, very much
J R Tarrag�

[EDUTECH]

RE: ALPHAB. ORDER= �, e, �, etc.

Hi Bill and Ira,
Thanks a lot, because I arrived successfully to a solution, with all your participation:
I created a function named no_acentos (I mix French, Spanish and English, sorry)
Then, a field named �verbe_s_a�, I maked it calculated type, with the expression : no_acentos(verbe)
And that�s all, each time I write a verb in the field �verbe�, automatically the field verbe_s_a writes the same verb, but without accents, no mater which place or haw many letters with accent exists.
The function wrote was (almost yours) :

function no_acentos as C(verbe)
' Replaces accented characters with unaccented equivalents for ordering '

tmpstr = ""
for i = 1 to len(verbe)
vChar = substr(verbe,i,1)
select
case between(vChar,"A","Z")
vChar = vChar
case vChar $ "��������"
vChar = "e"
case vChar $ "��"
vChar = "c"
case vChar $ "�����������"
vChar = "a"
Case vChar $ "��������"
vChar = "u"
Case vChar $ "����������"
vChar = "o"
Case vChar $ "��������"
vChar = "i"
Case vChar $ "���"
vChar = "y"
case else
vChar = vChar
end select
tmpstr = tmpstr+vChar
next
no_acentos = tmpstr
end function

Then I index by "verbe_s_a" and I finished

Thank you both, very much
J R Tarrag�