🚀 go-pugleaf

"Tommi Prami" <tommi.nospam.prami@poista.ecomond.com> wrote in message 
news:47fb1c46$1@newsgroups.borland.com...

> I was looking for very good way to force thousand separators
> to Integer (or float) formatted string (in Finlad thousand
> separators are not normally used in OS level).

What about using FormatFloat() for that?


Gambit 

Hello,

Can I force Thousand separators to it, even system has not set any??? 
Without changing locale settings etc...

-TP-

Remy Lebeau (TeamB) wrote:
> "Tommi Prami" <tommi.nospam.prami@poista.ecomond.com> wrote in message 
> news:47fb1c46$1@newsgroups.borland.com...
> 
>> I was looking for very good way to force thousand separators
>> to Integer (or float) formatted string (in Finlad thousand
>> separators are not normally used in OS level).
> 
> What about using FormatFloat() for that?
> 
> 
> Gambit 
> 
> 

"Tommi Prami" <tommi.nospam.prami@poista.ecomond.com> wrote in message 
news:47fc6cc4$1@newsgroups.borland.com...

> Can I force Thousand separators to it, even system has
> not set any??? Without changing locale settings etc...

Have you read the documentation for FormatFloat() yet?  It tells you how to 
indicate that you want thousand separators to appear in the output.  As for 
the actual separator, the global ThousandsSeparator variable is used.  If 
your system does not define any separator, then the VCL defaults to the ',' 
character.


Gambit 

> Have you read the documentation for FormatFloat() yet?  It tells you how to 
> indicate that you want thousand separators to appear in the output.  As for 
> the actual separator, the global ThousandsSeparator variable is used.  If 
> your system does not define any separator, then the VCL defaults to the ',' 
> character.


Just What I suspected, so I can't use that.

-TP-

"Tommi Prami" <tommi.nospam.prami@poista.ecomond.com> wrote in message
news:47fdc42d$1@newsgroups.borland.com...
> > Have you read the documentation for FormatFloat() yet?  It tells you how
to
> > indicate that you want thousand separators to appear in the output.  As
for
> > the actual separator, the global ThousandsSeparator variable is used.
If
> > your system does not define any separator, then the VCL defaults to the
','
> > character.
>
>
> Just What I suspected, so I can't use that.

Yes you can.
As for the decimal separator i do:
VAR
   kDecimalSeparator : Char ; // keep
...

   DecimalSeparator := '.' ; // in DK we normally use ','
   something := ' Stardate: ' + FormatFloat('#.0',Now) ; // now it uses '.'
as decimal separator
   DecimalSeparator := kDecimalSeparator ; // restore system settings.

   You just have to do the same with the ThousandsSeparator as Remy
mentions.


--
Med venlig hilsen/Best regards
Stig Johansen

>    kDecimalSeparator : Char ; // keep
> ....
> 
>    DecimalSeparator := '.' ; // in DK we normally use ','
>    something := ' Stardate: ' + FormatFloat('#.0',Now) ; // now it uses '.'
> as decimal separator
>    DecimalSeparator := kDecimalSeparator ; // restore system settings.
> 
>    You just have to do the same with the ThousandsSeparator as Remy
> mentions.

This is not acceptable, because software is heavily multi threaded and 
all other parts must not be affected with this.


And using critical section etc, is not an option here.

I think I need to use someting different. Code I send earlyer works, but 
is not very efficient, thats all...

-TP-

Hi,

FormatFloat is an overloaded function and can called with 
TFormatSettings. In TFormatSettings you can set the ThousandSeparator. 
This is a threadsave way to override the standard behavior.

Soeren

>>    kDecimalSeparator : Char ; // keep
>> ....
>>
>>    DecimalSeparator := '.' ; // in DK we normally use ','
>>    something := ' Stardate: ' + FormatFloat('#.0',Now) ; // now it 
>> uses '.'
>> as decimal separator
>>    DecimalSeparator := kDecimalSeparator ; // restore system settings.
>>
>>    You just have to do the same with the ThousandsSeparator as Remy
>> mentions.
> 
> This is not acceptable, because software is heavily multi threaded and 
> all other parts must not be affected with this.
> 
> 
> And using critical section etc, is not an option here.
> 
> I think I need to use someting different. Code I send earlyer works, but 
> is not very efficient, thats all...
> 
> -TP-

Ok,

i've useed it but never studied he internals...

Thanks...

btw... Is there (been) FastCode Chanllenge on format float (etc) functions).

-TP-

Little more Performance but full functional i hope:
No Insert String, nor delete String
See : Format('%.*n')
Change '.' to ',' and ',' to '.' or any other char u want


var MyThousandsSeparator:char='.';
    MyDecimalSeparator:char=',';
    MyCheckSeparators:byte=0;

function double2string(b:double; n:byte):string;
var r:double;
    v:double;
    z:integer;
begin
{
  for rounding currency values

  r:=0;
  case n of
    1:r:=0.001;
    2:r:=0.0001;
    3:r:=0.00001;
    4:r:=0.000001;
    5:r:=0.0000001;
    6:r:=0.00000001;
  end;
  if (b<0) then r := -r;
  v:=b+r;
}
  v:=b;
  if n=0 then begin   // NoDecimals Just integer
    result:=format('%d',[trunc(v)]);
    exit; // all done
  end else result:=format('%.*n',[n,v]);

  if MyCheckSeparators=0 then begin        // if you use always the same 
separators you might do this in initialization section

    MyCheckSeparators:=1;                        // This will be used ONCE
    if (MyThousandsSeparator <> ThousandSeparator) then 
MyCheckSeparators:=MyCheckSeparators or 2;
    if (MyDecimalSeparator <> DecimalSeparator) then 
MyCheckSeparators:=MyCheckSeparators or 4;
  end;

 // if you use always the same separators you might do this in 
initialization section


  case MyCheckSeparators of                    // This compiles to some 
additional sub and jnz ops
    1:; // Do nothing
    3:begin // Change thousend character only ( you may change result[z] to 
cP^ and use pointer ops
        for z:=1 to length(result) do begin
          if result[z]=ThousandSeparator then 
result[z]:=MyThousandsSeparator;
        end;
      end;
    5:begin // Chenge decimal only
        for z:=1 to length(result) do begin
          if result[z]=DecimalSeparator then result[z]:=MyDecimalSeparator;
        end;
      end;
    7:begin // change both
{
        MayBe Faster :

        cP:=@result[1];
        for z:=1 to length(result) do begin
           if cP^=DecimalSeparator then begin
                cP^:=MyDecimalSeparator;
           end else begin
                if cP^=ThousandSeparator then cP^:=MyThousandSeparator
           end;
           inc(cP);
        end;
}
         for z:=1 to length(result) do begin
          if result[z]=DecimalSeparator then begin
            result[z]:=MyDecimalSeparator;
          end else begin
            if result[z]=ThousandSeparator then 
result[z]:=MyThousandsSeparator;
          end;
        end;
      end;
  end;
end;

"Tommi Prami" <tommi.nospam.prami@poista.ecomond.com> schrieb im Newsbeitrag 
news:47fb1c46$1@newsgroups.borland.com...
> Hello,
>
> I was looking for very good way to force thousand separators to Integer 
> (or float) formatted string (in Finlad thousand separators are not 
> normally used in OS level).
>
> I came up with this (found it with google)
>
> function emdAddThousandSeparator(const S: string; Chr: Char = ' '): 
> string;
> var
>   I: Integer;
> begin
>   Result := S;
>   I := Length(S) - 2;
>   while I > 1 do
>   begin
>     Insert(Chr, Result, I);
>     I := I - 3;
>   end;
> end;
>
>
> Then I was thinking that it could be defined like this
>
> function IntToStrT(const AValue: Integer; AThousandsSeparator: Char = ' 
> '): string;
> (and one for float(s) maybe)
>
> Maybe this way it would be more efficient, to convert it directly to 
> string with forced thousand separator.
>
> Where I could find routine to modify, pure pascal would do, in which I 
> could implement thousand separator adding. Or does someone have optimized 
> code for this already (or good way to force separators with RTL/VCL, with 
> out changing the Thousand Separator char in ?system.pas/sysutils? (where 
> ever it is, can't remember)...
>
> -TP-