Converting WideChar/WideString to AnsiChar/AnsiString

[ckoegl]

Hi,

I cannot find anything definitive in the Delphi docs and in the
relevant newsgroups that answers this simple question:

What is the exact definition of the WideChar to AnsiChar conversion
(and in extension of the WideString to AnsiString conversion)?

I suppose that for all WideChar values that are valid AnsiChar values
(i.e. the ones with Ord-values of 0 to 255) are preserved in the
conversion, but what about the rest?

Cheers,

Christoph

[Rob Kennedy]

They're truncated, I think. Or it's a range-check error.

Converting a WideString to an AnsiString involves converting the
WideString to the current system code page, but I'm pretty sure that
effort is *not* used for converting WideChar to AnsiChar.

And if that's the case, then the conversion is nonsense. Don't do it.

--
Rob

[ckoegl]

Ok, in my case I simply cannot dodge the problem: I have (for whatever
reasons) text
data given as WideString and I need to call given procedures/functions
that only accept
String (which in my case means AnsiString). It is entirely acceptable
that some data
ist lost during the WideString to String conversion, but: (1) All
character information that
can be transferred should be transferred, and (2) I need to know quite
exactly what happens
in case unconvertible character data is encountered.

Well, I guess I could simply try out what happens, but I thought that
maybe you
folks already have deeper experience of what pitfalls may lie ahead of
me ...

One valuable information I already got is that the string conversion
is not simply a sequence
of single character conversions.

Thanks,

Christoph

[Maarten Wiltink]

"ckoegl" <ko...@heidenhain.de> wrote in message
news:e6eb3d02-23e4-4491...@k37g2000hsf.googlegroups.com...

> What is the exact definition of the WideChar to AnsiChar conversion
> (and in extension of the WideString to AnsiString conversion)?

That depends on the conversion you use. Windows has a built-in function
whose name escapes me at the moment; something like UnicodeToMultiByte.
It shouldn't be too hard to find. IIAMN, Delphi uses it when you assign
a WideString to an AnsiString.

> I suppose that for all WideChar values that are valid AnsiChar values
> (i.e. the ones with Ord-values of 0 to 255) are preserved in the
> conversion, but what about the rest?

The above-mentioned Windows function replaces them with question marks.
Which is not very good, but about the best you can do short of throwing
an exception.

Groetjes,
Maarten Wiltink

[Jamie]

I think Delphi does some converting on it's own how ever..
look at these

WideCharToMultiByte
MultiByteToWideChar
FoldString
IsDBCSLeadByte

These are in the windows help, they should get you started.
http://webpages.charter.net/jamie_5"

[Rudy Velthuis]

ckoegl wrote:

> Hi,
>
> I cannot find anything definitive in the Delphi docs and in the
> relevant newsgroups that answers this simple question:
>
> What is the exact definition of the WideChar to AnsiChar conversion
> (and in extension of the WideString to AnsiString conversion)?

The RTL basically uses this:

function CharFromWChar(CharDest: PChar; DestBytes: Integer; const
WCharSource: PWideChar; SrcChars: Integer): Integer;
begin
Result := WideCharToMultiByte(DefaultSystemCodePage, 0,
WCharSource, SrcChars,
CharDest, DestBytes, nil, nil);
end;

--
Rudy Velthuis http://rvelthuis.de

"Beware of bugs in the above code; I have only proven it correct,
not tried it." -- Donald E. Knuth.

[Ivan Levashew]

ckoegl пишет:

>
> Ok, in my case I simply cannot dodge the problem: I have (for
> whatever reasons) text data given as WideString and I need to call
> given procedures/functions that only accept String (which in my case
> means AnsiString). It is entirely acceptable that some data ist lost
> during the WideString to String conversion, but: (1) All character
> information that can be transferred should be transferred, and (2) I
> need to know quite exactly what happens in case unconvertible
> character data is encountered.
>

In the case these functions doesn't operate with the outer world (pure
string processing) you can probably use UTF-8.

--
If you want to get to the top, you have to start at the bottom