Malcolm,

<< Thanks James .. but I need it in code.  It now seems my only problem is
getting rid of the BOM ... >>

Yes, you need to strip out the BOM.  However, you can also use these
functions in the EDB edbcomps.pas unit (source provided with every version
of EDB) to translate files:

  { Text file encodings }

  ENCODING_UNKNOWN = 0;
  ENCODING_ANSI = 1;
  ENCODING_UNICODE = 2;

  procedure LoadStringsFromStream(Stream: TStream; Strings: TEDBStrings;
                                  Encoding: Integer=ENCODING_UNKNOWN);
  procedure LoadStringsFromFile(const FileName: TEDBString;
                                Strings: TEDBStrings;
                                Encoding: Integer=ENCODING_UNKNOWN);
  procedure SaveStringsToStream(Stream: TStream; Strings: TEDBStrings;
                                Encoding: Integer=ENCODING_UNKNOWN);
  procedure SaveStringsToFile(const FileName: TEDBString;
                              Strings: TEDBStrings;
                              Encoding: Integer=ENCODING_UNKNOWN);

The TEDBString and TEDBStrings are just mapped to
String/WideString/UnicodeString and TStrings/TWideStrings, respectively,
depending upon what version of Delphi/C++Builder and EDB you are using.
But, it won't matter for the purposes of doing a conversion like this unless
you're using < BDS 2006, which is required for TWideStrings support.

--
Tim Young
Elevate Software
www.elevatesoft.com

Malcolm,

Crap, I just noticed that these also write out a BOM at the start of the
file.  I'll see about adding a parameter for this for the next build.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim Young [Elevate Software] wrote:

> Malcolm,
>
> Crap, I just noticed that these also write out a BOM at the start
> of the file.  I'll see about adding a parameter for this for the
> next build.

Thanks Tim, I'm on D2009.

Everything I read says Delphi does not support BOM-less unicode files.
But EDB requires them and so does PHP and MySQL.

Unfortunately I can't use your export files directly as I need to
convert them to UTF8 for the web .. and guess what that does!

All this means that at the moment my first record imported into MySQL
has a bad first column value - not good as it is a key field!

At least with EDB such an import simply fails.

It seems to me that Delphi is going to need a DefuseBOM()
function/method before long - even if I have to write it.  

--

Malcolm,

<< Everything I read says Delphi does not support BOM-less unicode files.
But EDB requires them >>

This will be changing - the import/export is the last thing that is not
character set-agnostic, at least with respect to ANSI/Unicode (UTF16).
UTF-8 is still something that will come at a later time, most likely in
conjunction with the Linux support.

--
Tim Young
Elevate Software
www.elevatesoft.com

Tim Young [Elevate Software] wrote:

> Malcolm,
>
> << Everything I read says Delphi does not support BOM-less unicode
> files. But EDB requires them >>
>
> This will be changing - the import/export is the last thing that is
> not character set-agnostic, at least with respect to ANSI/Unicode
> (UTF16). UTF-8 is still something that will come at a later time,
> most likely in conjunction with the Linux support.

My need for UTF-8 is not related to EDB.  I will have to solve that
myself .. but it is the same issue.
I am just moaning!

--

Can anyone tell me what currently is the best way to perform an export/import from Ansi to Unicode?

I have tried it with creating a store and then export table/import table but also get the truncated/incomplete error.

F.J. Olthuijsen wrote:

> Can anyone tell me what currently is the best way to perform an
> export/import from Ansi to Unicode?
>
> I have tried it with creating a store and then export table/import
> table but also get the truncated/incomplete error.

It depends on exactly what you want to do.
In my case I am faced with a UTF-16 file including a BOM.
Until Tim provides a more elegant solution I am stripping the BOM
something like this:

[code]
PROCEDURE TImportForm.DefuseBOM(fname: STRING);
VAR
 fs:      INT64;
 MStream: TMemoryStream;
 FStream: TFileStream;
BEGIN  MStream := TMemoryStream.Create;
 TRY
   MStream.LoadFromFile(fname);
   fs := MStream.Size - 2;      // asumes a 2-byte BOM !!!!!!
   MStream.Position := 2;       // point at start of data
   FStream          := TFileStream.Create(fname, fmOpenWrite);
   TRY
     FStream.CopyFrom(MStream, fs);
     FStream.Size := fs;
   FINALLY
     FStream.Free;
   END;
 FINALLY
   MStream.Free;
 END;
END;
[/code]

You may first have to read the Ansi file and re-write it as a unicode
one, which will probably add a BOM.  Then you can strip it as shown
above.  You should really check for the actual size of the BOM unless
you have full control.

Malcolm
--

F.J.

<< Can anyone tell me what currently is the best way to perform an
export/import from Ansi to Unicode? >>

Are you trying to move an entire database from ANSI to Unicode, or just
doing an export/import for one table ?

--
Tim Young
Elevate Software
www.elevatesoft.com

Hi Tim,

I'd like to migrate the entire database.

Frank

Hi Malcolm,

thanks! Hopefully Tim has an "elegant" alternative. Otherwise I'll give this a try.

Frank

"Malcolm" wrote:

F.J. Olthuijsen wrote:

> Can anyone tell me what currently is the best way to perform an
> export/import from Ansi to Unicode?
>
> I have tried it with creating a store and then export table/import
> table but also get the truncated/incomplete error.

It depends on exactly what you want to do.
In my case I am faced with a UTF-16 file including a BOM.
Until Tim provides a more elegant solution I am stripping the BOM
something like this:

[code]
PROCEDURE TImportForm.DefuseBOM(fname: STRING);
VAR
 fs:      INT64;
 MStream: TMemoryStream;
 FStream: TFileStream;
BEGIN  MStream := TMemoryStream.Create;
 TRY
   MStream.LoadFromFile(fname);
   fs := MStream.Size - 2;      // asumes a 2-byte BOM !!!!!!
   MStream.Position := 2;       // point at start of data
   FStream          := TFileStream.Create(fname, fmOpenWrite);
   TRY
     FStream.CopyFrom(MStream, fs);
     FStream.Size := fs;
   FINALLY
     FStream.Free;
   END;
 FINALLY
   MStream.Free;
 END;
END;
[/code]

You may first have to read the Ansi file and re-write it as a unicode
one, which will probably add a BOM.  Then you can strip it as shown
above.  You should really check for the actual size of the BOM unless
you have full control.

Malcolm
--