Re: XML encoding - handling foreign lang chars (SOLUTION)

[Lou]

PROBLEM (restated):
==================
My client creates an XML doc in memory. It sends the xmldoc.xml string to
an ASP on the server. The server does xmldoc.load(Request) and gets an error:
"An invalid character was found in the text context." because the XML contains
a foreign language character.

SOLUTION
========
I tried having the client do an xmldoc.createProcessingInstruction and set
the encoding. Still got the error. Then I realized something.... xmldoc.xml
gives you a string that DOES NOT indicate the encoding!! SO, the encoding
was lost and the error continued.

Now I do this:
sXML = Replace(oXMLDoc.xml, "?>", " encoding=""ISO-8859-1""?>", 1, 1)

Which adds the encoding to the <?xml version="1.0" ?> statement at the beginning
of the doc (put there by the createProcessingInstruction). I send the sXML
string to the ASP and...

NOW IT WORKS!!! The xmldoc.load(Request) works without error!


SIDE NOTES
==========
1) xmldoc.save("somefile.xml") DOES indeed retain the encoding, but xmldoc.xml
does NOT.

2) I use the Internet Transfer Control to send my XML to the server because
the MSXML.XMLHTTPRequest requires IE and I don't want that requirement for
my users. I could not figure out which DLLs of IE are really needed by MSXML.XMLHTTPRequest
(its more than msxml.dll).

[Leon Mar]

"Lou" <l_kvitek@audiblemagic.com> wrote:
>
>PROBLEM (restated):
>==================
>My client creates an XML doc in memory. It sends the xmldoc.xml string
to
>an ASP on the server. The server does xmldoc.load(Request) and gets an
error:
> "An invalid character was found in the text context." because the XML contains
>a foreign language character.
>
>SOLUTION
>========
>I tried having the client do an xmldoc.createProcessingInstruction and set
>the encoding. Still got the error. Then I realized something.... xmldoc.xml
>gives you a string that DOES NOT indicate the encoding!! SO, the encoding
>was lost and the error continued.
>
>Now I do this:
>sXML = Replace(oXMLDoc.xml, "?>", " encoding=""ISO-8859-1""?>", 1, 1)
>
Would ISO-8859-1 be enough to cater for foriegn languages, like CJK's. I
don't think so.

ISO-8859-1 is only for Latin 1 ANSI. UTF-8 can deal with 64,000 code points.

XML DOM only deals with UTF-8 if it is loaded from file. with loadXML() it
is expecting a BSTR that is a UCS-2 string. So forcing a UTF string into
UCS-2 without going through MultiByteToWide(CP_UTF7, ...) and then handing
it to loadXML() will cause some strange thing to happen.