XSLT transform with MSXML doesn't use proper encoding

Question

I'm using IXMLDOMDocument::transformNode from MSXML 3.0 to apply XSLT transforms. Each of the transforms has an xsl:output directive that specifies UTF-8 as the encoding. For example,

<?xml version="1.0" encoding="UTF-8"?>
<xsl:stylesheet version="1.0"
                ...
                xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
                xmlns:str="http://exslt.org/strings"
                xmlns:math="http://exslt.org/math"
                extension-element-prefixes="str math">
  <xsl:output encoding="UTF-8" indent="yes" method="xml" />
  ...
</xsl:stylesheet>

Yet the transformed result is always UTF-16 (and the encoding attribute says UTF-16).

<?xml version="1.0" encoding="UTF-16"?>

Is this a bug in MSXML?

For various reasons, I'd really like to have UTF-8. Is there a workaround? Or do I have to convert the transformed result to UTF-8 myself and patch up the encoding attribute?

Update: I've worked around the problem by accepting the UTF-16 encoding and prepending a byte-order mark, which satisfies the downstream users of the transformed result, but I'm still be interested in how to get UTF-8 output.

Answer 1

You're probably sending the ouput to a DOM tree or to a character stream, not to a byte stream. If that's the case then it's not MSXML that's doing the encoding, and whatever does do the final encoding has no knowledge of the xsl:output directive (or indeed, of XSLT).

Answer 2

Supplementing what Michael Kay said (which is spot on, of course), here's a JScript example how to transform to a stream, using the XSLT serialization in the process:

// command line args
var args = WScript.Arguments;
if (args.length != 3) {
    WScript.Echo("usage: cscript msxsl.js in.xml ss.xsl out.xml");
    WScript.Quit();
}
xmlFile = args(0);
xslFile = args(1);
resFile = args(2);

// DOM objects
var xsl = new ActiveXObject("MSXML2.DOMDOCUMENT.6.0");
var xml = xsl.cloneNode(false);

// source document
xml.validateOnParse = false;
xml.async = false;
xml.load(xmlFile);
if (xml.parseError.errorCode != 0)
    WScript.Echo ("XML Parse Error : " + xml.parseError.reason);

// stylesheet document
xsl.validateOnParse = false;
xsl.async = false;
xsl.resolveExternals = true;
//xsl.setProperty("AllowDocumentFunction", true);
//xsl.setProperty("ProhibitDTD", false);
//xsl.setProperty("AllowXsltScript", true);
xsl.load(xslFile);
if (xsl.parseError.errorCode != 0)
    WScript.Echo ("XSL Parse Error : " + xsl.parseError.reason);

// output object, a stream
var stream = WScript.createObject("ADODB.Stream");
stream.open();
stream.type = 1;
xml.transformNodeToObject( xsl, stream );
stream.saveToFile( resFile );
stream.close();

You may test using this input:

<Urmel>
    <eins>Käse</eins>
    <deux>café</deux>
    <tre>supplì</tre>
</Urmel>

And this stylesheet:

<xsl:stylesheet version="1.0"
    xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <xsl:output encoding="UTF-8"/>
    <xsl:template match="@*|node()">
        <xsl:copy>
            <xsl:apply-templates select="@*|node()"/>
        </xsl:copy>
    </xsl:template>
</xsl:stylesheet>

I think it'll be easy for you to adapt the JScript example to C++.