asked on

URGENT: How to rewrite an xml file before saving it in ASP/MSXML

The part from the replace of & gives an error and I am not sure I am doing it right:

<%@ language=javascript %>
<%
Response.Expires = -1000;
// Load the posted XML document
var doc = Server.CreateObject("Microsoft.XMLDOM");
doc.load(Request);
doc.save(Server.MapPath("xxYY.xml"));
doc.save(Server.MapPath("../../xmldir1/xxYY.xml")); // WORKS UNTIL HERE
// I get an unspecified error somewhere from here on
contentNodes = doc.documentElement.selectNodes("/VARSFILE/VAR/CONTENT")
for (i=0;i<contentNodes.length;i++) {
val = contentNodes[i].text;
if (val.indexOf('&') !=-1) {
val=val.replace('&','!!!!!!!&!!!!'); // the !!! are to see it really worked
contentNodes[i].text=val;
changed=true;
}
}
doc.save(Server.MapPath("../../xmldir2/zzzzzz_YYxx.xml")); // did the contentNodes change at all?

%>

MCummings

MY first response is that you are generating some invalid XML by replacing the & entity with your string which includes the '&'. If you want the & without using & enclose your content in CDATA nodes.

rdcpro

That seems reasonable, but the thing is, if the input XML looks like:

<VARSFILE>
<VAR>
<CONTENT>Foo & Bar</CONTENT>
<CONTENT>snafu</CONTENT>
<CONTENT>tarfu</CONTENT>
</VAR>
</VARSFILE>

there won't be any entities like & in the parsed document object.

When an XML Document is parsed, XML markup entities (like &) are expanded. So the value of the first <CONTENT> node is actually: "Foo & Bar" where the "&" is the character for the ampersand. You will ONLY ever find "&" in the *parsed* document if the original XML looks like:

<CONTENT>Foo &amp; Bar</CONTENT>

or:

<CONTENT><![CDATA[Foo & Bar]]></CONTENT>

If you view the xml property of the DOM object, you'll see the "&" but that's because the DOM object is serialized, and all markup characters are encoded into their entity equivalents.

To see what I mean, copy this and save as test.html:

<html>
<script language = "Javascript">
var doc = new ActiveXObject("Msxml2.DomDocument");
doc.preserveWhiteSpace = true;
sXml = ""
sXml += "<VARSFILE>\n"
sXml += " <VAR>\n"
sXml += " <CONTENT>Foo & Bar</CONTENT>\n"
sXml += " <CONTENT>snafu</CONTENT>\n"
sXml += " <CONTENT>tarfu</CONTENT>\n"
sXml += " </VAR>\n"
sXml += "</VARSFILE>"

doc.loadXML(sXml);
contentNodes = doc.documentElement.selectNodes("/VARSFILE/VAR/CONTENT")
alert("This shows an entity, because the XML is serialized. \nTherefore the ampersand is encoded as an entity\n\n" + doc.xml)
for (i=0;i<contentNodes.length;i++) {
val = contentNodes[i].text;
if (val.indexOf('&') !=-1)
alert("This shows the entity has been expanded: \n\n" + val)
if (val.indexOf('&') !=-1) {
// This never tests true! No entities in the parsed XML.
alert("There's an entity in here");
val=val.replace('&','!!!!!!!&!!!!'); // the !!! are to see it really worked
contentNodes[i].text=val;
changed=true;
}
contentNodes[i].text = val + "&"
alert("The new Content node (contentNodes[i].text) isn't parsed, so \nit doesn't matter that I add an ampersand!\n\n" + contentNodes[i].text)

}
alert("Here's the modified XML. All ampersands have been encoded: \n\n" + doc.xml)
</script>
</html>

As far as the error you get....I don't see a problem The code I pasted here is fine.

However, DON'T use the progID "Microsoft.XMLDOM", expecially on the server!. This is MSXML 2.0. You should use at least MSXML 3. So don't do this:

var doc = Server.CreateObject("Microsoft.XMLDOM");

but do this:

var doc = Server.CreateObject("Msxml2.DomDocument");

Regards,
Mike Sharp

http://dev.rdcpro.com

rdcpro

I should like to add that if the server has MSXML 3, sp1 or later installed, then even if you *do* use

var doc = Server.CreateObject("Microsoft.XMLDOM");

it will actually be using MSXML 3 as though you typed:

var doc = Server.CreateObject("Msxml2.DomDocument");

MSXML 3, sp1 or later installs in replace mode. That doesn't mean you can leave the "Microsoft.XMLDOM" in there, because it's possible on some servers if there is legacy code, the MSXML 3 might be forcibly installed in side-by-side mode, and your code could have errors if it's referencing the wrong parser.

Regards,
Mike Sharp

rdcpro

Oh, and one more thing...though this should be obvious, using the save() method of the DOM object will escape the entity back into an ampersand, even if your original code had succeeded in finding an entity and replacing it with the "&". It must do this, because otherwise your saved XML is not well formed. You would have to use the CDATA section, as MCummings suggested, if you want to retain the "&". But this is almost never necessary, IMHO. If you think you need to do it, you've probably got something wrong elsewhere.

Regards,
Mike Sharp