Ori Doolman wrote:
Dave,
Thank you for your answer.
I did notice that it was a FAQ, but the FAQ is talking about
stylesheets, while I'm interested only in using the Xpath API in my code
to analyze a SOAP response.
I succeeded extracting the text I want using expressions like you have
suggested:
/*[local-name() = 'myLocalName' and namespace-uri()='myNamespaceURI']
However, I'm not sure it is the most efficient way of extracting the
text.
I would still like to use the specific node name instead of '*'.
According to the FAQ and other places, you can add the namespace
definition to the stylesheet, and "inform" the stylesheet about the
namespaces used in your xml document.
I wonder if you can do the same using the Xpath API.
For example, in my test, if I could "teach" the API that prefix 'sv' is
'http://www.SigValue.com/SVCrm', then I could write the expression:
/soap:Envelope/soap:Body/sv:CheckCustomerPINResponse/sv:CheckCustomerPIN
Result/text()
Yes, there is. Look at the SimpleXPathAPI sample, and you'll see it uses a
class called XalanDocumentPrefixResolver. You can use that class, if the
namespaces you're interested are declared in the incoming document, and you
can predict the prefixes.
Otherwise, you can decide on a set of namespace bindings that you expect,
and create your own derivative of Xalan-C's PrefixResolver class.
But actually, even the 'soap' prefix is used here only because the
document defines it as the prefix of
"://schemas.xmlsoap.org/soap/envelope/". In real life, I would not be
able to tell if the soap server defines it as "soap" or maybe "s" or
"soap12" etc. The only sure thing is the namespace uri and the local
node name.
That's why prefixes are irrelevant. The namespace URI is the only thing
that matters, once Xalan-C has compiled the XPath expression. But it needs
a PrefixResolver instance to do the mapping.
So now I have 3 options:
1) Use the API with expression containing '*' and local-name() and
namespace-uri() functions (not efficient???)
2) Use the API with expression containing myPrefix:myNodeName, but
"teach" the API first what is the namespace uri of 'myPrefix' (not sure
it's possible)
3) Use a stylesheet instead of the Xpath API. I will create the
stylesheet dynamically in my code with all the known namespaces uri and
prefixes. And then extract the text using xsl. (not sure that the result
of xsl can be a simple text and not xml, and also not sure it is more
efficient then the first method).
Use a custom implementation of PrefixResolver that knows all of the
prefix/namespace URI bindings in your XPath expressions, and ignore the
prefixes in the incoming document(s).
Dave
