RE: Xoath quote escaping

In XPath 1.0, without variables, you're basically stuck with the XPath
expression

concat("This is some ", '"', "quoted", '"', " content of Ted's")

and then of course you have to backslash-escape the quotes to get them
through Javascript.

Michael Kay
http://www.saxonica.com/


> -----Original Message-----
> From: Duane Hitz [mailto:dhitz@xxxxxxxxxxxxxxxx] 
> Sent: 28 April 2007 15:27
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject:  Xoath quote escaping
> 
> All,
> 
> I have a situation where I am building an XPath expression in 
> JavaScript to select nodes from an XML document.  The content 
> of the nodes could contain double quotes, single quotes or 
> both at the same time.  It is user-entered data, so I have no 
> control ultimately over what the content contains.
> 
> The content might be (brackets excluded): [This is some 
> "quoted" content of Ted's]
> 
> The dynamically built xpath expression might look like (where 
> subNodeValue is the above content):
> 
> 1) "//node["  + subNodeName + " = '" + subNodeValue + "']"
> 
> ... except for the single quote problem... or:
> 
> 2) "//node["  + subNodeName + " = \"" + subNodeValue + "\"]"
> 
> ... except for the double quote problem... or even:
> 
> 3) "//node["  + subNodeName + " = "" + subNodeValue + ""]"
> 
> ...except that this just doesn't work.
> 
> Any thoughts on how to make this work?  Is it possible?
> 
> I've read everything I can find on quote escaping (on this list and
> elsewhere) and there doesn't appear to be a solution that 
> covers this case (I know XPath 2 is supposed to allow "" 
> escaping - unfortunately, we're stuck on IE6+ & MSXML3).
> 
> We are dynamically displaying rows in a table that meet the 
> xpath criteria... setting a @showing attribute and then 
> re-transforming... so it really has to be in JavaScript.
> 
> One option would be to replace (s/"/'/g or s/"/''/g) all 
> double quotes on the server side when building the XML - and 
> then use option 2 above.  I don't really want to change 
> user-entered data. It is, sadly, what I've come down to.  I'm 
> hoping there's a more elegant solution to this.
> 
> Thanks for any help...
> 
> Duane