On Tue, 13 Dec 2005 10:06:00 -0700, Andrew Shafer wrote: > [...] > After the > rss = xmlParseMemory(msg->response.body, msg->response.length); > > The xml tree appears the same as the source file. The problem occurs in > the following situation: > > <description><![CDATA["One says definitely, the other says maybe. > At this point, the Nets will take whatever they can > get. . ." ]]></description> >
Ah, but that explains it all: the CDATA signals to the parser that the following characters (up to ']]>') are to be treated as _literals_. > After a the stylesheet is applied, the " are turned into > &quot;. Some of the CDATA has " which come out ok in the eventual > html. > Which is correct: the '&' character in a literal "&qout;" has to be escaped. This is (one of) the reason(s) why having kind-of-escaped HTML in an XML feed is really a bad idea [see for example: http://www.xml.com/pub/a/2003/08/20/embedded.html] CDATA is an unfortunate leftover from SGML where it served as a convenience for authors who entered text manually. > All the template do is grab the <title>,<link> and <description> then > throw the values into an html <table>. > >> So, where's the problem? > > Does that make it more clear? Yes, it does (exactly as i feared :). As a possible workaround have a look at the cdata-section-elements attribute of the xslt:output tag. HTH Ralf Mattes > Cheers, > > Andrew _______________________________________________ xslt mailing list, project page http://xmlsoft.org/XSLT/ [email protected] http://mail.gnome.org/mailman/listinfo/xslt
