J. S. Rawat wrote:
> Can anybody let me know about the mistake I am doing!!

Yes, you copied my untested code from last week verbatim. :)

So the mistaken impression that preceding::* includes ancestors was poor
recollection on my part.

But I'm glad you got it working.

>       @pg != preceding::*[1]/@pg

(I justified != rather than not(=) here because both sides of the
comparison are node-sets of at most one node.)