I'm using curl to get the page, piping it into a simple one-off Perl-script that produces a comma-separated file, which in turn is read by a tiny R script, which then produces the graph.