[kovidgoyal]

Looking at that log, most of the time seems to be taken up parsing the downloaded html. You could probably improve it by adding a few preprocess_regexps to the recipe to cleanup the markup before parsing.