I know you're not looking to go back through every page, but you don't really need to parse the whole page, just look for the html that always precedes an entry. From me just starting up google web history and doing some simple searches, if you look through a history page, each String that you've searched follows: <td style="padding:3px 0"><table id=bkmk_view_ class=noborder ><tr><td><table class="elem noborder"><tr><td class="grey" nowrap>Searched for </td><td nowrap><a title="http://www.google.com/search?q=
and is followed by &
(ampersand). This sequence of preceding html is unique on the page, only occuring when historical search terms are listed.
If you use two terms, you get a + in between the terms. Other conventions for different searching modes, I didn't go through them all.
It looks like if you use BalusC's method to pass parameters, then you could retreive the html, search the document for the string I mentioned(be sure to \" and other special characters), then copy the next String until you reach a & character. Then, all you need to do is parse your search term, not the whole page. Go through source code until you reach the end, then go to your next iteration in the loop.