class WS0607_TI3_ASSIGNMENTS(ScrapedNFed.ScrapedFeed): def __init__(self): ScrapedNFed.ScrapedFeed.__init__(self, "ws0607_ti3_assignments.pickle") self.rssFile = "output_xml/ws0607_ti3_assignments.xml" self.link = "http://www-depreciated.inf.fu-berlin.de/inst/ag-tech/teaching/WS0607/19513-V/index.html" self.title = u"TI III: Betriebssysteme und Rechnernetze: Assignments" def HTML2RSS(self, headers, body): soup = BeautifulSoup.BeautifulSoup(body) for i in soup.findAll(href=re.compile("Uebung_Aufgaben")): yield self.encloseItem(PyRSS2Gen.RSSItem( title = i.string, link = urlparse.urljoin(self.link, i["href"]), description = i.string, ))