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