コード例 #1
0
    def __str__(self):
        return ". Threads: " + str(self.thread) \


db = MongoDB()
db.query("labsk_merge")
db.insert("labsk_" + str(datetime.now()))

starttime = datetime.now()

listener = StdListener()
threads = ProcessThreads(db, MsgPageFactory())
threads.setListener(listener)
threads.setPageLimit(1)
threads.setMsgPageLimit(250)  # Nunca bajes este valor o perderas mensajes, al menos mantenlo igual

threads.scrapListOfURL(labsk_urls)
delta = datetime.now() - starttime

print "----------------------------------------------"
print "Total time: ", delta
print "Page limit ", threads.pagelimit, " Msg page limit ", threads.msgpagelimit
print str(listener)

mr = db.merge('link')
print str(mr)


print "Scrapping planeta ludico"
blogs = PlanetaLudicoScrap(db.blogs_collection())