Ejemplo n.º 1
0
def main(argv=sys.argv):
	sites = webinfo.getuncrawled()
	for i in sites:
		links = webinfo.getweblinks(i)
		for j in links:
			crawl(j)
        	webinfo.markcrawled(i)
Ejemplo n.º 2
0
def crawlfromdb():
        global webinfo,httputil
        if webinfo.getuncrawled() == False:
                metainfo = httputil.getmeta(url)
                links = httputil.getlinks(url)
                print "meta data size :"+str(len(metainfo))+" links size : "+str(len(links))
                webinfo.savemetainfo(url, metainfo)
                webinfo.saveweblinks(url, links)
		webinfo.markcrawled(url)