Пример #1
0
def mainspider():
    from sqlalchemy import create_engine
    from sqlalchemy.orm import sessionmaker
    engine = create_engine(
        MYSQL_URL,
        encoding="utf-8"
    )
    Session = sessionmaker(engine)
    session = Session()

    lasturl = wooyun.getlastbug()
    wooyun.savelastbug(lastbug())

    pagenum = wooyun.getpagenum()
    for pages in xrange(120, pagenum):
        buglist = wooyun.getwooyun(pages)
        print "page:" + str(pages)
        for bug in buglist:
            bugurl = bug[2]
            if bugurl == lasturl:
                print "You have Update ALL the News"
                return 0
            url = 'http://www.wooyun.org' + str(bug[2])
            content = getcontent.getContent(url)
            ucontent = getcontent.getimage(content, bug[0])
            entitywoo.pushContent(session, bug[2], ucontent)
            if len(bug):
                entitywoo.pushbuginfo(
                    session, bug[0],
                    bug[1], bug[2], bug[3]
                    )
            print bug[2]
    print "You have Update ALL the bugs"
    return 0
Пример #2
0
def lastbug():
    PAGE_NUM_1 = 1
    buglist = wooyun.getwooyun(PAGE_NUM_1)
    url = buglist[0][2]
    return url