email = args.user psw = args.psw uid = args.target #构建连接器 opener = build_opener() #尝试登录 success, gsid = simulate_login(opener, email, psw) print 'success: ', success print 'gsid : ', gsid print 'target : ', uid from sqla_db import DBSaver, DumpSaver saver = DBSaver(uid, debug=False) dump = DumpSaver(uid) def load_page(num): """ load target page back """ url = "http://weibo.cn/%s/profile?gsid=%s&page=%s" % (uid, gsid, num) r = opener.open(url) return r.read() def archive_page(page, ignore=False): soup = Soup(page) mids = get_mids(soup.prettify()) updated = False print 'mids len', len(mids)
break return fixing_page, fixed_count loader = Loader() #登录 success = loader.login(email, psw) print success print loader.gsid loader.set_target(uid) #初始化保存 archiver = Archiver(uid, debug=False) parser = create_page_parser(1) remote_count = parser.get_total_weibo_count() #TODO lock with this value durning archive total_page = parser.get_total_page_count() archived_count = archiver.get_count() print 'remote count :', remote_count print 'remote page count: ', total_page print 'archived_count :', archived_count is_success, archived_count = archive_page(start=1, end=total_page, is_update=True) print 'update success :', is_success print 'update count :', archived_count