def debug(self): url = "https://www.yify-torrent.org/latest-{0}.html" for i in range(0,257): yield request(url=url.format(str(i)),callback="_content") a = "https://www.yify-torrent.org/popular-{0}.html" for i in range(0,257): yield request(url=a.format(str(i)),callback="_content")
def start(self, url=None): yield request(url="http://bttiantangs.com/", callback="cc").cron("20 0 * * *") yield request(url="http://bttiantangs.com/list/dianying/index.html", callback="cc").cron("40 0 * * *") yield request(url="http://bttiantangs.com/list/dianying/index_2.html", callback="cc").cron("20 0 * * *") yield request(url="http://bttiantangs.com/list/dianying/index_3.html", callback="cc").cron("50 0 * * *") yield request(url="http://bttiantangs.com/top/update.html", callback="cc").cron("10 1 * * *")
def start(self): yield request( url="https://www.dygod.net/html/gndy/jddy/20181126/111507.html", callback='link') yield request( url="https://www.dygod.net/html/gndy/jddy/20181126/111507.html", callback='link') yield request( url="https://www.dygod.net/html/gndy/jddy/20181126/111507.html", callback='link')
def cc(self, response): if response.body: dd = response.body else: dd = 'erro' d = html.fromstring(dd) hre = d.xpath("//div[@class='article']//a/@href") for i in map(lambda x: self.url + x, hre): log.info("push url. {0}".format(i)) yield request(url=i, callback="content1")
def debug(self): for i in range(1, 50000): yield request( url="http://www.bttiantangs.com/movie/{0}.html".format(str(i)), callback="content1")
def _content(self,response): response.tree.make_links_absolute(response.url) for i in response.tree.xpath("//*[@class='mv']//a//@href"): yield request(url=i,callback='content')
def start(self,url=None): yield request(url="https://www.yify-torrent.org/latest.html",callback="_content").cron("30 0 * * *") yield request(url="https://www.yify-torrent.org/popular.html",callback="_content").cron("10 0 * * *")
def link(self, a): print "。。。。。。 。。。。 。。 " yield request( url="https://www.dygod.net/html/gndy/jddy/20181126/111507.html", callback='link')
def link(self, a): yield request( url="https://www.dygod.net/html/gndy/jddy/20181126/111502.html", callback='link2')