def start_requests(self):
 for i in range(1,101):
 page=i*2-1 #这里是构造请求url的page,表示奇数
 url=self.start_url+str(page)
 yield scrapy.Request(url,meta={'search_page':page+1},callback=self.parse_url) #这里使用meta想回调函数传入数据，回调函数使用response.meta['search-page']接受数据

def parse_url(self,response):
 if response.status==200: #判断是否请求成功
 # print response.url
 pids = set() #这个集合用于过滤和保存得到的id,用于作为后面的ajax请求的url构成
 try:
 all_goods = response.xpath("//div[@id='J_goodsList']/ul/li") #首先得到所有衣服的整个框架，然后从中抽取每一个框架
 for goods in all_goods: #从中解析每一个
 # scrapy.shell.inspect_response(response,self) #这是一个调试的方法，这里会直接打开调试模式
 items = JdSpiderItem() #定义要抓取的数据
 img_url_src = goods.xpath("div/div[1]/a/img/@src").extract() # 如果不存在就是一个空数组[]，因此不能在这里取[0]
 img_url_delay = goods.xpath(
  "div/div[1]/a/img/@data-lazy-img").extract() # 这个是没有加载出来的图片，这里不能写上数组取第一个[0]
 price = goods.xpath("div/div[3]/strong/i/text()").extract() #价格
 cloths_name = goods.xpath("div/div[4]/a/em/text()").extract()
 shop_id = goods.xpath("div/div[7]/@ data-shopid").extract()
 cloths_url = goods.xpath("div/div[1]/a/@href").extract()
 person_number = goods.xpath("div/div[5]/strong/a/text()").extract()
 pid = goods.xpath("@data-pid").extract()
 # product_id=goods.xpath("@data-sku").extract()
 if pid:
  pids.add(pid[0])
 if img_url_src: # 如果img_url_src存在
  print img_url_src[0]
  items['img_url'] = img_url_src[0]
 if img_url_delay: # 如果到了没有加载完成的图片，就取这个url
  print img_url_delay[0]
  items['img_url'] = img_url_delay[0] # 这里如果数组不是空的，就能写了
 if price:
  items['price'] = price[0]
 if cloths_name:
  items['cloths_name'] = cloths_name[0]
 if shop_id:
  items['shop_id'] = shop_id[0]
  shop_url = "https://mall.jd.com/index-" + str(shop_id[0]) + ".html"
  items['shop_url'] = shop_url
 if cloths_url:
  items['cloths_url'] = cloths_url[0]
 if person_number:
  items['person_number'] = person_number[0]
 # if product_id:
 # print "************************************csdjkvjfskvnk***********************"
 # print self.comments_url.format(str(product_id[0]),str(self.count))
 # yield scrapy.Request(url=self.comments_url.format(str(product_id[0]),str(self.count)),callback=self.comments)
 #yield scrapy.Request写在这里就是每解析一个键裤子就会调用回调函数一次
 yield items
 except Exception:
 print "********************************************ERROR**********************************************************************"
 yield scrapy.Request(url=self.search_url.format(str(response.meta['search_page']),",".join(pids)),callback=self.next_half_parse) #再次请求，这里是请求ajax加载的数据，必须放在这里，因为只有等到得到所有的pid才能构成这个请求，回调函数用于下面的解析

#分析异步加载的网页
 def next_half_parse(self,response):
 if response.status==200:
 print response.url
 items=JdSpiderItem()
 #scrapy.shell.inspect_response(response,self) #y用来调试的
 try:
 lis=response.xpath("//li[@class='gl-item']")
 for li in lis:
  cloths_url=li.xpath("div/div[1]/a/@href").extract()
  img_url_1=li.xpath("div/div[1]/a/img/@src").extract()
  img_url_2=li.xpath("div/div[1]/a/img/@data-lazy-img").extract()
  cloths_name=li.xpath("div/div[4]/a/em/text()").extract()
  price=li.xpath("div/div[3]/strong/i/text()").extract()
  shop_id=li.xpath("div/div[7]/@data-shopid").extract()
  person_number=li.xpath("div/div[5]/strong/a/text()").extract()
  if cloths_url:
  print cloths_url[0]
  items['cloths_url']=cloths_url[0]
  if img_url_1:
  print img_url_1[0]
  items['img_url']=img_url_1
  if img_url_2:
  print img_url_2[0]
  items['img_url']=img_url_2[0]
  if cloths_name:
  items['cloths_name']=cloths_name[0]
  if price:
  items['price']=price[0]
  if shop_id:
  items['shop_id']=shop_id[0]
  items['shop_url']="https://mall.jd.com/index-" + str(shop_id[0]) + ".html"
  if person_number:
  items['person_number']=person_number[0]
  yield items #又一次的生成，这里是完整的数据，因此可以yield items
 except Exception:
 print "**************************************************"