def crawlUserWeibo(self,url=None,pageHandler = None,threshold=0.3):
if url is None:
url = "http://weibo.com/"+str(self.userInfo["id"])
if pageHandler is None:
pageHandler = self.pageHandler_weibo
# ?????
self.driver.get(url)
time.sleep(2)
thresholdTime = time.time()+threshold
loopNum = 0
pageNum = 1
ifHandle = False
# ?????????????????
while True:
# ??????
self.driver.find_element_by_tag_name("body").send_keys(Keys.END)
# ??threshold???????????????
if time.time()>thresholdTime:
if u"???" in self.driver.page_source:
if ifHandle==False:
pageHandler(self.driver.page_source, pageNum)
ifHandle = True
try:
self.driver.find_element_by_partial_link_text("???").click()
loopNum = 0
ifHandle = False
pageNum = pageNum+1
print "page:"+str(pageNum)
except:
thresholdTime = time.time()+threshold
continue
else:
thresholdTime = time.time()+threshold
loopNum = loopNum + 1
continue
# ????????,????????????,????
if loopNum>10:
loopNum = 0
self.driver.refresh()
# ????????????????ID???????
评论列表
文章目录