def readlx(startpage, driver):
jumpage(startpage, driver)
flag = True
while (flag):
stopflag = readpage(driver)
if not stopflag:
break
try:
current = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
a = current
clickforname('/html/body/div[3]/div/section[2]/div/nav/ul/li[', 1,
']/a', '下一页', driver)
while (current == a):
sleep(3)
a = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
except:
break
def jumpage(startpage, driver):
currentindex = ''
stamp = 1
max = 5
while (currentindex != str(startpage)):
try:
currentindex = textXpath(
'/html/body/div[3]/div/section[2]/div/nav/ul/li[' +
str(stamp) + ']/a', driver)
except:
try:
current = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
a = current
clickforname('/html/body/div[3]/div/section[2]/div/nav/ul/li[',
1, ']/a', str(max), driver)
while (current == a):
sleep(3)
a = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
max += 5
stamp = 0
current = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
a = current
clickforname('/html/body/div[3]/div/section[2]/div/nav/ul/li[',
1, ']/a', '下一页', driver)
while (current == a):
sleep(3)
a = textXpath(
'/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',
driver)
except:
break
stamp += 1
def readtype(driver):
stamp = 1
global zzcount
while (True):
flag = readpage(driver)
if not flag:
zzcount = 0
return False
current = textXpath(
'/html/body/form/div[5]/div[2]/div/table/tbody/tr[1]/td[2]/a',
driver)
clickforname('/html/body/form/div[5]/div[2]/div/div/div/ul/li[', stamp,
']/a', '下一页', driver)
try:
while (current == textXpath(
'/html/body/form/div[5]/div[2]/div/table/tbody/tr[1]/td[2]/a',
driver, 30)):
sleep(3)
except:
zzcount = 0
return True
zzcount = 0
return True
driver.switch_to.window(handle)
if not flag:
return False
except:
driver.switch_to.window(handle)
break
return True
if __name__ == '__main__':
url = 'http://218.69.33.182/epr/Search/C_Common/CE/index.aspx'
driver = LoginUrlChrome(url)
print("input zznum:")
zznum = input()
stamp = 5
while (True):
flag = readpage(driver)
if not flag:
break
current = textXpath(
'/html/body/form/div[3]/div/div/table/tbody/tr/td/table[1]/tbody/tr[1]/td[2]',
driver)
clickforname(
'//*[@id="ASPxGridView2_ctl04"]/tbody/tr/td/table/tbody/tr/td[',
stamp, ']', '下一页', driver)
while (current == textXpath(
'/html/body/form/div[3]/div/div/table/tbody/tr/td/table[1]/tbody/tr[1]/td[2]',
driver, 3)):
sleep(3)
driver.quit()
stamp += 1
if not stopflag:
print(zzcount, zznum)
return False
except:
break
return True
if __name__ == '__main__':
driver= LoginUrl('http://gcxm.hunanjs.gov.cn/dataservice.html?queryType=0&keyword=')
print("*1、页面拖动验证条；2、先选好起始爬取页")
print("输入爬取资质数：")
zznum=input()
flag=True
stamp=1
while(flag):
stopflag=readpage(driver)
if not stopflag:
break
try:
current=textXpath('/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',driver)
a=current
clickforname('/html/body/div[3]/div/section[2]/div/nav/ul/li[',1,']/a','下一页',driver)
while(current==a):
sleep(3)
a=textXpath('/html/body/div[3]/div/section[1]/table/tbody/tr[1]/td[2]/a',driver)
except:
break
print(0)
driver.switch_to.window(newhandle)
flag = readzz()
driver.close()
if not flag:
return False
driver.switch_to.window(handle)
return True
if __name__ == '__main__':
url = 'http://dn4.gxzjt.gov.cn:1141/WebInfo/Person/Person.aspx'
driver = LoginUrl(url)
print('input zznum:')
zznum = input()
stamp = 1
'/html/body/form/div[6]/div[2]/div[2]/div/div/table/tbody/tr/td[2]/a[' ']'
while (True):
flag = readpage(driver)
if not flag:
break
current = textXpath(
'/html/body/form/div[6]/div[2]/div[2]/div/table/tbody/tr[2]/td[2]/a',
driver)
clickforname(
'/html/body/form/div[6]/div[2]/div[2]/div/div/table/tbody/tr/td[2]/a[',
stamp, ']', '下一页>', driver)
while (current == textXpath(
'/html/body/form/div[6]/div[2]/div[2]/div/table/tbody/tr[2]/td[2]/a',
driver)):
sleep(3)
driver.switch_to.window(newhandle)
flag = readzz()
driver.close()
if not flag:
return False
driver.switch_to.window(handle)
return True
if __name__ == '__main__':
driver = LoginUrl(
'http://110.16.70.26/nmjgpublisher/corpinfo/CorpInfoObtain.aspx#')
print('input zznum:')
zznum = input()
stamp = 1
while (True):
flag = readpage(driver)
if not flag:
break
current = textXpath(
'/html/body/form/div[4]/div[2]/div[1]/div[2]/div/div/table/tbody/tr[1]/td[2]',
driver)
clickforname(
'/html/body/form/div[4]/div[2]/div[1]/div[2]/div/div/div/div/ul/li[',
stamp, ']/a', '下一页', driver)
while (current == textXpath(
'/html/body/form/div[4]/div[2]/div[1]/div[2]/div/div/table/tbody/tr[1]/td[2]',
driver)):
sleep(3)
else:
_ = 1
except:
flag = False
break
return True
if __name__ == '__main__':
url = 'http://dn4.gxzjt.gov.cn:1141/WebInfo/Default.aspx'
driver = LoginUrIE(url)
print('input zznum:')
zznum = input()
flag = True
stamp = 1
while (flag):
if not readpage(driver):
break
current = textXpath(
'//*[@id="ContentPlaceHolder1_List_Datagrid1"]/tbody/tr[2]/td[2]/a',
driver)
clickforname(
'//*[@id="ContentPlaceHolder1_List_Pager"]/table/tbody/tr/td[2]/a[',
stamp, ']', '下一页>', driver)
while (current == textXpath(
'//*[@id="ContentPlaceHolder1_List_Datagrid1"]/tbody/tr[2]/td[2]/a',
driver, 3)):
sleep(3)