12306爬取基本车次信息(需下载chromedriver)

from selenium import webdriver
from selenium.webdriver.common.by import By
import re
import time
import csv
import pandas as pd
import sys
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.chrome.options import Options
import os

def request(fromCity,toCity):
def getDriver():
chrome_options = Options()
chrome_options.add_argument("--no-sandbox")#禁用沙盒模式
chrome_options.add_argument("--disable-dev-shm-usage")#禁用/dev/shm使用
#chrome_options.add_argument("--headless")#浏览器隐藏运行
#设置选项参数
service = Service("chromedriver.exe")
driver = webdriver.Chrome(service=service,options=chrome_options)#修改为自己电脑中chromedriver的路径
return driver

def downhandle(ccity):pattern = re.compile(r"[^\u4e00-\u9fa5]")#剔除非汉字模式for i in range(0,6):try:getCity = driver.find_element(By.ID,"citem_%d"%(i))#获取下拉框选定城市select=re.sub(pattern,"",getCity.text)#提取中文城市名if select==ccity:getCity.click()#获取城市名breakexcept Exception:pass##下拉框城市名获取处理
dateTime = time.strftime("%m-%d",time.localtime())#获取当前日期
url = "https://kyfw.12306.cn/otn/leftTicket/init?linktypeid=dc&fs=,SHH&ts=,TJP&date=2024-%s&flag=N,N,Y"%dateTime 
driver = getDriver()
driver.get(url)driver.find_element(By.ID,"fromStationText").clear()
driver.find_element(By.ID,"toStationText").clear()
##清除文本框driver.find_element(By.ID,"fromStationText").send_keys(fromCity)#发送出发城市
time.sleep(0.2)#等待下拉框加载选择
downhandle(fromCity)
driver.find_element(By.ID,"toStationText").send_keys(toCity)#发送目的城市
time.sleep(0.2)#..
downhandle(toCity)
time.sleep(0.5)
driver.find_element(By.ID,"query_ticket").click()#点击查询
time.sleep(0.2)
##城市车次查询allNumbers=[]#车次名列表
Starts=[]#起点站列表
Terminals=[]#到达站列表
startTimes=[]#出发时间列表
endTimes=[]#到达时间列表
totalTime=[]#历时列表
try:numbercount = len(driver.find_elements(By.CLASS_NAME,"number"))#获取车次总数i=0while(numbercount>i):time.sleep(0.2)number = driver.find_elements(By.XPATH,"//a[@class='number']")[i].text#获取车次名start = driver.find_elements(By.XPATH,"//div[@class='cdz']")[i].text#获取起点站名terminal = (driver.find_elements(By.XPATH,"//div[@class='cdz']")[i].text).replace(start,"")#获取到达站名startT = driver.find_elements(By.XPATH,"//div[@class='cds']")[i].text#获取出发时间endT = (driver.find_elements(By.XPATH,"//div[@class='cds']")[i].text).replace(startT,"")#获取到达时间total = driver.find_elements(By.XPATH,"//div[@class='ls']")[i].text#获取历时allNumbers.append(number)#添加车次名Starts.append(start)#添加起点站名Terminals.append(terminal)#添加到达站名startTimes.append(startT)#添加出发时间endTimes.append(endT)#添加到达时间totalTime.append(total)#添加历时i+=1
except Exception:passwith open("data.csv","w+",encoding="utf-8"):#创建csv文件pass
data = {"车次":allNumbers,"出发站":Starts,"到达站":Terminals,"出发时间":startTimes,"到达时间":endTimes,"历时:":totalTime}
dataFrame = pd.DataFrame(data)
dataFrame.to_csv(r"data.csv",encoding="utf_8_sig")
print("已生成车次信息xlsx表格！")
time.sleep(1)
sys.exit()##按列写入车次数据

def main():
while True:
fromCity=input("请输入起点站：")
toCity=input("请输入到达站：")
if fromCity != "" and toCity != "":
request(fromCity,toCity)
else:
print("站名不能为空！")
#获取城市名

if name == "main":
main()