抓取2021百科知识竞赛题库和答案

抓取2021百科知识竞赛题库和答案
刚学习PYTHON，希望路过的高手给个JSON方法。

from selenium import webdriver
import time
# 引入
driver = webdriver.Firefox()
# 用Firefox浏览器
driver.get("https://www.jhq8.cn/daan/tiku/3/")
time.sleep(10)
latest_window = driver.window_handles[-1]
driver.switch_to.window(latest_window)
time.sleep(35)
#list3=(8,27,39,43)  #只抓取部分网页，容易漏的网页
for i in range(1, 127):
#for i in list3:
    url = "https://www.jhq8.cn/daan/tiku/3/" + str(i) + ".html"
    try:
        latest_window = driver.window_handles[-1]
        driver.switch_to.window(latest_window)
        time.sleep(3)

        driver.get(url)
        time.sleep(20)
        with open("抓取的题库3774答案.txt", "a+") as f:
            f.write(url)
            f.write("\n")
        for j in range(1, 31):
            latest_window = driver.window_handles[-1]
            driver.switch_to.window(latest_window)
            time.sleep(2)
            tigan = "/html/body/div[1]/div/div/div[4]/div/div[1]/div/ul/li[" + str(j) + "]/a"
            driver.find_element_by_xpath(tigan).click()
            # 第一题

            latest_window = driver.window_handles[-1]
            driver.switch_to.window(latest_window)
            time.sleep(6)

            g1 = driver.find_element_by_xpath("/html/body/div[1]/div/div/div[3]/div/div/div[1]/div[2]/p").text.strip()
            time.sleep(2)
            with open("抓取的题库3774答案.txt", "a+") as f:
                f.write(tigan)
                f.write("\n")
                f.write(g1)
                f.write("\n")

            time.sleep(2)

            latest_window = driver.window_handles[-1]
            driver.switch_to.window(latest_window)
            driver.close()
            time.sleep(2)
    except Exception as e:
        print(e)
        #latest_window = driver.window_handles[-1]
        #driver.switch_to.window(latest_window)
        # driver.close()
        # time.sleep(2)