selenium 临时使用记录

截屏2021-08-12 11.32.22.png

def webdrive_get2(url: str):
    random_agent = USER_AGENTS[randint(0, len(USER_AGENTS) - 1)]
    op = Options()
    op.add_argument("--headless")
    op.add_argument("--disable-gpu")
    op.add_argument(f'user-agent="{random_agent}"')
    op.add_argument("--proxy-server=http://127.0.0.1:7890")
    dr = "/Users/rabbit/opt/selenium/webDriver/chromedriver"
    chromeA = webdriver.Chrome(options=op, executable_path=dr)
    chromeA.get(url)

    try:
        u = WebDriverWait(chromeA, 3, 2).until(EC.presence_of_all_elements_located((By.TAG_NAME, "td")))
        # print(u[1].text)
        for i in u:
            print(i.text)
    finally:
        # XPATH 测试
        a = chromeA.find_element(By.XPATH, "/html/body/table/tbody/tr[2]")
        print(a.text)
        chromeA.quit()

截屏2021-08-12 10.19.19.png

def webdriver_get(url: str):
    random_agent = USER_AGENTS[randint(0, len(USER_AGENTS) - 1)]
    op = Options()
    op.add_argument("--headless")
    op.add_argument("--disable-gpu")
    op.add_argument(f'user-agent="{random_agent}"')
    op.add_argument("--proxy-server=http://127.0.0.1:7890")
    dr = "/Users/ruiliang/opt/selenium/webDriver/chromedriver"
    chromeA = webdriver.Chrome(options=op, executable_path=dr)
    chromeA.get(url)

    print(f"Down: \t{chromeA.title} \n")

    u = WebDriverWait(chromeA, 5).until(lambda x: x.find_elements(By.XPATH, "/html/body/div/div[2]/div[4]/div/ul/li"))
    # print(f"Master => {chromeA.current_window_handle}")
    for i in track(range(len(u))):
        u = WebDriverWait(chromeA, 5).until(
            lambda x: x.find_elements(By.XPATH, "/html/body/div/div[2]/div[4]/div/ul/li"))
        ui = u[i].find_element_by_tag_name("a").get_attribute("href")

        chromeA.get(ui)
        # print(f"New Page => {chromeA.current_window_handle}")

        p = WebDriverWait(chromeA, 3).until(lambda x: x.find_element_by_class_name("fitVertical"))
        img = p.find_element_by_tag_name("img").get_attribute("src")

        filename = str(img).split("/")[-1]
        suffix = str(img).split(".")[-1]
        var1 = "%05d" % i
        full_name = str(var1) + "." + suffix
        print("Down: \t" + filename + "\t" + full_name)

        stream = send_request(img)
        write_file(full_name, stream)

        chromeA.back()
    chromeA.quit()

截屏2021-08-12 12.34.30.png