openi_online / sel.py
ACCA225's picture
Update sel.py
925507b verified
raw
history blame contribute delete
No virus
9.67 kB
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
import time
import json
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.common.keys import Keys
from urllib.parse import urlparse, urlunparse
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.chrome.options import Options
#选项
chrome_options=Options()
#设置chrome浏览器为无界面模式
chrome_options.add_argument('--headless')
#下面的两行代码一般是不需要的,至少经过我的测试在windows上是不需要的。但经过我的测试在Ubuntu上如果不加的话等下就会报错WebDriverException: unknown error: DevToolsActivePort file doesn't exist while trying to initiate Chrome Browser,所以要加上
chrome_options.add_argument('--no-sandbox')
chrome_options.add_argument('--disable-dev-shm-usage')
is_login = True
def main():
driver = webdriver.Chrome(options=chrome_options)
def login():
try:
driver.get("https://openi.pcl.ac.cn/user/login")
#driver.implicitly_wait(0.5)
text_box = driver.find_element(by=By.NAME, value="user_name")
password = driver.find_element(by=By.ID, value="input_password")
submit_button = driver.find_element(by=By.ID, value="submitId")
text_box.send_keys("184202787==")
password.send_keys("Qq2575044704")
submit_button.click()
driver.add_cookie({"name": "foo", "value": "bar"})
print(driver.get_cookie("foo"))
except Exception as e:
print(f"An error occurred: {e}")
# Function to perform the login and task creation
if is_login == True:
print("Logging in...")
login()
else:
print("跳过登录")
def stop():
driver.get("https://openi.pcl.ac.cn/asdfqwerty/sd2/grampus/onlineinfer")
# 等待可按按钮元素可见
button_element = WebDriverWait(driver, 100).until(
EC.visibility_of_element_located((By.XPATH, "//a[@class='']"))
)
# 检查按钮是否可点击
if button_element.is_enabled():
# 点击按钮
ActionChains(driver).move_to_element(button_element).click().perform()
#button_element.click()
print("停止按钮已点击")
else:
print("停止不可点击")
def perform_operations(driver):
if is_login == False:
driver.get("https://openi.pcl.ac.cn/index.html")
# 首先清除由于浏览器打开已有的cookies
driver.delete_all_cookies()
with open('cookies.txt','r') as f:
# 使用json读取cookies 注意读取的是文件 所以用load而不是loads
cookies_list = json.load(f)
for cookie in cookies_list:
driver.add_cookie(cookie)
else:
pass
# 打开新建任务网页
stop() #先停止
def fill_form(image_address, startup_file):
try:
# 使用显式等待等待输入框出现,这里等待10秒,可以根据实际情况调整
image_input = WebDriverWait(driver, 10).until(
EC.element_to_be_clickable((By.CLASS_NAME, 'el-input__inner'))
)
startup_input = WebDriverWait(driver, 10).until(
EC.element_to_be_clickable((By.CSS_SELECTOR, 'input[placeholder="请输入启动文件"].el-input__inner'))
)
# 填写表单
image_input.send_keys(image_address)
startup_input.send_keys(startup_file)
# 执行其他操作...
except Exception as e:
print(f"填写表单过程中发生错误: {str(e)}")
#填写表单
# Wait for the new task button to appear
def click_close_button():
driver.get("https://openi.pcl.ac.cn/asdfqwerty/sd2/grampus/onlineinfer/create")
try:
# 使用显式等待等待按钮出现,这里等待10秒,可以根据实际情况调整
close_button = WebDriverWait(driver, 10).until(
EC.presence_of_element_located((By.CSS_SELECTOR, 'button.el-button.el-button--primary.el-button--small'))
)
# 使用 ActionChains 模拟鼠标移动到按钮上并点击
actions = ActionChains(driver)
actions.move_to_element(close_button).click().perform()
# 执行其他操作...
print("点击关闭按钮成功")
time.sleep(2)
except Exception as e:
print(f"点击关闭按钮过程中发生错误: {str(e)}")
def click_create_task_button():
try:
# 使用显式等待等待按钮出现,这里等待10秒,可以根据实际情况调整
create_task_button = WebDriverWait(driver, 10).until(
EC.presence_of_element_located((By.CSS_SELECTOR, 'button.el-button.submit-btn'))
)
# 点击按钮
# 使用 ActionChains 模拟鼠标移动到按钮上并点击
actions = ActionChains(driver)
actions.move_to_element(create_task_button).click().perform()
# 执行其他操作...
print("创建任务按钮已按下")
except Exception as e:
print(f"点击按钮过程中发生错误: {str(e)}")
def click_online_inference_link():
global current_url
try:
driver.get("https://openi.pcl.ac.cn/asdfqwerty/sd2/grampus/onlineinfer")
# 使用显式等待等待链接出现,这里等待10秒,可以根据实际情况调整
online_inference_link = WebDriverWait(driver, 10).until(
EC.element_to_be_clickable((By.XPATH, "//div[@class='op-wrap']//a[contains(text(), '在线推理')]"))
)
actions = ActionChains(driver)
actions.move_to_element(online_inference_link).click().perform()
# 获取当前窗口句柄
current_window_handle = driver.current_window_handle
# 点击链接,在新标签页中打开
online_inference_link.send_keys(Keys.CONTROL + Keys.RETURN)
# 等待新标签页打开
WebDriverWait(driver, 10).until(EC.number_of_windows_to_be(2))
# 获取所有窗口句柄
all_window_handles = driver.window_handles
# 切换到新标签页
for window_handle in all_window_handles:
if window_handle != current_window_handle:
driver.switch_to.window(window_handle)
break
current_url = driver.current_url
print(f"当前浏览的网页URL: {current_url}")
except Exception as e:
print(f"点击链接过程中发生错误: {str(e)}")
def run():
if "is-disabled" in new_task_button.get_attribute("class"):
print("不可用,restart")
driver.quit()
main()
perform_operations(driver)
else:
# Click the new task button
new_task_button.click()
#run()
# Create a webdriver session
click_close_button()
fill_form("192.168.204.22:5000/default-workspace/99280a9940ae44ca8f5892134386fddb/image:sd_v10", "app.py")
click_create_task_button()
click_create_task_button()
click_create_task_button()
#!sleep 60
#click_online_inference_link()
# Main loop with a 30-minute interval
try:
perform_operations(driver)
# 去掉端口号
parsed_url = urlparse(current_url)
netloc_without_port = parsed_url.netloc.split(":")[0]
# 去掉token部分
new_path = parsed_url.path
if "?" in new_path:
new_path = new_path.split("?")[0]
# 构建新的URL,省略端口号和token部分
new_url = urlunparse((parsed_url.scheme, netloc_without_port, new_path,
parsed_url.params, parsed_url.query, parsed_url.fragment))
print(f"当前浏览的网页URL(去掉端口号): {new_url}")
print("完成")
time.sleep(200) # Sleep for 30 minutes (1800 seconds)
finally:
# Close the browser outside the loop
driver.quit()
import requests
def detect():
url = 'https://mouse-glowing-husky.ngrok-free.app/internal/ping'
headers = {'accept': 'application/json'}
response = requests.get(url, headers=headers)
if response.status_code == 200:
#main() #测试用
print("成功")
#print(response.json())
time.sleep(5)
detect()
else:
print("failed")
main()
detect()
while True:
main()