a selenium-based script to label data using Dialogflow CX console (dev environment)

cx_crawler.py

from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.chrome.options import Options
import time

path_to_data = "vi_hvac_train.txt"
path_to_save = "vi_hvac_train_res.txt"
email = "thesunkid19@gmail.com"
password = ""

options = Options()
options.binary_location = '/Applications/Brave Browser.app/Contents/MacOS/Brave Browser'
driver_path = '/Users/thesunkid/Downloads/chromedriver'
driver = webdriver.Chrome(options = options, executable_path = driver_path)

delay_time = 2

# Login google from stackoverflow to beat google's firewall
driver.get("https://stackoverflow.com/")

driver.maximize_window()
time.sleep(delay_time)

driver.find_element(By.XPATH, '/html/body/header/div/ol[2]/li[2]/a[1]').click()#Log in button in stackoverflow
time.sleep(delay_time)

driver.find_element(By.XPATH, '//*[@id="openid-buttons"]/button[1]').click()# Log in with Google button
time.sleep(delay_time)

driver.find_element(By.XPATH, '//*[@id="identifierId"]').send_keys(email)# Enter email address
time.sleep(delay_time)

driver.find_element(By.XPATH, '//*[@id="identifierNext"]/div/button/div[2]').click() # Click next button after entering email address
time.sleep(delay_time)

driver.find_element(By.XPATH, '//*[@id="password"]/div[1]/div/div[1]/input').send_keys(password)#Enter password
time.sleep(delay_time)

driver.find_element(By.XPATH, '//*[@id="passwordNext"]/div/button/div[2]').click()# Click on next button after entering the password
time.sleep(delay_time)

url = "https://dialogflow.cloud.google.com/cx/projects/vinfast-nlu/locations/global/agents/33118eff-91c9-4cb2-a9d0-76e8bc913b2d/flows/00000000-0000-0000-0000-000000000000/flow_creation"

driver.get(url)

time.sleep(8)

startBtn = driver.find_element_by_xpath('//*[@id="df-main-view-title-bar"]/df-custom-outlet/div/df-simulator/button')
startBtn.click()

lines = []
with open(path_to_data, 'r') as f:
    ls = f.readlines()
    for l in ls:
        lines.append(l.strip())

with open(path_to_save, 'a') as f_out:
    i = 0
    for line in lines[i:]:
        if "\n" == line:
            continue
        if "\n" in line:
            line = line[:-1]

        print(i,line)
        f_out.write(str(i) + "\t" + line + "\t")

        insert = driver.find_element_by_xpath('//*[@class="input-container"]/input')
        insert.send_keys(line)
        time.sleep(1)

        sendBtn = driver.find_element_by_id("button-send-wrapper")
        sendBtn.click()
        time.sleep(3)

        resultBtn = driver.find_element_by_xpath('//*[@id="response-info"]/div/div/button')
        resultBtn.click()
        time.sleep(1)

        result = driver.find_element_by_class_name("CodeMirror-lines")

        a = result.get_attribute('innerHTML')
        f_out.write(str(a).replace("\t", " ").replace("\n", "") + "\n")
        time.sleep(1)

        close = driver.find_element_by_xpath(
            '//*[@class="cdk-global-overlay-wrapper"]/div/mat-dialog-container/div/button[2]')
        close.click()
        time.sleep(2)

        i += 1