BRE/webapi/run_loader.py

57 lines
1.5 KiB
Python
Raw Normal View History

2024-06-07 19:50:21 +03:00
'''Run electron BRE loading'''
import sys
import time
import portal
START_ROUND = 1
NUMBER_OF_ROUNDS = 10
NUMBER_OF_ARTICLES = 10
CONTINUE_FLAG = False
START_ID = '5052657'
STOP_ID = ''
SAVE_TO_PATH = 'PATH_TO_SAVE_FOLDER'
LOGIN_NAME = 'LOGIN'
LOGIN_PW = 'PASSWORD'
# MAIN START
curr_id = START_ID
print("Start time is", time.strftime("%H") + ":" + time.strftime("%M"))
chrome_options = portal.Options()
chrome_options.add_experimental_option("prefs", {"download.default_directory": SAVE_TO_PATH, "download.prompt_for_download": False, })
chrome_options.add_argument('--headless')
chrome = portal.WebBrowser()
for round_number in range(NUMBER_OF_ROUNDS):
if curr_id == '':
sys.exit(0)
print("*********************************")
print("** round #" + str(round_number + START_ROUND) + " of " + str(START_ROUND + NUMBER_OF_ROUNDS - 1))
print("** start time is", time.strftime("%H") + ":" + time.strftime("%M"))
print("*********************************")
chrome.start_chrome(chrome_options)
processor = portal.GBDownloader(chrome, SAVE_TO_PATH, CONTINUE_FLAG, suffix_seed = round_number + START_ROUND)
if not processor.login(LOGIN_NAME, LOGIN_PW):
sys.exit(0)
curr_id = processor.scan(curr_id, STOP_ID, NUMBER_OF_ARTICLES)
time.sleep(2)
del processor
chrome.stop_chrome()
time.sleep(3)
print("** end time is", time.strftime("%H") + ":" + time.strftime("%M"))
print("")
print("next start article ID = " + curr_id)
print("next START_ROUND = " + str(START_ROUND + NUMBER_OF_ROUNDS))