Files
spider/core/spiders.py

32 lines
1.2 KiB
Python

import os
from utils import browsers
class Spiders:
def __init__(self, project_name: str, bitch: int = 5, worker_num: int = 1, switch_clash: bool = False, reflush_browser: bool = False):
self.bitch = bitch
self.worker_num = worker_num
self.switch_clash = switch_clash
self.reflush_browsers = reflush_browser
self.__data_path__ = f"data/{project_name}"
self.__database_path__ = f"database/{project_name}.db"
self.__urls_folder__ = f"{self.__data_path__}/urls"
self.__excels_folder__ = f"{self.__data_path__}/excels"
self.__errors_folder__ = f"{self.__data_path__}/errors"
if os.path.exists(self.__data_path__) == False:
os.makedirs(self.__data_path__)
os.makedirs(self.__urls_folder__)
os.makedirs(self.__excels_folder__)
os.makedirs(self.__errors_folder__)
self.browser = browsers.getBrowsers()
self.browser.set.timeouts(3)
self.tab = self.browser.get_tab(0)
self.tab.set.window.max()
def reflush_browser(self):
self.browser.quit()
self.browser = browsers.getBrowsers()
self.browser.set.timeouts(3)