|
@@ -4,7 +4,6 @@ import time
|
|
|
import requests.exceptions
|
|
|
from lxml.html import fromstring, HtmlElement, tostring
|
|
|
from lxml.html.clean import Cleaner
|
|
|
-from pymongo.errors import DuplicateKeyError
|
|
|
|
|
|
from crawler.check_utils import CheckText, CheckTask
|
|
|
from crawler.clean_html import cleaner
|
|
@@ -13,7 +12,6 @@ from crawler.login import login, load_login_cookies, login_check
|
|
|
from utils.databases import mongo_table, int2long
|
|
|
from utils.execptions import YbwCrawlError
|
|
|
from utils.log import logger
|
|
|
-from utils.socks5 import Proxy
|
|
|
|
|
|
|
|
|
def iter_node(element: HtmlElement):
|