PyBot/Dev_test.py

135 lines
4.9 KiB
Python
Raw Normal View History

2025-01-10 16:56:13 +08:00
# # -*- coding: utf-8 -*-
# """
# @Author: MasonLiu
# @Description: 测试用脚本,无需关注。
# """
# import schedule
# import os
# import signal
# import sys
# import time
# import yaml
# import requests
# from datetime import datetime, timedelta
# from SendCore.FeishuSendBot import SendToFeishu, gen_sign
# from SendCore.QiweiSendBot import SendToWX
# from spider.common import run, seebug_main, M_4hou_main, anquanke_main, sec_wiki_main, huawei_main, doonsec_main, qianxin_main
# from spider.freebuf import freebuf_main
# from spider.xianzhi import xianzhi_main
# from spider.sougou_wx import sougou_wx_main
# from spider.github import github_main, load_github_config
# from GotoSend.M_4hou import Src_4hou
# from GotoSend.anquanke import Src_anquanke
# from GotoSend.doonsec import Src_doonsec
# from GotoSend.xianzhi import Src_xianzhi
# from GotoSend.freebuf import Src_freebuf
# from GotoSend.qianxin import Src_qianxin
# from GotoSend.seebug import Src_seebug
# from GotoSend.sougou_wx import Src_sougou_wx
# from GotoSend.github import Src_github
# from config.check_config import get_core_config, get_debug_config, get_kewords_config
# from loguru import logger
# from baiduspider import BaiduSpider
# from pprint import pprint
# # 全局变量
# webhook_url_once, timestamp_once, sign_once = gen_sign()
# e_hour, time_choice, choice, fs_activate, wx_activate, ding_activate, lx_activate, url_web = get_core_config()
# Sogou_WX, Doonsec_switch, Doonsec = get_kewords_config()
# # print(f"当前配置信息Doonsec_switch{Doonsec_switch}")
# # if __name__ == "__main__":
# # result_doonsec_long = Src_doonsec(False, Doonsec_switch, Doonsec)
# # print(result_doonsec_long)
# # while True:
# # Sogou_WX, Doonsec_switch, Doonsec = get_kewords_config()
# # print(f"当前配置信息Doonsec_switch{Doonsec_switch}")
# # print(f"当前配置信息Sogou_WX{Sogou_WX}")
# # print(f"当前配置信息Doonsec{Doonsec}")
# # print("\n")
# # time.sleep(10)
# # pprint(BaiduSpider().search_web(input(), exclude=['all']))
# import requests
# def fetch_url(url, headers):
# try:
# response = requests.get(url, headers=headers)
# response.raise_for_status() # 如果响应状态码不是200会抛出异常
# return response.text
# except requests.RequestException as e:
# print(f"请求失败: {e}")
# return None
# # 示例使用
# if __name__ == "__main__":
# url = "https://www.baidu.com/s?tn=baidurt&cl=3&rn=20&ie=utf-8&rsv_bp=1&wd=齐鲁银行" # 替换为你要爬取的URL
# headers = {
# "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3",
# "Accept-Language": "en-US,en;q=0.9",
# # 可以在这里添加其他需要的头信息
# }
# content = fetch_url(url, headers)
# if content:
# print(content)
'''
<table cellpadding="0" cellspacing="0" class="result" id="8">
<tr>
<td class=f>
<h3 class="t">
<a
onmousedown="return ns_c({
'fm':'baidurt',
'F':'778317EA',
'F1':'9D73F1C4',
'F2':'4CA6DE6A',
'F3':'54E5243F',
'T':'1736494872',
'title':this.innerHTML,
'url':'https%3A%2F%2Fwww.aiqicha.com%2Fcompany_detail_82845142671484',
'p1':8,
'y':'CBDFF3F7'})"
href="https://www.aiqicha.com/company_detail_82845142671484"
target="_blank"
>
山东省<em>城市商业银行</em>合作<em>联盟</em>有限公司 - 爱企查
</a>
</h3>
<font size=-1>
<div class="realtime">
2024-11-29
</div>
目前,<em>联盟</em>共有15家<em>城商行</em>股东,注册资本10.45亿元,线上运营银行资产突破3万亿元公司自成立以来,以提升成员行信息科技支撑水平和风险治理水平为重点,为67家成员行提供了持续稳定强大的信息系统支撑,引领成员行通过科技创新推动业务发展和转型升级公司接受监管部门严格的信息科技风险评估和监管,被原中国银监会列入首...
<br>
<font color="#008000">
<!--{cmt 有些url是dx生成的, 需要对二级目录进行飘粗, dx保证长度, 这里不需要截断}-->
www.aiqicha.com/company_detail_82845...
&nbsp;</font>
-&nbsp;<a data-nolog href="http://cache.baiducontent.com/c?m=3acHSY3QSgkEw1sXW0UKzpyLjYru1huJJy0xUzV0yis6Cm9IHy_sypubAlV_Y-HWxpoUNk7WMqL-9pivNrI_Ok6p3R4wgYIQ3k0K1CMVMm5tRNn2c10xpHN5pXMeJY12&amp;p=93759a44d7d51de51cb5c7710f5f&amp;newp=89759a44d79217dd0be2963e615c92695d0fc20e3cd6d601298ffe0cc4241a1a1a3aecb833647d45848773305df9180a98&amp;s=1543843a4723ed2a&user=baidu&fm=sc&query=%25B3%25C7%25C9%25CC%25D0%25D0%25C1%25AA%25C3%25CB&qid=ced952bc00004e47&p1=8"
target="_blank"
class="m">
百度快照
</a> <br>
</font>
</td>
</tr>
</table><br>
'''