新的BOSS直聘脚本 新增AI分析岗位匹配度
"""
@Author: WY
@Date: 2025-02-08
@Description: NOW ENJOY IT
"""
import json
import re
from random import random
import loguru
from DrissionPage import Chromium, ChromiumOptions
from conf.conf import url_zp
# loguru配置
loguru.logger.add("log/boss.log", rotation="1 MB", encoding="utf-8", backtrace=True, diagnose=True)
class ZhiPuAi:
def __init__(self):
self.co1 = ChromiumOptions().set_paths(local_port=9111, user_data_path=r'D:\code_py\boss_auto\data')
self.url = url_zp
# 随机1-3秒
self.random_int = random() * 3 + 1
self.xpath_chat = "xpath://textarea[@class='scroll-display-none']"
self.xpath_send = "xpath://*[@id='search-input-box']/div[2]/div[3]"
self.xpath_data = 'xpath://div[@class="answer-content flex1"]'
self.judge_data_end_xpath = 'xpath://div[@class="panel"]'
def get_tab(self):
tab_new = Chromium(addr_or_opts=self.co1).get_tab()
loguru.logger.info('打开ZhipuAi')
tab_new.get(self.url)
tab_new.wait(self.random_int)
return tab_new
def get_search(self, tab_zp, job_data):
tab_zp.ele(self.xpath_chat).input(job_data)
tab_zp.ele(self.xpath_send).click()
# tab_zp.wait(120)
for i in range(20):
text_data1 = tab_zp.ele(self.xpath_data).text
tab_zp.wait(8)
text_data2 = tab_zp.ele(self.xpath_data).text
if text_data1 == text_data2:
return text_data1
else:
continue
class Boss:
def __init__(self):
self.co1 = ChromiumOptions().set_paths(local_port=9111, user_data_path=r'D:\code_py\boss_auto\data')
self.url = 'https://www.zhipin.com/'
self.sjkf = 'https://www.zhipin.com/web/geek/job?query=%E9%94%80%E5%94%AE&city=101010100&position=100508'
# 随机1-3秒
self.random_int = random() * 3 + 1
self.xpath_tag_list = "xpath://ul[@class='job-keyword-list']/li"
self.job_detail_xpath = "xpath://div[@class='job-sec-text']"
self.job_title_xpath = "xpath://div[@class='name']/h1"
self.job_bonus_xpath = "xpath://div[@class='job-tags']/span"
self.job_salary_xpath = "xpath://div[@class='name']/span[@class='salary']"
self.job_city_xpath = "xpath://div[@class='info-primary']/p/a"
self.job_year_xpath = "xpath://div[@class='info-primary']/p/span[1]"
self.job_degree_xpath = "xpath://div[@class='info-primary']/p/span[2]"
self.start_chat_xpath = "xpath://div[@class='btn-container']/div/a"
self.chat_close_xpath = "xpath://div[@class='dialog-title']/a/i"
self.chat_input_xpath = "xpath://div[@class='edit-area']/textarea"
self.chat_send_xpath = "xpath://div[@class='edit-area']/div"
self.next_page_xpath = "xpath://a/i[@class='ui-icon-arrow-right']"
def get_tab(self):
tab_new = Chromium(addr_or_opts=self.co1).get_tab()
loguru.logger.info('打开boss直聘')
tab_new.get(self.url)
tab_new.wait(self.random_int)
return tab_new
def get_latest_tab(self):
tab_new = Chromium(addr_or_opts=self.co1).latest_tab
loguru.logger.info('切换最新窗口')
tab_new.wait(self.random_int)
return tab_new
def judge_similar(self, tags_in, job_list_in, job_title_in, job_bonus_in, job_salary_in, job_city_in, job_year_in,
job_degree_in):
json_data = {
"tags": tags_in,
"job_list": job_list_in,
"job_title": job_title_in,
"job_bonus": job_bonus_in,
"job_salary": job_salary_in,
"job_city": job_city_in,
"job_year": job_year_in,
"job_degree": job_degree_in
}
print(json_data)
while True:
zhipu = ZhiPuAi()
tab_zp = zhipu.get_tab()
data = zhipu.get_search(tab_zp, json_data)
loguru.logger.info(data)
# 使用正则提取百分比
rate_list = re.findall(r'(\d+\.?\d*)%', data)
if rate_list and len(rate_list) == 1:
rate = rate_list[0]
loguru.logger.info('相似度：{}'.format(rate))
return int(rate)
elif rate_list and len(rate_list) > 1:
# todo 有待商榷，先取最大值
rate = max(rate_list)
loguru.logger.info('相似度：{}'.format(rate))
return int(rate)
else:
# 让zhipuai重新搜索
loguru.logger.warning('未找到相似度，重新搜索')
continue
# return rate
def get_search(self, tab_search, keyword):
loguru.logger.info('搜索关键词：{}'.format(keyword))
tab_search.ele('@class=ipt-search').input(keyword) # 输入搜索内容
# exit(0)
tab_search.wait(self.random_int)
tab_search.ele('@ka=search_box_index').click() # 点击搜索按钮
tab_search.wait(self.random_int)
loguru.logger.info('获取搜索结果')
tab_search.get(tab_search.url + '&experience=104&industry=100020&position=100508,100507,100511,100506,100104,100512,100514,100122')
for i in range(1, 10):
job_list = tab1.eles('@class=job-card-wrapper')
if len(job_list) == 0:
loguru.logger.info('未找到相关职位，网站更新了？')
exit(-1)
for job in job_list:
job.click()
tab_latest = self.get_latest_tab()
tags_in = self.get_tags(tab_latest)
job_list_in = self.get_job_detail(tab_latest)
job_title_in = self.get_job_title(tab_latest)
job_bonus_in = self.get_job_bonus(tab_latest)
job_salary_in = self.get_job_salary(tab_latest)
job_city_in = self.get_job_city(tab_latest)
job_year_in = self.get_job_year(tab_latest)
job_degree_in = self.get_job_degree(tab_latest)
# 快照！
job_url = tab_latest.url
rate_similar = self.judge_similar(tags_in, job_list_in, job_title_in, job_bonus_in, job_salary_in,
job_city_in,
job_year_in, job_degree_in)
if rate_similar >= 80:
loguru.logger.info('相似度大于80%，打招呼')
tab_latest.get(job_url)
self.get_start_chat(tab_latest)
self.chat_input(tab_latest,rate_similar)
self.chat_send(tab_latest)
self.chat_close(tab_latest)
tab_latest.close()
loguru.logger.info('第{}页'.format(i))
tab_search.ele(self.next_page_xpath).click()
loguru.logger.info('搜索完成10页')
def get_tags(self, tab_latest):
tags = tab_latest.eles(self.xpath_tag_list)
tag_list = [tag.text for tag in tags]
if tag_list:
loguru.logger.info('职位标签：{}'.format(tag_list))
return tag_list
else:
loguru.logger.warning('该职位没有标签，HR真懒啊')
return None
def get_job_detail(self, tab_latest):
job_detail = tab_latest.ele(self.job_detail_xpath).text
if job_detail:
loguru.logger.info('职位详情：{}'.format(job_detail))
return job_detail
else:
loguru.logger.warning('该职位没有详情???那还投个毛')
return None
def get_job_title(self, tab_latest):
job_title = tab_latest.ele(self.job_title_xpath).text
if job_title:
loguru.logger.info('职位名称：{}'.format(job_title))
return job_title
else:
loguru.logger.warning('该职位没有名称???那应该不可能吧')
return None
def get_job_bonus(self, tab_latest):
job_bonuss = tab_latest.eles(self.job_bonus_xpath)
job_bonus = [job_bonus1.text for job_bonus1 in job_bonuss]
if job_bonus:
loguru.logger.info('职位福利：{}'.format(job_bonus))
return job_bonus
else:
loguru.logger.warning('该职位没有福利???不太行啊')
return None
def get_job_salary(self, tab_latest):
job_salary = tab_latest.ele(self.job_salary_xpath).text
if job_salary:
loguru.logger.info('职位薪资：{}'.format(job_salary))
return job_salary
else:
loguru.logger.warning('该职位没有薪资???')
return None
def get_job_city(self, tab_latest):
job_city = tab_latest.ele(self.job_city_xpath).text
if job_city:
loguru.logger.info('职位城市：{}'.format(job_city))
return job_city
else:
loguru.logger.warning('该职位没有城市???')
return None
def get_job_year(self, tab_latest):
job_year = tab_latest.ele(self.job_year_xpath).text
if job_year:
loguru.logger.info('职位年限：{}'.format(job_year))
return job_year
else:
loguru.logger.warning('该职位没有年限???')
return None
def get_job_degree(self, tab_latest):
job_degree = tab_latest.ele(self.job_degree_xpath).text
if job_degree:
loguru.logger.info('职位学历：{}'.format(job_degree))
return job_degree
else:
loguru.logger.warning('该职位没有学历要求???')
return None
def get_start_chat(self, tab_latest):
start_chat = tab_latest.ele(self.start_chat_xpath).wait(self.random_int)
if start_chat:
loguru.logger.info('开始聊天')
start_chat.click()
else:
loguru.logger.warning('该职位没有开始聊天按钮???')
def chat_close(self, tab_latest):
chat_close = tab_latest.ele(self.chat_close_xpath).wait(self.random_int)
if chat_close:
loguru.logger.info('关闭聊天')
chat_close.click()
else:
loguru.logger.warning('该职位没有关闭聊天按钮???')
def chat_input(self, tab_latest,rate):
chat_input = tab_latest.ele(self.chat_input_xpath).wait(self.random_int)
if chat_input:
loguru.logger.info('输入聊天内容')
chat_input.input(f'你好，我对岗位很感兴趣，我想问可以发一份简历吗，谢谢（如果不满足可以跟我说一下不满足的方面吗，万分感谢）')
else:
loguru.logger.warning('该职位没有聊天输入框???')
def chat_send(self, tab_latest):
chat_send = tab_latest.ele(self.chat_send_xpath)
if chat_send:
loguru.logger.info('发送聊天内容')
chat_send.click()
else:
loguru.logger.warning('该职位没有发送聊天按钮???')
if __name__ == '__main__':
boss = Boss()
tab1 = boss.get_tab()
# exit(0)
boss.get_search(tab1, '数据开发')
# zhipu = ZhiPuAi()
# tab_zp = zhipu.get_tab()
# data=zhipu.get_search(tab_zp,"{'tags': ['Nginx', 'C++', 'Pandas', 'MySQL', 'Numpy', 'Python'], 'job_list': '* IT相关专业\n* 熟练掌握python编程技术，会使用numpy，pandas，matplotlib第三方库\n* 熟练掌握C++编程优先\n* 熟练掌握Qt开发优先\n* 有股票、期货、等金融基本知识,开发A股、期货专用策略具备量化交易经\n* 热爱软件开发\n职责：\n* 参与软件界面需求对接\n* 设计并实现软件界面功能\n* 编写软件需求文档\n* 参与软件测试', 'job_title': 'Python C++ 开发', 'job_bonus': ['股票期权', '五险一金', '股票期权', '五险一金'], 'job_salary': '7-12K', 'job_city': '北京', 'job_year': '3-5年', 'job_degree': '学历不限'}")
# print(data)