linkedin-cat 0.1.0

linkedin automation tool

Introduction

Install package

pip install linkedin_cat

Preparation

• create file linkedin_cookies.json
• please use Chrome Extension EditThisCookie to export linkedin cookies to linkedin_cookies.json

linkedin_cookies.json

[
{
    "domain": ".linkedin.com",
    "expirationDate": 1746423475,
    "hostOnly": false,
    "httpOnly": false,
    "name": "_gcl_au",
    "path": "/",
    "sameSite": "unspecified",
    "secure": false,
    "session": false,
    "storeId": "0",
    "value": "1.1.191481430.1738647475",
    "id": 1
},
{
    "domain": ".linkedin.com",
    "expirationDate": 1746089950.085703,
    ......

Example Usage:

- Linkedin Send Message

from linkedin_cat.message import LinkedinMessage

# linkedin cookie file path
linkedin_cookies_json='./linkedin_cookies.json'
# headless mode
headless = False
# Ininialized LinkedinMessage
bot = LinkedinMessage(linkedin_cookies_json,headless)

# message to sent, use FIRSTNAME or FULLNAME to customized the message
message = "Hello FIRSTNAME,hope you are doing well. Glad to e-meet with you on linkedin"

# Send single request by url
url = "https://www.linkedin.com/in/chandlersong/"
bot.send_single_request(url,message)

# Send multi request by linkedin url list 
urls_list = [    
    	"https://www.linkedin.com/in/chandlersong/",
    	"https://www.linkedin.com/in/chandlersong/",
        ]

bot.send_multi_request(urls_list,message)
    

- Python Command

cat_run.py

import argparse
from linkedin_cat.message import LinkedinMessage
import json
import time


class LocalStorageHelper:
    def __init__(self, storage_file_path):
        self.storage_file_path = storage_file_path
        self.storage = self.load_storage()

    def load_storage(self):
        try:
            with open(self.storage_file_path, "r", encoding="utf8") as f:
                return json.load(f)
        except FileNotFoundError:
            return {}

    def save_storage(self):
        with open(self.storage_file_path, "w", encoding="utf8") as f:
            json.dump(self.storage, f)

    def set(self, key, value):
        self.storage[key] = value
        self.save_storage()

    def get(self, key):
        return self.storage.get(key)


def get_message(message_file_path):
    with open(message_file_path, "r", encoding="utf8") as f:
        message = f.read()
    return message


def read_urls_list(urls_file_path):
    with open(urls_file_path, "r", encoding="utf8") as f:
        urls_list = f.readlines()
    urls_list = [url.strip() for url in urls_list]
    return urls_list


def send_messages(urls_list, message, storage_helper, bot):
    for url in urls_list:
        if storage_helper.get(url):
            continue
        bot.send_single_request(url, message)
        storage_helper.set(url, time.time())


def main():
    parser = argparse.ArgumentParser(description='Send LinkedIn messages to a list of URLs.')
    parser.add_argument('cookies', type=str, help='Path to the LinkedIn cookies JSON file (e.g., "cookies.json").')
    parser.add_argument('message', type=str, help='Path to the message file (e.g., "message.txt").')
    parser.add_argument('urls', type=str, help='Path to the URLs file (e.g., "urls.txt").')
    parser.add_argument('button_class', type=str, help="""Message Button Class: ieSHXhFfVTxQfadOJdXYOIDuVKsBXgPtjNxI (eg:<button aria-label="Invite XXXX to connect" id="ember840"
                                                            class="artdeco-button artdeco-button--2
                                                            artdeco-button--primary ember-view ieSHXhFfVTxQfadOJdXYOIDuVKsBXgPtjNxI"
                                                            type="button">)"""
                        )
    parser.add_argument('--headless', action='store_true',
                        help='Run the bot in headless mode (without opening a browser window).')
    parser.add_argument('--storage', type=str, default='processed_urls.json',
                        help='Path to the local storage file for processed URLs (e.g., "processed_urls.json"). This file keeps track of URLs that have already been processed to avoid sending duplicate messages.')

    args = parser.parse_args()

    message = get_message(args.message)
    urls_list = read_urls_list(args.urls)

    bot = LinkedinMessage(args.cookies, args.headless,button_class=args.button_class)

    storage_helper = LocalStorageHelper(args.storage)
    send_messages(urls_list, message, storage_helper, bot)




if __name__ == "__main__":
    main()

运行脚本：

python .\cat_run.py linkedin_cookies.json msg.txt urls.txt  gUfuZBuHecrFJGhcTDwkUSEhvpaAQjXWfwDwQ 

linkedin_cookies.json (EditThisCookies导出)

msg.txt

Hello FIRSTNAME, Hope you are doing well. Glad to e-meet with you on linkedin.

urls.txt

https://www.linkedin.com/in/xxxx/
https://www.linkedin.com/in/yyyyy/
https://www.linkedin.com/in/zzzzz/
......

Message/Connect Button Class (gUfuZBuHecrFJGhcTDwkUSEhvpaAQjXWfwDwQ)

<button aria-label="Invite xxx to connect" id="ember485" class="artdeco-button artdeco-button--2 artdeco-button--primary ember-view gUfuZBuHecrFJGhcTDwkUSEhvpaAQjXWfwDwQ" type="button">        <svg role="none" aria-hidden="true" class="artdeco-button__icon " xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 16 16" data-supported-dps="16x16" data-test-icon="connect-small">
<!---->    
    <use href="#connect-small" width="16" height="16"></use>
</svg>

<span class="artdeco-button__text">
    Connect
</span></button>

无Web界面化运行

python .\cat_run.py linkedin_cookies.json msg.txt urls.txt  gUfuZBuHecrFJGhcTDwkUSEhvpaAQjXWfwDwQ --headless

- Linkedin Search

linkedin_search_keyword.py 关键词搜索

from linkedin_cat.search import LinkedinSearch

# li_class
# <li class="AnzbBoSgWFPvyPjbkRjRKkrJVxiaC"
# title_div_class
# <div class="RTuDraFdKmgRAxtLKirgHvzAIXdijkkEQ t-14 t-black t-normal"
# location_div_class
# <div class="rnkdymaxWKCNeIKKaOIbrXkZKdCeIOFLMZ"
# intro_p_class
# <p class=KtEWEpbEqyHyYWJByRbdgYQzrISPzhXvdIM
# link_span_class
# <span class=OORPFqatrHRtkxtjkKqROvaiHZJKsVvncNCM

# 实例化bot
bot = LinkedinSearch(
         linkedin_cookies_json='linkedin_cookies.json',
         headless=False,
         li_class='AnzbBoSgWFPvyPjbkRjRKkrJVxiaC',
         title_div_class='RTuDraFdKmgRAxtLKirgHvzAIXdijkkEQ',
         location_div_class='rnkdymaxWKCNeIKKaOIbrXkZKdCeIOFLMZ',
         intro_p_class='KtEWEpbEqyHyYWJByRbdgYQzrISPzhXvdIM',
         link_span_class='OORPFqatrHRtkxtjkKqROvaiHZJKsVvncNCM'
         )

# 关键词搜索
results = bot.search_keywords('wang microsoft',wait=False)

# 返回 list 结果
for result in results:
    print(result)
    # {'name': '', 'title': '', 'location': '', 'introduction': '', 'linkedin_url': '', 'image_url': ''}

linkedin_search_filters.py 多条件搜索

from linkedin_cat.search import LinkedinSearch

# 实例化bot
bot = LinkedinSearch(
         linkedin_cookies_json='linkedin_cookies.json',
         headless=False,
         li_class='AnzbBoSgWFPvyPjbkRjRKkrJVxiaC',
         title_div_class='RTuDraFdKmgRAxtLKirgHvzAIXdijkkEQ',
         location_div_class='rnkdymaxWKCNeIKKaOIbrXkZKdCeIOFLMZ',
         intro_p_class='KtEWEpbEqyHyYWJByRbdgYQzrISPzhXvdIM',
         link_span_class='OORPFqatrHRtkxtjkKqROvaiHZJKsVvncNCM'
         )

# 按照搜索条件搜索, 自定义搜索条件
# 生成搜索的url
url = bot.generate_linkedin_search_url(keywords, company=None, title=None,school=None,first_name=None, last_name=None)
# 获取搜索页面的html
html = bot.open_linkedin_url(url)
# 解析html页面
results = bot.parse_linkedin_results(html)

for result in results:
    print(result)

save_profile_by_linkedin_url.py 保存Linkedin Profile

from linkedin_cat.search import LinkedinSearch
# 实例化bot
bot = LinkedinSearch(
         linkedin_cookies_json='linkedin_cookies.json',
         headless=False,
         )
# 搜索linkedin profile，并保存为json文件
save_folder = './linkedin'
url = "https://www.linkedin.com/in/chandlersong/"
bot.search_linkedin_profile(url,save_folder)

save_profile_by_linkedin_url_list.py

from linkedin_cat.search import LinkedinSearch
# 实例化bot
bot = LinkedinSearch(
         linkedin_cookies_json='linkedin_cookies.json',
         headless=False,
         )

# 批量搜索linkedin profile，并保存为json文件
save_folder = './linkedin'
url_list = [
    "https://www.linkedin.com/in/chandlersong/"
    "https://www.linkedin.com/in/chandlersong/"
    "https://www.linkedin.com/in/chandlersong/"
    ]

bot.search_linkedin_profile_list(url_list,save_folder)

- Recruiter LLM

import os
from jinja2 import Template
from cn_detect.detect import ChineseNameDetect
from linkedin_cat.search import LinkedinSearch # 继承了LinkedinMessage
from llm_cat.deepseek_api import deepseek_chat

# 添加自定义模块路径
# import sys
# sys.path.append(r'C:\Users\xxx\pyfunc')
# from llm import llm

# LinkedIn cookie文件路径
linkedin_cookies_json = "cookies.json"
headless = True
token = "sk-xxxxxxxxxxxxxxxxxxxxxx"

# 创建LinkedinSearch对象
search_bot = LinkedinSearch(linkedin_cookies_json,headless=headless)
detector = ChineseNameDetect()

def judge_skill_prompt(mini_profile):
    # 构建评估候选人的模板
    prompt_template = """
                            Please assess if the candidate in #### is a security engineer. 
                            If yes, return True; if not, return False. 
                            Provide the response in JSON format. 
                            example:
                            {
                                "result": True
                            }
                            \n\n 
                            ####{{ short_profile }}####
                        """
    template = Template(prompt_template)
    prompt = template.render(short_profile=mini_profile)

    # 使用LLM评估候选人
    # reply = llm(prompt)
    reply = deepseek_chat(prompt, token)
    print(reply)
    return reply


def search_recruiter(keywords, company=None, title=None, school=None, first_name=None, last_name=None,
                     chinese_detect=True, send_message=True, save_resume=True,
                     message=f"Hello FIRSTNAME, hope you are doing well. Glad to e-meet with you on LinkedIn."):
    page = 1
    while True:
        # 生成LinkedIn搜索URL
        url = search_bot.generate_linkedin_search_url(keywords, company, title, school, first_name, last_name)
        if page != 1:
            url += f'&page={page}'

        # 打开LinkedIn URL并解析结果
        html = search_bot.open_linkedin_url(url)
        results = search_bot.parse_linkedin_results(html)

        # 如果没有结果，退出循环
        if not results:
            print('No results')
            break

        print('Total Results:', len(results))
        for result in results:
            print(result)
            try:
                name = result['name']
                url = result['linkedin_url'].split('?')[0]

                mini_profile = {key: result[key] for key in ['title', 'introduction', 'location']}

                # 判断linkedin是否存在
                filename = search_bot.extract_username_from_linkedin_url(url) + '.json'
                if os.path.exists(os.path.join('./linkedin', filename)):
                    print(f'{filename} already exists, skipping...')
                    continue

                # 判断是否需要检测中文
                should_continue = not chinese_detect or detector.detect_chinese_word(name) or detector.detect_family_name(name)

                if should_continue:
                    reply = judge_skill_prompt(mini_profile)
                    if 'True' in reply:
                        # 发送消息
                        if send_message:
                            search_bot.send_single_request(url, message)
                        # 保存简历
                        if save_resume:
                            search_bot.search_linkedin_profile(url, thread_pool=True)

            except Exception as e:
                print(f"Error: {e}")

        # 翻页
        page += 1

# 使用示例
search_recruiter('security', company='google',last_name='jiang',send_message=False,save_resume=True)