turtlebasket
/
JobSpy
mirror of https://github.com/Bunsly/JobSpy
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

created defualt handler and another for indeed. 

on bot i can do findAll and find one by one with the command of the site name
pull/231/head
Yariv Menachem 2024-12-26 17:53:31 +02:00
parent 34e85b9261
commit 3fcc87e9b7
4 changed files with 46 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

67
src/main.py
View File

@ -3,11 +3,9 @@ import os
from telegram import Update
from telegram.ext import Application, CommandHandler
from src.jobspy import Site, scrape_jobs
from src.jobspy.db.job_repository import JobRepository
from src.jobspy import Site
from src.jobspy.scrapers.utils import create_logger
from src.telegram_bot import TelegramBot
from src.telegram_handler import TelegramAllHandler,TelegramGoozaliHandler
from src.telegram_handler import TelegramIndeedHandler, TelegramDefaultHandler
logger = create_logger("Main")
title_filters: list[str] = ["test", "qa", "Lead", "Full-Stack", "Full Stack", "Fullstack", "Frontend", "Front-end",
@ -15,46 +13,39 @@ title_filters: list[str] = ["test", "qa", "Lead", "Full-Stack", "Full Stack", "F
"automation", "BI", "Principal", "Architect", "Android", "Machine Learning", "Student",
"Data Engineer", "DevSecOps"]
async def main():
telegramBot = TelegramBot()
jobRepository = JobRepository()
# sites_to_scrap = [Site.LINKEDIN, Site.GLASSDOOR, Site.INDEED, Site.GOOZALI]
sites_to_scrap = [Site.GOOZALI]
# sites_to_scrap = [Site.GOOZALI]
jobs = scrape_jobs(
site_name=sites_to_scrap,
search_term="software engineer",
locations=["Tel Aviv, Israel", "Ramat Gan, Israel",
"Central, Israel", "Rehovot ,Israel"],
results_wanted=200,
hours_old=48,
country_indeed='israel'
)
logger.info(f"Found {len(jobs)} jobs")
newJobs = jobRepository.insertManyIfNotFound(jobs)
for newJob in newJobs:
await telegramBot.sendJob(newJob)
# Run the async main function
if __name__ == "__main__":
logger.info("Starting initialize ")
_api_token = os.getenv("TELEGRAM_API_TOKEN")
search_term = "software engineer"
locations = ["Tel Aviv, Israel", "Ramat Gan, Israel", "Central, Israel", "Rehovot ,Israel"]
application = Application.builder().token(_api_token).build()
tg_handler_all = TelegramAllHandler(sites=[Site.LINKEDIN, Site.GLASSDOOR, Site.INDEED, Site.GOOZALI],
locations=locations,
title_filters=title_filters,
search_term=search_term)
tg_handler_all = TelegramDefaultHandler(sites=[Site.LINKEDIN, Site.GLASSDOOR, Site.INDEED, Site.GOOZALI],
locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler("findAll", tg_handler_all.handle))
tg_handler_goozali = TelegramGoozaliHandler(locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler("goozali", tg_handler_goozali.handle))
# application.add_handler(CommandHandler("galssdoor", find_glassdoor))
# application.add_handler(CommandHandler("linkedin", find_linkedin))
# application.add_handler(CommandHandler("indeed", find_indeed))
# Goozali
tg_handler_goozali = TelegramDefaultHandler(sites=[Site.GOOZALI],
locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler(Site.GOOZALI.value, tg_handler_goozali.handle))
# GlassDoor
tg_handler_glassdoor = TelegramDefaultHandler(sites=[Site.GLASSDOOR],
locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler(Site.GLASSDOOR.value, tg_handler_glassdoor.handle))
# LinkeDin
tg_handler_linkedin = TelegramDefaultHandler(sites=[Site.LINKEDIN],
locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler(Site.LINKEDIN.value, tg_handler_linkedin.handle))
# Indeed
tg_handler_indeed = TelegramIndeedHandler(locations=locations,
title_filters=title_filters,
search_term=search_term)
application.add_handler(CommandHandler(Site.INDEED.value, tg_handler_indeed.handle))
logger.info("Run polling from telegram")
application.run_polling(allowed_updates=Update.ALL_TYPES)

4
src/telegram_handler/__init__.py
View File

@ -1,2 +1,2 @@
from .telegram_all_handler import TelegramAllHandler
from .telegram_goozali_handler import TelegramGoozaliHandler
from .telegram_default_handler import TelegramDefaultHandler
from .telegram_indeed_handler import TelegramIndeedHandler

17
src/telegram_handler/telegram_all_handler.py → src/telegram_handler/telegram_default_handler.py
View File

@ -9,31 +9,32 @@ from src.jobspy.scrapers.utils import create_logger
from src.telegram_bot import TelegramBot
from src.telegram_handler.telegram_handler import TelegramHandler
logger = create_logger("TelegramAllHandler")
class TelegramAllHandler(TelegramHandler):
def __init__(self, sites: list[Site], locations: list[str], title_filters: list[str],search_term:str):
class TelegramDefaultHandler(TelegramHandler):
def __init__(self, sites: list[Site], locations: list[str], title_filters: list[str], search_term: str):
self.sites_to_scrap = sites
self.locations = locations
self.search_term = search_term
self.title_filters = title_filters
self.telegramBot = TelegramBot()
self.jobRepository = JobRepository()
if len(sites) == 1:
self.logger = create_logger(f"Telegram{sites[0].name.title()}Handler")
else:
self.logger = create_logger("TelegramAllHandler")
async def handle(self, update: Update, context: ContextTypes.DEFAULT_TYPE):
logger.info("start handling")
self.logger.info("start handling")
jobs = scrape_jobs(
site_name=self.sites_to_scrap,
search_term=self.search_term,
locations=self.locations,
results_wanted=200,
hours_old=48,
country_indeed='israel',
filter_by_title=self.title_filters
)
logger.info(f"Found {len(jobs)} jobs")
self.logger.info(f"Found {len(jobs)} jobs")
new_jobs = self.jobRepository.insertManyIfNotFound(jobs)
for newJob in new_jobs:
await self.telegramBot.sendJob(newJob)
logger.info("finished handling")
self.logger.info("finished handling")

13
src/telegram_handler/telegram_goozali_handler.py → src/telegram_handler/telegram_indeed_handler.py
View File

@ -9,20 +9,19 @@ from src.jobspy.scrapers.utils import create_logger
from src.telegram_bot import TelegramBot
from src.telegram_handler.telegram_handler import TelegramHandler
logger = create_logger("TelegramAllHandler")
class TelegramGoozaliHandler(TelegramHandler):
class TelegramIndeedHandler(TelegramHandler):
def __init__(self, locations: list[str], title_filters: list[str], search_term: str):
self.sites_to_scrap = [Site.GOOZALI]
self.sites_to_scrap = [Site.INDEED]
self.locations = locations
self.search_term = search_term
self.title_filters = title_filters
self.telegramBot = TelegramBot()
self.jobRepository = JobRepository()
self.logger = create_logger(f"Telegram{self.sites_to_scrap[0].name.title()}Handler")
async def handle(self, update: Update, context: ContextTypes.DEFAULT_TYPE):
logger.info("start handling")
self.logger.info("start handling")
jobs = scrape_jobs(
site_name=self.sites_to_scrap,
search_term=self.search_term,
@ -32,8 +31,8 @@ class TelegramGoozaliHandler(TelegramHandler):
country_indeed='israel',
filter_by_title=self.title_filters
)
logger.info(f"Found {len(jobs)} jobs")
self.logger.info(f"Found {len(jobs)} jobs")
new_jobs = self.jobRepository.insertManyIfNotFound(jobs)
for newJob in new_jobs:
await self.telegramBot.sendJob(newJob)
logger.info("finished handling")
self.logger.info("finished handling")