#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Author: Ken git : https://github.com/PlayingNumbers/ds_salary_proj Tutorial(youtube) : https://www.youtube.com/watch?v=GmW4F6MHqqs&list=PL2zq7klxX5ASFejJj80ob9ZAnBHdz5O1t Inspired by (Tutorial) : https://towardsdatascience.com/selenium-tutorial-scraping-glassdoor-com-in-10-minutes-3d0915c6d905 Disclaimer: I don't own the copyrights of the code , It was written and coded as followed on the youtube channel as mentioned above.please take a note that I have changed the names of the column and files according to my need , if you are copy pasting this code you have to look for syntax errors in names of files and data-columns that are used in tutorial. """ import glassdoor_scrapper as gs import pandas as pd # path = "FirefoxDriver/geckodriver" path = "ChromeDriver/chromedriver" # ChromeDriver is Folder, inside the fodler is file "chromedriver" df = gs.get_jobs('data scientist', 100, False, path, 10) df.to_csv('glsDr_jobs.csv', index=False)
# -*- coding: utf-8 -*- """ Created on Sun Nov 29 23:24:42 2020 @author: kevin """ import glassdoor_scrapper as gs import pandas as pd path = "K:/Personal/Data Analyst salary prediction/chromedriver" df = gs.get_jobs("data analyst", 950, False, path, 15)
# -*- coding: utf-8 -*- """ Created on Fri Feb 5 10:17:40 2021 @author: Katpadi """ import glassdoor_scrapper as gs import pandas as pd path = "C:/Users/Katpadi/Documents/DataScience/Ken Jee/sal_proj/chromedriver" slp_time = 15 df = gs.get_jobs("data scientist", 40, False, path, slp_time)
# -*- coding: utf-8 -*- """ Created on Tue Oct 13 01:02:18 2020 @author: madpa """ import glassdoor_scrapper as gs import pandas as pd ds_job_data_df = gs.get_jobs('data scientist', 10, False)
# -*- coding: utf-8 -*- """ Created on Sat Jan 23 22:25:10 2021 @author: PRANIT """ from glassdoor_scrapper import get_jobs path = 'C:/Users/PRANIT/Desktop/Coding/selenium/web_drivers/chromedriver.exe' df = get_jobs('data Scientist', 5, False, path, 10)
# -*- coding: utf-8 -*- """ Created on Tue Jun 16 10:11:51 2020 @author: kapil """ import glassdoor_scrapper as gs import pandas as py path = "C:\\Users\\kapil\\Downloads\\chromedriver_win32\\chromedriver.exe" df = gs.get_jobs(5, False, path, 2)
#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Created on Fri May 8 15:44:57 2020 @author: lilpkstud """ import glassdoor_scrapper as gs import pandas as pd path = "/Users/lilpkstud/Desktop/github_repo/glassdoor_datascience/chromedriver" df = gs.get_jobs('project manager', 100, False, path, 15) df.to_csv('glassdoor_jobs.csv', index=False)
# -*- coding: utf-8 -*- """ Created on Wed Jan 13 13:32:55 2021 @author: shobhit deshwal """ import glassdoor_scrapper as gs import pandas as pd path = "C:/Users/shobhit deshwal/Desktop/Ken_Jee_salary_pred/chromedriver_win32/chromedriver" df = gs.get_jobs('data scientist', 75, False, path, 15) df.to_csv('glassdoor_jobs.csv', index=False)
# -*- coding: utf-8 -*- """ Created on Mon Jun 22 17:28:23 2020 @author: suyogya """ import glassdoor_scrapper as sc import pandas as pd path = "C:/Users/suy/Documents/Github Backup/Glassdoor_Salary_Project/chromedriver" frame = int(input("Enter the number of iterations : \n")) dataframe = sc.get_jobs("data scientist", frame, False, path, 2) dataframe.to_csv("glassdoor_data.csv", index=False)
#$ cd Documents #$ mkdir ds_salary_proj #$ cd ds_salary_proj #$ git init #then create a new reposatory in your account from browser, now after doing all the work link your folder with this repo as below. #(make sure the project folder name and repo name is same) import glassdoor_scrapper as gs #import pandas as pd path ="E:/OneDrive/DS Journey/DS_Projects/ds_salary_proj/chromedriver" sleep_time = 5 location = 'us' df = gs.get_jobs("Data-Scientist",1000,False,path,location,sleep_time) df.to_csv("glassdoor_jobs2.csv") #$ cd ds_salary_proj #echo "# ds_salary_proj" >> README.md #git add . #git commit -m "uploaded scrapper and run code" #git remote add origin https://github.com/rmsharma5989/ds_salary_proj.git #git push -u origin master ####################### extra info ########################## #git push -u origin master
# -*- coding: utf-8 -*- """ Created on Sat Jun 6 22:01:06 2020 @author: hafizuddin """ import glassdoor_scrapper as gs import pandas as pd path = "C:/Users/hafizuddin/Documents/Data-Science-Project1/chromedriver" df = gs.get_jobs(False, 150, False, path, 15) df.to_csv( r'C:/Users/hafizuddin/Documents/Data-Science-Project1/salary_150.csv', index=False)
# -*- coding: utf-8 -*- """ Created on Fri Dec 11 12:37:03 2020 @author: Meghna """ import glassdoor_scrapper as gs import pandas as pd path="C:/Users/Meghna/OneDrive/Documents/ML AI Projects/ds_salary_proj/chromedriver.exe" df=gs.get_jobs("data scientist",15,False,path,15)