Skip to content

Latest commit

 

History

History
104 lines (81 loc) · 4.18 KB

README.md

File metadata and controls

104 lines (81 loc) · 4.18 KB
Raw

Scrape resumes from indeed

Setup

Install requirements.txt for your Python environment e.g

pip install -r requirements.txt

or

conda install --file requirements.txt

A Mac version geckodriver is installed. If you use Mac and have Firefox you can use it. If not please, follow the driver installations for your platform and desired browser (you can use either Chrome or Firefox) as mentioned here.

You would need to put it to some directory that is in your PATH e.g /usr/local/bin. Alertnatively you can put the drivers into this directory and add this directory into your path:

PATH=$(pwd):$PATH

You can also modify the PATH variable in files like .bashrc and .bash_profile to persist the change (though you would need to statically set the directory path of where you put the scraper folder and not use $(pwd))

Running script

usage: indeed-resume-scraper.py [-h] -q query --name name [-l location]
                                [-si start] [-ei end] [--processes processes]
                                [--override] [--driver {firefox,chrome}]
                                [--login] [--simulate-user] [--headless]

Scrape Indeed Resumes

optional arguments:
  -h, --help            show this help message and exit
  -l location           location scope for search (default: canada)
  -si start             starting index (multiples of 50) (default: 0)
  -ei end               ending index (multiples of 50) (default: 1050)
  --processes processes
                        # of processes to run (max 4) (default: 1)
  --override            override existing result if any (default: False)
  --driver {firefox,chrome}
  --login               Simulate logging in as a user (read README further for
                        details) (default: False)
  --simulate-user       Whether to simulate user clicks or not (slower)
                        (default: False)
  --headless            Run browsers in headless mode (default: False)

required arguments:
  -q query              search query to run on indeed e.g software engineer
                        (default: None)
  --name name           name of search (used to save files, lowercased and
                        spaces turned to "-") (default: None)

Simulating logging in

It seems that Indeed blocks non-loggedin user to get resume results above a certain point (so far the ceiling seems to be 1050 resumes). However logging in seems to circumvent this.

The --login option allows you to do this. You would need to set the environemnt variables INDEED_RESUME_USER and INDEED_RESUME_PASSWORD. If not, the program will exit telling you that you need to set those environment variables.

So you do the following

export INDEED_RESUME_USER=<your indeed login user>
export INDEED_RESUME_PASSWORD=<your indeed user password>

You can do it for the current terminal session, or put it somewhere so it persists over sessions e.g your .bashrc or .bash_profile file.

Due to the restriction that Indeed imposes for non login scraping, -si and -ei options are automatically constrained to a minimum 0 and maximum of 1050 respectively when --login option is not specified

Simulating user behaviour

User behaviour can be simulated using the --simulate-user option. Using this, it seems that throttling is severely mitigated, and overall helps for a smoother scraping experience

Multiprocessing

By default the program runs with one process, but the --processes option can be given to indicate the number of processes to use. The maximum number of processes allowed is 4.

NOTE: If more than 1 process is to be used it is highly encouraged to turn on --simulate-user due to throttling mechanisms. Higher number of processes also may incur further throttling so be careful.

Example

Scrape 100 resumes (1st - 100th resume) for software engineering in Canada

python indeed-scraper.py -q 'software engineer'  --name software-canada -ei 100

Multiple queries

The script.sh can be run with a file that has a job title per line

./script.sh <filename>

Please read script.sh for some more details (you may modify as needed)