Scraper for prnt.sc

Introduction

The website LightShot or prnt.sc is a public image sharing website which is most well known for its quick and easy downloadable sharing utility activated by pressing the PrtScn key. It's a very useful tool, however I noticed that it stores images based on a sequential 6 digit code, meaning the 1.3 billion or so images uploaded there can be indexed programmatically quite easily. That is what this utility does.

Pre-downloaded Dataset

I have already published a set of around 13,000 images downloaded from the site on Kaggle which may be of use if you do not want to use the scraper utility.

Pre-requisites

This script was tested on the following python modules, however earlier/later versions may work fine:

- python 3.6
- requests 2.8.1
- beautifulsoup4 4.6.0
- lxml 3.8.0
- faker 8.11.0

Using the Script

The script takes 4 arguments as follows:

• --start_code: 6 or 7 character string made up of lowercase letters and numbers which is where the scraper will start.
  • e.g. 'lj9me9'
• --count: The number of images to scrape.
  • e.g. '200'
• --output_path: The path where images will be stored.
  • e.g. 'output_001/
• --resume_from_last :If files allready exist in the output get last created/modified and resume from there, default=True
  • e.g. True
• --max_files_per_destination :The maximum number of files to place in a directory prior to creating a new output directory, default='32766'
  • e.g. '10000'

parser.add_argument( '--max_files_per_destination', help='The maximum number of files to place in a directory prior to creating a new output directory', default='32766')

Uses/Explanation

It can be very interesting to see what people upload to these sites, generally having sequential IDs of any type is bad, and the same applies here. People might not be aware that what they are uploading is visible to others, however prnt.sc/lightshot have not shown any inclination in wanting to change their site design.

As a result, this provides a useful way to create datasets of real world images. A useful/interesting use case is building a machine learning algorithm to classify images into categories, which requires some manual classification, but is nonetheless interesting and a good learning task.

Sequential is a poor practice

 --start_code values on prnt.cs is sequential, so from image context...
 14akf6 ~ Oct 2013
 999997 ~ Jan 2015
 a9998j ~ Feb 2016
 h4akgb ~ Oct 2017
 sp2gna ~ May 2020 ;sp2nuo=2020-05-27;sp2v18=2020-05-27;sp2v7o=2020-05-28
 z4akga ~ Feb 2021
 10000am ~ Feb 18, 2021; 100001g=2021-02-18;10000rt=2021-02-18-194850
 1qrrrav ~ Aug 30, 2021
 1tekpzg ~ Sep 23, 2021

Licensing

This project is released under the MIT license, see LICENSE.md for more details.