This project contains Node.js scripts to generate a database of Sudoku puzzles with solutions. Each puzzle includes a difficulty rating, and the number of clues (or givens) is also included.
The difficulty rating is open-ended, and is based on the search depth required to solve the puzzle. Since the difficulty is estimated by an algorithm, it does not always reflect the difficulty for a human solver. Most puzzles have difficulties between 0 and 2, but a few puzzles have difficulty ratings of 7 or higher.
Most puzzles have between 23 and 26 clues, but I have encountered puzzles with as few as 20 clues and as many as 29 clues. It is known that 17 is the least possible number of clues for a proper Sudoku puzzle having a unique solution, but these are hard to generate.
It is not possible with these scripts to specify the difficulty level or the number of clues when generating a Sudoku puzzle. The scripts generate Sudoku puzzles completely at random, and then estimate the difficulty afterwards.
This project requires Node.js and MySQL. I am using Node.js version 8.10.0 and MySQL version 5.7.28. Later versions of Node.js and MySQL (or MariaDB) should also work. Some earlier versions will also work, but I do not know the minimum requirements.
-
Clone this repository.
-
Make a copy of the file
secrets.template.jsand name itsecrets.js. -
Edit
secrets.jsto include your MySQL database credentials. -
Enter the command
npm installon the command line. -
Enter the command
node create-table.jsto create the table. Only run this script once, because it will drop the table if it already exists.
To run the script, type the command
$ node generate-puzzles.js [number] [batchSize]
where [number] is the number of puzzles to generate, and [batchSize] is the number of puzzles that are saved in each commit. The default values are 1000000 and 1000. With these values, the script runs for about 12 hours on my computer. The script can be safely interrupted by pressing Control-C, and the script can be run repeatedly to enlarge the database.
Note: The script uses recursion to process the batches, so you may experience a stack overflow if the number of batches is greater than 10,000. The number of batches is equal to the number of puzzles divided by the batch size, rounded up to the nearest integer. If you don't like this, please submit a pull request to fix it.
Contributions are welcome. Please read the Code of Conduct before contributing.
David Radcliffe (dradcliffe@gmail.com)
This project is licensed under the terms of the MIT license.
This project is based on the excellent sudoku project by Blagovest Dachev.