I am a computer science student looking for unsolved sudoku games that are valid and available in a comma separated text file. Does such a prepared data set exist and where can I find such a data set? If not, does anyone know of an open source solution that can perform the OCR from a picture of a sudoku game, and convert it into some array or text file? The array or text file must be more conducive to testing out sudoku solving algorithms that I want to implement by writing my own programs.
From the SO answer to the essentially same question (test cases for Sudoku solver):
You can find some large datasets for Sudoku benchmarking and testing in this project: https://github.com/t-dillon/tdoku
See data.zip for the puzzles.
See https://github.com/t-dillon/tdoku/blob/master/benchmarks/README.md for descriptions of the datasets, their sources, and their difficulties.
In addition to that, if you want some puzzles that require a specific solving technique, you can check out some Sudoku collections on SudoCue (check the Downloads page). You can also find the docs for their file formats on the same site.