Sudoku Gen

sudoku-gen generates and solves Sudoku puzzles across 9×9, 12×12, and 16×16 grids, including jigsaw (irregular-region) variants, with a constraint-propagation solver fast enough to validate uniqueness while a puzzle is still being carved out of a filled grid. The web frontend runs generation in a Web Worker so dragging the difficulty slider or requesting a fresh batch never blocks the UI thread, with a database of generated puzzles and a batch-print view for producing a stack of puzzles at once. Alongside the web app, a separate CLI client drives the same generation core directly from the command line: it fans the search across goroutines, one per worker, to brute-force-generate and validate large batches of puzzles locally far faster than a single-threaded generator — useful for pre-seeding the puzzle database or stress-testing the solver against pathological grids.

graph TD
  CLI["sudoku-gen CLI"] -->|size, type, difficulty| Engine["sudoku_gen_go
(submodule)"]
  Engine --> Workers["N worker goroutines
(threadID 0..threads-1)"]
  Workers -->|backtracking + MRV heuristic| Solve["solve() per worker"]
  Solve -->|first valid grid wins| ResultChan[("resultChan")]
  Workers -.close on success.-> StopChan[("stopChan")]
  ResultChan --> Output["Puzzle + Solution grid"]

for i := 0; i < g.threads; i++ {
	wg.Add(1)
	go func(threadID int) {
		defer wg.Done()
		for attempt := 0; attempt < attemptsPerThread; attempt++ {
			select {
			case <-stopChan:
				return
			default:
			}

			grid := types.NewGrid(g.size, g.sudokuType)
			// ... build subgrids, then solve ...
			if solved := g.solve(grid, startTime, maxTime); solved {
				g.removeNumbers(grid)
				select {
				case resultChan <- grid:
					close(stopChan) // Signal other goroutines to stop
					return
				case <-stopChan:
					return
				}
			}
		}
	}(i)
}