Solving the maze

Print the shortest way through as a trail of markers, or watch the search spread out and find it.

--solve prints the shortest way from the entrance to the exit as a trail of . markers laid over the maze:

python -m py_maze -d easy --seed 2024 --solve

Output:

start
*.***********
*.*     *   *
*.*** * *** *
*...* *     *
***.* *******
*...*       *
*.*** ***** *
*...* *   * *
***.* * *** *
*...*   *...*
*.*******.*.*
*.........*.*
***********.*
end
seed: 2024

The solver is a breadth-first search, so the route it draws is always the shortest one. Mazes built by recursive backtracking have exactly one route between any two points anyway, which makes the shortest route the only route.

Watching the Search

--animate steps that same search across the screen before the solved maze is printed. Each frame is one wave further from the entrance:

python -m py_maze -d easy --seed 2024 --animate

One frame partway through:

Solving...
start
*~***********
*~*     *   *
*~*** * *** *
*~~~* *     *
***~* *******
*~~~*       *
*~*** ***** *
*?  * *   * *
*** * * *** *
*   *   *   *
* ******* * *
*         * *
*********** *
end
frontier ?   explored ~   solution .

? marks the frontier the search is about to grow from, ~ the cells it has already explored, and . the finished path on the last frame.

Animating needs a screen to draw over. When the output is piped or redirected there is nothing to animate, so the maze is solved without the frames and only the solved maze is written:

python -m py_maze --animate > solved.txt

A loaded maze can lack a way through, which a generated one never does. When one does, the run exits with a status code saying so, tabled under Scripting py_maze.