A speed-coded NaNoGenMo of cryptic crossword grids

[mmoskowitz]

I'm attempting to do NaNoGenMo in just a few days by also using it as an opportunity to release some code I've had sitting on my hard drive for handling cryptic crossword grids. My plan is to create a reference with a filled-out sample grid for all 6,364 contiguous black-square cryptic crossword grids of size 15x15 with word lengths of 5, 7, or 9.

[mmoskowitz]

Experimental investigation indicates that the actual number of grids is 6,170.

[mmoskowitz]

Data generation started Sun Nov 28 21:03:39 EST 2021 ended Mon Nov 29 16:40:40 EST 2021 duration 19:37:01

[mmoskowitz]

Completed!

This is sort of an imaginary 1970's project from IBM or some similar company, to produce a sample word setting for all possible grids for cryptic crosswords within a set of constraints:

1. Grids are 15x15 and use black squares rather than bars
2. Every even letter is unchecked
3. Words may be of 5, 7, or 9 letters
4. Grids must be contiguous
5. Grids have rotational symmetry

This leads to every grid being definable by the lengths of the four Across and four Down entries in the top left corner, which is what this project uses as a code to distinguish grids.

Code is at https://github.com/mmoskowitz/NaNoGenMo2021 with three volumes, each of which has over 60,000 words, at:
https://github.com/mmoskowitz/NaNoGenMo2021/blob/main/volI.pdf
https://github.com/mmoskowitz/NaNoGenMo2021/blob/main/volII.pdf
https://github.com/mmoskowitz/NaNoGenMo2021/blob/main/volIII.pdf

The code makes a lot of use of the grimdrake library at https://github.com/mmoskowitz/grimdrake which I've added to github and will be writing a real release for as part of this project.