Table of Contents

1. Demo

2. Documentation

3. News

• Algot's Puzzles: http://runeman.org/embroidery/

Demo

This was recorded with asciinema.

• Octans v0.2.3: Visualized Output

• Octans v0.1.0 - finds 10 solutions to a puzzle: https://asciinema.org/a/385598

• Octans v0.1.0: https://asciinema.org/a/385500

• Octans 2021-01-14: https://asciinema.org/a/384464

• alt-link (download): https://andinus.nand.sh/static/octans

Documentation

Implementation

Initially it went over the list of words & checked if they exist in the grid. This was very slow.

Currently it walks the grid & checks if the current string exist in the dictionary. This is faster for these reasons:

• The dictionary is sorted, we perform binary range search on the dictionary to return the list of all words that start with specific string.

• Starting positions are limited.

If the dictionary wasn't sorted then this probably would've been slower than previous implementation.

The neighbors subroutine (lib/Octans/Neighbors.rakumod) was taken from my AoC (Advent of Code) 2020's day-11 solution.

Options

dict

Octans's default dictionary file is /usr/share/dict/words, use --dict flag to change the dictionary. The words in dictionary must be seperated by a newline (\n) & sorted alphabetically.

visualize

Generates a video solution for the puzzle.

length

Minimum word length to look for. Default is 7.

path

The path to be passed must be a readable file or an url in either format:

• Link when you view it from your local instance: https://tilde.zone/web/statuses/105531207939242077

• Link from Algot's profile: https://mastodon.art/@Algot/105333136907848390

verbose

This will increase verbosity.

News

v0.2.5 - 2023-07-02

• Fixed puzzle output with visualize turned off.

  Earlier the program would output a bunch of new lines, etc. The issue was that word-search subroutine was passed :visualize which sets that flag to always true and that caused weird behaviour, this has been fixed.

v0.2.3 - 2022-01-12

• Add visualize option.

  --visualize now generates a video solution for the puzzle using Cairo and ffmpeg.

v0.2.0 - 2021-03-04

• Removed sample option

  --sample will not solve the sample puzzle.

• Removed shorthand for verbose option

  -v won't work in v0.2.0.

• Change representation of visited squares

  When --verbose was passed, it would print the solved puzzle with visited squares replaced with fancy characters. Now it marks them by adding:

  • * to visited + gray squares (start positions)

  • / to visited squares

• Removed time taken

  Time taken won't be printed anymore.

  It was misleading because the time printed was not the time taken to find that specific word, it was the time taken to find all the words upto that word. It would reset for each starting position.

v0.1.4 - 2021-02-19

• Fixed the puzzle parsing regex

  The older regex fails to parse this puzzle: https://mastodon.art/@Algot/105690195742318751.

v0.1.3 - 2021-01-24

• Added an option to specify minimum word length.

v0.1.2 - 2021-01-20

• Input puzzle can now be of any size & not restricted to 4x4 grid.

v0.1.1 - 2021-01-20

• Read puzzle from a file.

v0.1.0 - 2021-01-19

• Improved performance by using a better algorithm to find words in the grid.