squareword_gen

This is an experiment to try generatating valid NxN squares of words, where both columns and row are words from a given dictionary.

It was inspired by the squareword.com game, and an algorithm from @annalee

It outputs things like this:

WordSquare 17 from word 4252/5000: double-word-square, unique
  gnomes
  repent
  outage
  origin
  moment
  snares

Usage

usage: squareword_gen.py [-h] [--freq_csv_file FREQ_CSV_FILE]
                         [--scrabble_words_file SCRABBLE_WORDS_FILE]
                         [--top_n TOP_N] [--word_len WORD_LEN]
                         [--double_squares_only] [--log_details]
                         [--just_benchmark]

optional arguments:
  -h, --help            show this help message and exit
  --freq_csv_file FREQ_CSV_FILE
                        CSV file where the first row is words, sorted in most
                        popular first
  --scrabble_words_file SCRABBLE_WORDS_FILE
                        Text file of valid words to use
  --top_n TOP_N         Cutoff for N most popular words to use
  --word_len WORD_LEN   Len of words to use
  --double_squares_only
                        Print only valid double squares
  --log_details
  --just_benchmark

TODO

• Get a better word-frequency list. The one here seems to lean very heavily towards words seen online, rather than in books. The Wiktionary:Frequency lists/English/TV and Movie Scripts (2006) looks good, but I haven't found it in machine-readable form. Another possibility is Leipzig Corpora Collection.

Attribution:

• unigram_freq.txt: from https://www.kaggle.com/datasets/rtatman/english-word-frequency/
• scrabble_words.txt: is from https://raw.githubusercontent.com/redbo/scrabble/master/dictionary.txt (https://github.com/redbo/scrabble)