cromulence

A small, puzzlehunt-tuned, word frequency library. Basically a port of the wordlist part of Solvertools. The default wordlist is 713 kB gzipped; the library itself is 5 kB.

Usage

Install with npm install cromulence. This is an ESM only library.

``ts
import { Cromulence, loadWordlist } from "cromulence";

const cromulence = Cromulence(await loadWordlist());

cromulence.cromulence("carl_t_on_fisk"); // 9.9

cromulence.info("carl_t_on_fisk");
// {
// cromulence: 9.9,
// logProb: -28.344822494756706,
// slug: 'carltonfisk',
// spacing: 'CARLTON FISK',
// zipf: -3.3,
// }
`

$3

The Cromulence constructor takes wordlist, an object with slug keys and Zipf frequency values:

- A slug is a string of lowercase English letters.
- The Zipf frequency of a text is the base-10 logarithm of the number of times it appears per billion words; for single words it usually ranges from 0 to 8. cromulence always rounds Zipf frequencies to one decimal place.

The default wordlist, data/wordlist, are the entries in the Solvertools combined wordlist with no spaces and weighted frequency at least one million.

It can be loaded with loadWordlist. On Node, it will use the Node APIs to open the wordlist distributed with the package. On the browser, it will use fetch to download the wordlist from jsDelivr. If you're using cromulence in a webapp, you may want to host the wordlist on your own servers.

See the blog post cromulence: a small word frequency library for more information on how the wordlist was chosen.

$3

Instances of Cromulence have a wordlist property, which is the wordlist it was constructed with.

They also have the following methods, each which take a single string parameter text:

- cromulence: Returns the cromulence of text, using the Solvertools algorithm. Cromulence is a measure of how likely the text is to be a puzzlehunt answer, and it usually ranges from -45 to 35, with positive cromulences corresponding to real answers.
- logProb: The log probability of text appearing when words are drawn independently from the wordlist.
- zipf: The log probability of text, as a Zipf frequency. The Zipf frequency of a phrase is the base-10 logarithm of the number of times it appears per billion words; for single words it usually ranges from 0 to 8.

There's also the info method, which includes each of these, along with the slug of text, and the likeliest spacing.

$3

Other exported utility functions:

- slugify: Converts a string to a slug.
- logProbToZipf, zipfToLogProb: Converts between log probabilities and Zipf frequencies.
- zipfToCromulence`: Converts a Zipf frequency to cromulence. Cromulence takes into account the length of the text.