• adj-noun
• NLTK.org

• SecureRandom seeding?
• e.g. Spicy::Proton.new(seed: 1234).noun
• Instance only, or support in class methods, too? (maybe Spicy::Proton.seed(1234)) before other calls

Right now, random words must be picked from a specific category (noun, adjective, ...). There should be a way to get a random word across all categories.

• Support class and instance methods: Spicy::Proton.word, gen.word
• Support constraints: gen.word(min: 2, max: 5)
• Support formatters: Spicy::Proton.format('%w %w %w')
• Support word list: gen.words gives everything in the corpus
• Update readme

I'm trying to override rand in Spicy::Seek to allow for the flexibility of making the function deterministic based on an incoming seed. Before submitting a PR, I wanted to propose a couple of ways to implement this.

1. The easiest is to allow the options hash to be forwarded to rand. Then I can extend the classes and override rand with my own custom implementation:

  def seek(opts = {}, &found)
  ...
-      index = rand(rand_min, rand_max)
+      index = rand(rand_min, rand_max, opts)
  ...
  end

- def rand(low, high)
+ def rand(low, high, opts = {})
    range = high - low + 1
    low + SecureRandom.random_number(range)
  end

2. Allow passing a lambda as :rand option:

  def seek(opts = {}, &found)
  ...
-      index = rand(rand_min, rand_max)
+      index = opts[:rand] ? opts[:rand].call(rand_min, rand_max) : rand(rand_min, rand_max)

3. Any suggestions how I can make this work without otherwise.

Let's discuss below, and I can submit a PR with the preferred implementation.

• Currently YAML
• Can just be null-separated words
• File.read('adjectives.txt').split("\0")
• Parsing should be faster than YAML

Can we ensure that the generated string is SFW?

• Verb
• Adverb
• Prepostion? Interjection? Conjunction? Pronoun?