jesmarkov.js

'use strict';

class Markov {
  constructor() {
    this.minLength = 6;
    this.maxLength = 30;
    this.dict = {};
  }

  addSentence(text) {
    // '\0' is a sentinel marking the beginning + end of sentences
    text = "\0 \0 " + text + " \0";
    let words = text.match(/\S+\s*/g).map(function(x) { return x.trim(); });
    for (let i = 2; i < text.length; i++) {
        let k = words[i-2] + ' ' + words[i-1];
        if (!this.dict[k])
            this.dict[k] = [];
        this.dict[k].push(words[i]);
    }
  }

  getRandomValue(key) {
    if (!this.dict[key]) {
        console.log("error: encountered a state we can't proceed from: " + key);
        return '\0';
    }

    return this.dict[key][Math.floor(Math.random() * this.dict[key].length)];
  }

  generateSentence(seedword) {
    // try up to 100 times to generate a sentence that is less than maxLength
    for (let i = 0; i < 100; i++) {
        let sentence = ['\0', '\0'];

        if (seedword) {
            let words = seedword.match(/\S+\s*/g).map(function(x) { return x.trim(); });
            sentence = sentence.concat(words);
            let search = sentence.slice(-2).join(' ');
            if (!this.dict[search]) {
                return "Can't work with " + seedword + " m9";
            }
        }

        while (1) {
          let search = sentence.slice(-2).join(' ');
          let found = this.getRandomValue(search);

          if (found == '\0')
            break;

          sentence.push(found);

          if (sentence.length > this.maxLength+2) {
            sentence.pop();
            break;
          }
        }

        if (i == 99 || (sentence.length < this.maxLength+2 && sentence.length >= this.minLength+2)) {
            sentence.shift(); sentence.shift();
            return sentence.join(' ');
        }
    }
  }

  generate(sentenceCount) {
    let sentences = [];

    while (sentenceCount--) {
      sentences.push(this.generateSentence());
    }

    return sentences.join(' ');
  }
}

module.exports = Markov;