CMU pronouncing dictionary in Typescript

README.md

cmudict-0.7b.txt

$ npm install
$ npx ts-node cmudict.ts hello world
hh-ah0-l-ow1 w-er1-l-d

cmudict.ts

const assert = require('assert')
const fs = require('fs')
const AVLTree = require('binary-search-tree').AVLTree
const minimist = require('minimist')
const split = require('split')

const load = async (): Promise<any> => {
  const cmudict = new AVLTree({ unique: true })
  return new Promise<any>((resolve: any) => {
    fs.createReadStream('cmudict-0.7b.txt')
      .pipe(split())
      .on('data', (line: string) => {
        if (/(^$|^[^A-Z']|\(\d\))/.test(line)) return
        var [word, pronunciation] = line.split('  ', 2)
        pronunciation = pronunciation.replace(/ /g, '-')
        cmudict.insert(word.toLowerCase(), pronunciation.toLowerCase())
      })
      .on('end', () => resolve(cmudict))
  })
}

const main = async (words: string[]) => {
  const cmudict = await load()
  for (const word of words) {
    process.stdout.write(cmudict.search(word.toLowerCase())[0])
    process.stdout.write(' ')
  }
  console.log()
}

if (require.main === module) {
  var argv = minimist(process.argv.slice(2))
  assert(argv._.length > 0)
  main(argv._)
}

package.json

{
  "devDependencies": {
    "@types/minimist": "^1.2.2",
    "@types/split": "^1.0.0",
    "npx": "^10.2.2",
    "ts-node": "^10.4.0"
  },
  "dependencies": {
    "binary-search-tree": "^0.2.6",
    "minimist": "^1.2.5",
    "split": "^1.0.1"
  }
}