config.sample.json

{
  "user": "metadb",
  "host": "localhost",
  "database": "metadb",
  "password": ""
}

convert.js

const fs = require('fs')
const path = require('path')
const { Client } = require('pg')
const config = require('./config.json')
const client = new Client(config)

const OUTPUT_DIR = path.join(__dirname, 'authorities')

client.connect()

const query = "SELECT * FROM controlled_vocab WHERE vocab_name != '' AND contents != ''"
client.query(query, (err, res) => {
  client.end()

  // get all of the vocabularies w/ values
  const vocabs = res.rows

  for (let i = 0; i < vocabs.length; i++) {
    vocab = vocabs[i]
    const name = vocab['vocab_name'].toLowerCase().replace(/\./g, '_')
    let terms = vocab['contents'].split(';')

    if (name.indexOf('subject_ocm') > -1) {
      terms = terms.sort((a, b) => {
        const anum = parseOcmNumber(a)
        const bnum = parseOcmNumber(b)

        return anum < bnum ? -1 : 1
      })
    }

    const body = terms.reduce((str, term) => {
      let t = term

      if (/[#|:]/g.test(t) && t.indexOf('"')) {
        t = '"' + t + '"'
      }
      
      str += '  - ' + t + '\n'
      return str
    }, '')


    const outputPath = path.join(OUTPUT_DIR, name + '.yml')
    fs.writeFileSync(outputPath, '# ' + vocab['vocab_name'] + '\nterms:\n' + body)
  }
})

function parseOcmNumber (str) {
  const match = str.match(/^\d+/g)

  if (!match) {
    return str
  }

  return parseInt(match[0], 10)
}

package.json

{
  "name": "metadb-authorities-to-hyrax-yaml",
  "version": "0.0.0",
  "description": "",
  "main": "convert.js",
  "keywords": [],
  "author": "Adam Malantonio <[email protected]>",
  "license": "GPL-3.0",
  "dependencies": {
    "pg": "^7.3.0"
  }
}