extract_headwords.js

'use strict';
/******************************************************************************/
const fpath = process.argv[2];
if (!fpath) {
  console.error('Add the file path, please.');
  process.exit(1);
}
/******************************************************************************/
const fs = require('fs');

const inCoding = guessEncoding(fpath);
if (!inCoding) process.exit(1);

const outCoding = 'utf8';

const rl = require('readline').createInterface({
  input: fs.createReadStream(fpath, inCoding),
  terminal: false,
  historySize: 0
});

const outfile =  fs.openSync(fpath.replace(/(\.[^.]+)?$/, '.hw.txt'), 'w');
fs.writeSync(outfile, '\uFEFF', null, outCoding);

const bomRE = /^\uFEFF/;

const hwRE = /^[^#\s]/;

let   lineNumber = 0;
/******************************************************************************/
rl.on('line', line => {
  if (++lineNumber === 1) line = line.replace(bomRE, '');
  if (hwRE.test(line)) {
    fs.writeSync(outfile, `${line.trim()}\n`, null, outCoding);
  }
});
/******************************************************************************/
function guessEncoding(fpath) {
  try  {
    const fd = fs.openSync(fpath, 'r');
    const bf = Buffer.alloc(2);
    fs.readSync(fd, bf, 0, 2, 0);
    fs.closeSync(fd);
    return bf[0] === 0xFF && bf[1] === 0xFE ? 'utf16le' : 'utf8';
  } catch(e) {
    console.error(`Error: ${e.message}.`);
    return null;
  }
}
/******************************************************************************/