Creates an object from a html string obtained from an index of apache directory. Translates an APACHE default directory listing, to javascript object.

directoryparse.js

/**
* Requires JQuery >= 1.8
* @param {info} str string-based data of the 
* Author: @thblckjkr | Theo G
* Ported to JQuery from https://github.com/juliangruber/parse-apache-directory-index
* Usage:
*   $.get("/a/directory/path", function(data, status){
*     var files = directoryparse(data);
*     console.log(files);
*  })
*/
function bytes(str){
  var m = /^\s*([0-9.]+)([A-Z]*)\s*$/.exec(str);
  if (!m) return null;
  var num = Number(m[1]);
  var suf = m[2];
  return suf === 'K' ? num * 1024
  : suf === 'M' ? num * 1024 * 1024
  : suf === 'G' ? num * 1024 * 1024 * 1024
  : num;
};
  
function directoryparse(src){
  var dir = '/';
  var files = [];
  
  // The fifth element is always the table?
  $document = $.parseHTML(src);
  $table = $($document[5]);
  $rows = $table.find('tr').toArray();
  
  // Figure out the order of the columns, by looking at the header row.
  // eg { 'Name': 0, 'Last modified': 1, 'Size': 2 }
  var fieldCols = $($rows[0])
    .children('th')
    .toArray()
    .reduce((fieldCols, th, i) =>
    Object.assign(fieldCols, {
      [$(th).text().trim()]: i
    }),
    {});
  
  // Make sure we at least found a "Name" column
  if (fieldCols['Name'] === undefined) {
    throw new Error('Unable to parse apache index html: cannot identify "Name" column.');
  }
  
  // Parse fields
  $rows
  // Ignore the header row
  .slice(1)
  .forEach((tr) => {
    var $tds = $(tr).find('td');
    var getCol = label => fieldCols[label] === undefined ? null : $tds.eq(fieldCols[label]);
    var getColText = label => getCol(label) && getCol(label).text().trim();
    
    var path = getCol('Name').children().eq(0).attr('href');
    var name = getColText('Name');
    
    // Ignore 'Parent Directory' row
    if (name === 'Parent Directory' || !name) return;
    
    files.push({
      type: path.endsWith('/')
      ? 'directory'
      : 'file',
      name: name,
      path: dir + path,
      lastModified: getCol('Last modified') && new Date(getColText('Last modified')),
      size: getCol('Size') && bytes(getColText('Size')),
      description: getColText('Description')
    });
  });
    
  return { dir, files };
};