Gets micro-data from html

node-microdata.js

// required stuff
var _ = require('lodash'),
  cheerio = require('cheerio');

// exporting outside
module.exports = {
  /**
   * Converts a html code to the object
   * @param {String} html
   * @returns {Object}
   */
  getMicrodata: function (html) {
    var $ = cheerio.load(html, {ignoreWhitespace: true}),
      scopeSelector = '[itemscope][itemtype]:not([itemscope] [itemscope])',
      propSelector = '[itemprop]:not([itemscope] [itemprop]):not([itemscope])',
      self = this, result = {};

    $(scopeSelector).each(function () {
      var itemType = this.attr('itemtype'), props = {};

      $(propSelector, $(this).html()).each(function () {
        var propName = this.attr('itemprop'), value;

        switch (this[0].name) {
          case 'audio':
          case 'embed':
          case 'iframe':
          case 'img':
          case 'source':
          case 'track':
          case 'video':
            value = this.attr('src');
            break;
          case 'a':
          case 'area':
          case 'link':
            value = this.attr('href');
            break;
          case 'object':
            value = this.attr('data');
            break;
          case 'meta':
            value = this.attr('content');
            break;
          default:
            value = this.text().trim();
            break;
        }

        if (!_.has(props, propName)) {
          props[propName] = value;
          return;
        }

        if (_.isArray(props[propName])) {
          props[propName].push(value);
          return;
        }

        props[propName] = [props[propName], value];
      });

      var tmp = {}, keyName = itemType.replace('http://schema.org/', '');
      tmp[keyName] = _.extend(props, self.getMicrodata($(this).html()));

      _.merge(result, tmp);
    });

    return result;
  }
};

Requires

lodash and the latest version of cheerio

Code

console.log(myVar.getMicrodata('html goes here'));

Example