Question

I’m trying to extract email info from HTML files in my hard drive.

If I load the file in firefox and run jQuerify bookmarklet I can use successfully the following selector/function

window.jQuery("a.iEmail").each(function(el) {
  console.log(window.jQuery(this).attr('href'))
});

But using this in Node.js is not working

var document = require("jsdom").jsdom(),
  script = document.createElement("script"),
  fs = require('fs');

fs.readFile('file_1.html', 'utf-8', function(err, data){
  if (err) {
    throw err;
  }

  // This output the document
  //console.log(data)

  var window = document.createWindow(data);

  script.src = 'http://code.jquery.com/jquery-1.4.2.js';
  script.onload = function() {
    console.log(window.jQuery.fn.jquery);
    // outputs: 1.4.2
    //console.log(window.jQuery);

    /*
     * This line works if i load the local file in firefox and execute
     * the jQuerify bookmarlet
     */
    window.jQuery("a.iEmail").each(function(el) {
      console.log(window.jQuery(this).attr('href'))
    });
  };
  document.head.appendChild(script);
});

Answer 1

I now know what the problem is.

The html data, must be passed in the document creation call, so the code look like this:

var jsdom = require("jsdom"),
    fs = require('fs');

fs.readFile('file_1.html', 'utf-8', function(err, data){
  if (err) {
    throw err;
  }

  // This output the document
  //console.log(data)

  // HTML data should be in document creation call
  var document = jsdom.jsdom(data); // data is the html content
  var script = document.createElement("script");

  // HTML data SHOULD NOT be in window creation call
  var window = document.createWindow();

  script.src = 'http://code.jquery.com/jquery-1.4.2.js';
  script.onload = function() {
    console.log(window.jQuery.fn.jquery);
    // outputs: 1.4.2
    //console.log(window.jQuery);

    /*
     * This line works if i load the local file in firefox and execute
     * the jQuerify bookmarlet
     */
    window.jQuery("a.iEmail").each(function(el) {
      console.log(window.jQuery(this).attr('href'))
    });
  };
  document.head.appendChild(script);
});