joelongstreet/buttWipe.js

## buttWipe.js
// Pasting from Microsoft word is an ABSOLUTE DISASTER
// this method removes the endless gobs of garbage produced
// by the world's worst, yet most popular, text editor
// Requires jQuery
var cleanHTML = function(pastedString){

  // If this looks like some kind of raunchy ass microsoft bull shit,
  // rip off it's effing <HEAD></HEAD>
  var headRegex = new RegExp("<head[\\d\\D]*?\/head>", "g");
  pastedString  = pastedString.replace(headRegex, '');

  // Take the body, and ELMINIATE GARBAGE
  var pattern   = /<body[^>]*>((.|[\n\r])*)<\/body>/im;
  var cleaned   = pattern.exec(pastedString);
  if(cleaned && cleaned.length && cleaned[1]) pastedString = cleaned[1]

  // Remove meaningless HTML Comments... you have no idea how
  // meaningless these comments are. NO IDEA...
  var commentRegex = /<!--[\s\S]*?-->/g;
  pastedString     = pastedString.replace(commentRegex, '');

  // Remove whatever <o:p> and </o:p> is... JUST WHATEVER!!!
  pastedString  = pastedString.replace(/<o:p>/g, '');
  pastedString  = pastedString.replace(/<\/o:p>/g, '');

  // Take Microsoft's atrocious formatting and correcting it by
  // creating a jQuery object, wrapping it in a series of some random
  // div(s) and then finally making it HTML again
  var tranny  = $('<div>' + pastedString + '</div>').html();
  tranny      = '<div><div>' + tranny.replace(/&quot;/g, '') + '</div></div>';
  tranny      = $(tranny).find('*').attr('style', '').html();

  return tranny
};


// ------ Sample Usage ------ //
var $editor = $('textarea');
$editor[0].onpaste = function(e){
  var pasteContent = e.clipboardData.getData('text/html');
  if(pasteContent == '') pasteContent = e.clipboardData.getData('text/plain');
  var cleanContent = cleanHTML(pasteContent);

  $editor.html(cleanContent);
  return false
};
	// Pasting from Microsoft word is an ABSOLUTE DISASTER
	// this method removes the endless gobs of garbage produced
	// by the world's worst, yet most popular, text editor
	// Requires jQuery
	var cleanHTML = function(pastedString){

	// If this looks like some kind of raunchy ass microsoft bull shit,
	// rip off it's effing <HEAD></HEAD>
	var headRegex = new RegExp("<head[\\d\\D]*?\/head>", "g");
	pastedString = pastedString.replace(headRegex, '');

	// Take the body, and ELMINIATE GARBAGE
	var pattern = /<body[^>]>((.\|[\n\r]))<\/body>/im;
	var cleaned = pattern.exec(pastedString);
	if(cleaned && cleaned.length && cleaned[1]) pastedString = cleaned[1]

	// Remove meaningless HTML Comments... you have no idea how
	// meaningless these comments are. NO IDEA...
	var commentRegex = /<!--[\s\S]*?-->/g;
	pastedString = pastedString.replace(commentRegex, '');

	// Remove whatever <o:p> and </o:p> is... JUST WHATEVER!!!
	pastedString = pastedString.replace(/<o:p>/g, '');
	pastedString = pastedString.replace(/<\/o:p>/g, '');

	// Take Microsoft's atrocious formatting and correcting it by
	// creating a jQuery object, wrapping it in a series of some random
	// div(s) and then finally making it HTML again
	var tranny = $('<div>' + pastedString + '</div>').html();
	tranny = '<div><div>' + tranny.replace(/"/g, '') + '</div></div>';
	tranny = $(tranny).find('*').attr('style', '').html();

	return tranny
	};


	// ------ Sample Usage ------ //
	var $editor = $('textarea');
	$editor[0].onpaste = function(e){
	var pasteContent = e.clipboardData.getData('text/html');
	if(pasteContent == '') pasteContent = e.clipboardData.getData('text/plain');
	var cleanContent = cleanHTML(pasteContent);

	$editor.html(cleanContent);
	return false
	};