August 16, 2009

Spellcheck HTML Inline with GoogieSpell

GoogieSpell screenshot

Orangoo Labs' GoogieSpell widget is a really clever way of spellchecking input fields or textareas on a Web form, completely inline (think Microsoft Word) and without that nasty popup window you usually see that loops through all of your typos serially. As a bonus, by default it makes a call to Google's spellchecking web service so you don't have to install or maintain your own dictionary (although you can easily extend it to do so). Also, did I mention it's FREE?

I recently worked on an online workflow for writing, and one of our primary goals was to curb the amount of spelling mistakes we were seeing from the authors. Our solution was to spellcheck their content in an elegant but mandatory way. Enter GoogieSpell.

But, nobody is satisfied with just plain text anymore, right? Much of this content happens to be HTML from a popular 3rd party rich text editor. This presented a more complicated problem than it sounds (or maybe it sounds just about as complicated as it is?) and I'm going to walk through my solution: making GoogieSpell play nicely with both HTML and plain text.

You can download my modified version of GoogieSpell below; I've modified both the original GoogieSpell JavaScript (which instantiates one spellchecker per DOM element) as well as GoogieSpell's later-released "Multiple" version, which can check multiple elements through one spellchecker instance. I have also modified the original demo page to include an HTML spellchecking example.

Download files

First, we add a new GoogieSpell property, GoogieSpell.isHTML, along with getter/setter methods. This acts as a flag for whether you need to spellcheck plain text (input or textarea tags) or HTML (anything else).

    GoogieSpell HTML support / switch between modes
GoogieSpell.isHTML = false;
GoogieSpell.prototype.setIsHTML = function(el) {
    var origSupport = new RegExp('input|textarea', 'i');
    this.isHTML = origSupport.test(el.nodeName) ? false : true;
GoogieSpell.prototype.getIsHTML = function() {
    return this.isHTML;

Second, we need to wire this new property into GoogieSpell's getValue and setValue methods, and tell it to use either the innerHTML or value attribute.


return (this.getIsHTML()) ? ta.innerHTML : ta.value;


if (this.getIsHTML()) {
    ta.innerHTML = value;
} else {
    ta.value = value;

Up to now, this has all been pretty straightforward, but once we get to GoogieSpell's showErrorsInIframe method, we start to do some 'hacking'. For anyone who has already attempted using GoogieSpell on HTML, you know the challenge is in GoogieSpell's custom error object/elements that have many custom properties important to GoogieSpell's proper functioning, and cannot simply be appended via innerHTML; they will lose all of this custom information in the process. So in order to retain this information, we add a step before placing the error objects where error object placeholders are appended using string manipulation (fully compatible with innerHTML) and follow that by placing the real error objects using DOM manipulation. Problem solved!

Phase 1: String manipulation. A string representation of the output called outputBuffer is appended with placeholders for the real GoogieSpell error objects:

outputBuffer += '<var class="err_hook">&nbsp;</var>';

Phase 2: DOM manipulation. A new function, insertErrLinks, is created that will read through outputBuffer and place GoogieSpell's custom err_link objects in their correct spots, using the DOM method replaceChild, preserving all error object information:

GoogieSpell.prototype.insertErrLinks = function(output, err_links) {
    var all_vars = output.getElementsByTagName('var');
    var all_vars_len = all_vars.length;
    var err_count = 0;
    var err_len = err_links.length;
    for (var i = 0; i < all_vars_len; i++) {
        if (all_vars[i].className == 'err_hook') {
            //  Replace &nbsp; with span.    is needed for IE implementation of innerHTML
            all_vars[i].replaceChild(err_links[err_count], all_vars[i].firstChild);
            if (err_count == err_len) break;

Minor note: I insert the var tag now, which some browsers give a default styling too. I reset this styling in googiespell.css. If you think that's crazy then by all means modify the script to suit whatever tag you'd prefer to use! I like var becaues it's likely nobody is using it in the HTML you're checking, so it makes the DOM manipulation slightly more efficient.

Try it yourself in the paragraph below. Note that you can correct the spelling but I'm not saving it in any way so it will reappear when this page is reloaded.

Heere is a parargaph with bould and underlined text that has some mistaykes

  1. And here is an orderred list
  2. With evne more typpos and misspellings!

There is one edge condition with this that I've run across: Google's dictionary (GoogieSpell's "out of the box" back-end) does not seem to work perfectly when you feed it HTML. Apparently it will return spelling suggestions that assume some closing tags are part of the words themselves. I've only seen this happen in lists on li tags, and only when the last word is misspelled with no spacing afterwards. We replaced Google with our own dictionary on the back-end and were able to fully support HTML afterwards.

So, what next? Well, after getting this far, why not put GoogieSpell inside the rich text editor itself? I can safely say that after sufficient banging of head against keyboard, we got this to work too. Maybe I'll go through that in a future post, but it ended up being a very specific, customized integration with another 3rd party tool so I'll leave this discussion as agnostic as possible. The work remaining, for those interested, involves adding a button to the toolbar of your editor du jour, and dealing with the never-ending nesting of iframes these tools typically use.

Happy spellchecking!

Update: By popular demand, Spellcheck Rich Text with GoogieSpell explains how to use GoogieSpell with FCKEditor