mike's Blog

Well – we’ve had the need for a spell checker for our blog, comment and review text areas for a while – we initially integrated FCKEditor but I found it’s interface a bit garish and it didn’t work in Safari at all so I went and took a look into TinyMCE again.   They have a very simple and nicely implemented (google gmail style) spellchecker that works as a plugin to the TinyMCE framework.   

The problem with the plugin is that out of the box it requires PHP.  Initially we recompiled lighttpd to enable it to serve up PHP files but the dueling app servers didn’t seem like a good idea for production.   So I set out to write a rails implementation against aspell which is one of the supported methods that the plugin uses.

I had already written some homegrown code to talk to aspell and decided that this could be refactored to use the XMLHttpRequest method that TinyMCE used. 

It was actually very simple.   Just add the spelling.rb module to your /lib directory in Rails which looks like this:

---

module Spelling

  ASPELL_WORD_DATA_REGEX = Regexp.new(/&sw+sd+sd+(.*)$/)
  ASPELL_PATH = “aspell-0.60”

  def check_spelling(spell_check_text, command, lang)
    xml_response_values = Array.new
    spell_check_response = `echo ”#{spell_check_text}” | #{ASPELL_PATH} -a -l #{lang}`
    if (spell_check_response != ’’)
      spelling_errors = spell_check_response.split(” ).slice(1..-1)
      if (command 'spell')
        for error in spelling_errors
          error.strip!
          if (match_data = error.match(ASPELL_WORD_DATA_REGEX))
            arr = match_data[0].split(' ')
            xml_response_values << arr[1]
          end
        end
      elsif (command ‘suggest’)
        for error in spelling_errors
          error.strip!
          if (match_data = error.match(ASPELL_WORD_DATA_REGEX))
            xml_response_values << error.split((”, ”, ““)
          end
        end
      end
    end
    return xml_response_values.join(““)
  end
end

then in your application.rb (or wherever you feel like putting it) add this spellcheck action:

---

def spellcheck
  @headers[‘Content-Type’] = ‘text/xml’
  @headers[‘charset’] = ‘utf-8’
  suggestions = check_spelling(params[:check], params[:cmd], params[:lang])
  xml = ”<?xml version=”1.0” encoding=”utf-8” ?>#{suggestions}“
  render :text => xml
  return
end


Note in the spellcheck action there’s a parameter called :tinymce_id.   That’s gets created in a route because tinymce actually passes this variable with the “id” which confuses Rails since it has special meaning. 

Here at gusto we don’t use routes.rb because the amount of URL rewriting we do required it’s own controller that we call RoutingController.  Here’s the code that renders the spellcheck action:

---

params[:tinymce_id]  = params[:id]
render_component(:controller => ‘gusto’, :action => “spellcheck”, :params => params)


You will need to also either trap for the /tinyspell.php request or do what we did and modify editor_plugin.js to point to a method called ”/spellcheck” which we trap for in our routing controller.   Either way should work fine. 

Also I had to fix some bugs in the spellchecker and TinyMCE to get this all to work in IE6 and IE7 (gotta love Internet Explorer for wasting your time if you’re web developer).   You can read about that here:

http://tinymce.moxiecode.com/punbb/viewtopic.php?id=5055


All in all though I think the combination of TinyMCE and Ruby on Rails provides an excellent spell checker solution that works in all the major browsers and except for the IE nonsense was trivial to integrate.

Feel free to ask questions if you need help getting it setup. 

Mike