Mathias Brandewinder on .NET, F#, VSTO and Excel development, and quantitative analysis / machine learning.
by Mathias 19. September 2010 16:48

fail-owned-translation-failI have been involved recently with a project for a foreign company with offices in the US; the US team receives lots of documents in a foreign language which require translation, and they were interested in a way to speed up the process, using Bing Translator and Google Translate.

Why not just pick one? Both are awesome, but if you have tried them, you know the results can be somewhat random. Having the option to compare their respective results is a good way to make sure you end up with something which makes sense.

This sounded like a perfect case for the Strategy Pattern. I started by defining a common interface:

public interface ITranslator
{
  string TranslateToHindi(string original);
  string TranslateToEnglish(string original);
}

… and implemented two versions, the BingTranslator and the GoogleTranslator, using the API for each of these services. So far so good, but when I started working on the user interface, I ran into a small problem. My user interface has just 2 buttons, “translate to English”, and “Translate to Hindi”, and Ideally, I would have liked to  just pass the specific language pair to use, along these lines:

private void toEnglishButton_Click(object sender, RibbonControlEventArgs e)
{
  ITranslator translator = this.GetTranslator();
  this.TranslateStuff(translator, stuffToTranslate, “en”, “hi”);
}

More...

by Mathias 7. September 2010 16:56

In my previous posts, I explored how to identify text in a PowerPoint slide and use the Google .NET API to translate it; let’s put it all together in a simple VSTO add-in for PowerPoint 2007.

The translation functionality is displayed in a Custom Task Pane, where you can pick the language of origin, and the language to translate to. I used the same general design I presented in my Excel add-in tutorial, using WPF controls in the task pane with the MVVM pattern, leveraging the small yet very useful MVVM foundation framework. When running, this is how the add-in looks like:

TranslatorStart

TranslatorEnd

I added only 3 languages in there, but it is fairly easy to modify the code and get it to work with any language pair supported by Google translate.

Download code sample

by Mathias 2. September 2010 17:04

The beauty of working with a framework like .NET is that when you have a problem, chances are, someone else did before you, and might have even resolved it for you. In our last post, I explained how to find text in a PowerPoint slide using C#. My goal was to translate it using the Google Translate service, and I intended to write my own code to call the web service and retrieve the translation. Turns out, there is a .NET API for Google Translate, ready to use, which does it for you already (of course, I found that out already after rolling out my own code, which wasn’t nearly as good).

Building up where we left off, I quickly wrote this class, which will translate the slide currently in view; I think the code is self-explanatory: get the slide, create a Google translator, pass it the language of origin and of destination, and translate every chunk of text you find! The only thing I needed to do was to download GoogleTranslateAPI_0.3.1, add the dll to the project as a reference.

namespace ClearLines.PowerPointTranslator
{
   using Google.API.Translate;
   using PowerPoint = Microsoft.Office.Interop.PowerPoint;

   public class SlideTranslator
   {
      public static void TranslateSlide(Language from, Language to)
      {
         var googleTranslator = new TranslateClient("http://www.clear-lines.com");

         var powerpoint = Globals.ThisAddIn.Application;
         var slide = (PowerPoint.Slide)powerpoint.ActiveWindow.View.Slide;

         foreach (PowerPoint.Shape shape in slide.Shapes)
         {
            if (shape.HasTextFrame == Microsoft.Office.Core.MsoTriState.msoTrue)
            {
               var textFrame = shape.TextFrame;
               var textRange = textFrame.TextRange;
               TranslateTextRange(textRange, googleTranslator, from, to);
            }
         }
      }

      public static void TranslateTextRange(
         PowerPoint.TextRange textRange, 
         TranslateClient translator, 
         Language from,
         Language to)
      {
         var paragraphs = textRange.Paragraphs(-1, -1);
         foreach (PowerPoint.TextRange paragraph in paragraphs)
         {
            var text = paragraph.Text;
            text = text.Replace("\r", "");
            paragraph.Text = translator.Translate(text, from, to);
         }
      }
   }
}

Later this week, after I do some code scrubbing, I’ll post the entire VSTO solution. Until then, have fun – it’s almost the week-end!

by Mathias 8. August 2010 17:30

I am currently on a project which involves creating a PowerPoint VSTO add-in. I have very limited experience with PowerPoint automation, so before committing to the project, I thought it would be a good idea to explore a bit the object model, to gauge how difficult things could get, and I set to write a small PowerPoint add-in which would automatically translate slides. Sounds like a simple enough project, how difficult could it be?

Turns out, not too difficult, but not completely trivial either. I discovered quickly that the PowerPoint object model, unlike most Office applications, doesn’t have much (any?) documentation for the .Net developer; the best I found is the VBA PowerPoint 2007 developer reference, which gives a decent starting point to figure out what the objects are about. So I thought I would share my exploration of the PowerPoint jungle, and hopefully spare some trouble to other .Net developers.

The plan

The objective is simple: write an add-in which allows the user to

  • select a language to translate from, and a language to translate to,
  • create a duplicate of the current slide, translating all the text and keeping the layout

The plan will be to use Google Translate to perform the translation. In order to do that, we will nedd to extract out all pieces of text that require translating.

Finding all the text in a slide

Lets’ start by identifying where we have text in the current slide. Let’s first create a PowerPoint 2007 Add-in project in Visual Studio. To keep things simple for now, we will add a Ribbon control with a button, and when that button is clicked, we’ll start working on the current slide:

RibbonWithButton

Double-click on the Button (I renamed my button translateButton) to generate an event handler for the Click event, and get the current Slide:

private void translateButton_Click(object sender, RibbonControlEventArgs e)
{
   var powerpoint = Globals.ThisAddIn.Application;
   if (powerpoint.ActivePresentation.Slides.Count > 0)
   {
      var slide = (PowerPoint.Slide)powerpoint.ActiveWindow.View.Slide as PowerPoint.Slide;
   }
}

More...

Comments

Comment RSS