I downloaded pywikipedia yesterday and am using it on the Assamese Wikipedia.  Thanks for a great product! 

Nevertheless, I have a little difficulty trying to make it do exactly what I want.   I am using it to correct some unicode encoding issues.  In particular, I am trying to replace some unicode characters by some others.  They are very short: at most 3 characters long. 

But I have been unable to avoid picking up matches inside wikilinks (internal as well as inter-language).  Is there a way to do so without employing unicode regularization?