> à vo -- cê uma can -- ção legal > > I don't know how to change it, so that > http://lsr.di.unimi.it/LSR/Snippet?id=600 > shows the same as in Han-Wen's patch here.
This seems to be a bug on LSR webpage: What you see is double-encoded UTF-8 (see https://stackoverflow.com/questions/11436594/how-to-fix-double-encoded-utf8-characters-in-an-utf-8-table for a similar mysql issue) – UTF-8 encoded characters get interpreted as Latin-1, which in turn get re-interpreted as UTF-8. Note that in the LSR database itself the problem doesn't happen. Please contact Sebastiano so that he can comment and probably implement a fix. https://codereview.appspot.com/571640044/