Basic Classifier implimentation

I want an implementation of a basic sentiment classifier on a set of text files. The attached files represent news, speeches, whatever. A whole bunch of text. For example, on Sept. 13, there is one speech transcript. The output would be something like the following: 13 Sept 12: 65% Where the percentage is some characterization of the positive or negative communication inherent in the document. This would probably be done by a binary read of the sentences. The weather looks good. The weather outlook is bad. 13 Sept 12: 50% ... would be the result from those two sentences. I want to do that for each sentence within each document. I want a summary by document in a spreadsheet. The implemented algorithm should be either Sentaword net )[login to view URL]), OpinionFinder ([login to view URL]) or both.