I've introduced some changes to make this tokenizer work for python 3.x and improved a little bit sentence segmentation:

-added lists with abbreviations and endings as the global variables so as to make it work not only for Corpus class, but for Text also;
-added the function get_sentence_segmentation for Text class to improve sentence segmentation a little bit and make the segmentation result returnable; you can still write out the results just using write_out function if you wish;
-added the opportunity to change input type for Text class: now you can enter not only a path to a file, but a text string also.

The code is available here, the lists with abbreviations etc. are available here.