This source code is a demo which extracts phrase pairs from given corpus. Corpuses are referenced from Taetoba, "https://tatoeba.org/eng/". For tokenization, tokenizer of mosesdecoder, "https://github.com/moses-smt/mosesdecoder/tree/master/scripts/tokenizer", and for aligning words, Chris Dyer's fast_align "https://github.com/clab/fast_align" is used.