def read_file(input, max_rows=None): count = 0 vocab = Vocabulary() lines = [] with open(input, 'r') as infile: while True: if max_rows is not None: if count > max_rows: break count += 1 line = infile.readline() if not line: print('EOF reached') break line = normalize_line(line) vocab.add_sentence(line) lines.append(line) return lines, vocab