def modifyShakespeare(file_in, file_out): builder = FileBuilder(file_in, file_out) builder.remove_lines(line_last, len(builder)) builder.remove_lines(0, line_first) builder.remove_repeated_empty_lines() builder.remove_lines_within("<<", ">>") builder.remove_number_lines(6) builder.sub('[<}`\r]', '') builder.sub('[|_]', ' ') char_words = [ '!', '"', '&', '(', ')', ',', '-', '.', ':', ';', '?', '[', ']', '\n' ] builder.to_words(char_words) builder.to_words_apostrophes() builder.to_lowercase() builder.strip_lines()
def test_remove_number_lines(): builder = FileBuilder("../test_text/number_lines.txt", outfile) builder.remove_number_lines(4) assert builder[0] == 'abc\n' assert builder[1] == '\n' assert builder[2] == ' \n' assert builder[3] == 'six\n' assert builder[4] == '12345\n' assert builder[5] == '1234\n' assert builder[6] == 'q\n' builder = FileBuilder("../test_text/number_lines.txt", outfile) builder.remove_number_lines() assert builder[0] == 'abc\n' assert builder[1] == '\n' assert builder[2] == ' \n' assert builder[3] == 'six\n' assert builder[4] == 'q\n'