火炬之光2 翻译文件清理

translation_dup_clean.py

from tl2_utils import dat_format


def remove_duplicates_ordered(input_list):
    seen = set()
    output_list = []
    for item in input_list:
        if item not in seen:
            seen.add(item)
            output_list.append(item)
    return output_list


translation_file = 'E:\Program Files (x86)\Steam\steamapps\common\Torchlight II\mods\IMBA-符文觉醒-暗黑传奇\MEDIA\TRANSLATIONS\MANDARIN\TRANSLATION.DAT'

list_original = []
list_translation = []

PATTERN_ORI = '<STRING>ORIGINAL:'
PATTERN_TRA = '<STRING>TRANSLATION:'

with open(translation_file, 'r', encoding='utf_16_le') as translation_file_handler:
    translation_block_lines = translation_file_handler.readlines()[1:-1]
    for idx, line in enumerate(translation_block_lines):
        if PATTERN_ORI in line:
            list_original.append(line.strip().replace(PATTERN_ORI, ''))
            if PATTERN_TRA not in translation_block_lines[idx + 1]:
                print("TRANSLATION not found in the following line: " + idx)
        if PATTERN_TRA in line:
            list_translation.append(line.strip().replace(PATTERN_TRA, ''))
            if PATTERN_ORI not in translation_block_lines[idx - 1]:
                print("ORIGINAL not found in the upper line: " + idx)

assert len(list_original) == len(list_translation)

clean_o = remove_duplicates_ordered(list_original)

translation_file_new = 'E:\Program Files (x86)\Steam\steamapps\common\Torchlight II\mods\IMBA-符文觉醒-暗黑传奇\MEDIA\TRANSLATIONS\MANDARIN\TRANSLATION_NEW.DAT'

with open(translation_file_new, 'w', encoding="utf-16") as translation_file_new_handler:
    translation_file_new_handler.write(dat_format('[TRANSLATIONS]', 0))

    for orig in clean_o:
        text_o = orig
        text_t = list_translation[list_original.index(orig)]
        translation_file_new_handler.write(dat_format('[TRANSLATION]', 1))
        translation_file_new_handler.write(dat_format('<STRING>ORIGINAL:' + text_o, 2))
        translation_file_new_handler.write(dat_format('<STRING>TRANSLATION:' + text_t, 2))
        translation_file_new_handler.write(dat_format('[/TRANSLATION]', 1))

    translation_file_new_handler.write(dat_format('[/TRANSLATIONS]', 0))