parse_nl.py 文件源码

def parse_translation_table(self, table):
        """ Overrides GeneralParser's method.
        :param table: a Tag object. Not necessary a table; can be a div.
        :return: (translation, language_name, language_code)
        """

        # go through all "li" elements in a table
        for li in table.find_all('li'):
            if not isinstance(li, Tag):
                continue
            text = li.get_text().split(':')
            if len(text) < 2:
                continue

            # language name is before ":"
            lang_name = text[0]

            # language code is usually in super script
            lang_code = li.find(class_="trad-sup-code")
            if lang_code:
                lang_code = lang_code.text.strip()[1:-1]
            else:
                lang_code = ""

            # There are two functions that removes parentheses. Not sure which one to use.
            t = remove_parenthesis(text[1])
            trans_list = re.split(COMMA_OR_SEMICOLON, t)
                # each "trans" is: translation <sup>(lang_code)</sup> (transliteration)
                # lang_code and transliteration may not exist
            for trans in trans_list:
                translation = trans.split('(')[0].strip()
                yield (translation, lang_name.strip(), lang_code)