import pandas as pd

df = pd.read_parquet(r'E:\tmp\互动小说\0001.parquet')


def clearT():
    import unicodedata
    from opencc import OpenCC

    def full2half(input_str):
        return ''.join([unicodedata.normalize('NFKC', char) for char in input_str])

    cc = OpenCC('t2s')  # 't2s'表示繁体转简体

    def _clearT(s):
        s = cc.convert(full2half(s))
        return s.strip().replace('\n', '\\n')

    return _clearT


clearT = clearT()

import json

z = 2767
for a, b, c, d, e, f, g in df.itertuples(index=False, name=None):
    z += 1
    print(z)
    with open(r'E:\tmp\互动小说\tmp' + f'\\{z:05}_' + a + '.txt', 'w', encoding='utf-8') as file:
        tmp = [a, b, list(c), d, e, f]
        tmp = json.dumps(tmp, ensure_ascii=False)
        file.write(tmp + '\n')
        for h in g:
            file.write('旁白：章节:' + clearT(h['chapter_title']) + '\n')
            m = h['content']
            for n in m:
                if n['role']:
                    file.write(clearT(n['role']) + '：' + clearT(n['content']) + '\n')
                else:
                    file.write(clearT(n['content_tag']) + '：' + clearT(n['content']) + '\n')