if __name__ == "__main__":
'''
预处理 并保存
'''
files = os.listdir(RFDIR)
fdf = list(map(info, files))
fdf = pd.DataFrame(data=fdf,
columns=['date', 'rank', 'cid',
'fname']).sort_values(['date', 'rank'])
fdf = fdf[fdf['rank'] <= 15]
fdf['fpath'] = fdf.apply(lambda x: os.path.join(RFDIR, x['fname']), axis=1)
texts = dict()
for date, frame in tqdm(fdf.groupby('date')['fpath']):
text = []
for fpath in frame:
text.extend(Cleaner.preprocess_danmu(fpath))
write(os.path.join(WFDIR, date + '_danmu.txt'), text)
texts[date] = text
# '''
# 预处理后，可直接读取
# '''
# texts = dict()
# files = os.listdir(WFDIR)
# for f in tqdm(files):
# fpath = os.path.join(WFDIR, f)
# date = f.split('_')[0]
# text = read(fpath)
# texts[date] = text
####################################################
