from opencc import OpenCC
cc = OpenCC('t2s')  # 't2s'表示繁体转简体
import unicodedata
import json

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

def clearT(s):
    s = cc.convert(fullwidth_to_halfwidth(s))
    return s.lstrip('n').strip().strip(r'\n')
def startwithany(s, l):
    return any(s.startswith(x) for x in l)

f = open(r'E:\tmp\Renpy\script.rpy', encoding='utf-8')
start = False
keys = [
    '$',
    'stop',
    'scene',
    'with',
    'pause',
    'play',
    'show',
    'voice',
    'hide',
    'nvl ',
    '#',
    'yalign',
    'linear',
    'anchor',
    'if',
    'return'
]
_n = {
  "z": "老师",
  "y": "叶雨潇",
  "nvl_narrator": "旁白",
  "l": "梁芷柔",
  "a": "同学A",
  "b": "同学B",
  "c": "其他同学",
  "who": "?",
  "az": "医生",
  "k": "护士",
  "kfc": "服务员",
  "d": "店员",
  "bird": "候鸟",
  "yl": "叶雨潇&梁芷柔",
  "loli": "小梁芷柔",
  "e": "同学们",
  "f": "女同学",
  "g": "同学",
  "ab": "同学A&B",
  "m": "广播",
  "h": "工作人员",
  "i": "保安",
  "j": "李金凡",
  "bird2": "海鸟",
  "phone": "电话"
}
sc = []
b = open(r'D:\datasets\b-corpus\视觉小说\format\transient\transient.txt','w',encoding='utf-8')
for line in f:
    if line.startswith('label start:'):
        start = True
        continue
    if not start:
        continue
    line = line.strip()
    if not line:
        continue
    if startwithany(line, keys):
        continue
    if line.startswith('"'):
        n = '旁白'
    if startwithany(line, _n.keys()) and ' ' in line:
        idx = line.index(' ')
        n = line[:idx]
        n = _n[n]
        line = line[idx:].strip()
    左引号 = line.index('"')
    右引号 = line.rindex('"') + 1
    d = line[左引号:右引号]
    sc.append(n + '：' + clearT(d))
b.write('\n'.join(sc))
f.close()
b.close()
tmp = json.dumps(_n, ensure_ascii=False, indent=2)