# 科技類(lèi)文本
# 用阿拉伯?dāng)?shù)字連續(xù)編號(hào)，不同層次的數(shù)字之間加下圓點(diǎn)相隔（即圓點(diǎn)加在數(shù)字的右下角），最后數(shù)字后面不加標(biāo)點(diǎn)，如“1”“1.1”“1.1.1”……
# 人文類(lèi)文本
# 第一層用“一、”，第二層用“（一）”，第三層用“1．”，第四層用“（1）”，第五層用 “ ① ”
class Typesetting():
def __init__(self,text):
self.text = ''.join(text)
def division(self,i,t):
if i < len(self.text)-1:
# A. a.
if t.encode('utf-8').isalpha() == True and self.text[i+1] == '.' :
return True
# A、 a、
elif t.encode('utf-8').isalpha() == True and self.text[i+1] == '、' :
return True
# 1、 一、
elif self.text[i - 1].isnumeric() == False and t.isnumeric() == True and self.text[i + 1] == '、':
return True
# 11、 一十、
elif t.isnumeric() == True and self.text[i+1].isnumeric() == True and self.text[i+2] == '、' :
return True
# 1. 一.
elif self.text[i - 1] != '.' and self.text[i - 1].isnumeric() == False and t.isnumeric() == True and \
self.text[i + 1] == '.' and self.text[i + 2].isnumeric() == False:
return True
# 11. 一十.
elif self.text[i-1] != '.' and t.isnumeric() == True and self.text[i+1].isnumeric() == True and self.text[i+2] == '.' :
return True
# 1.1.1.1
elif t.isnumeric() == True and \
self.text[i + 1] == '.' and \
self.text[i + 2].isnumeric() == True and \
self.text[i + 3] == '.' and \
self.text[i + 4].isnumeric() == True and \
self.text[i + 5] == '.' and \
self.text[i + 6].isnumeric() == True:
return True
# 11.1.1.1
elif t.isnumeric() == True and \
self.text[i + 1].isnumeric() == True and \
self.text[i + 2] == '.' and \
self.text[i + 3].isnumeric() == True and \
self.text[i + 4] == '.' and \
self.text[i + 5].isnumeric() == True and \
self.text[i + 6] == '.' and \
self.text[i + 7].isnumeric() == True:
return True
# 11.11.1.1
elif t.isnumeric() == True and \
self.text[i + 1].isnumeric() == True and \
self.text[i + 2] == '.' and \
self.text[i + 3].isnumeric() == True and \
self.text[i + 4].isnumeric() == True and \
self.text[i + 5] == '.' and \
self.text[i + 6].isnumeric() == True and \
self.text[i + 7] == '.' and \
self.text[i + 8].isnumeric() == True:
return True
# 11.11.11.1
elif t.isnumeric() == True and \
self.text[i + 1].isnumeric() == True and \
self.text[i + 2] == '.' and \
self.text[i + 3].isnumeric() == True and \
self.text[i + 4].isnumeric() == True and \
self.text[i + 5] == '.' and \
self.text[i + 6].isnumeric() == True and \
self.text[i + 7].isnumeric() == True and \
self.text[i + 8] == '.' and \
self.text[i + 9].isnumeric() == True:
return True
# 1.1.1
elif self.text[i-1] != '.' and t.isnumeric() == True and self.text[i+1] == '.' and self.text[i+2].isnumeric() == True and self.text[i+3] == '.' and self.text[i+4].isnumeric() == True :
return True
# 11.1.1
elif self.text[i - 1] != '.' and \
t.isnumeric() == True and \
self.text[i + 1].isnumeric() == True and \
self.text[i + 2] == '.' and \
self.text[i + 3].isnumeric() == True and \
self.text[i + 4] == '.' and \
self.text[i + 5].isnumeric() == True:
return True
# 11.11.1
elif self.text[i - 1] != '.' and \
t.isnumeric() == True and \
self.text[i + 1].isnumeric() == True and \
self.text[i + 2] == '.' and \
self.text[i + 3].isnumeric() == True and \
self.text[i + 4].isnumeric() == True and \
self.text[i + 5] == '.' and \
self.text[i + 6].isnumeric() == True:
return True
# 1.1
elif (self.text[i-1] == '。' or self.text[i-1] == '，') and self.text[i-1] != '.' and t.isnumeric() == True and self.text[i+1] == '.' and self.text[i+2].isnumeric() == True :
return True
# 11.1
elif (self.text[i - 1] == '。' or self.text[i - 1] == '，') and self.text[i - 1] != '.' and t.isnumeric() == True and self.text[i + 2].isnumeric() == True and self.text[i + 3] == '.' and self.text[i + 4].isnumeric() == True:
return True
# （1） （一）
elif t == '（' and self.text[i+1].isnumeric() == True and self.text[i+2] == '）' :
return True
# （11） （一十）
elif t == '（' and self.text[i + 1].isnumeric() == True and self.text[i + 2].isnumeric() and self.text[i + 3] == '）':
return True
# ① 暫未解決 無(wú)法用isnumeric()函數(shù) 雖然對(duì)此符號(hào)可識(shí)別但是也對(duì)其他數(shù)字識(shí)別所以會(huì)造成混亂
elif t in ['①','②','③','④','⑤','⑥','⑦','⑧','⑨','⑩',
'⑴','⑵','⑶','⑷','⑸','⑹','⑺','⑻','⑼','⑽','⑾','⑿','⒀','⒁','⒂','⒃','⒄','⒅','⒆','⒇',
'⒈', '⒉', '⒊', '⒋', '⒌', '⒍', '⒎', '⒏', '⒐', '⒑', '⒒', '⒓', '⒔', '⒕', '⒖', '⒗', '⒘', '⒙', '⒚', '⒛',
] :
return True
# 第1章 第一章
elif t == '第' and self.text[i+1].isnumeric() == True and self.text[i+2] == '章' :
return True
# 第11章 第一十章
elif t == '第' and self.text[i + 1].isnumeric() == True and self.text[i + 2].isnumeric() == True and self.text[i + 3] == '章':
return True
# 第1節(jié) 第一節(jié)
elif t == '第' and self.text[i+1].isnumeric() == True and self.text[i+2] == '節(jié)' :
return True
# 第11節(jié) 第一十節(jié)
elif t == '第' and self.text[i + 1].isnumeric() == True and self.text[i + 2].isnumeric() == True and self.text[i + 3] == '節(jié)':
return True
# 1）二）
elif self.text[i-1] != '（' and t.isnumeric() == True and self.text[i+1] == '）':
return True
# 11）二十）
elif self.text[i - 1] != '（' and t.isnumeric() == True and self.text[i + 1].isnumeric() == True and self.text[i + 2] == '）':
return True
# 1）、 一）、
elif t.isnumeric() == True and self.text[i+1] == '）' and self.text[i+2] == '、':
return True
# 11）、 一十）、
elif t.isnumeric() == True and self.text[i + 1] == '）' and self.text[i + 2].isnumeric() == True and self.text[i + 3] == '、':
return True
def start(self):
str = ''
for i,t in enumerate(self.text):
if self.division(i,t) == True :
str = '%s\n%s' % (str,t)
# print(str)
else:
str = '%s%s' % (str, t)
return str
text = open('text.txt','r',encoding='utf-8').read()
#調(diào)用編譯模塊
a = Typesetting(text).start()
print(a)
# text = open('text.txt','w',encoding='utf-8')
# text.write(a)
浙公網(wǎng)安備 33010602011771號(hào)