增加了auto切分

This commit is contained in:
XTer 2024-03-10 21:17:27 +08:00
parent ed3abd5fd5
commit c89f4511da
2 changed files with 34 additions and 1 deletions

View File

@ -118,6 +118,39 @@ def cut5(inp):
opt = "\n".join(mergeitems)
return opt
@register_method("auto_cut")
def auto_cut(inp):
# if not re.search(r'[^\w\s]', inp[-1]):
# inp += '。'
inp = inp.strip("\n")
split_punds = r'[?!。?!~]'
if inp[-1] not in split_punds:
inp+=""
items = re.split(f'({split_punds})', inp)
items = ["".join(group) for group in zip(items[::2], items[1::2])]
def process_commas(text):
separators = ['', ',', '', '——', '']
count = 0
processed_text = ""
for char in text:
processed_text += char
if char in separators:
if count > 12:
processed_text += '\n'
count = 0
else:
count += 1 # 对于非分隔符字符,增加计数
return processed_text
final_items=[process_commas(item) for item in items]
final_items = [item for item in final_items if item.strip()]
return "\n".join(final_items)
if __name__ == '__main__':

@ -1 +1 @@
Subproject commit 39118a4e07c66690e33be6ec8a90652b7dd2a432
Subproject commit 5c8314db790f57c488c2d69f129c475440879ba6