[python] 把中文按字分，英文按词分

def split_char(str):
	english = 'abcdefghijklmnopqrstuvwxyz0123456789'
	output = []
	buffer = ''
	for s in str:
		if s in english or s in english.upper(): #英文或数字
			buffer += s
		else: #中文
			if buffer:
				output.append(buffer)
			buffer = ''
			output.append(s)
	if buffer:
		output.append(buffer)
	return output

str = '今天是我喜欢Python的第123天'
print(split_char(str))

 

['今','天','是','我','喜','欢','Python','的','第','123','天']