83 lines
2.6 KiB
Python
83 lines
2.6 KiB
Python
import re
|
||
from datetime import datetime
|
||
|
||
|
||
# 处理金额类数据
|
||
def handle_decimal(string):
|
||
if not string:
|
||
return ""
|
||
string = re.sub(r'[^0-9.]', '', string)
|
||
if "." not in string:
|
||
return string
|
||
front, back = string.rsplit('.', 1)
|
||
front = front.replace(".", "")
|
||
front = front[-16:]
|
||
if back:
|
||
back = "." + back
|
||
return front + back
|
||
|
||
|
||
# 处理日期类数据
|
||
def handle_date(string):
|
||
if not string:
|
||
return ""
|
||
|
||
string = string.replace("年", "-").replace("月", "-").replace("日", "").replace("/", "-").replace(".", "-")
|
||
string = re.sub(r'[^0-9-]', '', string)
|
||
length = len(string)
|
||
if length > 8 and "-" not in string:
|
||
string = string[:8]
|
||
elif length > 10 and "-" in string:
|
||
string = string[:10]
|
||
if is_valid_date_format(string):
|
||
return string
|
||
else:
|
||
return ""
|
||
|
||
|
||
# 判断是否是合法的日期格式
|
||
def is_valid_date_format(date_str):
|
||
if len(date_str) < 6:
|
||
return False
|
||
|
||
# 定义可能的日期格式
|
||
formats = [
|
||
# yyyy-MM-dd
|
||
'%Y-%m-%d',
|
||
# yy-MM-dd
|
||
'%y-%m-%d',
|
||
# yyyyMMdd
|
||
'%Y%m%d',
|
||
# yyMMdd
|
||
'%y%m%d',
|
||
]
|
||
|
||
# 遍历所有格式，尝试解析日期
|
||
for fmt in formats:
|
||
try:
|
||
datetime.strptime(date_str, fmt)
|
||
return True
|
||
except ValueError:
|
||
pass
|
||
|
||
return False
|
||
|
||
|
||
def handle_department(string):
|
||
result = []
|
||
if not string:
|
||
return result
|
||
result.append(string)
|
||
string_without_num = re.sub(r'\d|一|二|三|四|五|六|七|八|九|十', '', string)
|
||
if string_without_num != string:
|
||
result.append(string_without_num)
|
||
string_without_brackets = re.sub(r'\([^()]*\)|\[[^\[\]]*\]|\{[^\{\}]*\}|（[^（）]*）', "", string_without_num)
|
||
if string_without_brackets != string_without_num:
|
||
result.append(string_without_brackets)
|
||
pure_string = string_without_brackets.split("科")[0] + "科"
|
||
if pure_string != string_without_brackets:
|
||
result.append(pure_string)
|
||
return result
|
||
|
||
if __name__ == '__main__':
|
||
print(handle_decimal("~202312167,214.83金额（元）1,920.008,888.38740.00交三医专用章广东省医疗压院收票据（电子）政策性游其他支付：0.00医保类型：跨省异地预缴金额：11564.15备注病历号：2165996治疗费手术费护理费中成药费项目名称政电电子“东省结算方式：个人账户支付：0.00医保编号：补缴金额：0.00住院号：216599612,971.56金额（元）7,483.062,192.0211.96复核人：（小写）68,214.38")) |