优化费用总额的解析

util/data_util.py

@@ -1,6 +1,9 @@
+import logging
 import re
 from datetime import datetime
 
+from util import util
+
 
 # 处理金额类数据
 def handle_decimal(string):
@@ -8,21 +11,28 @@ def handle_decimal(string):
     if not string:
         return ""
     if "." not in string:
-        front = string
-        back = ""
+        if len(string) > 2:
+            result = string[:-2] + "." + string[-2:]
+        else:
+            result = string
     else:
         front, back = string.rsplit('.', 1)
         front = front.replace(".", "")
+        if back:
+            back = "." + back[:2]
+        result = front + back
+    return result[:16]
 
-    front = front[-16:]
-    if back:
-        back = "." + back
-    result = float(front + back)
-    # 金额较大的暂且交给人工确认
-    if result > 100000:
-        return ""
-    else:
-        return front + back
+
+def parse_money(capital_num, num):
+    if capital_num:
+        try:
+            money = util.chinese_money_to_number(capital_num)
+            return capital_num, money
+        except Exception as e:
+            logging.warning("大写金额解析失败", exc_info=e)
+
+    return num, handle_decimal(num)
 
 
 # 处理日期类数据
@@ -161,7 +171,3 @@ def handle_age(string):
     string = string.split("岁")[0]
     num = re.sub(r'\D', '', string)
     return num[-3:]
-
-
-if __name__ == '__main__':
-    print(handle_decimal(" "))