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(" "))

util/string_util.py

@@ -0,0 +1,7 @@
+def blank(string):
+    """
+    判断字符串是否为空或者纯空格
+    :param string: 字符串
+    :return: 字符串是否为空或者纯空格
+    """
+    return not string or string.isspace()

util/util.py

@@ -2,6 +2,8 @@ import logging
 import os
 from datetime import datetime
 
+from util import string_util
+
 
 # 获取yyyy-MM-dd HH:mm:ss格式的当前时间
 def get_default_datetime():
@@ -76,3 +78,130 @@ def zoom_rectangle(rectangle, ratio):
     x2 = round(x2 + x2 * ratio)
     y2 = round(y2 + y2 * ratio)
     return [x1, y1, x2, y2]
+
+
+def chinese_to_money_unit(chinese):
+    if chinese in ["拾", "十"]:
+        return 10, False
+    elif chinese in ["佰", "百"]:
+        return 100, False
+    elif chinese in ["仟", "千"]:
+        return 1000, False
+    elif chinese == "万":
+        return 10000, True
+    elif chinese == "亿":
+        return 100000000, True
+    else:
+        return None, False
+
+
+def chinese_char_to_number(chinese):
+    if chinese == "零":
+        return 0
+    elif chinese in ["一", "壹"]:
+        return 1
+    elif chinese in ["二", "贰"]:
+        return 2
+    elif chinese in ["三", "叁"]:
+        return 3
+    elif chinese in ["四", "肆"]:
+        return 4
+    elif chinese in ["五", "伍"]:
+        return 5
+    elif chinese in ["六", "陆"]:
+        return 6
+    elif chinese in ["七", "柒"]:
+        return 7
+    elif chinese in ["八", "捌"]:
+        return 8
+    elif chinese in ["九", "玖"]:
+        return 9
+    else:
+        return -1
+
+
+def chinese_to_number(chinese):
+    length = len(chinese)
+    result = 0
+    section = 0
+    number = 0
+    unit = [None, False]
+    for i in range(length):
+        c = chinese[i]
+        num = chinese_char_to_number(c)
+        if num >= 0:
+            if num == 0:
+                if number > 0 and unit[0] != None:
+                    section += number * (unit[0] / 10)
+                unit = [None, False]
+            elif number > 0:
+                raise ValueError(f"Bad number '{chinese[i - 1]}{c}' at: {i}")
+            number = num
+        else:
+            unit = chinese_to_money_unit(c)
+            if unit[0] == None:
+                raise ValueError(f"Unknown unit '{c}' at: {i}")
+            if unit[1]:
+                section = (section + number) * unit[0]
+                result += section
+                section = 0
+            else:
+                unitNumber = number
+                if number == 0 and i == 0:
+                    unitNumber = 1
+
+                section += unitNumber * unit[0]
+            number = 0
+
+    if number > 0 and unit[0] != None:
+        number *= unit[0] / 10
+    return result + section + number
+
+
+def chinese_money_to_number(chinese_money_amount):
+    if string_util.blank(chinese_money_amount):
+        return None
+    yi = chinese_money_amount.find("元")
+    if yi == -1:
+        yi = chinese_money_amount.find("圆")
+    ji = chinese_money_amount.find("角")
+    fi = chinese_money_amount.find("分")
+    y_str = None
+    if yi > 0:
+        y_str = chinese_money_amount[0:yi]
+
+    j_str = None
+    if ji > 0:
+        if yi >= 0:
+            if ji > yi:
+                j_str = chinese_money_amount[yi + 1:ji]
+        else:
+            j_str = chinese_money_amount[0:ji]
+
+    f_str = None
+    if fi > 0:
+        if ji >= 0:
+            if fi > ji:
+                f_str = chinese_money_amount[ji + 1:fi]
+        elif yi > 0:
+            if fi > yi:
+                f_str = chinese_money_amount[yi + 1:fi]
+        else:
+            f_str = chinese_money_amount[0: fi]
+
+    y = 0
+    j = 0
+    f = 0
+    if not string_util.blank(y_str):
+        y = chinese_to_number(y_str)
+
+    if not string_util.blank(j_str):
+        j = chinese_to_number(j_str)
+
+    if not string_util.blank(f_str):
+        f = chinese_to_number(f_str)
+
+    amount = y
+    amount += j / 10
+    amount += f / 100
+    return round(amount, 2)