Python实现中文大写金额转阿拉伯数字

def chinese_to_number(chinese_str):
    result = 0
    temp = 0
    unit = 1
    for char in chinese_str:
        if char in chinese_to_arabic:
            temp += chinese_to_arabic[char] * unit
            if unit < 10: 
                unit = 1  # 重置单位为个位
        elif char in unit_map:
            unit = unit_map[char]
            temp = 1 if temp == 0 else temp  # 处理"拾"等单位前无数字的情况
            result += temp * unit
            temp = 0
            unit = 1
    return result + temp

def extract_parts_yi_wan(s):
    result = {"亿": "", "万": "", "一": ""}
    if "亿" in s:
        yi_index = s.index("亿")
        result["亿"] = s[:yi_index]
        after_yi = s[yi_index+1:]
        if "万" in after_yi:
            wan_index = after_yi.index("万")
            result["万"] = after_yi[:wan_index]
            result["一"] = after_yi[wan_index+1:]
        else:
            result["一"] = after_yi
    elif "万" in s:
        wan_index = s.index("万")
        result["万"] = s[:wan_index]
        result["一"] = s[wan_index+1:]
    else:
        result["一"] = s
    return result

def convert_amounts(amount_list):
    pattern = re.compile(
        r'([零壹贰叁肆伍陆柒捌玖拾佰仟万亿]+)元?'
        r'((?:零|[壹贰叁肆伍陆柒捌玖]+)角)?'
        r'((?:零|[壹贰叁肆伍陆柒捌玖]+)分)?'
    )
    converted = []
    for amount in amount_list:
        match = pattern.findall(amount)
        if not match:
            converted.append("无法转换")
            continue
            
        yuan, jiao, fen = match[0]
        parts = extract_parts_yi_wan(yuan)
        total = 0
        
        # 处理亿、万、个位部分
        for unit, value in parts.items():
            if not value:
                continue
            num = chinese_to_number(value)
            if unit == "亿":
                num *= 100000000
            elif unit == "万":
                num *= 10000
            total += num
        
        # 处理角分
        if jiao and jiao != '零角':
            total += chinese_to_arabic[jiao[:-1]] * 0.1
        if fen and fen != '零分':
            total += chinese_to_arabic[fen[:-1]] * 0.01
        
        converted.append(f"{total:.2f}")
    return converted

amounts = [
    '贰拾捌万壹仟柒佰伍拾伍元壹角玖分',
    '贰拾伍万捌仟肆佰玖拾壹元'
]
print(convert_amounts(amounts)) 

import re

# 定义中文数字到阿拉伯数字的映射
chinese_to_arabic = {
    '零': 0, '壹': 1, '贰': 2, '叁': 3, '肆': 4,
    '伍': 5, '陆': 6, '柒': 7, '捌': 8, '玖': 9,
}

unit_map = {'拾': 10, '佰': 100, '仟': 1000}


def chinese_to_number(chinese_str):
    result = 0
    temp_result = 0
    unit = 1
    for char in chinese_str:
        if char in chinese_to_arabic:
            temp_result += chinese_to_arabic[char] * unit
            if unit < 10:  # 当前处理的是个位数，需要重置unit
                unit = 1
        elif char in unit_map:
            unit = unit_map[char]
            temp_result = 1 if temp_result == 0 else temp_result
            result += temp_result * unit
            temp_result = 0
            unit = 1
        elif char in ['元', '角', '分']:
            break
    result += temp_result
    return result


def extract_parts_yi_wan(s):
    # 初始化结果字典
    result = {"亿": "", "万": "", "一": ""}

    # 检查字符串是否包含“亿”
    if "亿" in s:
        yi_index = s.index("亿")
        result["亿"] = s[:yi_index]

        # 检查亿后面的部分是否包含“万”
        after_yi = s[yi_index + 1:]
        if "万" in after_yi:
            wan_index = after_yi.index("万")
            result["万"] = after_yi[:wan_index]
            result["一"] = after_yi[wan_index + 1:]
        else:
            # 如果没有“万”，则亿与万中间的字符串为空，万后面的字符串为亿后面的所有内容
            result["万"] = ""
            result["一"] = after_yi
    else:
        # 如果没有“亿”，检查是否包含“万”
        if "万" in s:
            wan_index = s.index("万")
            result["一"] = s[wan_index + 1:]
            result["万"] = s[:wan_index]
        else:
            # 如果没有“万”，则万后面的字符串为空，亿前面的字符串为整个字符串
            result["一"] = s
            result["万"] = ""

    return result


def convert_amounts(amount_list):
    pattern = re.compile(
        r'([壹贰叁肆伍陆柒捌玖零拾佰仟万亿]+)元?((?:零|[壹贰叁肆伍陆柒捌玖]+)角)?((?:零|[壹贰叁肆伍陆柒捌玖]+)分)?')

    converted_list = []
    for amount in amount_list:
        number = 0
        match = pattern.findall(amount)
        if match:
            yuan, jiao, fen = match[0]
            ret_yuan = extract_parts_yi_wan(yuan)
            for k, v in ret_yuan.items():
                new_number = 0
                if v:
                    new_number = chinese_to_number(v)
                    if k == "亿":
                        new_number *= 100000000
                    elif k == "万":
                        new_number *= 10000
                number += new_number

            if jiao and jiao != '零角':
                number += chinese_to_arabic[jiao.replace('角', '')] * 0.1
            if fen and fen != '零分':
                number += chinese_to_arabic[fen.replace('分', '')] * 0.01

            converted_list.append(f"{number:.2f}")
        else:
            converted_list.append("无法转换")
    return converted_list


amounts = ['贰拾捌万壹仟柒佰伍拾伍元壹角玖分', '贰拾伍万捌仟肆佰玖拾壹元']
converted_amounts = convert_amounts(amounts)
print(converted_amounts)