Databricks自定义Python UDF报错：charMonth变量未赋值

在Databricks/Python环境中，尝试创建自定义UDF CreateBloombergSymbol，将日期字符串pctym的月份转换为彭博对应字符符号，但运行时抛出错误：UnboundLocalError: local variable 'charMonth' referenced before assignment。

函数定义代码：

from pyspark.sql.functions import col, substr
def CreateBloombergSymbol(pctym):
    digitMonth = pctym[6:2:1]
    match digitMonth:
        case "01":
            charMonth = "F"
        case "02":
            charMonth = "G"    
        case "03":
            charMonth = "H"  
        case "04":
            charMonth = "J"  
        case "05":
            charMonth = "K"  
        case "06":
            charMonth = "M"  
        case "07":
            charMonth = "N"  
        case "08":
            charMonth = "Q"  
        case "09":
            charMonth = "U"  
        case "10":
            charMonth = "V" 
        case "11":
            charMonth = "X" 
        case "12":
            charMonth = "Z"                            
    return charMonth

注册并调用UDF代码：

varMyBloombergFunction = spark.udf.register("CreateBloombergSymbol", CreateBloombergSymbol)
from pyspark.sql.functions import col
dfPos.select(
    col('*'),
    varMyBloombergFunction(col('pctym')).alias('mynewcolFuncBB')
).display()

1. 未覆盖所有分支：match语句只处理了01-12的月份值，当digitMonth是其他值时，charMonth从未被赋值，但函数最后直接返回该变量，触发未绑定局部变量的错误。
2. 切片逻辑错误：pctym[6:2:1]不符合Python切片规则（格式为[起始索引:结束索引:步长]），如果日期字符串是YYYYMMDD格式，正确提取月份的写法应该是pctym[4:6]，当前写法会得到空字符串或错误子串，这也是导致digitMonth不匹配任何case的常见原因。

方案1：完善match分支，添加默认处理 #

给match语句增加默认分支，处理所有非预期的月份值：

from pyspark.sql.functions import col, substr
def CreateBloombergSymbol(pctym):
    # 修正切片逻辑，提取YYYYMMDD格式中的月份部分
    digitMonth = pctym[4:6]
    match digitMonth:
        case "01":
            charMonth = "F"
        case "02":
            charMonth = "G"    
        case "03":
            charMonth = "H"  
        case "04":
            charMonth = "J"  
        case "05":
            charMonth = "K"  
        case "06":
            charMonth = "M"  
        case "07":
            charMonth = "N"  
        case "08":
            charMonth = "Q"  
        case "09":
            charMonth = "U"  
        case "10":
            charMonth = "V" 
        case "11":
            charMonth = "X" 
        case "12":
            charMonth = "Z"
        case _:  # 处理所有非预期情况
            charMonth = ""  # 也可返回"INVALID"等标记值                         
    return charMonth

方案2：用字典映射替代match（更简洁） #

用字典存储月份与彭博符号的对应关系，通过get方法自动处理默认情况：

from pyspark.sql.functions import col, substr
def CreateBloombergSymbol(pctym):
    # 修正切片逻辑
    digitMonth = pctym[4:6]
    month_map = {
        "01": "F", "02": "G", "03": "H", "04": "J",
        "05": "K", "06": "M", "07": "N", "08": "Q",
        "09": "U", "10": "V", "11": "X", "12": "Z"
    }
    # 不存在对应值时返回空字符串
    return month_map.get(digitMonth, "")

额外建议：增加格式校验 #

如果pctym可能存在非标准日期格式，可在函数开头添加校验：

def CreateBloombergSymbol(pctym):
    # 校验日期字符串长度是否符合YYYYMMDD格式
    if len(pctym) != 8:
        return ""
    digitMonth = pctym[4:6]
    month_map = {
        "01": "F", "02": "G", "03": "H", "04": "J",
        "05": "K", "06": "M", "07": "N", "08": "Q",
        "09": "U", "10": "V", "11": "X", "12": "Z"
    }
    return month_map.get(digitMonth, "")

内容的提问来源于stack exchange，提问作者solarissf