PySpark数据帧将异常字符串格式转换为时间戳

from pyspark.sql import SparkSession

spark = SparkSession.builder.appName("example").getOrCreate()

data = [("2023-04-31 12:34:56",), ("2023-05-01 09:10:11",), ("invalid_date",)]
columns = ["date_string"]

df = spark.createDataFrame(data, columns)
df.show()

+-------------------+
|         date_string|
+-------------------+
|2023-04-31 12:34:56|
|2023-05-01 09:10:11|
|       invalid_date|
+-------------------+

from pyspark.sql.functions import to_timestamp, col, when, lit

# 定义一个函数来处理无效日期
def handle_invalid_date(date_str):
    try:
        return to_timestamp(date_str, "yyyy-MM-dd HH:mm:ss")
    except:
        return lit(None)

# 注册 UDF
handle_invalid_date_udf = spark.udf.register("handle_invalid_date", handle_invalid_date)

# 使用 UDF 转换日期
df = df.withColumn("timestamp", handle_invalid_date_udf(col("date_string")))

df.show()

+-------------------+----------+
|         date_string| timestamp|
+-------------------+----------+
|2023-04-31 12:34:56|      null|
|2023-05-01 09:10:11|2023-05-01 09:10:11|
|       invalid_date|      null|
+-------------------+----------+