1. 目的：数据行专列，列转行
2. 数据准备：

df = spark.createDataFrame([(1,"age","23|45|67|32"),(3,"score","90|91|92|93")]).toDF("id","typ","ls")

列转行：

df2 = df.withColumn("xx",explode(split("ls","\\|"))).drop("ls")
df2.show()

Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
+---+-----+-----------+
| id|  typ|         ls|
+---+-----+-----------+
|  1|  age|23|45|67|32|
|  3|score|90|91|92|93|
+---+-----+-----------+

+---+-----+---+
| id|  typ| xx|
+---+-----+---+
|  1|  age| 23|
|  1|  age| 45|
|  1|  age| 67|
|  1|  age| 32|
|  3|score| 90|
|  3|score| 91|
|  3|score| 92|
|  3|score| 93|
+---+-----+---+

行专列：

df2.groupBy("id","typ").agg(concat_ws("|",collect_set("xx")).alias("ls2")).show()
+---+-----+-----------+
| id|  typ|        ls2|
+---+-----+-----------+
|  3|score|92|93|91|90|
|  1|  age|23|45|67|32|
+---+-----+-----------+

完整代码：

# coding=utf-8
from pyspark.sql import SparkSession
from pyspark.sql.functions import *


if __name__ == "__main__":
    spark = SparkSession.builder.master("local").getOrCreate()
    #数据准备
    df = spark.createDataFrame([(1,"age","23|45|67|32"),(3,"score","90|91|92|93")]).toDF("id","typ","ls")
    df.show()
    #列转行
    df2 = df.withColumn("xx",explode(split("ls","\\|"))).drop("ls")
    df2.show()

    #行转列
    df2.groupBy("id","typ").agg(concat_ws("|",collect_set("xx")).alias("ls2")).show()