是的，你可以 unnest 数组并传递给 partitionBy 条款。 Example: spark scala公司：

df.show()
//+---+----+
//| id|dept|
//+---+----+
//|  1|   a|
//|  2|   a|
//|  3|   b|
//|  4|   c|
//+---+----+

import org.apache.spark.sql.expressions.Window
val cols=df.columns
val w_spec=Window.partitionBy(cols.head,cols.tail:_*)

df.withColumn("cc",avg("id").over(w_spec)).show()
//+---+----+---+
//| id|dept| cc|
//+---+----+---+
//|  2|   a|2.0|
//|  4|   c|4.0|
//|  3|   b|3.0|
//|  1|   a|1.0|
//+---+----+---+
``` `Pyspark:` ```
df.show()

# +---+----+

# | id|dept|

# +---+----+

# |  1|   a|

# |  2|   a|

# |  3|   b|

# |  4|   c|

# +---+----+

from pyspark.sql import window
cols=df.columns
w_spec=Window.partitionBy(*[cols])

df.withColumn("cc",avg("id").over(w_spec)).show()

# +---+----+---+

# | id|dept| cc|

# +---+----+---+

# |  2|   a|2.0|

# |  4|   c|4.0|

# |  1|   a|1.0|

# |  3|   b|3.0|

# +---+----+---+

这是你如何通过，而不是 Arrays.stream(df.columns()) 可以替换为字符串列表（列名）

List<Column> arrayList = Arrays.stream(df.columns())
        .map(functions::col).collect(Collectors.toList());

WindowSpec window = Window.partitionBy(arrayList.toArray(new Column[arrayList.size()]));