我建议你分两步来做：首先，您必须过滤重复的值并 rank 它们以给予递增的索引号，然后合并非重复的值，如下所示：

from pyspark.sql.window import Window
from pyspark.sql.functions import col, count, lit, dense_rank 

df = spark.createDataFrame([(0,), (0,), (0,), (0,), (1000231,), (1000236,), (1000363,), (1000363,)], "DUNS_NUMBER: int")

count_window = Window.partitionBy("DUNS_NUMBER")
order_window = Window.partitionBy().orderBy("DUNS_NUMBER")

df_with_count = df.withColumn("dup_COUNT", count(col("DUNS_NUMBER")).over(count_window))

df_with_count \
    .filter("dup_COUNT > 1") \
    .withColumn("dup_DUNS_NUMBER", dense_rank().over(order_window)) \
    .union( \
        df_with_count.filter("dup_COUNT = 1").withColumn("dup_DUNS_NUMBER", lit("NaN"))) \
    .select(["DUNS_NUMBER", "dup_DUNS_NUMBER"]) \
    .orderBy("DUNS_NUMBER") \
    .show()

+-----------+---------------+
|DUNS_NUMBER|dup_DUNS_NUMBER|
+-----------+---------------+
|          0|              1|
|          0|              1|
|          0|              1|
|          0|              1|
|    1000231|            NaN|
|    1000236|            NaN|
|    1000363|              2|
|    1000363|              2|
+-----------+---------------+