如何在pyspark中将列表列表合并为单个列表

Question

在 Spark 数据框中，我有 1 列包含列表列表作为行。我想将字符串列表合并为一个。

INPUT DATAFRAME:
+-------+--------------------+
| name  |friends             |
+-------+--------------------+
| Jim   |[["C","A"]["B","C"]]|
+-------+--------------------+
| Bill  |[["E","A"]["F","L"]]|
+-------+--------------------+
| Kim   |[["C","K"]["L","G"]]| 
+-------+--------------------+

OUTPUT DATAFRAME:  

+-------+--------------------+
| name  |friends             |
+-------+--------------------+
| Jim   |["C","A","B"]       |
+-------+--------------------+
| Bill  |["E","A","F","L"]   |
+-------+--------------------+
| Kim   |["C","K","L","G"]   | 
+-------+--------------------+

我想将列表列表合并为单个列表并删除重复项。提前致谢

Answer 1

我认为您可以依靠组合explode来解构列表并collect_set重建它们：

import pyspark
from pyspark.sql import SparkSession
from pyspark import SparkContext
import pandas as pd
from pyspark.sql import functions as F
from pyspark.sql import Window

sc = SparkContext.getOrCreate()
spark = SparkSession(sc)

columns = ['name', 'friends']
data = [("Jim", [["C","A"], ["B","C"]]), ("Bill", [["E","A"], ["F","L"]]), ("Kim", [["C","K"], ["L","G"]])]
pd_data = pd.DataFrame.from_records(data=data, columns=columns)
spark_data = spark.createDataFrame(pd_data)

first_explode = spark_data.withColumn("first_explode", F.explode((F.col("friends"))))
first_explode.show()

+----+----------------+-------------+
|name|         friends|first_explode|
+----+----------------+-------------+
| Jim|[[C, A], [B, C]]|       [C, A]|
| Jim|[[C, A], [B, C]]|       [B, C]|
|Bill|[[E, A], [F, L]]|       [E, A]|
|Bill|[[E, A], [F, L]]|       [F, L]|
| Kim|[[C, K], [L, G]]|       [C, K]|
| Kim|[[C, K], [L, G]]|       [L, G]|
+----+----------------+-------------+

第一层已解构。现在来说第二个：

second_explode = first_explode.withColumn("second_explode", F.explode(F.col("first_explode")))
second_explode.show()

+----+----------------+-------------+--------------+
|name|         friends|first_explode|second_explode|
+----+----------------+-------------+--------------+
| Jim|[[C, A], [B, C]]|       [C, A]|             C|
| Jim|[[C, A], [B, C]]|       [C, A]|             A|
| Jim|[[C, A], [B, C]]|       [B, C]|             B|
| Jim|[[C, A], [B, C]]|       [B, C]|             C|
|Bill|[[E, A], [F, L]]|       [E, A]|             E|
|Bill|[[E, A], [F, L]]|       [E, A]|             A|
|Bill|[[E, A], [F, L]]|       [F, L]|             F|
|Bill|[[E, A], [F, L]]|       [F, L]|             L|
| Kim|[[C, K], [L, G]]|       [C, K]|             C|
| Kim|[[C, K], [L, G]]|       [C, K]|             K|
| Kim|[[C, K], [L, G]]|       [L, G]|             L|
| Kim|[[C, K], [L, G]]|       [L, G]|             G|
+----+----------------+-------------+--------------+

重建列表，丢弃重复项：

grouped = second_explode.groupBy("name").agg(F.collect_set(F.col("second_explode")).alias("friends"))
grouped.show()

+----+------------+
|name|     friends|
+----+------------+
| Jim|   [C, B, A]|
|Bill|[F, E, A, L]|
| Kim|[K, C, G, L]|
+----+------------+