java - 将 JavaObject `scala.collection.Map<Object,RDD<?>>` 转换为 python 字典

Question

在pyspark中，调用getPersistentRDDs() Java SparkContext 方法返回 scala.collection.Map<Object,RDD<?>> 的 JavaObject 实例。

from pyspark.sql import SparkSession
from pyspark import StorageLevel

spark = SparkSession.builder.master('yarn').getOrCreate()
sc = spark.sparkContext

df = spark.range(0, 25000000, 1)
df.persist(StorageLevel.MEMORY_ONLY)
df.limit(1).count()

sc._jsc.sc().getPersistentRDDs()

返回JavaObject id=o477

如何转换 scala.collection.Map<Object,RDD<?>> 的 JavaObject进入Python字典？

Answer 1

from pyspark import RDD

scala_map = sc._jsc.sc().getPersistentRDDs()
py_dict = {e._1(): RDD(e._2().toJavaRDD(), sc) for e in [scala_map.toList().apply(i) for i in range(scala_map.size())]}