用于数组的pyspark输入读取模式和用于读取json的结构

from pyspark import SparkContext

sc = SparkContext()

# 读取数组
array_rdd = sc.wholeTextFiles("path/to/directory")

# 打印数组内容
for file_path, content in array_rdd.collect():
    print("File:", file_path)
    print("Content:", content)

from pyspark.sql import SparkSession

spark = SparkSession.builder.getOrCreate()

# 读取JSON文件
json_df = spark.read.json("path/to/json/file")

# 显示DataFrame内容
json_df.show()