Since Spark 2.3, the queries from raw JSON/CSV files are disallowed when the referenced columns only include the internal corrupt record column

You may try either of these two ways.

Option-1: JSON in single line as answered above by @Avishek Bhattacharya.

Option-2: Add option to read multi line JSON in the code as follows. You could read the nested attribute also as shown below.

val df = spark.read.option("multiline","true").json("C:\\data\\nested-data.json")
df.select("a.b").show()

Here is the output for Option-2.

20/07/29 23:14:35 INFO DAGScheduler: Job 1 finished: show at NestedJsonReader.scala:23, took 0.181579 s
+---+
|  b|
+---+
|  1|
+---+