0

Here is my JSON

[
    {"string":"string1","int":1,"array":[1,2,3],"dict": {"key": "value1"}},
    {"string":"string2","int":2,"array":[2,4,6],"dict": {"key": "value2"}}
]

Here is my parse code:

val mdf = sparkSession.read.option("multiline", "true").json("multi2.json")
mdf.show(false)

This outputs:

+---------------+---------+--------+----+-------+
|_corrupt_record|array    |dict    |int |string |
+---------------+---------+--------+----+-------+
|[              |null     |null    |null|null   |
|null           |[1, 2, 3]|[value1]|1   |string1|
|null           |[2, 4, 6]|[value2]|2   |string2|
|]              |null     |null    |null|null   |
+---------------+---------+--------+----+-------+

Why do I have a _corrupt_record, everything looks ok? Why does the dict column only give the values and not the keys?

Thanks

Alper t. Turker
  • 34,230
  • 9
  • 83
  • 115
More Than Five
  • 9,959
  • 21
  • 77
  • 127

2 Answers2

1

It is a typo. Option name is multiLine not multiline.

user9592924
  • 11
  • 1
1

"multiLine" option is supported from Spark 2.2.0 onwards.
Contrast it with 2.1.0 documentation

With > 2.2.0, your example code with the data , works.

Regarding the dict column, it will still show the values only but the schema is preserved. You can verify with:

scala> mdf.printSchema
root
 |-- array: array (nullable = true)
 |    |-- element: long (containsNull = true)
 |-- dict: struct (nullable = true)
 |    |-- key: string (nullable = true)
 |-- int: long (nullable = true)
 |-- string: string (nullable = true)

EDIT I realized, much of the info is already here

sujit
  • 2,258
  • 1
  • 15
  • 24