[jira] [Commented] (SPARK-42118) Wrong result when parsing a multiline JSON file with differing types for same column

["Hyukjin Kwon (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SPARK-42118?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17680419#comment-17680419 ] 

Hyukjin Kwon commented on SPARK-42118:
--------------------------------------

As a workaround you can do:

{code}
val df = spark.read.format("json").option("multiLine", true).load("/tmp/json")
val newDF = spark.createDataFrame(df.rdd, df.schema)
{code}

then 

{code}
df.show(false)
df.count
{code}

will show the consistent output.

> Wrong result when parsing a multiline JSON file with differing types for same column
> ------------------------------------------------------------------------------------
>
>                 Key: SPARK-42118
>                 URL: https://issues.apache.org/jira/browse/SPARK-42118
>             Project: Spark
>          Issue Type: Improvement
>          Components: SQL
>    Affects Versions: 3.2.1
>            Reporter: Dilip Biswal
>            Priority: Major
>
> Here is a simple reproduction of the problem. We have a JSON file whose content looks like following and is in multiLine format.
> {code}
> [{"name":""},{"name":123.34}]
> {code}
> Here is the result of spark query when we read the above content.
> scala> val df = spark.read.format("json").option("multiLine", true).load("/tmp/json")
> df: org.apache.spark.sql.DataFrame = [name: double]
> scala> df.show(false)
> +----+
> |name|
> +----+
> |null|
> +----+
> scala> df.count
> res5: Long = 2
> This is quite a serious problem for us as it's causing us to master corrupt data in lake. If there is some issue with parsing the input, we expect spark set the "_corrupt_record" so that we can act on it. Please note that df.count is reporting 2 rows where as df.show only reports 1 row with null value.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org