[
https://issues.apache.org/jira/browse/SPARK-14231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15215577#comment-15215577
]
Hyukjin Kwon commented on SPARK-14231:
--------------------------------------
Sorry for adding many comment but maybe would this better just infers them as
{{DoubleType}}?
I just noticed JSON data source produces {{DoubleType}} when it does not fix in
a decimal during trying to find a compatible type.
If {{prefersDecimal}} does not infer all floating types must be
{{DecimalTypes}} then, I feel it might be okay to make this {{DoubleType}} when
it does not fit..
> JSON data source fails to infer floats as decimal when precision is bigger
> than 38 or scale is bigger than precision.
> ---------------------------------------------------------------------------------------------------------------------
>
> Key: SPARK-14231
> URL: https://issues.apache.org/jira/browse/SPARK-14231
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Reporter: Hyukjin Kwon
> Priority: Minor
> Fix For: 2.0.0
>
>
> Currently, JSON data source supports {{floatAsBigDecimal}} option, which
> reads floats as {{DecimalType}}.
> I noticed there are several restrictions in Spark {{DecimalType}} below:
> 1. The precision cannot be bigger than 38.
> 2. scale cannot be bigger than precision.
> However, with the option above, it reads {{BigDecimal}} which does not follow
> the conditions above.
> This could be observed as below:
> {code}
> def simpleFloats: RDD[String] =
> sqlContext.sparkContext.parallelize(
> """{"a": 0.01}""" ::
> """{"a": 0.02}""" :: Nil)
> val jsonDF = sqlContext.read
> .option("floatAsBigDecimal", "true")
> .json(simpleFloats)
> jsonDF.printSchema()
> {code}
> throws an exception below:
> {code}
> org.apache.spark.sql.AnalysisException: Decimal scale (2) cannot be greater
> than precision (1).;
> at org.apache.spark.sql.types.DecimalType.<init>(DecimalType.scala:44)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$.org$apache$spark$sql$execution$datasources$json$InferSchema$$inferField(InferSchema.scala:144)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$.org$apache$spark$sql$execution$datasources$json$InferSchema$$inferField(InferSchema.scala:108)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$$anonfun$1$$anonfun$apply$1$$anonfun$apply$3.apply(InferSchema.scala:59)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$$anonfun$1$$anonfun$apply$1$$anonfun$apply$3.apply(InferSchema.scala:57)
> at org.apache.spark.util.Utils$.tryWithResource(Utils.scala:2249)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$$anonfun$1$$anonfun$apply$1.apply(InferSchema.scala:57)
> at
> org.apache.spark.sql.execution.datasources.json.InferSchema$$anonfun$1$$anonfun$apply$1.apply(InferSchema.scala:55)
> at scala.collection.Iterator$$anon$12.hasNext(Iterator.scala:396)
> at scala.collection.Iterator$class.foreach(Iterator.scala:742)
> ...
> {code}
> Since JSON data source infers {{DataType}} as {{StringType}} when it fails to
> infer, it might have to be inferred as {{StringType}} or maybe just simply
> {{DoubleType}}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
