[SPARK-34564] DateTimeUtils.fromJavaDate fails for very late dates during casting to Int

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Not A Problem
• Affects Version/s: 3.0.1, 3.1.2, 3.2.0
• Component/s: SQL
• Assignee: Unassigned
• Reporter: kondziolka9ld
• Created: 27/Feb/21 19:56
• Updated: 12/Mar/21 17:59
• Resolved: 12/Mar/21 17:59

Description

Please consider a following scenario on spark-3.0.1: 

scala> List(("some date", new Date(Int.MaxValue)), ("some corner case date", new Date(Long.MaxValue))).toDF

java.lang.RuntimeException: Error while encoding: java.lang.ArithmeticException: integer overflow
staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, knownnotnull(assertnotnull(input[0, scala.Tuple2, true]))._1, true, false) AS _1#0
staticinvoke(class org.apache.spark.sql.catalyst.util.DateTimeUtils$, DateType, fromJavaDate, knownnotnull(assertnotnull(input[0, scala.Tuple2, true]))._2, true, false) AS _2#1
  at org.apache.spark.sql.catalyst.encoders.ExpressionEncoder$Serializer.apply(ExpressionEncoder.scala:215)
  at org.apache.spark.sql.SparkSession.$anonfun$createDataset$1(SparkSession.scala:466)
  at scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:238)
  at scala.collection.immutable.List.foreach(List.scala:392)
  at scala.collection.TraversableLike.map(TraversableLike.scala:238)
  at scala.collection.TraversableLike.map$(TraversableLike.scala:231)
  at scala.collection.immutable.List.map(List.scala:298)
  at org.apache.spark.sql.SparkSession.createDataset(SparkSession.scala:466)
  at org.apache.spark.sql.SQLContext.createDataset(SQLContext.scala:353)
  at org.apache.spark.sql.SQLImplicits.localSeqToDatasetHolder(SQLImplicits.scala:231)
  ... 51 elided
Caused by: java.lang.ArithmeticException: integer overflow
  at java.lang.Math.toIntExact(Math.java:1011)
  at org.apache.spark.sql.catalyst.util.DateTimeUtils$.fromJavaDate(DateTimeUtils.scala:111)
  at org.apache.spark.sql.catalyst.util.DateTimeUtils.fromJavaDate(DateTimeUtils.scala)
  at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
  at org.apache.spark.sql.catalyst.encoders.ExpressionEncoder$Serializer.apply(ExpressionEncoder.scala:211)
  ... 60 more

 In opposition to spark-2.4.7 where it is possible to create dataframe with such values:  

scala> val df = List(("some date", new Date(Int.MaxValue)), ("some corner case date", new Date(Long.MaxValue))).toDF
df: org.apache.spark.sql.DataFrame = [_1: string, _2: date]scala> df.show
+--------------------+-------------+
|                  _1|           _2|
+--------------------+-------------+
|           some date|   1970-01-25|
|some corner case ...|1701498-03-18|
+--------------------+-------------+

Anyway, I am aware of the fact that during collecting these data I will got another result: 

scala> df.collect
res10: Array[org.apache.spark.sql.Row] = Array([some date,1970-01-25], [some corner case date,?498-03-18])

what seems to be natural because of behaviour of java.sql.Date:

scala> new java.sql.Date(Long.MaxValue)
res1: java.sql.Date = ?994-08-17

  

---

When it comes to easier reproduction, please consider: 

scala> org.apache.spark.sql.catalyst.util.DateTimeUtils.fromJavaDate(new java.sql.Date(Long.MaxValue))
java.lang.ArithmeticException: integer overflow
  at java.lang.Math.toIntExact(Math.java:1011)
  at org.apache.spark.sql.catalyst.util.DateTimeUtils$.fromJavaDate(DateTimeUtils.scala:111)
  ... 47 elided

 However, the question is even if such late dates are not supported, could it fail in more gentle way?