Skip to content

[SPARK-13056][SQL] map column would throw NPE if value is null #10964

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
adrian-wang wants to merge 7 commits into from
Closed

[SPARK-13056][SQL] map column would throw NPE if value is null #10964

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
5b2d942
map column would throw NPE if value is null
adrian-wang Jan 28, 2016
2244999
address comments
adrian-wang Jan 29, 2016
bf429ca
address comments
adrian-wang Jan 29, 2016
21f29a5
move test
adrian-wang Feb 1, 2016
898cf20
use withTempTable
adrian-wang Feb 1, 2016
d77013f
modify test case
adrian-wang Feb 1, 2016
5b83626
add comment
adrian-wang Feb 1, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 9 additions & 6 deletions ...lyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/complexTypeExtractors.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -218,7 +218,7 @@ case class GetArrayItem(child: Expression, ordinal: Expression)
protected override def nullSafeEval(value: Any, ordinal: Any): Any = {
val baseValue = value.asInstanceOf[ArrayData]
val index = ordinal.asInstanceOf[Number].intValue()
if (index >= baseValue.numElements() || index < 0) {
if (index >= baseValue.numElements() || index < 0 || baseValue.isNullAt(index)) {
null
} else {
baseValue.get(index, dataType)
Copy link
Contributor Author

@adrian-wang adrian-wang Jan 29, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just realized it is ok to let it return null here, without the condition above.

Copy link
Contributor

@cloud-fan cloud-fan Jan 29, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yea, it's ok, but having it makes the interpreted version and codegen version more consistent, which is good.

Expand Down Expand Up @@ -267,6 +267,7 @@ case class GetMapValue(child: Expression, key: Expression)
val map = value.asInstanceOf[MapData]
val length = map.numElements()
val keys = map.keyArray()
val values = map.valueArray()

var i = 0
var found = false
Expand All @@ -278,10 +279,10 @@ case class GetMapValue(child: Expression, key: Expression)
}
}

if (!found) {
if (!found || values.isNullAt(i)) {
null
} else {
map.valueArray().get(i, dataType)
values.get(i, dataType)
}
}

Expand All @@ -291,10 +292,12 @@ case class GetMapValue(child: Expression, key: Expression)
val keys = ctx.freshName("keys")
val found = ctx.freshName("found")
val key = ctx.freshName("key")
val values = ctx.freshName("values")
nullSafeCodeGen(ctx, ev, (eval1, eval2) => {
s"""
final int $length = $eval1.numElements();
final ArrayData $keys = $eval1.keyArray();
final ArrayData $values = $eval1.valueArray();

int $index = 0;
boolean $found = false;
Expand All @@ -307,10 +310,10 @@ case class GetMapValue(child: Expression, key: Expression)
}
}

if ($found) {
${ev.value} = ${ctx.getValue(eval1 + ".valueArray()", dataType, index)};
} else {
if (!$found || $values.isNullAt($index)) {
${ev.isNull} = true;
} else {
${ev.value} = ${ctx.getValue(values, dataType, index)};
}
"""
})
Expand Down
7 changes: 7 additions & 0 deletions sql/core/src/test/scala/org/apache/spark/sql/SQLQuerySuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2056,4 +2056,11 @@ class SQLQuerySuite extends QueryTest with SharedSQLContext {
)
}
}

test("SPARK-13056: Null in map value causes NPE") {
val df = Seq(1 -> Map("abc" -> "somestring", "cba" -> null)).toDF("key", "value")
df.registerTempTable("maptest")
Copy link
Contributor

@cloud-fan cloud-fan Feb 1, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

use withTempTable to delete the table after test.

checkAnswer(sql("SELECT value['abc'] FROM maptest"), Row("somestring"))
checkAnswer(sql("SELECT value['cba'] FROM maptest"), Row(null))
Copy link
Contributor

@tejasapatil tejasapatil Feb 1, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Have you verified if this throws NPE without this fix ? The test runs fine over trunk:

scala> val df = Seq(1 -> Map("abc" -> "somestring", "cba" -> null)).toDF("key", "value")
df: org.apache.spark.sql.DataFrame = [key: int, value: map<string,string>]

scala> df.registerTempTable("maptest")

scala> sqlContext.sql("SELECT value['cba'] FROM maptest").collect()
res28: Array[org.apache.spark.sql.Row] = Array([null])

scala> sqlContext.sql("SELECT value['cba'] FROM maptest").foreach(println)
[null]

Copy link
Contributor

@cloud-fan cloud-fan Feb 1, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we need to use RDD instead of Seq to build the DataFrame, or the local optimization will evaluate it directly, without codegen.

Copy link
Contributor

@tejasapatil tejasapatil Feb 1, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You need to modify the test case:

scala> sqlContext.sql("SELECT value['cba'] FROM maptest WHERE key = 1").collect()
16/01/31 21:22:13 ERROR Executor: Exception in task 15.0 in stage 2.0 (TID 47)
java.lang.NullPointerException
    at org.apache.spark.sql.catalyst.expressions.UnsafeRowWriters$UTF8StringWriter.getSize(UnsafeRowWriters.java:90)
    at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificUnsafeProjection.apply(Unknown Source)
    at org.apache.spark.sql.execution.TungstenProject$$anonfun$3$$anonfun$apply$3.apply(basicOperators.scala:90)
    at org.apache.spark.sql.execution.TungstenProject$$anonfun$3$$anonfun$apply$3.apply(basicOperators.scala:88)
    at scala.collection.Iterator$$anon$11.next(Iterator.scala:328)
    at scala.collection.Iterator$$anon$11.next(Iterator.scala:328)
    at scala.collection.Iterator$class.foreach(Iterator.scala:727)
    at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
    at scala.collection.generic.Growable$class.$plus$plus$eq(Growable.scala:48)
    at scala.collection.mutable.ArrayBuffer.$plus$plus$eq(ArrayBuffer.scala:103)....
....

Copy link
Contributor Author

@adrian-wang adrian-wang Feb 1, 2016

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@tejasapatil OK, actually I used a udf to generate the map and got your exception, later I changed to this following the guide from @cloud-fan but didn't verify it myself. I'll modify the test case here, Thanks!

}
}