Github user kiszk commented on a diff in the pull request:
https://github.com/apache/spark/pull/13680#discussion_r74038625
--- Diff:
sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/util/UnsafeArraySuite.scala
---
@@ -18,27 +18,131 @@
package org.apache.spark.sql.catalyst.util
import org.apache.spark.SparkFunSuite
+import org.apache.spark.sql.catalyst.encoders.ExpressionEncoder
import org.apache.spark.sql.catalyst.expressions.UnsafeArrayData
+import org.apache.spark.unsafe.Platform
class UnsafeArraySuite extends SparkFunSuite {
- test("from primitive int array") {
- val array = Array(1, 10, 100)
- val unsafe = UnsafeArrayData.fromPrimitiveArray(array)
- assert(unsafe.numElements == 3)
- assert(unsafe.getSizeInBytes == 4 + 4 * 3 + 4 * 3)
- assert(unsafe.getInt(0) == 1)
- assert(unsafe.getInt(1) == 10)
- assert(unsafe.getInt(2) == 100)
+ val booleanArray = Array(false, true)
+ val shortArray = Array(1.toShort, 10.toShort, 100.toShort)
+ val intArray = Array(1, 10, 100)
+ val longArray = Array(1.toLong, 10.toLong, 100.toLong)
+ val floatArray = Array(1.1.toFloat, 2.2.toFloat, 3.3.toFloat)
+ val doubleArray = Array(1.1, 2.2, 3.3)
+ val stringArray = Array("1", "10", "100")
--- End diff --
@davies thanks. I understand what I should do. While I specified the scheme
as follows, the generated code still uses 38 and 18.
When I checked code generation, [this
code](https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/codegen/GenerateUnsafeProjection.scala#L300)
gets data type from serializer instead of schema. If I am correct, [this
code](https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/encoders/ExpressionEncoder.scala#L61)
generates a serializer in `ExpressionEncoder[T]` based on `[T]`, not a schema.
When I replaced
['DecimalType.SYSTEM_DEFAULT'](https://github.com/apache/spark/blob/master/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/ScalaReflection.scala#L511
) with 'DecimalType(4,1)', the generated code uses 4 and 1.
Would it be possible to let me know how to specify specific dataType by
using schema?
```
val decimalArray = Array(BigDecimal("123").setScale(1,
BigDecimal.RoundingMode.FLOOR))
test("read array") {
val unsafeDecimal = ExpressionEncoder[Array[BigDecimal]].copy(schema =
new StructType()
.add("value", ArrayType(DataTypes.createDecimalType(4, 1), true),
true), true)
.resolveAndBind().toRow(decimalArray).getArray(0)
decimalArray.zipWithIndex.map { case (e, i) =>
assert(unsafeDecimal.getDecimal(i, e.precision, e.scale) == e)
}
}
```
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
