1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
|
package org.apache.spark.sql.catalyst
import scala.reflect.macros.blackbox.Context
import scala.reflect.api.Universe
import scala.reflect.ClassTag
import org.apache.spark.sql.Encoder
import org.apache.spark.sql.catalyst.encoders.ExpressionEncoder
//import org.apache.spark.sql.catalyst.{InternalRow, JavaTypeInference, ScalaReflection}
import org.apache.spark.sql.catalyst.analysis.{Analyzer, GetColumnByOrdinal, SimpleAnalyzer, UnresolvedAttribute, UnresolvedExtractValue}
import org.apache.spark.sql.catalyst.expressions._
import org.apache.spark.sql.catalyst.expressions.codegen.{GenerateSafeProjection, GenerateUnsafeProjection}
import org.apache.spark.sql.catalyst.expressions.objects.{AssertNotNull, Invoke, NewInstance}
import org.apache.spark.sql.catalyst.optimizer.SimplifyCasts
import org.apache.spark.sql.catalyst.plans.logical.{CatalystSerde, DeserializeToObject, LocalRelation}
import org.apache.spark.sql.types.{BooleanType, ObjectType, StructField, StructType}
import org.apache.spark.util.Utils
/** Type parameter is required for a workaround described here http://docs.scala-lang.org/overviews/macros/overview.html
* TODO: remove the type parameter once Scala 2.10 is dropped */
class MacrosHelper[C <: Context](val context: C) extends ScalaReflection {
val universe: context.universe.type = context.universe
val mirror: universe.Mirror = context.mirror
import universe._
def generate[T: context.WeakTypeTag]: context.Tree = {
val tag = implicitly[WeakTypeTag[T]]
val tpe = tag.tpe
val flat = !definedByConstructorParams(tpe)
val inputObject = BoundReference(0, dataTypeFor[T](tag), nullable = true)
val nullSafeInput = if (flat) {
inputObject
} else {
// For input object of non-flat type, we can't encode it to row if it's null, as Spark SQL
// doesn't allow top-level row to be null, only its columns can be null.
AssertNotNull(inputObject, Seq("top level non-flat input object"))
}
val serializer = serializerFor[T](nullSafeInput)(tag)
val deserializer = deserializerFor[T](tag)
val schema = schemaFor[T](tag) match {
case Schema(s: StructType, _) => s
case Schema(dt, nullable) => new StructType().add("value", dt, nullable)
}
q"""new _root_.org.apache.spark.sql.catalyst.encoders.ExpressionEncoder[T](
${reify(schema)},
${reify(flat)},
${reify(serializer.flatten)},
${reify(deserializer)},
classTag[$tpe])"""
}
}
import scala.language.experimental.macros
object ScalaMacros {
final def newEncoder[T]: ExpressionEncoder[T] = macro newEncoderImpl[T]
def newEncoderImpl[T: c.WeakTypeTag](c: Context): c.Expr[ExpressionEncoder[T]] = {
val helper = new MacrosHelper[c.type](c)
c.Expr[ExpressionEncoder[T]](helper.generate)
}
}
|
