-
Notifications
You must be signed in to change notification settings - Fork 28k
/
UserDefinedTypeSuite.scala
302 lines (254 loc) · 10.7 KB
/
UserDefinedTypeSuite.scala
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.spark.sql
import java.time.Year
import java.util.Arrays
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.catalyst.CatalystTypeConverters
import org.apache.spark.sql.catalyst.expressions.{Cast, CodegenObjectFactoryMode, ExpressionEvalHelper, Literal}
import org.apache.spark.sql.execution.datasources.parquet.ParquetTest
import org.apache.spark.sql.functions._
import org.apache.spark.sql.internal.SQLConf
import org.apache.spark.sql.test.SharedSparkSession
import org.apache.spark.sql.types._
private[sql] case class MyLabeledPoint(label: Double, features: TestUDT.MyDenseVector) {
def getLabel: Double = label
def getFeatures: TestUDT.MyDenseVector = features
}
private[sql] case class FooWithDate(year: Year, s: String, i: Int)
private[sql] class YearUDT extends UserDefinedType[Year] {
override def sqlType: DataType = IntegerType
override def serialize(obj: Year): Int = {
obj.getValue
}
def deserialize(datum: Any): Year = datum match {
case value: Int => Year.of(value)
}
override def userClass: Class[Year] = classOf[Year]
private[spark] override def asNullable: YearUDT = this
}
class UserDefinedTypeSuite extends QueryTest with SharedSparkSession with ParquetTest
with ExpressionEvalHelper {
import testImplicits._
private lazy val pointsRDD = Seq(
MyLabeledPoint(1.0, new TestUDT.MyDenseVector(Array(0.1, 1.0))),
MyLabeledPoint(0.0, new TestUDT.MyDenseVector(Array(0.2, 2.0)))).toDF()
private lazy val pointsRDD2 = Seq(
MyLabeledPoint(1.0, new TestUDT.MyDenseVector(Array(0.1, 1.0))),
MyLabeledPoint(0.0, new TestUDT.MyDenseVector(Array(0.3, 3.0)))).toDF()
test("SPARK-32090: equal") {
val udt1 = new ExampleBaseTypeUDT
val udt2 = new ExampleSubTypeUDT
val udt3 = new ExampleSubTypeUDT
assert(udt1 !== udt2)
assert(udt2 !== udt1)
assert(udt2 === udt3)
assert(udt3 === udt2)
}
test("SPARK-32090: acceptsType") {
val udt1 = new ExampleBaseTypeUDT
val udt2 = new ExampleSubTypeUDT
assert(udt1.acceptsType(udt2))
assert(!udt2.acceptsType(udt1))
}
test("register user type: MyDenseVector for MyLabeledPoint") {
val labels: RDD[Double] = pointsRDD.select($"label").rdd.map { case Row(v: Double) => v }
val labelsArrays: Array[Double] = labels.collect()
assert(labelsArrays.length === 2)
assert(labelsArrays.contains(1.0))
assert(labelsArrays.contains(0.0))
val features: RDD[TestUDT.MyDenseVector] =
pointsRDD.select($"features").rdd.map { case Row(v: TestUDT.MyDenseVector) => v }
val featuresArrays: Array[TestUDT.MyDenseVector] = features.collect()
assert(featuresArrays.length === 2)
assert(featuresArrays.contains(new TestUDT.MyDenseVector(Array(0.1, 1.0))))
assert(featuresArrays.contains(new TestUDT.MyDenseVector(Array(0.2, 2.0))))
}
test("UDTs and UDFs") {
withTempView("points") {
spark.udf.register("testType",
(d: TestUDT.MyDenseVector) => d.isInstanceOf[TestUDT.MyDenseVector])
pointsRDD.createOrReplaceTempView("points")
checkAnswer(
sql("SELECT testType(features) from points"),
Seq(Row(true), Row(true)))
}
}
testStandardAndLegacyModes("UDTs with Parquet") {
withTempPath { dir =>
val path = dir.getCanonicalPath
pointsRDD.write.parquet(path)
checkAnswer(
spark.read.parquet(path),
Seq(
Row(1.0, new TestUDT.MyDenseVector(Array(0.1, 1.0))),
Row(0.0, new TestUDT.MyDenseVector(Array(0.2, 2.0)))))
}
}
testStandardAndLegacyModes("Repartition UDTs with Parquet") {
withTempPath { dir =>
val path = dir.getCanonicalPath
pointsRDD.repartition(1).write.parquet(path)
checkAnswer(
spark.read.parquet(path),
Seq(
Row(1.0, new TestUDT.MyDenseVector(Array(0.1, 1.0))),
Row(0.0, new TestUDT.MyDenseVector(Array(0.2, 2.0)))))
}
}
// Tests to make sure that all operators correctly convert types on the way out.
test("Local UDTs") {
val vec = new TestUDT.MyDenseVector(Array(0.1, 1.0))
val df = Seq((1, vec)).toDF("int", "vec")
assert(vec === df.collect()(0).getAs[TestUDT.MyDenseVector](1))
assert(vec === df.take(1)(0).getAs[TestUDT.MyDenseVector](1))
checkAnswer(df.limit(1).groupBy($"int").agg(first($"vec")), Row(1, vec))
checkAnswer(df.orderBy($"int").limit(1).groupBy($"int")
.agg(first($"vec")), Row(1, vec))
}
test("UDTs with JSON") {
val data = Seq(
"{\"id\":1,\"vec\":[1.1,2.2,3.3,4.4]}",
"{\"id\":2,\"vec\":[2.25,4.5,8.75]}"
)
val schema = StructType(Seq(
StructField("id", IntegerType, false),
StructField("vec", new TestUDT.MyDenseVectorUDT, false)
))
val jsonRDD = spark.read.schema(schema).json(data.toDS())
checkAnswer(
jsonRDD,
Row(1, new TestUDT.MyDenseVector(Array(1.1, 2.2, 3.3, 4.4))) ::
Row(2, new TestUDT.MyDenseVector(Array(2.25, 4.5, 8.75))) ::
Nil
)
}
test("UDTs with JSON and Dataset") {
val data = Seq(
"{\"id\":1,\"vec\":[1.1,2.2,3.3,4.4]}",
"{\"id\":2,\"vec\":[2.25,4.5,8.75]}"
)
val schema = StructType(Seq(
StructField("id", IntegerType, false),
StructField("vec", new TestUDT.MyDenseVectorUDT, false)
))
val jsonDataset = spark.read.schema(schema).json(data.toDS())
.as[(Int, TestUDT.MyDenseVector)]
checkDataset(
jsonDataset,
(1, new TestUDT.MyDenseVector(Array(1.1, 2.2, 3.3, 4.4))),
(2, new TestUDT.MyDenseVector(Array(2.25, 4.5, 8.75)))
)
}
test("SPARK-10472 UserDefinedType.typeName") {
assert(IntegerType.typeName === "integer")
assert(new TestUDT.MyDenseVectorUDT().typeName === "mydensevector")
}
test("Catalyst type converter null handling for UDTs") {
val udt = new TestUDT.MyDenseVectorUDT()
val toScalaConverter = CatalystTypeConverters.createToScalaConverter(udt)
assert(toScalaConverter(null) === null)
val toCatalystConverter = CatalystTypeConverters.createToCatalystConverter(udt)
assert(toCatalystConverter(null) === null)
}
test("SPARK-15658: Analysis exception if Dataset.map returns UDT object") {
// call `collect` to make sure this query can pass analysis.
pointsRDD.as[MyLabeledPoint].map(_.copy(label = 2.0)).collect()
}
test("SPARK-19311: UDFs disregard UDT type hierarchy") {
UDTRegistration.register(classOf[IExampleBaseType].getName,
classOf[ExampleBaseTypeUDT].getName)
UDTRegistration.register(classOf[IExampleSubType].getName,
classOf[ExampleSubTypeUDT].getName)
// UDF that returns a base class object
sqlContext.udf.register("doUDF", (param: Int) => {
new ExampleBaseClass(param)
}: IExampleBaseType)
// UDF that returns a derived class object
sqlContext.udf.register("doSubTypeUDF", (param: Int) => {
new ExampleSubClass(param)
}: IExampleSubType)
// UDF that takes a base class object as parameter
sqlContext.udf.register("doOtherUDF", (obj: IExampleBaseType) => {
obj.field
}: Int)
// this worked already before the fix SPARK-19311:
// return type of doUDF equals parameter type of doOtherUDF
checkAnswer(sql("SELECT doOtherUDF(doUDF(41))"), Row(41) :: Nil)
// this one passes only with the fix SPARK-19311:
// return type of doSubUDF is a subtype of the parameter type of doOtherUDF
checkAnswer(sql("SELECT doOtherUDF(doSubTypeUDF(42))"), Row(42) :: Nil)
}
test("except on UDT") {
checkAnswer(
pointsRDD.except(pointsRDD2),
Seq(Row(0.0, new TestUDT.MyDenseVector(Array(0.2, 2.0)))))
}
test("SPARK-23054 Cast UserDefinedType to string") {
val udt = new TestUDT.MyDenseVectorUDT()
val vector = new TestUDT.MyDenseVector(Array(1.0, 3.0, 5.0, 7.0, 9.0))
val data = udt.serialize(vector)
val ret = Cast(Literal(data, udt), StringType, None)
checkEvaluation(ret, "(1.0, 3.0, 5.0, 7.0, 9.0)")
}
test("SPARK-28497 Can't up cast UserDefinedType to string") {
val udt = new TestUDT.MyDenseVectorUDT()
assert(!Cast.canUpCast(udt, StringType))
}
test("typeof user defined type") {
val schema = new StructType().add("a", new TestUDT.MyDenseVectorUDT())
val data = Arrays.asList(
RowFactory.create(new TestUDT.MyDenseVector(Array(1.0, 3.0, 5.0, 7.0, 9.0))))
checkAnswer(spark.createDataFrame(data, schema).selectExpr("typeof(a)"),
Seq(Row("array<double>")))
}
test("SPARK-30993: UserDefinedType matched to fixed length SQL type shouldn't be corrupted") {
def concatFoo(a: FooWithDate, b: FooWithDate): FooWithDate = {
FooWithDate(b.year, a.s + b.s, a.i)
}
UDTRegistration.register(classOf[Year].getName, classOf[YearUDT].getName)
val year = Year.now()
val inputDS = List(FooWithDate(year, "Foo", 1), FooWithDate(year, "Foo", 3),
FooWithDate(year, "Foo", 3)).toDS()
val agg = inputDS.groupByKey(x => x.i).mapGroups((_, iter) => iter.reduce(concatFoo))
val result = agg.collect()
assert(result.toSet === Set(FooWithDate(year, "FooFoo", 3), FooWithDate(year, "Foo", 1)))
}
test("Test unwrap_udt function") {
val unwrappedFeatures = pointsRDD.select(unwrap_udt(col("features")))
.rdd.map { (row: Row) => row.getAs[Seq[Double]](0).toArray }
val unwrappedFeaturesArrays: Array[Array[Double]] = unwrappedFeatures.collect()
assert(unwrappedFeaturesArrays.length === 2)
java.util.Arrays.equals(unwrappedFeaturesArrays(0), Array(0.1, 1.0))
java.util.Arrays.equals(unwrappedFeaturesArrays(1), Array(0.2, 2.0))
}
test("SPARK-46289: UDT ordering") {
val settings = Seq(
("true", CodegenObjectFactoryMode.CODEGEN_ONLY.toString),
("false", CodegenObjectFactoryMode.NO_CODEGEN.toString))
withTempView("v1") {
pointsRDD.createOrReplaceTempView("v1")
for ((wsSetting, cgSetting) <- settings) {
withSQLConf(SQLConf.WHOLESTAGE_CODEGEN_ENABLED.key -> wsSetting,
SQLConf.CODEGEN_FACTORY_MODE.key -> cgSetting) {
val df = sql("select label from v1 order by features")
checkAnswer(df, Row(1.0) :: Row(0.0) :: Nil)
}
}
}
}
}