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
|
package spark
import scala.collection.mutable.HashMap
import org.scalatest.FunSuite
import spark.SparkContext._
import spark.rdd.{CoalescedRDD, PartitionPruningRDD}
class RDDSuite extends FunSuite with LocalSparkContext {
test("basic operations") {
sc = new SparkContext("local", "test")
val nums = sc.makeRDD(Array(1, 2, 3, 4), 2)
assert(nums.collect().toList === List(1, 2, 3, 4))
val dups = sc.makeRDD(Array(1, 1, 2, 2, 3, 3, 4, 4), 2)
assert(dups.distinct().count() === 4)
assert(dups.distinct.count === 4) // Can distinct and count be called without parentheses?
assert(dups.distinct.collect === dups.distinct().collect)
assert(dups.distinct(2).collect === dups.distinct().collect)
assert(nums.reduce(_ + _) === 10)
assert(nums.fold(0)(_ + _) === 10)
assert(nums.map(_.toString).collect().toList === List("1", "2", "3", "4"))
assert(nums.filter(_ > 2).collect().toList === List(3, 4))
assert(nums.flatMap(x => 1 to x).collect().toList === List(1, 1, 2, 1, 2, 3, 1, 2, 3, 4))
assert(nums.union(nums).collect().toList === List(1, 2, 3, 4, 1, 2, 3, 4))
assert(nums.glom().map(_.toList).collect().toList === List(List(1, 2), List(3, 4)))
assert(nums.collect({ case i if i >= 3 => i.toString }).collect().toList === List("3", "4"))
assert(nums.keyBy(_.toString).collect().toList === List(("1", 1), ("2", 2), ("3", 3), ("4", 4)))
val partitionSums = nums.mapPartitions(iter => Iterator(iter.reduceLeft(_ + _)))
assert(partitionSums.collect().toList === List(3, 7))
val partitionSumsWithSplit = nums.mapPartitionsWithSplit {
case(split, iter) => Iterator((split, iter.reduceLeft(_ + _)))
}
assert(partitionSumsWithSplit.collect().toList === List((0, 3), (1, 7)))
val partitionSumsWithIndex = nums.mapPartitionsWithIndex {
case(split, iter) => Iterator((split, iter.reduceLeft(_ + _)))
}
assert(partitionSumsWithIndex.collect().toList === List((0, 3), (1, 7)))
intercept[UnsupportedOperationException] {
nums.filter(_ > 5).reduce(_ + _)
}
}
test("SparkContext.union") {
sc = new SparkContext("local", "test")
val nums = sc.makeRDD(Array(1, 2, 3, 4), 2)
assert(sc.union(nums).collect().toList === List(1, 2, 3, 4))
assert(sc.union(nums, nums).collect().toList === List(1, 2, 3, 4, 1, 2, 3, 4))
assert(sc.union(Seq(nums)).collect().toList === List(1, 2, 3, 4))
assert(sc.union(Seq(nums, nums)).collect().toList === List(1, 2, 3, 4, 1, 2, 3, 4))
}
test("aggregate") {
sc = new SparkContext("local", "test")
val pairs = sc.makeRDD(Array(("a", 1), ("b", 2), ("a", 2), ("c", 5), ("a", 3)))
type StringMap = HashMap[String, Int]
val emptyMap = new StringMap {
override def default(key: String): Int = 0
}
val mergeElement: (StringMap, (String, Int)) => StringMap = (map, pair) => {
map(pair._1) += pair._2
map
}
val mergeMaps: (StringMap, StringMap) => StringMap = (map1, map2) => {
for ((key, value) <- map2) {
map1(key) += value
}
map1
}
val result = pairs.aggregate(emptyMap)(mergeElement, mergeMaps)
assert(result.toSet === Set(("a", 6), ("b", 2), ("c", 5)))
}
test("basic checkpointing") {
import java.io.File
val checkpointDir = File.createTempFile("temp", "")
checkpointDir.delete()
sc = new SparkContext("local", "test")
sc.setCheckpointDir(checkpointDir.toString)
val parCollection = sc.makeRDD(1 to 4)
val flatMappedRDD = parCollection.flatMap(x => 1 to x)
flatMappedRDD.checkpoint()
assert(flatMappedRDD.dependencies.head.rdd == parCollection)
val result = flatMappedRDD.collect()
Thread.sleep(1000)
assert(flatMappedRDD.dependencies.head.rdd != parCollection)
assert(flatMappedRDD.collect() === result)
checkpointDir.deleteOnExit()
}
test("basic caching") {
sc = new SparkContext("local", "test")
val rdd = sc.makeRDD(Array(1, 2, 3, 4), 2).cache()
assert(rdd.collect().toList === List(1, 2, 3, 4))
assert(rdd.collect().toList === List(1, 2, 3, 4))
assert(rdd.collect().toList === List(1, 2, 3, 4))
}
test("caching with failures") {
sc = new SparkContext("local", "test")
val onlySplit = new Partition { override def index: Int = 0 }
var shouldFail = true
val rdd = new RDD[Int](sc, Nil) {
override def getPartitions: Array[Partition] = Array(onlySplit)
override val getDependencies = List[Dependency[_]]()
override def compute(split: Partition, context: TaskContext): Iterator[Int] = {
if (shouldFail) {
throw new Exception("injected failure")
} else {
return Array(1, 2, 3, 4).iterator
}
}
}.cache()
val thrown = intercept[Exception]{
rdd.collect()
}
assert(thrown.getMessage.contains("injected failure"))
shouldFail = false
assert(rdd.collect().toList === List(1, 2, 3, 4))
}
test("coalesced RDDs") {
sc = new SparkContext("local", "test")
val data = sc.parallelize(1 to 10, 10)
val coalesced1 = data.coalesce(2)
assert(coalesced1.collect().toList === (1 to 10).toList)
assert(coalesced1.glom().collect().map(_.toList).toList ===
List(List(1, 2, 3, 4, 5), List(6, 7, 8, 9, 10)))
// Check that the narrow dependency is also specified correctly
assert(coalesced1.dependencies.head.asInstanceOf[NarrowDependency[_]].getParents(0).toList ===
List(0, 1, 2, 3, 4))
assert(coalesced1.dependencies.head.asInstanceOf[NarrowDependency[_]].getParents(1).toList ===
List(5, 6, 7, 8, 9))
val coalesced2 = data.coalesce(3)
assert(coalesced2.collect().toList === (1 to 10).toList)
assert(coalesced2.glom().collect().map(_.toList).toList ===
List(List(1, 2, 3), List(4, 5, 6), List(7, 8, 9, 10)))
val coalesced3 = data.coalesce(10)
assert(coalesced3.collect().toList === (1 to 10).toList)
assert(coalesced3.glom().collect().map(_.toList).toList ===
(1 to 10).map(x => List(x)).toList)
// If we try to coalesce into more partitions than the original RDD, it should just
// keep the original number of partitions.
val coalesced4 = data.coalesce(20)
assert(coalesced4.collect().toList === (1 to 10).toList)
assert(coalesced4.glom().collect().map(_.toList).toList ===
(1 to 10).map(x => List(x)).toList)
}
test("zipped RDDs") {
sc = new SparkContext("local", "test")
val nums = sc.makeRDD(Array(1, 2, 3, 4), 2)
val zipped = nums.zip(nums.map(_ + 1.0))
assert(zipped.glom().map(_.toList).collect().toList ===
List(List((1, 2.0), (2, 3.0)), List((3, 4.0), (4, 5.0))))
intercept[IllegalArgumentException] {
nums.zip(sc.parallelize(1 to 4, 1)).collect()
}
}
test("partition pruning") {
sc = new SparkContext("local", "test")
val data = sc.parallelize(1 to 10, 10)
// Note that split number starts from 0, so > 8 means only 10th partition left.
val prunedRdd = new PartitionPruningRDD(data, splitNum => splitNum > 8)
assert(prunedRdd.partitions.size === 1)
val prunedData = prunedRdd.collect()
assert(prunedData.size === 1)
assert(prunedData(0) === 10)
}
}
|
