aboutsummaryrefslogtreecommitdiff
path: root/core/src/test/scala/spark/FileSuite.scala
blob: e61ff7793d20cd343839f1774fa0079ca3c515a6 (plain) (blame)
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
package spark

import java.io.{FileWriter, PrintWriter, File}

import scala.io.Source

import com.google.common.io.Files
import org.scalatest.FunSuite
import org.apache.hadoop.io._
import org.apache.hadoop.io.compress.{DefaultCodec, CompressionCodec, GzipCodec}


import SparkContext._

class FileSuite extends FunSuite with LocalSparkContext {

  test("text files") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 4)
    nums.saveAsTextFile(outputDir)
    // Read the plain text file and check it's OK
    val outputFile = new File(outputDir, "part-00000")
    val content = Source.fromFile(outputFile).mkString
    assert(content === "1\n2\n3\n4\n")
    // Also try reading it in as a text file RDD
    assert(sc.textFile(outputDir).collect().toList === List("1", "2", "3", "4"))
  }

  test("text files (compressed)") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val normalDir = new File(tempDir, "output_normal").getAbsolutePath
    val compressedOutputDir = new File(tempDir, "output_compressed").getAbsolutePath
    val codec = new DefaultCodec()

    val data = sc.parallelize("a" * 10000, 1)
    data.saveAsTextFile(normalDir)
    data.saveAsTextFile(compressedOutputDir, classOf[DefaultCodec])

    val normalFile = new File(normalDir, "part-00000")
    val normalContent = sc.textFile(normalDir).collect
    assert(normalContent === Array.fill(10000)("a"))

    val compressedFile = new File(compressedOutputDir, "part-00000" + codec.getDefaultExtension)
    val compressedContent = sc.textFile(compressedOutputDir).collect
    assert(compressedContent === Array.fill(10000)("a"))

    assert(compressedFile.length < normalFile.length)
  }

  test("SequenceFiles") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (x, "a" * x)) // (1,a), (2,aa), (3,aaa)
    nums.saveAsSequenceFile(outputDir)
    // Try reading the output back as a SequenceFile
    val output = sc.sequenceFile[IntWritable, Text](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("SequenceFile (compressed)") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val normalDir = new File(tempDir, "output_normal").getAbsolutePath
    val compressedOutputDir = new File(tempDir, "output_compressed").getAbsolutePath
    val codec = new DefaultCodec()

    val data = sc.parallelize(Seq.fill(100)("abc"), 1).map(x => (x, x))
    data.saveAsSequenceFile(normalDir)
    data.saveAsSequenceFile(compressedOutputDir, Some(classOf[DefaultCodec]))

    val normalFile = new File(normalDir, "part-00000")
    val normalContent = sc.sequenceFile[String, String](normalDir).collect
    assert(normalContent === Array.fill(100)("abc", "abc"))

    val compressedFile = new File(compressedOutputDir, "part-00000" + codec.getDefaultExtension)
    val compressedContent = sc.sequenceFile[String, String](compressedOutputDir).collect
    assert(compressedContent === Array.fill(100)("abc", "abc"))

    assert(compressedFile.length < normalFile.length)
  }

  test("SequenceFile with writable key") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (new IntWritable(x), "a" * x)) 
    nums.saveAsSequenceFile(outputDir)
    // Try reading the output back as a SequenceFile
    val output = sc.sequenceFile[IntWritable, Text](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("SequenceFile with writable value") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (x, new Text("a" * x)))
    nums.saveAsSequenceFile(outputDir)
    // Try reading the output back as a SequenceFile
    val output = sc.sequenceFile[IntWritable, Text](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("SequenceFile with writable key and value") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (new IntWritable(x), new Text("a" * x)))
    nums.saveAsSequenceFile(outputDir)
    // Try reading the output back as a SequenceFile
    val output = sc.sequenceFile[IntWritable, Text](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("implicit conversions in reading SequenceFiles") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (x, "a" * x)) // (1,a), (2,aa), (3,aaa)
    nums.saveAsSequenceFile(outputDir)
    // Similar to the tests above, we read a SequenceFile, but this time we pass type params
    // that are convertable to Writable instead of calling sequenceFile[IntWritable, Text]
    val output1 = sc.sequenceFile[Int, String](outputDir)
    assert(output1.collect().toList === List((1, "a"), (2, "aa"), (3, "aaa")))
    // Also try having one type be a subclass of Writable and one not
    val output2 = sc.sequenceFile[Int, Text](outputDir)
    assert(output2.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
    val output3 = sc.sequenceFile[IntWritable, String](outputDir)
    assert(output3.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("object files of ints") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 4)
    nums.saveAsObjectFile(outputDir)
    // Try reading the output back as an object file
    val output = sc.objectFile[Int](outputDir)
    assert(output.collect().toList === List(1, 2, 3, 4))
  }

  test("object files of complex types") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (x, "a" * x))
    nums.saveAsObjectFile(outputDir)
    // Try reading the output back as an object file
    val output = sc.objectFile[(Int, String)](outputDir)
    assert(output.collect().toList === List((1, "a"), (2, "aa"), (3, "aaa")))
  }

  test("write SequenceFile using new Hadoop API") {
    import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (new IntWritable(x), new Text("a" * x)))
    nums.saveAsNewAPIHadoopFile[SequenceFileOutputFormat[IntWritable, Text]](
        outputDir)
    val output = sc.sequenceFile[IntWritable, Text](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("read SequenceFile using new Hadoop API") {
    import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val outputDir = new File(tempDir, "output").getAbsolutePath
    val nums = sc.makeRDD(1 to 3).map(x => (new IntWritable(x), new Text("a" * x)))
    nums.saveAsSequenceFile(outputDir)
    val output =
        sc.newAPIHadoopFile[IntWritable, Text, SequenceFileInputFormat[IntWritable, Text]](outputDir)
    assert(output.map(_.toString).collect().toList === List("(1,a)", "(2,aa)", "(3,aaa)"))
  }

  test("file caching") {
    sc = new SparkContext("local", "test")
    val tempDir = Files.createTempDir()
    val out = new FileWriter(tempDir + "/input")
    out.write("Hello world!\n")
    out.write("What's up?\n")
    out.write("Goodbye\n")
    out.close()
    val rdd = sc.textFile(tempDir + "/input").cache()
    assert(rdd.count() === 3)
    assert(rdd.count() === 3)
    assert(rdd.count() === 3)
  }
}