/**
* Test {@link CompressionEmulationUtil#isCompressionEmulationEnabled(
* org.apache.hadoop.conf.Configuration)}.
*/
@Test
public void testIsCompressionEmulationEnabled() {
Configuration conf = new Configuration();
// Check default values
assertTrue(CompressionEmulationUtil.isCompressionEmulationEnabled(conf));
// Check disabled
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, false);
assertFalse(CompressionEmulationUtil.isCompressionEmulationEnabled(conf));
// Check enabled
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
assertTrue(CompressionEmulationUtil.isCompressionEmulationEnabled(conf));
}
/** Test compressible {@link GridmixRecord}. */
@Test
public void testCompressibleGridmixRecord() throws IOException {
JobConf conf = new JobConf();
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
CompressionEmulationUtil.setInputCompressionEmulationEnabled(conf, true);
FileSystem lfs = FileSystem.getLocal(conf);
int dataSize = 1024 * 1024 * 10; // 10 MB
float ratio = 0.357F;
// define the test's root temp directory
Path rootTempDir =
new Path(System.getProperty("test.build.data", "/tmp"))
.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
Path tempDir = new Path(rootTempDir, "TestPossiblyCompressibleGridmixRecord");
lfs.delete(tempDir, true);
// define a compressible GridmixRecord
GridmixRecord record = new GridmixRecord(dataSize, 0);
record.setCompressibility(true, ratio); // enable compression
conf.setClass(FileOutputFormat.COMPRESS_CODEC, GzipCodec.class, CompressionCodec.class);
org.apache.hadoop.mapred.FileOutputFormat.setCompressOutput(conf, true);
// write the record to a file
Path recordFile = new Path(tempDir, "record");
OutputStream outStream =
CompressionEmulationUtil.getPossiblyCompressedOutputStream(recordFile, conf);
DataOutputStream out = new DataOutputStream(outStream);
record.write(out);
out.close();
outStream.close();
// open the compressed stream for reading
Path actualRecordFile = recordFile.suffix(".gz");
InputStream in =
CompressionEmulationUtil.getPossiblyDecompressedInputStream(actualRecordFile, conf, 0);
// get the compressed file size
long compressedFileSize = lfs.listStatus(actualRecordFile)[0].getLen();
GridmixRecord recordRead = new GridmixRecord();
recordRead.readFields(new DataInputStream(in));
assertEquals(
"Record size mismatch in a compressible GridmixRecord", dataSize, recordRead.getSize());
assertTrue(
"Failed to generate a compressible GridmixRecord",
recordRead.getSize() > compressedFileSize);
// check if the record can generate data with the desired compression ratio
float seenRatio = ((float) compressedFileSize) / dataSize;
assertEquals(
CompressionEmulationUtil.standardizeCompressionRatio(ratio),
CompressionEmulationUtil.standardizeCompressionRatio(seenRatio),
1.0D);
}
/** Test {@link RandomTextDataMapper} via {@link CompressionEmulationUtil}. */
@Test
public void testRandomCompressedTextDataGenerator() throws Exception {
int wordSize = 10;
int listSize = 20;
long dataSize = 10 * 1024 * 1024;
Configuration conf = new Configuration();
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
CompressionEmulationUtil.setInputCompressionEmulationEnabled(conf, true);
// configure the RandomTextDataGenerator to generate desired sized data
conf.setInt(RandomTextDataGenerator.GRIDMIX_DATAGEN_RANDOMTEXT_LISTSIZE, listSize);
conf.setInt(RandomTextDataGenerator.GRIDMIX_DATAGEN_RANDOMTEXT_WORDSIZE, wordSize);
conf.setLong(GenerateData.GRIDMIX_GEN_BYTES, dataSize);
FileSystem lfs = FileSystem.getLocal(conf);
// define the test's root temp directory
Path rootTempDir =
new Path(System.getProperty("test.build.data", "/tmp"))
.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
Path tempDir = new Path(rootTempDir, "TestRandomCompressedTextDataGenr");
lfs.delete(tempDir, true);
runDataGenJob(conf, tempDir);
// validate the output data
FileStatus[] files = lfs.listStatus(tempDir, new Utils.OutputFileUtils.OutputFilesFilter());
long size = 0;
long maxLineSize = 0;
for (FileStatus status : files) {
InputStream in =
CompressionEmulationUtil.getPossiblyDecompressedInputStream(status.getPath(), conf, 0);
BufferedReader reader = new BufferedReader(new InputStreamReader(in));
String line = reader.readLine();
if (line != null) {
long lineSize = line.getBytes().length;
if (lineSize > maxLineSize) {
maxLineSize = lineSize;
}
while (line != null) {
for (String word : line.split("\\s")) {
size += word.getBytes().length;
}
line = reader.readLine();
}
}
reader.close();
}
assertTrue(size >= dataSize);
assertTrue(size <= dataSize + maxLineSize);
}
/**
* Test if {@link RandomTextDataGenerator} can generate random text data with the desired
* compression ratio. This involves - using {@link CompressionEmulationUtil} to configure the MR
* job for generating the random text data with the desired compression ratio - running the MR job
* - test {@link RandomTextDataGenerator}'s output and match the output size (compressed) with the
* expected compression ratio.
*/
private void testCompressionRatioConfigure(float ratio) throws Exception {
long dataSize = 10 * 1024 * 1024;
Configuration conf = new Configuration();
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
CompressionEmulationUtil.setInputCompressionEmulationEnabled(conf, true);
conf.setLong(GenerateData.GRIDMIX_GEN_BYTES, dataSize);
float expectedRatio = CompressionEmulationUtil.DEFAULT_COMPRESSION_RATIO;
if (ratio > 0) {
// set the compression ratio in the conf
CompressionEmulationUtil.setMapInputCompressionEmulationRatio(conf, ratio);
expectedRatio = CompressionEmulationUtil.standardizeCompressionRatio(ratio);
}
// invoke the utility to map from ratio to word-size
CompressionEmulationUtil.setupDataGeneratorConfig(conf);
FileSystem lfs = FileSystem.getLocal(conf);
// define the test's root temp directory
Path rootTempDir =
new Path(System.getProperty("test.build.data", "/tmp"))
.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
Path tempDir = new Path(rootTempDir, "TestCustomRandomCompressedTextDataGenr");
lfs.delete(tempDir, true);
runDataGenJob(conf, tempDir);
// validate the output data
FileStatus[] files = lfs.listStatus(tempDir, new Utils.OutputFileUtils.OutputFilesFilter());
long size = 0;
for (FileStatus status : files) {
size += status.getLen();
}
float compressionRatio = ((float) size) / dataSize;
float stdRatio = CompressionEmulationUtil.standardizeCompressionRatio(compressionRatio);
assertEquals(expectedRatio, stdRatio, 0.0D);
}
/**
* Test of {@link FileQueue} can identify compressed file and provide readers to extract
* uncompressed data only if input-compression is enabled.
*/
@Test
public void testFileQueueDecompression() throws IOException {
JobConf conf = new JobConf();
FileSystem lfs = FileSystem.getLocal(conf);
String inputLine = "Hi Hello!";
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
CompressionEmulationUtil.setInputCompressionEmulationEnabled(conf, true);
org.apache.hadoop.mapred.FileOutputFormat.setCompressOutput(conf, true);
org.apache.hadoop.mapred.FileOutputFormat.setOutputCompressorClass(conf, GzipCodec.class);
// define the test's root temp directory
Path rootTempDir =
new Path(System.getProperty("test.build.data", "/tmp"))
.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
Path tempDir = new Path(rootTempDir, "TestFileQueueDecompression");
lfs.delete(tempDir, true);
// create a compressed file
Path compressedFile = new Path(tempDir, "test");
OutputStream out =
CompressionEmulationUtil.getPossiblyCompressedOutputStream(compressedFile, conf);
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(out));
writer.write(inputLine);
writer.close();
compressedFile = compressedFile.suffix(".gz");
// now read back the data from the compressed stream using FileQueue
long fileSize = lfs.listStatus(compressedFile)[0].getLen();
CombineFileSplit split =
new CombineFileSplit(new Path[] {compressedFile}, new long[] {fileSize});
FileQueue queue = new FileQueue(split, conf);
byte[] bytes = new byte[inputLine.getBytes().length];
queue.read(bytes);
queue.close();
String readLine = new String(bytes);
assertEquals("Compression/Decompression error", inputLine, readLine);
}
/**
* Test {@link CompressionEmulationUtil#getPossiblyDecompressedInputStream(Path, Configuration,
* long)} and {@link CompressionEmulationUtil#getPossiblyCompressedOutputStream(Path,
* Configuration)}.
*/
@Test
public void testPossiblyCompressedDecompressedStreams() throws IOException {
JobConf conf = new JobConf();
FileSystem lfs = FileSystem.getLocal(conf);
String inputLine = "Hi Hello!";
CompressionEmulationUtil.setCompressionEmulationEnabled(conf, true);
CompressionEmulationUtil.setInputCompressionEmulationEnabled(conf, true);
conf.setBoolean(FileOutputFormat.COMPRESS, true);
conf.setClass(FileOutputFormat.COMPRESS_CODEC, GzipCodec.class, CompressionCodec.class);
// define the test's root temp directory
Path rootTempDir =
new Path(System.getProperty("test.build.data", "/tmp"))
.makeQualified(lfs.getUri(), lfs.getWorkingDirectory());
Path tempDir = new Path(rootTempDir, "TestPossiblyCompressedDecompressedStreams");
lfs.delete(tempDir, true);
// create a compressed file
Path compressedFile = new Path(tempDir, "test");
OutputStream out =
CompressionEmulationUtil.getPossiblyCompressedOutputStream(compressedFile, conf);
BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(out));
writer.write(inputLine);
writer.close();
// now read back the data from the compressed stream
compressedFile = compressedFile.suffix(".gz");
InputStream in =
CompressionEmulationUtil.getPossiblyDecompressedInputStream(compressedFile, conf, 0);
BufferedReader reader = new BufferedReader(new InputStreamReader(in));
String readLine = reader.readLine();
assertEquals("Compression/Decompression error", inputLine, readLine);
reader.close();
}
