List of usage examples for org.apache.hadoop.fs FileContext getLocalFSFileContext
public static FileContext getLocalFSFileContext() throws UnsupportedFileSystemException
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testScannerPartitioning() throws Exception { DirectoryScanner scanner = new DirectoryScanner(); scanner.setFilePatternRegexp(".*partition([\\d]*)"); Path path = new Path(new File(testMeta.dir).getAbsolutePath()); FileContext.getLocalFSFileContext().delete(path, true); for (int file = 0; file < 4; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), ""); }/* w ww .j a va2 s . co m*/ FileSystem fs = FileSystem.get(FileContext.getLocalFSFileContext().getDefaultFileSystem().getUri(), new Configuration()); List<DirectoryScanner> partitions = scanner.partition(2); Set<Path> allFiles = Sets.newHashSet(); for (DirectoryScanner partition : partitions) { Set<Path> files = partition.scan(fs, path, Sets.<String>newHashSet()); Assert.assertEquals("", 3, files.size()); allFiles.addAll(files); } Assert.assertEquals("Found all files " + allFiles, 5, allFiles.size()); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testPartitioning() throws Exception { LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.getScanner().setFilePatternRegexp(".*partition([\\d]*)"); oper.setDirectory(new File(testMeta.dir).getAbsolutePath()); Path path = new Path(new File(testMeta.dir).getAbsolutePath()); FileContext.getLocalFSFileContext().delete(path, true); for (int file = 0; file < 4; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), ""); }//from w w w. jav a 2s.c o m List<Partition<AbstractFileInputOperator<String>>> partitions = Lists.newArrayList(); partitions.add(new DefaultPartition<AbstractFileInputOperator<String>>(oper)); Collection<Partition<AbstractFileInputOperator<String>>> newPartitions = oper.definePartitions(partitions, new PartitioningContextImpl(null, 2)); Assert.assertEquals(2, newPartitions.size()); Assert.assertEquals(1, oper.getCurrentPartitions()); // partitioned() wasn't called for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { Assert.assertNotSame(oper, p.getPartitionedInstance()); Assert.assertNotSame(oper.getScanner(), p.getPartitionedInstance().getScanner()); Set<String> consumed = Sets.newHashSet(); LinkedHashSet<Path> files = p.getPartitionedInstance().getScanner() .scan(FileSystem.getLocal(new Configuration(false)), path, consumed); Assert.assertEquals("partition " + files, 3, files.size()); } }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
/** * Test for testing dynamic partitioning. * - Create 4 file with 3 records each.//w w w .j a v a2 s .co m * - Create a single partition, and read all records, populating pending files in operator. * - Split it in two operators * - Try to emit records again, expected result is no record is emitted, as all files are * processed. * - Create another 4 files with 3 records each * - Try to emit records again, expected result total record emitted 4 * 3 = 12. */ @Test public void testPartitioningStateTransfer() throws Exception { LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.getScanner().setFilePatternRegexp(".*partition([\\d]*)"); oper.setDirectory(new File(testMeta.dir).getAbsolutePath()); oper.setScanIntervalMillis(0); LineByLineFileInputOperator initialState = new Kryo().copy(oper); // Create 4 files with 3 records each. Path path = new Path(new File(testMeta.dir).getAbsolutePath()); FileContext.getLocalFSFileContext().delete(path, true); int file; for (file = 0; file < 4; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), "a\nb\nc\n"); } CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); int wid = 0; // Read all records to populate processedList in operator. oper.setup(testMeta.context); for (int i = 0; i < 10; i++) { oper.beginWindow(wid); oper.emitTuples(); oper.endWindow(); wid++; } Assert.assertEquals("All tuples read ", 12, sink.collectedTuples.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); initialState.setPartitionCount(2); StatsListener.Response rsp = initialState.processStats(null); Assert.assertEquals(true, rsp.repartitionRequired); // Create partitions of the operator. List<Partition<AbstractFileInputOperator<String>>> partitions = Lists.newArrayList(); partitions.add(new DefaultPartition<AbstractFileInputOperator<String>>(oper)); // incremental capacity controlled partitionCount property Collection<Partition<AbstractFileInputOperator<String>>> newPartitions = initialState .definePartitions(partitions, new PartitioningContextImpl(null, 0)); Assert.assertEquals(2, newPartitions.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); Map<Integer, Partition<AbstractFileInputOperator<String>>> m = Maps.newHashMap(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { m.put(m.size(), p); } initialState.partitioned(m); Assert.assertEquals(2, initialState.getCurrentPartitions()); /* Collect all operators in a list */ List<AbstractFileInputOperator<String>> opers = Lists.newArrayList(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { LineByLineFileInputOperator oi = (LineByLineFileInputOperator) p.getPartitionedInstance(); oi.setup(testMeta.context); oi.output.setSink(sink); opers.add(oi); } sink.clear(); for (int i = 0; i < 10; i++) { for (AbstractFileInputOperator<String> o : opers) { o.beginWindow(wid); o.emitTuples(); o.endWindow(); } wid++; } // No record should be read. Assert.assertEquals("No new tuples read ", 0, sink.collectedTuples.size()); // Add four new files with 3 records each. for (; file < 8; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), "a\nb\nc\n"); } for (int i = 0; i < 10; i++) { for (AbstractFileInputOperator<String> o : opers) { o.beginWindow(wid); o.emitTuples(); o.endWindow(); } wid++; } // If all files are processed only once then number of records emitted should // be 12. Assert.assertEquals("All tuples read ", 12, sink.collectedTuples.size()); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
/** * Test for testing dynamic partitioning. * - Create 4 file with 3 records each./* ww w . j a va 2 s . c o m*/ * - Create a single partition, and read some records, populating pending files in operator. * - Split it in two operators * - Try to emit the remaining records. */ @Test public void testPartitioningStateTransferInterrupted() throws Exception { LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.getScanner().setFilePatternRegexp(".*partition([\\d]*)"); oper.setDirectory(new File(testMeta.dir).getAbsolutePath()); oper.setScanIntervalMillis(0); oper.setEmitBatchSize(2); LineByLineFileInputOperator initialState = new Kryo().copy(oper); // Create 4 files with 3 records each. Path path = new Path(new File(testMeta.dir).getAbsolutePath()); FileContext.getLocalFSFileContext().delete(path, true); int file; for (file = 0; file < 4; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), "a\nb\nc\n"); } CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); int wid = 0; //Read some records oper.setup(testMeta.context); for (int i = 0; i < 5; i++) { oper.beginWindow(wid); oper.emitTuples(); oper.endWindow(); wid++; } Assert.assertEquals("Partial tuples read ", 6, sink.collectedTuples.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); initialState.setPartitionCount(2); StatsListener.Response rsp = initialState.processStats(null); Assert.assertEquals(true, rsp.repartitionRequired); // Create partitions of the operator. List<Partition<AbstractFileInputOperator<String>>> partitions = Lists.newArrayList(); partitions.add(new DefaultPartition<AbstractFileInputOperator<String>>(oper)); // incremental capacity controlled partitionCount property Collection<Partition<AbstractFileInputOperator<String>>> newPartitions = initialState .definePartitions(partitions, new PartitioningContextImpl(null, 0)); Assert.assertEquals(2, newPartitions.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); Map<Integer, Partition<AbstractFileInputOperator<String>>> m = Maps.newHashMap(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { m.put(m.size(), p); } initialState.partitioned(m); Assert.assertEquals(2, initialState.getCurrentPartitions()); /* Collect all operators in a list */ List<AbstractFileInputOperator<String>> opers = Lists.newArrayList(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { LineByLineFileInputOperator oi = (LineByLineFileInputOperator) p.getPartitionedInstance(); oi.setup(testMeta.context); oi.output.setSink(sink); opers.add(oi); } sink.clear(); for (int i = 0; i < 10; i++) { for (AbstractFileInputOperator<String> o : opers) { o.beginWindow(wid); o.emitTuples(); o.endWindow(); } wid++; } Assert.assertEquals("Remaining tuples read ", 6, sink.collectedTuples.size()); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
/** * Test for testing dynamic partitioning interrupting ongoing read. * - Create 4 file with 3 records each./* w ww . jav a2 s . co m*/ * - Create a single partition, and read some records, populating pending files in operator. * - Split it in two operators * - Try to emit the remaining records. */ @Test public void testPartitioningStateTransferFailure() throws Exception { LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.getScanner().setFilePatternRegexp(".*partition([\\d]*)"); oper.setDirectory(new File(testMeta.dir).getAbsolutePath()); oper.setScanIntervalMillis(0); oper.setEmitBatchSize(2); LineByLineFileInputOperator initialState = new Kryo().copy(oper); // Create 4 files with 3 records each. Path path = new Path(new File(testMeta.dir).getAbsolutePath()); FileContext.getLocalFSFileContext().delete(path, true); int file; for (file = 0; file < 4; file++) { FileUtils.write(new File(testMeta.dir, "partition00" + file), "a\nb\nc\n"); } CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); int wid = 0; //Read some records oper.setup(testMeta.context); for (int i = 0; i < 5; i++) { oper.beginWindow(wid); oper.emitTuples(); oper.endWindow(); wid++; } Assert.assertEquals("Partial tuples read ", 6, sink.collectedTuples.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); initialState.setPartitionCount(2); StatsListener.Response rsp = initialState.processStats(null); Assert.assertEquals(true, rsp.repartitionRequired); // Create partitions of the operator. List<Partition<AbstractFileInputOperator<String>>> partitions = Lists.newArrayList(); partitions.add(new DefaultPartition<AbstractFileInputOperator<String>>(oper)); // incremental capacity controlled partitionCount property Collection<Partition<AbstractFileInputOperator<String>>> newPartitions = initialState .definePartitions(partitions, new PartitioningContextImpl(null, 0)); Assert.assertEquals(2, newPartitions.size()); Assert.assertEquals(1, initialState.getCurrentPartitions()); Map<Integer, Partition<AbstractFileInputOperator<String>>> m = Maps.newHashMap(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { m.put(m.size(), p); } initialState.partitioned(m); Assert.assertEquals(2, initialState.getCurrentPartitions()); /* Collect all operators in a list */ List<AbstractFileInputOperator<String>> opers = Lists.newArrayList(); for (Partition<AbstractFileInputOperator<String>> p : newPartitions) { LineByLineFileInputOperator oi = (LineByLineFileInputOperator) p.getPartitionedInstance(); oi.setup(testMeta.context); oi.output.setSink(sink); opers.add(oi); } sink.clear(); for (int i = 0; i < 10; i++) { for (AbstractFileInputOperator<String> o : opers) { o.beginWindow(wid); o.emitTuples(); o.endWindow(); } wid++; } // No record should be read. Assert.assertEquals("Remaining tuples read ", 6, sink.collectedTuples.size()); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testRecoveryWithFailedFile() throws Exception { FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true); List<String> allLines = Lists.newArrayList(); HashSet<String> lines = Sets.newHashSet(); for (int line = 0; line < 5; line++) { lines.add("f0" + "l" + line); }// www . j a v a 2s . co m allLines.addAll(lines); File testFile = new File(testMeta.dir, "file0"); FileUtils.write(testFile, StringUtils.join(lines, '\n')); LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.scanner = null; oper.failedFiles.add(new AbstractFileInputOperator.FailedFile(testFile.getAbsolutePath(), 1)); CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); oper.setDirectory(testMeta.dir); oper.setup(testMeta.context); oper.beginWindow(0); oper.emitTuples(); oper.endWindow(); oper.teardown(); Assert.assertEquals("number tuples", 4, queryResults.collectedTuples.size()); Assert.assertEquals("lines", allLines.subList(1, allLines.size()), new ArrayList<String>(queryResults.collectedTuples)); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testRecoveryWithUnfinishedFile() throws Exception { FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true); List<String> allLines = Lists.newArrayList(); HashSet<String> lines = Sets.newHashSet(); for (int line = 0; line < 5; line++) { lines.add("f0" + "l" + line); }// w w w . j a v a 2 s .c o m allLines.addAll(lines); File testFile = new File(testMeta.dir, "file0"); FileUtils.write(testFile, StringUtils.join(lines, '\n')); LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.scanner = null; oper.unfinishedFiles.add(new AbstractFileInputOperator.FailedFile(testFile.getAbsolutePath(), 2)); CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); oper.setDirectory(testMeta.dir); oper.setup(testMeta.context); oper.beginWindow(0); oper.emitTuples(); oper.endWindow(); oper.teardown(); Assert.assertEquals("number tuples", 3, queryResults.collectedTuples.size()); Assert.assertEquals("lines", allLines.subList(2, allLines.size()), new ArrayList<String>(queryResults.collectedTuples)); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testRecoveryWithPendingFile() throws Exception { FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true); List<String> allLines = Lists.newArrayList(); HashSet<String> lines = Sets.newHashSet(); for (int line = 0; line < 5; line++) { lines.add("f0" + "l" + line); }//w w w.j a v a2 s. c o m allLines.addAll(lines); File testFile = new File(testMeta.dir, "file0"); FileUtils.write(testFile, StringUtils.join(lines, '\n')); LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.scanner = null; oper.pendingFiles.add(testFile.getAbsolutePath()); CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); oper.setDirectory(testMeta.dir); oper.setup(testMeta.context); oper.beginWindow(0); oper.emitTuples(); oper.endWindow(); oper.teardown(); Assert.assertEquals("number tuples", 5, queryResults.collectedTuples.size()); Assert.assertEquals("lines", allLines, new ArrayList<String>(queryResults.collectedTuples)); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testRecoveryWithCurrentFile() throws Exception { FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true); List<String> allLines = Lists.newArrayList(); HashSet<String> lines = Sets.newHashSet(); for (int line = 0; line < 5; line++) { lines.add("f0" + "l" + line); }//from ww w . j a v a2 s. c o m allLines.addAll(lines); File testFile = new File(testMeta.dir, "file0"); FileUtils.write(testFile, StringUtils.join(lines, '\n')); LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); oper.scanner = null; oper.currentFile = testFile.getAbsolutePath(); oper.offset = 1; CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); @SuppressWarnings({ "unchecked", "rawtypes" }) CollectorTestSink<Object> sink = (CollectorTestSink) queryResults; oper.output.setSink(sink); oper.setDirectory(testMeta.dir); oper.setup(testMeta.context); oper.beginWindow(0); oper.emitTuples(); oper.endWindow(); oper.teardown(); Assert.assertEquals("number tuples", 4, queryResults.collectedTuples.size()); Assert.assertEquals("lines", allLines.subList(1, allLines.size()), new ArrayList<String>(queryResults.collectedTuples)); }
From source file:org.apache.apex.malhar.lib.io.fs.AbstractFileInputOperatorTest.java
License:Apache License
@Test public void testIdempotency() throws Exception { FileContext.getLocalFSFileContext().delete(new Path(new File(testMeta.dir).getAbsolutePath()), true); List<String> allLines = Lists.newArrayList(); for (int file = 0; file < 2; file++) { List<String> lines = Lists.newArrayList(); for (int line = 0; line < 2; line++) { lines.add("f" + file + "l" + line); }// www. j av a2s. c o m allLines.addAll(lines); FileUtils.write(new File(testMeta.dir, "file" + file), StringUtils.join(lines, '\n')); } LineByLineFileInputOperator oper = new LineByLineFileInputOperator(); FSWindowDataManager manager = new FSWindowDataManager(); manager.setStatePath(testMeta.dir + "/recovery"); oper.setWindowDataManager(manager); CollectorTestSink<String> queryResults = new CollectorTestSink<String>(); TestUtils.setSink(oper.output, queryResults); oper.setDirectory(testMeta.dir); oper.getScanner().setFilePatternRegexp(".*file[\\d]"); oper.setup(testMeta.context); for (long wid = 0; wid < 3; wid++) { oper.beginWindow(wid); oper.emitTuples(); oper.endWindow(); } oper.teardown(); List<String> beforeRecovery = Lists.newArrayList(queryResults.collectedTuples); queryResults.clear(); //idempotency part oper.setup(testMeta.context); for (long wid = 0; wid < 3; wid++) { oper.beginWindow(wid); oper.endWindow(); } Assert.assertEquals("number tuples", 4, queryResults.collectedTuples.size()); Assert.assertEquals("lines", beforeRecovery, queryResults.collectedTuples); oper.teardown(); }