List of usage examples for org.apache.hadoop.mapreduce RecordReader getCurrentValue
public abstract VALUEIN getCurrentValue() throws IOException, InterruptedException;
From source file:org.apache.mnemonic.mapreduce.MneMapreduceBufferDataTest.java
License:Apache License
@Test(enabled = true, dependsOnMethods = { "testWriteBufferData" })
public void testReadBufferData() throws Exception {
long reccnt = 0L;
long tsize = 0L;
byte[] buf;//from ww w .j a v a 2s .co m
Checksum cs = new CRC32();
cs.reset();
File folder = new File(m_workdir.toString());
File[] listfiles = folder.listFiles();
for (int idx = 0; idx < listfiles.length; ++idx) {
if (listfiles[idx].isFile()
&& listfiles[idx].getName().startsWith(MneConfigHelper.getBaseOutputName(m_conf, null))
&& listfiles[idx].getName().endsWith(MneConfigHelper.DEFAULT_FILE_EXTENSION)) {
m_partfns.add(listfiles[idx].getName());
}
}
Collections.sort(m_partfns); // keep the order for checksum
for (int idx = 0; idx < m_partfns.size(); ++idx) {
System.out.println(String.format("Verifying : %s", m_partfns.get(idx)));
FileSplit split = new FileSplit(new Path(m_workdir, m_partfns.get(idx)), 0, 0L, new String[0]);
InputFormat<NullWritable, MneDurableInputValue<DurableBuffer<?>>> inputFormat = new MneInputFormat<MneDurableInputValue<DurableBuffer<?>>, DurableBuffer<?>>();
RecordReader<NullWritable, MneDurableInputValue<DurableBuffer<?>>> reader = inputFormat
.createRecordReader(split, m_tacontext);
MneDurableInputValue<DurableBuffer<?>> dbufval = null;
while (reader.nextKeyValue()) {
dbufval = reader.getCurrentValue();
assert dbufval.getValue().getSize() == dbufval.getValue().get().capacity();
dbufval.getValue().get().clear();
buf = new byte[dbufval.getValue().get().capacity()];
dbufval.getValue().get().get(buf);
cs.update(buf, 0, buf.length);
tsize += dbufval.getValue().getSize();
++reccnt;
}
reader.close();
}
AssertJUnit.assertEquals(m_reccnt, reccnt);
AssertJUnit.assertEquals(m_totalsize, tsize);
AssertJUnit.assertEquals(m_checksum, cs.getValue());
System.out.println(String.format("The checksum of buffer is %d", m_checksum));
}
From source file:org.apache.mnemonic.mapreduce.MneMapreduceChunkDataTest.java
License:Apache License
@Test(enabled = true, dependsOnMethods = { "testWriteChunkData" })
public void testReadChunkData() throws Exception {
List<String> partfns = new ArrayList<String>();
long reccnt = 0L;
long tsize = 0L;
Checksum cs = new CRC32();
cs.reset();/* w ww.j ava2s . c om*/
File folder = new File(m_workdir.toString());
File[] listfiles = folder.listFiles();
for (int idx = 0; idx < listfiles.length; ++idx) {
if (listfiles[idx].isFile()
&& listfiles[idx].getName().startsWith(MneConfigHelper.getBaseOutputName(m_conf, null))
&& listfiles[idx].getName().endsWith(MneConfigHelper.DEFAULT_FILE_EXTENSION)) {
partfns.add(listfiles[idx].getName());
}
}
Collections.sort(partfns); // keep the order for checksum
for (int idx = 0; idx < partfns.size(); ++idx) {
System.out.println(String.format("Verifying : %s", partfns.get(idx)));
FileSplit split = new FileSplit(new Path(m_workdir, partfns.get(idx)), 0, 0L, new String[0]);
InputFormat<NullWritable, MneDurableInputValue<DurableChunk<?>>> inputFormat = new MneInputFormat<MneDurableInputValue<DurableChunk<?>>, DurableChunk<?>>();
RecordReader<NullWritable, MneDurableInputValue<DurableChunk<?>>> reader = inputFormat
.createRecordReader(split, m_tacontext);
MneDurableInputValue<DurableChunk<?>> dchkval = null;
while (reader.nextKeyValue()) {
dchkval = reader.getCurrentValue();
byte b;
for (int j = 0; j < dchkval.getValue().getSize(); ++j) {
b = unsafe.getByte(dchkval.getValue().get() + j);
cs.update(b);
}
tsize += dchkval.getValue().getSize();
++reccnt;
}
reader.close();
}
AssertJUnit.assertEquals(m_reccnt, reccnt);
AssertJUnit.assertEquals(m_totalsize, tsize);
AssertJUnit.assertEquals(m_checksum, cs.getValue());
System.out.println(String.format("The checksum of chunk is %d", m_checksum));
}
From source file:org.apache.mnemonic.mapreduce.MneMapreduceLongDataTest.java
License:Apache License
@Test(enabled = true, dependsOnMethods = { "testWriteLongData" })
public void testReadLongData() throws Exception {
long sum = 0L;
long reccnt = 0L;
File folder = new File(m_workdir.toString());
File[] listfiles = folder.listFiles();
for (int idx = 0; idx < listfiles.length; ++idx) {
if (listfiles[idx].isFile()
&& listfiles[idx].getName().startsWith(MneConfigHelper.getBaseOutputName(m_conf, null))
&& listfiles[idx].getName().endsWith(MneConfigHelper.DEFAULT_FILE_EXTENSION)) {
System.out.println(String.format("Verifying : %s", listfiles[idx].getName()));
FileSplit split = new FileSplit(new Path(m_workdir, listfiles[idx].getName()), 0, 0L,
new String[0]);
InputFormat<NullWritable, MneDurableInputValue<Long>> inputFormat = new MneInputFormat<MneDurableInputValue<Long>, Long>();
RecordReader<NullWritable, MneDurableInputValue<Long>> reader = inputFormat
.createRecordReader(split, m_tacontext);
MneDurableInputValue<Long> mdval = null;
while (reader.nextKeyValue()) {
mdval = reader.getCurrentValue();
sum += mdval.getValue();
++reccnt;/* w w w.ja v a 2s . c o m*/
}
reader.close();
}
}
AssertJUnit.assertEquals(m_sum, sum);
AssertJUnit.assertEquals(m_reccnt, reccnt);
System.out.println(String.format("The checksum of long data is %d", sum));
}
From source file:org.apache.mnemonic.mapreduce.MneMapreducePersonDataTest.java
License:Apache License
@Test(enabled = true, dependsOnMethods = { "testWritePersonData" })
public void testReadPersonData() throws Exception {
long sumage = 0L;
long reccnt = 0L;
File folder = new File(m_workdir.toString());
File[] listfiles = folder.listFiles();
for (int idx = 0; idx < listfiles.length; ++idx) {
if (listfiles[idx].isFile()
&& listfiles[idx].getName().startsWith(MneConfigHelper.getBaseOutputName(m_conf, null))
&& listfiles[idx].getName().endsWith(MneConfigHelper.DEFAULT_FILE_EXTENSION)) {
System.out.println(String.format("Verifying : %s", listfiles[idx].getName()));
FileSplit split = new FileSplit(new Path(m_workdir, listfiles[idx].getName()), 0, 0L,
new String[0]);
InputFormat<NullWritable, MneDurableInputValue<Person<Long>>> inputFormat = new MneInputFormat<MneDurableInputValue<Person<Long>>, Person<Long>>();
RecordReader<NullWritable, MneDurableInputValue<Person<Long>>> reader = inputFormat
.createRecordReader(split, m_tacontext);
MneDurableInputValue<Person<Long>> personval = null;
while (reader.nextKeyValue()) {
personval = reader.getCurrentValue();
AssertJUnit.assertTrue(personval.getValue().getAge() < 51);
sumage += personval.getValue().getAge();
++reccnt;/*from w ww . j a va2s .c o m*/
}
reader.close();
}
}
AssertJUnit.assertEquals(m_reccnt, reccnt);
AssertJUnit.assertEquals(m_sumage, sumage);
System.out.println(String.format("The checksum of ages is %d", sumage));
}
From source file:org.apache.orc.mapreduce.TestMapreduceOrcOutputFormat.java
License:Apache License
@Test public void testPredicatePushdown() throws Exception { TaskAttemptID id = new TaskAttemptID("jt", 0, TaskType.MAP, 0, 0); TaskAttemptContext attemptContext = new TaskAttemptContextImpl(conf, id); final String typeStr = "struct<i:int,s:string>"; OrcConf.MAPRED_OUTPUT_SCHEMA.setString(conf, typeStr); conf.set("mapreduce.output.fileoutputformat.outputdir", workDir.toString()); conf.setInt(OrcConf.ROW_INDEX_STRIDE.getAttribute(), 1000); conf.setBoolean(OrcOutputFormat.SKIP_TEMP_DIRECTORY, true); OutputFormat<NullWritable, OrcStruct> outputFormat = new OrcOutputFormat<OrcStruct>(); RecordWriter<NullWritable, OrcStruct> writer = outputFormat.getRecordWriter(attemptContext); // write 4000 rows with the integer and the binary string TypeDescription type = TypeDescription.fromString(typeStr); OrcStruct row = (OrcStruct) OrcStruct.createValue(type); NullWritable nada = NullWritable.get(); for (int r = 0; r < 4000; ++r) { row.setFieldValue(0, new IntWritable(r)); row.setFieldValue(1, new Text(Integer.toBinaryString(r))); writer.write(nada, row);/*from w w w.j a va 2 s. c o m*/ } writer.close(attemptContext); OrcInputFormat.setSearchArgument(conf, SearchArgumentFactory.newBuilder() .between("i", PredicateLeaf.Type.LONG, new Long(1500), new Long(1999)).build(), new String[] { null, "i", "s" }); FileSplit split = new FileSplit(new Path(workDir, "part-m-00000.orc"), 0, 1000000, new String[0]); RecordReader<NullWritable, OrcStruct> reader = new OrcInputFormat<OrcStruct>().createRecordReader(split, attemptContext); // the sarg should cause it to skip over the rows except 1000 to 2000 for (int r = 1000; r < 2000; ++r) { assertEquals(true, reader.nextKeyValue()); row = reader.getCurrentValue(); assertEquals(r, ((IntWritable) row.getFieldValue(0)).get()); assertEquals(Integer.toBinaryString(r), row.getFieldValue(1).toString()); } assertEquals(false, reader.nextKeyValue()); }
From source file:org.apache.orc.mapreduce.TestMapreduceOrcOutputFormat.java
License:Apache License
@Test public void testColumnSelection() throws Exception { String typeStr = "struct<i:int,j:int,k:int>"; OrcConf.MAPRED_OUTPUT_SCHEMA.setString(conf, typeStr); conf.set("mapreduce.output.fileoutputformat.outputdir", workDir.toString()); conf.setInt(OrcConf.ROW_INDEX_STRIDE.getAttribute(), 1000); conf.setBoolean(OrcOutputFormat.SKIP_TEMP_DIRECTORY, true); TaskAttemptID id = new TaskAttemptID("jt", 0, TaskType.MAP, 0, 1); TaskAttemptContext attemptContext = new TaskAttemptContextImpl(conf, id); OutputFormat<NullWritable, OrcStruct> outputFormat = new OrcOutputFormat<OrcStruct>(); RecordWriter<NullWritable, OrcStruct> writer = outputFormat.getRecordWriter(attemptContext); // write 4000 rows with the integer and the binary string TypeDescription type = TypeDescription.fromString(typeStr); OrcStruct row = (OrcStruct) OrcStruct.createValue(type); NullWritable nada = NullWritable.get(); for (int r = 0; r < 3000; ++r) { row.setFieldValue(0, new IntWritable(r)); row.setFieldValue(1, new IntWritable(r * 2)); row.setFieldValue(2, new IntWritable(r * 3)); writer.write(nada, row);// ww w .j a v a 2s .c o m } writer.close(attemptContext); conf.set(OrcConf.INCLUDE_COLUMNS.getAttribute(), "0,2"); FileSplit split = new FileSplit(new Path(workDir, "part-m-00000.orc"), 0, 1000000, new String[0]); RecordReader<NullWritable, OrcStruct> reader = new OrcInputFormat<OrcStruct>().createRecordReader(split, attemptContext); // the sarg should cause it to skip over the rows except 1000 to 2000 for (int r = 0; r < 3000; ++r) { assertEquals(true, reader.nextKeyValue()); row = reader.getCurrentValue(); assertEquals(r, ((IntWritable) row.getFieldValue(0)).get()); assertEquals(null, row.getFieldValue(1)); assertEquals(r * 3, ((IntWritable) row.getFieldValue(2)).get()); } assertEquals(false, reader.nextKeyValue()); }
From source file:org.apache.parquet.hadoop.thrift.TestParquetToThriftReadWriteAndProjection.java
License:Apache License
private <T extends TBase<?, ?>> void shouldDoProjection(Configuration conf, T recordToWrite, T exptectedReadResult, Class<? extends TBase<?, ?>> thriftClass) throws Exception { final Path parquetFile = new Path("target/test/TestParquetToThriftReadWriteAndProjection/file.parquet"); final FileSystem fs = parquetFile.getFileSystem(conf); if (fs.exists(parquetFile)) { fs.delete(parquetFile, true);/* ww w . jav a 2s . c o m*/ } //create a test file final TProtocolFactory protocolFactory = new TCompactProtocol.Factory(); final TaskAttemptID taskId = new TaskAttemptID("local", 0, true, 0, 0); final ThriftToParquetFileWriter w = new ThriftToParquetFileWriter(parquetFile, ContextUtil.newTaskAttemptContext(conf, taskId), protocolFactory, thriftClass); final ByteArrayOutputStream baos = new ByteArrayOutputStream(); final TProtocol protocol = protocolFactory.getProtocol(new TIOStreamTransport(baos)); recordToWrite.write(protocol); w.write(new BytesWritable(baos.toByteArray())); w.close(); final ParquetThriftInputFormat<T> parquetThriftInputFormat = new ParquetThriftInputFormat<T>(); final Job job = new Job(conf, "read"); job.setInputFormatClass(ParquetThriftInputFormat.class); ParquetThriftInputFormat.setInputPaths(job, parquetFile); final JobID jobID = new JobID("local", 1); List<InputSplit> splits = parquetThriftInputFormat .getSplits(ContextUtil.newJobContext(ContextUtil.getConfiguration(job), jobID)); T readValue = null; for (InputSplit split : splits) { TaskAttemptContext taskAttemptContext = ContextUtil.newTaskAttemptContext( ContextUtil.getConfiguration(job), new TaskAttemptID(new TaskID(jobID, true, 1), 0)); final RecordReader<Void, T> reader = parquetThriftInputFormat.createRecordReader(split, taskAttemptContext); reader.initialize(split, taskAttemptContext); if (reader.nextKeyValue()) { readValue = reader.getCurrentValue(); LOG.info(readValue); } } assertEquals(exptectedReadResult, readValue); }
From source file:org.apache.rya.accumulo.mr.GraphXEdgeInputFormatTest.java
License:Apache License
@SuppressWarnings("rawtypes") @Test//from w w w.j a v a 2 s.c om public void testInputFormat() throws Exception { RyaStatement input = RyaStatement.builder().setSubject(new RyaURI("http://www.google.com")) .setPredicate(new RyaURI("http://some_other_uri")).setObject(new RyaURI("http://www.yahoo.com")) .setColumnVisibility(new byte[0]).setValue(new byte[0]).build(); apiImpl.add(input); Job jobConf = Job.getInstance(); GraphXEdgeInputFormat.setMockInstance(jobConf, instance.getInstanceName()); GraphXEdgeInputFormat.setConnectorInfo(jobConf, username, password); GraphXEdgeInputFormat.setTableLayout(jobConf, TABLE_LAYOUT.SPO); GraphXEdgeInputFormat.setInputTableName(jobConf, table); GraphXEdgeInputFormat.setInputTableName(jobConf, table); GraphXEdgeInputFormat.setScanIsolation(jobConf, false); GraphXEdgeInputFormat.setLocalIterators(jobConf, false); GraphXEdgeInputFormat.setOfflineTableScan(jobConf, false); GraphXEdgeInputFormat inputFormat = new GraphXEdgeInputFormat(); JobContext context = new JobContextImpl(jobConf.getConfiguration(), jobConf.getJobID()); List<InputSplit> splits = inputFormat.getSplits(context); Assert.assertEquals(1, splits.size()); TaskAttemptContext taskAttemptContext = new TaskAttemptContextImpl(context.getConfiguration(), new TaskAttemptID(new TaskID(), 1)); RecordReader reader = inputFormat.createRecordReader(splits.get(0), taskAttemptContext); RecordReader ryaStatementRecordReader = (RecordReader) reader; ryaStatementRecordReader.initialize(splits.get(0), taskAttemptContext); List<Edge> results = new ArrayList<Edge>(); while (ryaStatementRecordReader.nextKeyValue()) { Edge writable = (Edge) ryaStatementRecordReader.getCurrentValue(); long srcId = writable.srcId(); long destId = writable.dstId(); RyaTypeWritable rtw = null; Object text = ryaStatementRecordReader.getCurrentKey(); Edge<RyaTypeWritable> edge = new Edge<RyaTypeWritable>(srcId, destId, rtw); results.add(edge); System.out.println(text); } System.out.println(results.size()); System.out.println(results); Assert.assertTrue(results.size() == 2); }
From source file:org.apache.tinkerpop.gremlin.hadoop.structure.io.RecordReaderWriterTest.java
License:Apache License
private static void validateFileSplits(final List<FileSplit> fileSplits, final Configuration configuration, final Class<? extends InputFormat<NullWritable, VertexWritable>> inputFormatClass, final Optional<Class<? extends OutputFormat<NullWritable, VertexWritable>>> outFormatClass) throws Exception { final InputFormat inputFormat = ReflectionUtils.newInstance(inputFormatClass, configuration); final TaskAttemptContext job = new TaskAttemptContextImpl(configuration, new TaskAttemptID(UUID.randomUUID().toString(), 0, TaskType.MAP, 0, 0)); int vertexCount = 0; int outEdgeCount = 0; int inEdgeCount = 0; final OutputFormat<NullWritable, VertexWritable> outputFormat = outFormatClass.isPresent() ? ReflectionUtils.newInstance(outFormatClass.get(), configuration) : null;/*from w w w . java2s.c om*/ final RecordWriter<NullWritable, VertexWritable> writer = null == outputFormat ? null : outputFormat.getRecordWriter(job); boolean foundKeyValue = false; for (final FileSplit split : fileSplits) { logger.info("\treading file split {}", split.getPath().getName() + " ({}", split.getStart() + "..." + (split.getStart() + split.getLength()), "{} {} bytes)"); final RecordReader reader = inputFormat.createRecordReader(split, job); float lastProgress = -1f; while (reader.nextKeyValue()) { //System.out.println("" + reader.getProgress() + "> " + reader.getCurrentKey() + ": " + reader.getCurrentValue()); final float progress = reader.getProgress(); assertTrue(progress >= lastProgress); assertEquals(NullWritable.class, reader.getCurrentKey().getClass()); final VertexWritable vertexWritable = (VertexWritable) reader.getCurrentValue(); if (null != writer) writer.write(NullWritable.get(), vertexWritable); vertexCount++; outEdgeCount = outEdgeCount + (int) IteratorUtils.count(vertexWritable.get().edges(Direction.OUT)); inEdgeCount = inEdgeCount + (int) IteratorUtils.count(vertexWritable.get().edges(Direction.IN)); // final Vertex vertex = vertexWritable.get(); assertEquals(Integer.class, vertex.id().getClass()); if (vertex.value("name").equals("SUGAR MAGNOLIA")) { foundKeyValue = true; assertEquals(92, IteratorUtils.count(vertex.edges(Direction.OUT))); assertEquals(77, IteratorUtils.count(vertex.edges(Direction.IN))); } lastProgress = progress; } } assertEquals(8049, outEdgeCount); assertEquals(8049, inEdgeCount); assertEquals(outEdgeCount, inEdgeCount); assertEquals(808, vertexCount); assertTrue(foundKeyValue); if (null != writer) { writer.close(new TaskAttemptContextImpl(configuration, job.getTaskAttemptID())); for (int i = 1; i < 10; i++) { final File outputDirectory = new File( new URL(configuration.get("mapreduce.output.fileoutputformat.outputdir")).toURI()); final List<FileSplit> splits = generateFileSplits( new File(outputDirectory.getAbsoluteFile() + "/_temporary/0/_temporary/" + job.getTaskAttemptID().getTaskID().toString().replace("task", "attempt") + "_0" + "/part-m-00000"), i); validateFileSplits(splits, configuration, inputFormatClass, Optional.empty()); } } }
From source file:org.apache.vxquery.metadata.VXQueryCollectionOperatorDescriptor.java
License:Apache License
@Override public IOperatorNodePushable createPushRuntime(IHyracksTaskContext ctx, IRecordDescriptorProvider recordDescProvider, int partition, int nPartitions) throws HyracksDataException { final FrameTupleAccessor fta = new FrameTupleAccessor(ctx.getFrameSize(), recordDescProvider.getInputRecordDescriptor(getActivityId(), 0)); final int fieldOutputCount = recordDescProvider.getOutputRecordDescriptor(getActivityId(), 0) .getFieldCount();//from ww w .j a v a 2 s. c o m final ByteBuffer frame = ctx.allocateFrame(); final FrameTupleAppender appender = new FrameTupleAppender(ctx.getFrameSize(), fieldOutputCount); final short partitionId = (short) ctx.getTaskAttemptId().getTaskId().getPartition(); final ITreeNodeIdProvider nodeIdProvider = new TreeNodeIdProvider(partitionId, dataSourceId, totalDataSources); final String nodeId = ctx.getJobletContext().getApplicationContext().getNodeId(); final DynamicContext dCtx = (DynamicContext) ctx.getJobletContext().getGlobalJobData(); final String collectionName = collectionPartitions[partition % collectionPartitions.length]; final XMLParser parser = new XMLParser(false, nodeIdProvider, nodeId, frame, appender, childSeq, dCtx.getStaticContext()); return new AbstractUnaryInputUnaryOutputOperatorNodePushable() { @Override public void open() throws HyracksDataException { appender.reset(frame, true); writer.open(); hdfs = new HDFSFunctions(); } @Override public void nextFrame(ByteBuffer buffer) throws HyracksDataException { fta.reset(buffer); String collectionModifiedName = collectionName.replace("${nodeId}", nodeId); if (!collectionModifiedName.contains("hdfs:/")) { File collectionDirectory = new File(collectionModifiedName); //check if directory is in the local file system if (collectionDirectory.exists()) { // Go through each tuple. if (collectionDirectory.isDirectory()) { for (int tupleIndex = 0; tupleIndex < fta.getTupleCount(); ++tupleIndex) { Iterator<File> it = FileUtils.iterateFiles(collectionDirectory, new VXQueryIOFileFilter(), TrueFileFilter.INSTANCE); while (it.hasNext()) { File xmlDocument = it.next(); if (LOGGER.isLoggable(Level.FINE)) { LOGGER.fine( "Starting to read XML document: " + xmlDocument.getAbsolutePath()); } parser.parseElements(xmlDocument, writer, fta, tupleIndex); } } } else { throw new HyracksDataException("Invalid directory parameter (" + nodeId + ":" + collectionDirectory.getAbsolutePath() + ") passed to collection."); } } } else { // Else check in HDFS file system // Get instance of the HDFS filesystem FileSystem fs = hdfs.getFileSystem(); if (fs != null) { collectionModifiedName = collectionModifiedName.replaceAll("hdfs:/", ""); Path directory = new Path(collectionModifiedName); Path xmlDocument; if (tag != null) { hdfs.setJob(directory.getName(), tag); tag = "<" + tag + ">"; Job job = hdfs.getJob(); InputFormat inputFormat = hdfs.getinputFormat(); try { hdfs.scheduleSplits(); ArrayList<Integer> schedule = hdfs .getScheduleForNode(InetAddress.getLocalHost().getHostName()); List<InputSplit> splits = hdfs.getSplits(); List<FileSplit> fileSplits = new ArrayList<FileSplit>(); for (int i : schedule) { fileSplits.add((FileSplit) splits.get(i)); } FileSplitsFactory splitsFactory = new FileSplitsFactory(fileSplits); List<FileSplit> inputSplits = splitsFactory.getSplits(); ContextFactory ctxFactory = new ContextFactory(); int size = inputSplits.size(); InputStream stream; String value; RecordReader reader; TaskAttemptContext context; for (int i = 0; i < size; i++) { //read split context = ctxFactory.createContext(job.getConfiguration(), i); try { reader = inputFormat.createRecordReader(inputSplits.get(i), context); reader.initialize(inputSplits.get(i), context); while (reader.nextKeyValue()) { value = reader.getCurrentValue().toString(); //Split value if it contains more than one item with the tag if (StringUtils.countMatches(value, tag) > 1) { String items[] = value.split(tag); for (String item : items) { if (item.length() > 0) { item = START_TAG + tag + item; stream = new ByteArrayInputStream( item.getBytes(StandardCharsets.UTF_8)); parser.parseHDFSElements(stream, writer, fta, i); } } } else { value = START_TAG + value; //create an input stream to the file currently reading and send it to parser stream = new ByteArrayInputStream( value.getBytes(StandardCharsets.UTF_8)); parser.parseHDFSElements(stream, writer, fta, i); } } } catch (InterruptedException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } } } catch (IOException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } catch (ParserConfigurationException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } catch (SAXException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } } else { try { //check if the path exists and is a directory if (fs.exists(directory) && fs.isDirectory(directory)) { for (int tupleIndex = 0; tupleIndex < fta.getTupleCount(); ++tupleIndex) { //read every file in the directory RemoteIterator<LocatedFileStatus> it = fs.listFiles(directory, true); while (it.hasNext()) { xmlDocument = it.next().getPath(); if (fs.isFile(xmlDocument)) { if (LOGGER.isLoggable(Level.FINE)) { LOGGER.fine("Starting to read XML document: " + xmlDocument.getName()); } //create an input stream to the file currently reading and send it to parser InputStream in = fs.open(xmlDocument).getWrappedStream(); parser.parseHDFSElements(in, writer, fta, tupleIndex); } } } } else { throw new HyracksDataException("Invalid HDFS directory parameter (" + nodeId + ":" + directory + ") passed to collection."); } } catch (FileNotFoundException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } catch (IOException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } } try { fs.close(); } catch (IOException e) { if (LOGGER.isLoggable(Level.SEVERE)) { LOGGER.severe(e.getMessage()); } } } } } @Override public void fail() throws HyracksDataException { writer.fail(); } @Override public void close() throws HyracksDataException { // Check if needed? fta.reset(frame); if (fta.getTupleCount() > 0) { FrameUtils.flushFrame(frame, writer); } writer.close(); } }; }