Example usage for org.apache.hadoop.mapreduce RecordReader getCurrentKey

Introduction

In this page you can find the example usage for org.apache.hadoop.mapreduce RecordReader getCurrentKey.

Prototype

public abstract KEYIN getCurrentKey() throws IOException, InterruptedException;

Source Link

Document

Get the current key

Usage

From source file:org.commoncrawl.hadoop.io.mapreduce.ArcFileInputFormatTests.java

License:Apache License

static void validateArcFileItemSplit(FileSystem fs, InputSplit split, List<Pair<Path, List<TestRecord>>> splits,
        RecordReader<Text, ArcFileItem> reader) throws IOException, InterruptedException {

    int splitDataIndex = getIndexOfSplit(splits, split);

    Assert.assertTrue(splitDataIndex != -1);

    List<TestRecord> records = splits.get(splitDataIndex).e1;

    int itemIndex = 0;
    // iterate and validate stuff ...
    while (reader.nextKeyValue()) {

        Text key = reader.getCurrentKey();
        ArcFileItem value = reader.getCurrentValue();

        TestRecord testRecord = records.get(itemIndex++);

        // get test key bytes as utf-8 bytes ... 
        byte[] testKeyBytes = testRecord.url.getBytes(Charset.forName("UTF-8"));
        // compare against raw key bytes to validate key is the same (Text's utf-8 mapping code replaces invalid characters 
        // with ?, which causes our test case (which does use invalid characters to from the key, to break.
        Assert.assertTrue(ArcFileReaderTests.compareTo(testKeyBytes, 0, testKeyBytes.length, key.getBytes(), 0,
                key.getLength()) == 0);/*from   ww  w  .  j av  a 2s.  co  m*/
        // retured bytes represent the header(encoded in utf-8), terminated by a \r\n\r\n. The content follows this terminator
        // we search for this specific byte pattern to locate start of content, then compare it against source ... 
        Assert.assertTrue(ArcFileReaderTests.compareTo(testRecord.data, 0, testRecord.data.length,
                value.getContent().getReadOnlyBytes(), value.getContent().getOffset(),
                value.getContent().getCount()) == 0);
        NIOHttpHeaders headers = ArcFileItemUtils.buildHeaderFromArcFileItemHeaders(value.getHeaderItems());
        // validate metadata 
        Assert.assertEquals("text/html", headers.findValue(Constants.ARCFileHeader_ARC_MimeType));
        Assert.assertEquals(value.getArcFilePos(), testRecord.streamPos);
        Assert.assertEquals(value.getArcFileSize(), testRecord.rawSize);
        Assert.assertEquals("test-value", headers.findValue("test"));
        Assert.assertEquals(value.getArcFileName(), ((FileSplit) split).getPath().getName());

    }
    reader.close();

    Assert.assertEquals(itemIndex, ArcFileReaderTests.BASIC_TEST_RECORD_COUNT);

    splits.remove(splitDataIndex);

}

---

From source file:org.msgpack.hadoop.mapreduce.input.TestMessagePackInputFormat.java

License:Apache License

void checkFormat(Job job) throws Exception {
    TaskAttemptContext attemptContext = new TaskAttemptContext(job.getConfiguration(),
            new TaskAttemptID("123", 0, false, 1, 2));

    MessagePackInputFormat format = new MessagePackInputFormat();
    FileInputFormat.setInputPaths(job, workDir);

    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size());//from ww  w  .j  a v  a 2  s.  com
    for (int j = 0; j < splits.size(); j++) {
        RecordReader<LongWritable, MessagePackWritable> reader = format.createRecordReader(splits.get(j),
                attemptContext);
        reader.initialize(splits.get(j), attemptContext);

        int count = 0;
        try {
            while (reader.nextKeyValue()) {
                LongWritable key = reader.getCurrentKey();
                Value val = reader.getCurrentValue().get();
                assertEquals(count, val.asIntegerValue().getLong());
                count++;
            }
        } finally {
            reader.close();
        }
    }
}

---

From source file:org.tensorflow.hadoop.io.TFRecordFileTest.java

License:Open Source License

@Test
public void testInputOutputFormat() throws Exception {
    Configuration conf = new Configuration();
    Job job = Job.getInstance(conf);/*from w ww .j a v a 2  s  .c om*/

    Path outdir = new Path(System.getProperty("test.build.data", "/tmp"), "tfr-test");

    TFRecordFileOutputFormat.setOutputPath(job, outdir);

    TaskAttemptContext context = MapReduceTestUtil.createDummyMapTaskAttemptContext(job.getConfiguration());
    OutputFormat<BytesWritable, NullWritable> outputFormat = new TFRecordFileOutputFormat();
    OutputCommitter committer = outputFormat.getOutputCommitter(context);
    committer.setupJob(job);
    RecordWriter<BytesWritable, NullWritable> writer = outputFormat.getRecordWriter(context);

    // Write Example with random numbers
    Random rand = new Random();
    Map<Long, Long> records = new TreeMap<Long, Long>();
    try {
        for (int i = 0; i < RECORDS; ++i) {
            long randValue = rand.nextLong();
            records.put((long) i, randValue);
            Int64List data = Int64List.newBuilder().addValue(i).addValue(randValue).build();
            Feature feature = Feature.newBuilder().setInt64List(data).build();
            Features features = Features.newBuilder().putFeature("data", feature).build();
            Example example = Example.newBuilder().setFeatures(features).build();
            BytesWritable key = new BytesWritable(example.toByteArray());
            writer.write(key, NullWritable.get());
        }
    } finally {
        writer.close(context);
    }
    committer.commitTask(context);
    committer.commitJob(job);

    // Read and compare
    TFRecordFileInputFormat.setInputPaths(job, outdir);
    InputFormat<BytesWritable, NullWritable> inputFormat = new TFRecordFileInputFormat();
    for (InputSplit split : inputFormat.getSplits(job)) {
        RecordReader<BytesWritable, NullWritable> reader = inputFormat.createRecordReader(split, context);
        MapContext<BytesWritable, NullWritable, BytesWritable, NullWritable> mcontext = new MapContextImpl<BytesWritable, NullWritable, BytesWritable, NullWritable>(
                job.getConfiguration(), context.getTaskAttemptID(), reader, null, null,
                MapReduceTestUtil.createDummyReporter(), split);
        reader.initialize(split, mcontext);
        try {
            while (reader.nextKeyValue()) {
                BytesWritable bytes = reader.getCurrentKey();
                Example example = Example.parseFrom(bytes.getBytes());
                Int64List data = example.getFeatures().getFeatureMap().get("data").getInt64List();
                Long key = data.getValue(0);
                Long value = data.getValue(1);
                assertEquals(records.get(key), value);
                records.remove(key);
            }
        } finally {
            reader.close();
        }
    }
    assertEquals(0, records.size());
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013MultiSheetHeaderLowFootPrint()
        throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "multisheetheader.xlsx";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);
    // set locale to the one of the test data
    conf.set("hadoopoffice.locale.bcp47", "us");
    conf.set("hadoopoffice.read.header.read", "true");
    conf.set("hadoopoffice.read.header.skipheaderinallsheets", "true");
    conf.set("hadoopoffice.read.lowFootprint", "true");

    conf.set("hadoopoffice.read.lowFootprint.parser", "sax");
    Job job = Job.getInstance(conf);//from  w ww .j  a va 2s.  c  o  m
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    assertEquals("column1", ((ExcelRecordReader) reader).getOfficeReader().getCurrentParser().getHeader()[0],
            " header column 1 correctly read");
    assertEquals("column2", ((ExcelRecordReader) reader).getOfficeReader().getCurrentParser().getHeader()[1],
            " header column 2 correctly read");
    assertEquals("column3", ((ExcelRecordReader) reader).getOfficeReader().getCurrentParser().getHeader()[2],
            " header column 3 correctly read");
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();

    // First Sheet
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A2 = \"1\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B2 = \"test1\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C2 = \"10\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A3 = \"2\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B3 = \"test3\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C3 = \"15\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");

    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A4 = \"10\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B4 = \"test2\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("20", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C4 = \"20\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C4", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");

    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1 (Second Sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("50", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A2 = \"50\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B2 = \"test1\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("80", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C2 = \"80\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");

    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2 (Second Sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("60", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A3 = \"60\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B3 = \"test3\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("90", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C3 = \"90\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");

    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3 (Second Sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("70", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A4 = \"70\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B4 = \"test2\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("10000", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "C4 = \"10000\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C4", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");
    assertFalse(reader.nextKeyValue(), "Input Split for Excel file contains no further row");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2003SingleSheetLowFootPrint() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2003test.xls";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);
    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");
    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");
    Job job = Job.getInstance(conf);//from   ww w  . j  a  va2s.c o m
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2003test.xls]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2003test.xls]Sheet1!A1\"");
    assertEquals(4, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 4 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
    assertEquals("test4", ((SpreadSheetCellDAO) spreadSheetValue.get()[3]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 4 == \"test4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 2 with 1 column");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(5, spreadSheetValue.get().length, "Input Split for Excel file contains row 3 with 5 columns");
    assertEquals("31/12/99", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 1 == \"31/12/99\"");
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 2 == \"5\"");
    assertNull(spreadSheetValue.get()[2], "Input Split for Excel file contains row 3 with cell 3 == null");
    assertNull(spreadSheetValue.get()[3], "Input Split for Excel file contains row 3 with cell 4 == null");
    assertEquals("null", ((SpreadSheetCellDAO) spreadSheetValue.get()[4]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 5 == \"null\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 4");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 4 with 1 column");
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 4 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 5");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 5 with 3 columns");
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 1 == \"2\"");
    assertEquals("6", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 2== \"6\"");
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 3== \"10\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 6");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 6 with 3 columns");
    assertEquals("3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 1 == \"3\"");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 2== \"4\"");
    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"15\"");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2003MultiSheetAllLowFootPrint() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2003testmultisheet.xls";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);

    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");

    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");
    Job job = Job.getInstance(conf);//w  w  w .java2  s .c om
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2003testmultisheet.xls]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2003testmultisheet.xls]Sheet1!A1\"");
    assertEquals(4, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 4 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
    assertEquals("test4", ((SpreadSheetCellDAO) spreadSheetValue.get()[3]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 4 == \"test4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 2 with 1 column");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(5, spreadSheetValue.get().length, "Input Split for Excel file contains row 3 with 5 columns");
    assertEquals("31/12/99", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 1 == \"31/12/99\"");
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 2 == \"5\"");
    assertNull(spreadSheetValue.get()[2], "Input Split for Excel file contains row 3 with cell 3 == null");
    assertNull(spreadSheetValue.get()[3], "Input Split for Excel file contains row 3 with cell 4 == null");
    assertEquals("null", ((SpreadSheetCellDAO) spreadSheetValue.get()[4]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 5 == \"null\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 4 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 4 with 1 column");
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 4 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 5 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 5 with 3 columns");
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 1 == \"2\"");
    assertEquals("6", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 2== \"6\"");
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 3== \"10\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 6 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 6 with 3 columns");
    assertEquals("3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 1 == \"3\"");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 2== \"4\"");
    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"15\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 7 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("8", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 7 with cell 1 == \"8\"");
    assertEquals("99", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 7 with cell 2 == \"99\"");
    assertEquals(2, spreadSheetValue.get().length, "Input Split for Excel file contains row 7 with 2 columns");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 8 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 8 with 1 column");
    assertEquals("test", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 8 with cell 1 == \"test\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 9 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 9 with 3 columns");
    assertNull(spreadSheetValue.get()[0], "Input Split for Excel file contains row 9 with cell 1 == null");
    assertNull(spreadSheetValue.get()[1], "Input Split for Excel file contains row 9 with cell 2 == null");
    assertEquals("seven", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 9 with cell 3 == \"seven\"");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013SingleSheetLowFootPrint() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2013test.xlsx";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);
    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");
    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");
    conf.set("hadoopoffice.read.lowFootprint.parser", "sax");
    Job job = Job.getInstance(conf);/*from  w  ww . jav a2s.co  m*/
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2013test.xlsx]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2013test.xlsx]Sheet1!A1\"");
    assertEquals(4, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 4 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
    assertEquals("test4", ((SpreadSheetCellDAO) spreadSheetValue.get()[3]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 4 == \"test4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 2 with 1 column");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(5, spreadSheetValue.get().length, "Input Split for Excel file contains row 3 with 5 columns");
    assertEquals("31/12/99", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 1 == \"31/12/99\"");
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 2 == \"5\"");
    assertNull(spreadSheetValue.get()[2], "Input Split for Excel file contains row 3 with cell 3 == null");
    assertNull(spreadSheetValue.get()[3], "Input Split for Excel file contains row 3 with cell 4 == null");
    assertEquals("null", ((SpreadSheetCellDAO) spreadSheetValue.get()[4]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 5 == \"null\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 4");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 4 with 1 column");
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 4 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 5");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 5 with 3 columns");
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 1 == \"2\"");
    assertEquals("6", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 2== \"6\"");
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 3== \"10\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 6");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 6 with 3 columns");
    assertEquals("3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 1 == \"3\"");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 2== \"4\"");
    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"15\"");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013MultiSheetAllLowFootPrint() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2013testmultisheet.xlsx";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);

    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");

    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");
    conf.set("hadoopoffice.read.lowFootprint.parser", "sax");
    Job job = Job.getInstance(conf);/* w  w w  .  j av a  2 s  .c  om*/
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2013testmultisheet.xlsx]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2013testmultisheet.xlsx]Sheet1!A1\"");
    assertEquals(4, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 4 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
    assertEquals("test4", ((SpreadSheetCellDAO) spreadSheetValue.get()[3]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 4 == \"test4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 2 with 1 column");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(5, spreadSheetValue.get().length, "Input Split for Excel file contains row 3 with 5 columns");
    assertEquals("31/12/99", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 1 == \"31/12/99\"");
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 2 == \"5\"");
    assertNull(spreadSheetValue.get()[2], "Input Split for Excel file contains row 3 with cell 3 == null");
    assertNull(spreadSheetValue.get()[3], "Input Split for Excel file contains row 3 with cell 4 == null");
    assertEquals("null", ((SpreadSheetCellDAO) spreadSheetValue.get()[4]).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 5 == \"null\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 4 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 4 with 1 column");
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 4 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 5 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 5 with 3 columns");
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 1 == \"2\"");
    assertEquals("6", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 2== \"6\"");
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 3== \"10\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 6 (first sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 6 with 3 columns");
    assertEquals("3", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 1 == \"3\"");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 2== \"4\"");
    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"15\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 7 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("8", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 7 with cell 1 == \"8\"");
    assertEquals("99", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 7 with cell 2 == \"99\"");
    assertEquals(2, spreadSheetValue.get().length, "Input Split for Excel file contains row 7 with 2 columns");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 8 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(1, spreadSheetValue.get().length, "Input Split for Excel file contains row 8 with 1 column");
    assertEquals("test", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 8 with cell 1 == \"test\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 9 (second sheet)");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 9 with 3 columns");
    assertNull(spreadSheetValue.get()[0], "Input Split for Excel file contains row 9 with cell 1 == null");
    assertNull(spreadSheetValue.get()[1], "Input Split for Excel file contains row 9 with cell 2 == null");
    assertEquals("seven", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 9 with cell 3 == \"seven\"");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2003SingleSheetEncryptedPositiveLowFootprint()
        throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2003encrypt.xls";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);

    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");

    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");

    // for decryption simply set the password
    conf.set("hadoopoffice.read.security.crypt.password", "test");
    Job job = Job.getInstance(conf);/*from  www  .ja  v  a 2  s  . c om*/
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2003encrypt.xls]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2003encrypt.xls]Sheet1!A1\"");
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 3 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
}

---

From source file:org.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintSAXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013SingleSheetEncryptedPositiveLowFootprint()
        throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2013encrypt.xlsx";
    String fileNameSpreadSheet = classLoader.getResource(fileName).getFile();
    Path file = new Path(fileNameSpreadSheet);

    // set locale to the one of the test data
    conf.set("hadoopoffice.read.locale.bcp47", "de");
    // low footprint
    conf.set("hadoopoffice.read.lowFootprint", "true");
    conf.set("hadoopoffice.read.lowFootprint.parser", "sax");
    // for decryption simply set the password
    conf.set("hadoopoffice.read.security.crypt.password", "test");
    Job job = Job.getInstance(conf);/*from w  w w.j  ava  2s .  c om*/
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelFileInputFormat format = new ExcelFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, ArrayWritable> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    ArrayWritable spreadSheetValue = new ArrayWritable(SpreadSheetCellDAO.class);
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 1");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("[excel2013encrypt.xlsx]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2013encrypt.xlsx]Sheet1!A1\"");
    assertEquals(3, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 3 columns");
    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
}

---

• «
• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• »