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.zuinnote.hadoop.office.format.mapreduce.OfficeFormatHadoopExcelLowFootPrintStaXTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013MultiSheetSkipWithHeaderLowFootprint()
        throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "skipsheet.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.sheet.skiplines.num", "5");
    conf.set("hadoopoffice.read.sheet.skiplines.allsheets", "true");
    conf.set("hadoopoffice.read.lowFootprint", "true");
    conf.set("hadoopoffice.read.lowFootprint.parser", "stax");
    Job job = Job.getInstance(conf);/*from w w w.  ja v  a 2 s.  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("20", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A7 = \"20\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A7", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B7 = \"test2\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B7", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C7 = \"5\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C7", ((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("30", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A8 = \"30\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A8", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B8 = \"test1\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B8", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C8 = \"10\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C8", ((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("40", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A9 = \"40\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A9", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B9 = \"test3\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B9", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C9 = \"15\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C9", ((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("90", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A7 = \"90\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A7", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B7 = \"test2\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B7", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("230", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C7 = \"230\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C7", ((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("200", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A8 = \"200\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A8", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B8 = \"test1\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B8", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("240", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C8 = \"240\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C8", ((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("101", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(), "A9 = \"101\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormula(), "Empty formula");
    assertEquals("A9", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getSheetName(), "Correct sheet");

    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "B9 = \"test3\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormula(), "Empty formula");
    assertEquals("B9", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getSheetName(), "Correct sheet");

    assertEquals("250", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(), "C9 = \"250\"");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getComment(), "Empty Comment");
    assertEquals("", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormula(), "Empty formula");
    assertEquals("C9", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getAddress(), "Correct address");
    assertEquals("Sheet2", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getSheetName(), "Correct sheet");
    // third sheet is skipped because it does not contain enough rows
    assertFalse(reader.nextKeyValue(), "Input Split for Excel file contains no further row");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2003Empty() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2003empty.xls";
    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", "de");
    Job job = Job.getInstance(conf);/*from  w  ww  .  j a v  a 2  s .  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(0, spreadSheetValue.get().length, "Input Split for Excel file contain row 1 and is empty");
    assertFalse(reader.nextKeyValue(), "Input Split for Excel file contains no further row");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2013Empty() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2013empty.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", "de");
    Job job = Job.getInstance(conf);/*from w  ww . j  a 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(0, spreadSheetValue.get().length, "Input Split for Excel file contain row 1 and is empty");
    assertFalse(reader.nextKeyValue(), "Input Split for Excel file contains no further row");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2003SingleSheet() 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");
    Job job = Job.getInstance(conf);/* w w w.j  a  v  a 2 s  . 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("[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.OfficeFormatHadoopExcelNormalTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013SingleSheetEncryptedPositive()
        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");
    // for decryption simply set the password
    conf.set("hadoopoffice.read.security.crypt.password", "test");
    Job job = Job.getInstance(conf);/*  ww w.  j  a  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("[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\"");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2003SingleSheetEncryptedPositive()
        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");
    // for decryption simply set the password
    conf.set("hadoopoffice.read.security.crypt.password", "test");
    Job job = Job.getInstance(conf);//from   w  w  w  . j a va  2  s.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("[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.OfficeFormatHadoopExcelNormalTest.java

License:Apache License

@Test
public void readExcelInputFormatExcel2013EmptyRows() throws IOException, InterruptedException {
    Configuration conf = new Configuration(defaultConf);
    ClassLoader classLoader = getClass().getClassLoader();
    String fileName = "excel2013testemptyrows.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");
    Job job = Job.getInstance(conf);//  w  w  w  . j a  v a  2s. 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("[excel2013testemptyrows.xlsx]Sheet1!A1", spreadSheetKey.toString(),
            "Input Split for Excel file has keyname == \"[excel2013testemptyrows.xlsx]Sheet1!A1\"");
    assertEquals(0, spreadSheetValue.get().length, "Input Split for Excel file contains row 1 with 0 columns");

    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 2");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(2, spreadSheetValue.get().length, "Input Split for Excel file contains row 2 with 2 columns");
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue.get()[0]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue.get()[1]).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains row 3");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals(0, spreadSheetValue.get().length, "Input Split for Excel file contains row 3 with 0 columns");

    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("10", ((SpreadSheetCellDAO) spreadSheetValue.get()[2]).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"10\"");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2013SingleSheet() 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");
    Job job = Job.getInstance(conf);//  w  w w .  j ava2 s  . 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("[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.OfficeFormatHadoopExcelNormalTest.java

License:Apache License

@Test
public void readExcelCellInputFormatExcel2013SingleSheet() 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");
    Job job = Job.getInstance(conf);/*from  w w w. j av a2 s .co m*/
    FileInputFormat.setInputPaths(job, file);
    TaskAttemptContext context = new TaskAttemptContextImpl(conf, new TaskAttemptID());
    ExcelCellFileInputFormat format = new ExcelCellFileInputFormat();
    List<InputSplit> splits = format.getSplits(job);
    assertEquals(1, splits.size(), "Only one split generated for Excel file");
    RecordReader<Text, SpreadSheetCellDAO> reader = format.createRecordReader(splits.get(0), context);
    assertNotNull(reader, "Format returned  null RecordReader");
    reader.initialize(splits.get(0), context);
    Text spreadSheetKey = new Text();
    SpreadSheetCellDAO spreadSheetValue = new SpreadSheetCellDAO();
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    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("test1", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 1 == \"test1\"");
    assertEquals("Sheet1", ((SpreadSheetCellDAO) spreadSheetValue).getSheetName(),
            "Input Split for Excel file contains row 1 with cell 1 sheetname == \"Sheet1\"");
    assertEquals("A1", ((SpreadSheetCellDAO) spreadSheetValue).getAddress(),
            "Input Split for Excel file contains row 1 with cell 1 address == \"A1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("test2", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 2 == \"test2\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("test3", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 3 == \"test3\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("test4", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 1 with cell 4 == \"test4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 2 with cell 1 == \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("31/12/99", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 1 == \"31/12/99\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("5", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 2 == \"5\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("null", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 3 with cell 5 == \"null\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("1", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 4 with cell 1 == \"1\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("2", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 1 == \"2\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("6", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 2== \"6\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("10", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 5 with cell 3== \"10\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("3", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 1 == \"3\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("4", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 2== \"4\"");
    assertTrue(reader.nextKeyValue(), "Input Split for Excel file contains cell");
    spreadSheetKey = reader.getCurrentKey();
    spreadSheetValue = reader.getCurrentValue();
    assertEquals("15", ((SpreadSheetCellDAO) spreadSheetValue).getFormattedValue(),
            "Input Split for Excel file contains row 6 with cell 3== \"15\"");
    assertFalse(reader.nextKeyValue(), "No more cells");
}

---

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

License:Apache License

@Test
public void readExcelInputFormatExcel2013MultiSheetAll() 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");
    Job job = Job.getInstance(conf);/*  w  w w. j  av  a 2 s .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 (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\"");
}

---

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