Examples with TsvParser org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser used on opensource projects

Example 1 with TsvParser

use of org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser in project hbase by apache.

the class TestImportTsvParser method testTsvParserSpecParsing.

@Test
public void testTsvParserSpecParsing() {
    TsvParser parser;
    parser = new TsvParser("HBASE_ROW_KEY", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertFalse(parser.hasTimestamp());
    parser = new TsvParser("HBASE_ROW_KEY,col1:scol1", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(1));
    assertBytesEquals(Bytes.toBytes("scol1"), parser.getQualifier(1));
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertFalse(parser.hasTimestamp());
    parser = new TsvParser("HBASE_ROW_KEY,col1:scol1,col1:scol2", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(1));
    assertBytesEquals(Bytes.toBytes("scol1"), parser.getQualifier(1));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(2));
    assertBytesEquals(Bytes.toBytes("scol2"), parser.getQualifier(2));
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertFalse(parser.hasTimestamp());
    parser = new TsvParser("HBASE_ROW_KEY,col1:scol1,HBASE_TS_KEY,col1:scol2", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(1));
    assertBytesEquals(Bytes.toBytes("scol1"), parser.getQualifier(1));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(3));
    assertBytesEquals(Bytes.toBytes("scol2"), parser.getQualifier(3));
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertTrue(parser.hasTimestamp());
    assertEquals(2, parser.getTimestampKeyColumnIndex());
    parser = new TsvParser("HBASE_ROW_KEY,col1:scol1,HBASE_TS_KEY,col1:scol2,HBASE_ATTRIBUTES_KEY", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(1));
    assertBytesEquals(Bytes.toBytes("scol1"), parser.getQualifier(1));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(3));
    assertBytesEquals(Bytes.toBytes("scol2"), parser.getQualifier(3));
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertTrue(parser.hasTimestamp());
    assertEquals(2, parser.getTimestampKeyColumnIndex());
    assertEquals(4, parser.getAttributesKeyColumnIndex());
    parser = new TsvParser("HBASE_ATTRIBUTES_KEY,col1:scol1,HBASE_TS_KEY,col1:scol2,HBASE_ROW_KEY", "\t");
    assertNull(parser.getFamily(0));
    assertNull(parser.getQualifier(0));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(1));
    assertBytesEquals(Bytes.toBytes("scol1"), parser.getQualifier(1));
    assertBytesEquals(Bytes.toBytes("col1"), parser.getFamily(3));
    assertBytesEquals(Bytes.toBytes("scol2"), parser.getQualifier(3));
    assertEquals(4, parser.getRowKeyColumnIndex());
    assertTrue(parser.hasTimestamp());
    assertEquals(2, parser.getTimestampKeyColumnIndex());
    assertEquals(0, parser.getAttributesKeyColumnIndex());
}

Example 2 with TsvParser

use of org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser in project hbase by apache.

the class TestImportTsvParser method testTsvParseAttributesKey.

@Test
public void testTsvParseAttributesKey() throws BadTsvLineException {
    TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a,HBASE_TS_KEY,HBASE_ATTRIBUTES_KEY", "\t");
    assertEquals(0, parser.getRowKeyColumnIndex());
    byte[] line = Bytes.toBytes("rowkey\tval_a\t1234\tkey=>value");
    ParsedLine parse = parser.parse(line, line.length);
    assertEquals(18, parse.getAttributeKeyOffset());
    assertEquals(3, parser.getAttributesKeyColumnIndex());
    String[] attributes = parse.getIndividualAttributes();
    assertEquals(attributes[0], "key=>value");
    try {
        line = Bytes.toBytes("rowkey\tval_a\t1234");
        parser.parse(line, line.length);
        fail("Should get BadTsvLineException on empty rowkey.");
    } catch (BadTsvLineException b) {
    }
    parser = new TsvParser("HBASE_ATTRIBUTES_KEY,col_a,HBASE_ROW_KEY,HBASE_TS_KEY", "\t");
    assertEquals(2, parser.getRowKeyColumnIndex());
    line = Bytes.toBytes("key=>value\tval_a\trowkey\t1234");
    parse = parser.parse(line, line.length);
    assertEquals(0, parse.getAttributeKeyOffset());
    assertEquals(0, parser.getAttributesKeyColumnIndex());
    attributes = parse.getIndividualAttributes();
    assertEquals(attributes[0], "key=>value");
    try {
        line = Bytes.toBytes("val_a");
        ParsedLine parse2 = parser.parse(line, line.length);
        fail("Should get BadTsvLineException when number of columns less than rowkey position.");
    } catch (BadTsvLineException b) {
    }
    parser = new TsvParser("col_a,HBASE_ATTRIBUTES_KEY,HBASE_TS_KEY,HBASE_ROW_KEY", "\t");
    assertEquals(3, parser.getRowKeyColumnIndex());
    line = Bytes.toBytes("val_a\tkey0=>value0,key1=>value1,key2=>value2\t1234\trowkey");
    parse = parser.parse(line, line.length);
    assertEquals(1, parser.getAttributesKeyColumnIndex());
    assertEquals(6, parse.getAttributeKeyOffset());
    String[] attr = parse.getIndividualAttributes();
    int i = 0;
    for (String str : attr) {
        assertEquals(("key" + i + "=>" + "value" + i), str);
        i++;
    }
}

Example 3 with TsvParser

use of org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser in project hbase by apache.

the class TestImportTsvParser method testTsvParserWithCellVisibilityCol.

@Test
public void testTsvParserWithCellVisibilityCol() throws BadTsvLineException {
    TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a,HBASE_TS_KEY,HBASE_ATTRIBUTES_KEY,HBASE_CELL_VISIBILITY", "\t");
    assertEquals(0, parser.getRowKeyColumnIndex());
    assertEquals(4, parser.getCellVisibilityColumnIndex());
    byte[] line = Bytes.toBytes("rowkey\tval_a\t1234\tkey=>value\tPRIVATE&SECRET");
    ParsedLine parse = parser.parse(line, line.length);
    assertEquals(18, parse.getAttributeKeyOffset());
    assertEquals(3, parser.getAttributesKeyColumnIndex());
    String[] attributes = parse.getIndividualAttributes();
    assertEquals(attributes[0], "key=>value");
    assertEquals(29, parse.getCellVisibilityColumnOffset());
}

Example 4 with TsvParser

use of org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser in project hbase by apache.

the class TestImportTsvParser method testTsvParserBadTsvLineOnlyKey.

@Test(expected = BadTsvLineException.class)
public void testTsvParserBadTsvLineOnlyKey() throws BadTsvLineException {
    TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a", "\t");
    byte[] line = Bytes.toBytes("key_only");
    parser.parse(line, line.length);
}

Example 5 with TsvParser

use of org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser in project hbase by apache.

the class TestImportTsvParser method testTsvParserNoTimestampValue.

@Test(expected = BadTsvLineException.class)
public void testTsvParserNoTimestampValue() throws BadTsvLineException {
    TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a,HBASE_TS_KEY", "\t");
    assertEquals(2, parser.getTimestampKeyColumnIndex());
    byte[] line = Bytes.toBytes("rowkey\tval_a");
    parser.parse(line, line.length);
}