org.apache.hadoop.hbase.mapreduce

Class ImportTsv.TsvParser

java.lang.Object

org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser

Enclosing class:

ImportTsv

public static class ImportTsv.TsvParser
extends Object

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	ImportTsv.TsvParser.BadTsvLineException
(package private) class	ImportTsv.TsvParser.ParsedLine

Field Summary

Fields

Modifier and Type	Field and Description
static String	ATTRIBUTES_COLUMN_SPEC
private int	attrKeyColumnIndex
static String	CELL_TTL_COLUMN_SPEC
static String	CELL_VISIBILITY_COLUMN_SPEC
private int	cellTTLColumnIndex
private int	cellVisibilityColumnIndex
static int	DEFAULT_ATTRIBUTES_COLUMN_INDEX
static int	DEFAULT_CELL_TTL_COLUMN_INDEX
static int	DEFAULT_CELL_VISIBILITY_COLUMN_INDEX
static int	DEFAULT_TIMESTAMP_COLUMN_INDEX
private byte[][]	families Column families and qualifiers mapped to the TSV columns
private int	maxColumnCount
private byte[][]	qualifiers
static String	ROWKEY_COLUMN_SPEC
private int	rowKeyColumnIndex
private byte	separatorByte
static String	TIMESTAMPKEY_COLUMN_SPEC
private int	timestampKeyColumnIndex

Constructor Summary

Constructors

Constructor and Description
TsvParser(String columnsSpecification, String separatorStr)

Method Summary

Modifier and Type	Method and Description
int	getAttributesKeyColumnIndex()
int	getCellTTLColumnIndex()
int	getCellVisibilityColumnIndex()
byte[]	getFamily(int idx)
byte[]	getQualifier(int idx)
int	getRowKeyColumnIndex()
int	getTimestampKeyColumnIndex()
boolean	hasAttributes()
boolean	hasCellTTL()
boolean	hasCellVisibility()
boolean	hasTimestamp()
ImportTsv.TsvParser.ParsedLine	parse(byte[] lineBytes, int length)
Pair<Integer,Integer>	parseRowKey(byte[] lineBytes, int length) Return starting position and length of row key from the specified line bytes.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

families

private final byte[][] families

Column families and qualifiers mapped to the TSV columns

qualifiers

private final byte[][] qualifiers

separatorByte

private final byte separatorByte

rowKeyColumnIndex

private int rowKeyColumnIndex

maxColumnCount

private int maxColumnCount

DEFAULT_TIMESTAMP_COLUMN_INDEX

public static final int DEFAULT_TIMESTAMP_COLUMN_INDEX

See Also:

Constant Field Values

timestampKeyColumnIndex

private int timestampKeyColumnIndex

ROWKEY_COLUMN_SPEC

public static final String ROWKEY_COLUMN_SPEC

See Also:

Constant Field Values

TIMESTAMPKEY_COLUMN_SPEC

public static final String TIMESTAMPKEY_COLUMN_SPEC

See Also:

Constant Field Values

ATTRIBUTES_COLUMN_SPEC

public static final String ATTRIBUTES_COLUMN_SPEC

See Also:

Constant Field Values

CELL_VISIBILITY_COLUMN_SPEC

public static final String CELL_VISIBILITY_COLUMN_SPEC

See Also:

Constant Field Values

CELL_TTL_COLUMN_SPEC

public static final String CELL_TTL_COLUMN_SPEC

See Also:

Constant Field Values

attrKeyColumnIndex

private int attrKeyColumnIndex

DEFAULT_ATTRIBUTES_COLUMN_INDEX

public static final int DEFAULT_ATTRIBUTES_COLUMN_INDEX

See Also:

Constant Field Values

DEFAULT_CELL_VISIBILITY_COLUMN_INDEX

public static final int DEFAULT_CELL_VISIBILITY_COLUMN_INDEX

See Also:

Constant Field Values

DEFAULT_CELL_TTL_COLUMN_INDEX

public static final int DEFAULT_CELL_TTL_COLUMN_INDEX

See Also:

Constant Field Values

cellVisibilityColumnIndex

private int cellVisibilityColumnIndex

cellTTLColumnIndex

private int cellTTLColumnIndex

Constructor Detail

TsvParser

public TsvParser(String columnsSpecification,
                 String separatorStr)

Parameters:

columnsSpecification - the list of columns to parser out, comma separated. The row key should be the special token TsvParser.ROWKEY_COLUMN_SPEC n

Method Detail

hasTimestamp

public boolean hasTimestamp()

getTimestampKeyColumnIndex

public int getTimestampKeyColumnIndex()

hasAttributes

public boolean hasAttributes()

hasCellVisibility

public boolean hasCellVisibility()

hasCellTTL

public boolean hasCellTTL()

getAttributesKeyColumnIndex

public int getAttributesKeyColumnIndex()

getCellVisibilityColumnIndex

public int getCellVisibilityColumnIndex()

getCellTTLColumnIndex

public int getCellTTLColumnIndex()

getRowKeyColumnIndex

public int getRowKeyColumnIndex()

getFamily

public byte[] getFamily(int idx)

getQualifier

public byte[] getQualifier(int idx)

parse

public ImportTsv.TsvParser.ParsedLine parse(byte[] lineBytes,
                                            int length)
                                     throws ImportTsv.TsvParser.BadTsvLineException

Throws:

ImportTsv.TsvParser.BadTsvLineException

parseRowKey

public Pair<Integer,Integer> parseRowKey(byte[] lineBytes,
                                         int length)
                                  throws ImportTsv.TsvParser.BadTsvLineException

Return starting position and length of row key from the specified line bytes. nn * @return Pair of row key offset and length. n

Throws:

ImportTsv.TsvParser.BadTsvLineException