@Test public void testTsvParser() throws BadTsvLineException { TsvParser parser = new TsvParser("col_a,col_b:qual,HBASE_ROW_KEY,col_d", "\t"); assertBytesEquals(Bytes.toBytes("col_a"), parser.getFamily(0)); assertBytesEquals(HConstants.EMPTY_BYTE_ARRAY, parser.getQualifier(0)); assertBytesEquals(Bytes.toBytes("col_b"), parser.getFamily(1)); assertBytesEquals(Bytes.toBytes("qual"), parser.getQualifier(1)); assertNull(parser.getFamily(2)); assertNull(parser.getQualifier(2)); assertEquals(2, parser.getRowKeyColumnIndex()); assertEquals(TsvParser.DEFAULT_TIMESTAMP_COLUMN_INDEX, parser.getTimestampKeyColumnIndex()); byte[] line = Bytes.toBytes("val_a\tval_b\tval_c\tval_d"); ParsedLine parsed = parser.parse(line, line.length); checkParsing(parsed, Splitter.on("\t").split(Bytes.toString(line))); }
@Test public void testTsvParserWithTimestamp() throws BadTsvLineException { TsvParser parser = new TsvParser("HBASE_ROW_KEY,HBASE_TS_KEY,col_a,", "\t"); assertNull(parser.getFamily(0)); assertNull(parser.getQualifier(0)); assertNull(parser.getFamily(1)); assertNull(parser.getQualifier(1)); assertBytesEquals(Bytes.toBytes("col_a"), parser.getFamily(2)); assertBytesEquals(HConstants.EMPTY_BYTE_ARRAY, parser.getQualifier(2)); assertEquals(0, parser.getRowKeyColumnIndex()); assertEquals(1, parser.getTimestampKeyColumnIndex()); byte[] line = Bytes.toBytes("rowkey\t1234\tval_a"); ParsedLine parsed = parser.parse(line, line.length); assertEquals(1234l, parsed.getTimestamp(-1)); checkParsing(parsed, Splitter.on("\t").split(Bytes.toString(line))); }
@Test public void testTsvParser() throws BadTsvLineException { TsvParser parser = new TsvParser("col_a,col_b:qual,HBASE_ROW_KEY,col_d", "\t"); assertBytesEquals(Bytes.toBytes("col_a"), parser.getFamily(0)); assertBytesEquals(HConstants.EMPTY_BYTE_ARRAY, parser.getQualifier(0)); assertBytesEquals(Bytes.toBytes("col_b"), parser.getFamily(1)); assertBytesEquals(Bytes.toBytes("qual"), parser.getQualifier(1)); assertNull(parser.getFamily(2)); assertNull(parser.getQualifier(2)); assertEquals(2, parser.getRowKeyColumnIndex()); assertEquals(TsvParser.DEFAULT_TIMESTAMP_COLUMN_INDEX, parser .getTimestampKeyColumnIndex()); byte[] line = Bytes.toBytes("val_a\tval_b\tval_c\tval_d"); ParsedLine parsed = parser.parse(line, line.length); checkParsing(parsed, Splitter.on("\t").split(Bytes.toString(line))); }
/** * Handles initializing this class with objects specific to it (i.e., the parser). Common * initialization that might be leveraged by a subsclass is done in <code>doSetup</code>. Hence a * subclass may choose to override this method and call <code>doSetup</code> as well before * handling it's own custom params. * @param context */ @Override protected void setup(Context context) throws IOException { doSetup(context); Configuration conf = context.getConfiguration(); parser = new TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator); if (parser.getRowKeyColumnIndex() == -1) { throw new RuntimeException("No row key column specified"); } String tableName = conf.get(TableInputFormat.INPUT_TABLE); HTable hTable = null; try { hTable = new HTable(conf, tableName); this.startKeys = hTable.getStartKeys(); byte[] indexBytes = hTable.getTableDescriptor().getValue(Constants.INDEX_SPEC_KEY); if (indexBytes != null) { TableIndices tableIndices = new TableIndices(); tableIndices.readFields(indexBytes); this.indices = tableIndices.getIndices(); } } finally { if (hTable != null) hTable.close(); } }
@Test public void testTsvParser() throws BadTsvLineException { TsvParser parser = new TsvParser("col_a,col_b:qual,HBASE_ROW_KEY,col_d", "\t"); assertBytesEquals(Bytes.toBytes("col_a"), parser.getFamily(0)); assertBytesEquals(HConstants.EMPTY_BYTE_ARRAY, parser.getQualifier(0)); assertBytesEquals(Bytes.toBytes("col_b"), parser.getFamily(1)); assertBytesEquals(Bytes.toBytes("qual"), parser.getQualifier(1)); assertNull(parser.getFamily(2)); assertNull(parser.getQualifier(2)); assertEquals(2, parser.getRowKeyColumnIndex()); assertEquals(ImportTsv.TsvParser.DEFAULT_TIMESTAMP_COLUMN_INDEX, parser.getTimestampKeyColumnIndex()); byte[] line = Bytes.toBytes("val_a\tval_b\tval_c\tval_d"); ParsedLine parsed = parser.parse(line, line.length); checkParsing(parsed, Splitter.on("\t").split(Bytes.toString(line))); }
/** * Test cases that throw BadTsvLineException */ @Test(expected = BadTsvLineException.class) public void testTsvParserBadTsvLineExcessiveColumns() throws BadTsvLineException { TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a", "\t"); byte[] line = Bytes.toBytes("val_a\tval_b\tval_c"); parser.parse(line, line.length); }
@Test(expected = BadTsvLineException.class) public void testTsvParserInvalidTimestamp() throws BadTsvLineException { TsvParser parser = new TsvParser("HBASE_ROW_KEY,HBASE_TS_KEY,col_a,", "\t"); assertEquals(1, parser.getTimestampKeyColumnIndex()); byte[] line = Bytes.toBytes("rowkey\ttimestamp\tval_a"); ParsedLine parsed = parser.parse(line, line.length); assertEquals(-1, parsed.getTimestamp(-1)); checkParsing(parsed, Splitter.on("\t").split(Bytes.toString(line))); }
@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); }
@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()); }
/** * Test cases that throw BadTsvLineException */ @Test(expected=BadTsvLineException.class) public void testTsvParserBadTsvLineExcessiveColumns() throws BadTsvLineException { TsvParser parser = new TsvParser("HBASE_ROW_KEY,col_a", "\t"); byte[] line = Bytes.toBytes("val_a\tval_b\tval_c"); parser.parse(line, line.length); }
