use of org.apache.accumulo.core.data.ColumnUpdate in project hive by apache.
the class TestAccumuloRowSerializer method testBinarySerialization.
@Test
public void testBinarySerialization() throws IOException, SerDeException {
List<String> columns = Arrays.asList("row", "cq1", "cq2", "cq3");
List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.stringTypeInfo);
List<String> typeNames = new ArrayList<String>(types.size());
for (TypeInfo type : types) {
typeNames.add(type.getTypeName());
}
Properties tableProperties = new Properties();
tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,cf:cq1#b,cf:cq2#b,cf:cq3");
tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(typeNames));
AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
LazySimpleStructObjectInspector oi = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(columns, types, serDeParams.getSeparators(), serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), new ColumnVisibility(), accumuloSerDeParams.getRowIdFactory());
// Create the LazyStruct from the LazyStruct...Inspector
LazyStruct obj = (LazyStruct) LazyFactory.createLazyObject(oi);
ByteArrayRef byteRef = new ByteArrayRef();
byteRef.setData(new byte[] { 'r', 'o', 'w', '1', ' ', '1', '0', ' ', '2', '0', ' ', 'v', 'a', 'l', 'u', 'e' });
obj.init(byteRef, 0, byteRef.getData().length);
Mutation m = (Mutation) serializer.serialize(obj, oi);
Assert.assertArrayEquals("row1".getBytes(), m.getRow());
List<ColumnUpdate> updates = m.getUpdates();
Assert.assertEquals(3, updates.size());
ByteArrayOutputStream baos = new ByteArrayOutputStream();
DataOutputStream out = new DataOutputStream(baos);
ColumnUpdate update = updates.get(0);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq1", new String(update.getColumnQualifier()));
out.writeInt(10);
Assert.assertArrayEquals(baos.toByteArray(), update.getValue());
update = updates.get(1);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq2", new String(update.getColumnQualifier()));
baos.reset();
out.writeInt(20);
Assert.assertArrayEquals(baos.toByteArray(), update.getValue());
update = updates.get(2);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq3", new String(update.getColumnQualifier()));
Assert.assertEquals("value", new String(update.getValue()));
}
use of org.apache.accumulo.core.data.ColumnUpdate in project hive by apache.
the class TestAccumuloSerDe method testStructOfMapSerialization.
@Test
public void testStructOfMapSerialization() throws IOException, SerDeException {
List<String> columns = Arrays.asList("row", "col");
List<String> structColNames = Arrays.asList("map1", "map2");
TypeInfo mapTypeInfo = TypeInfoFactory.getMapTypeInfo(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
// struct<map1:map<string,string>,map2:map<string,string>>,string
List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.getStructTypeInfo(structColNames, Arrays.asList(mapTypeInfo, mapTypeInfo)), TypeInfoFactory.stringTypeInfo);
Properties tableProperties = new Properties();
tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,cf:cq");
// Use the default separators [0, 1, 2, 3, ..., 7]
tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
byte[] seps = serDeParams.getSeparators();
// struct<map<k:v,k:v>_map<k:v,k:v>>>
TypeInfo stringTypeInfo = TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME);
LazyStringObjectInspector stringOI = (LazyStringObjectInspector) LazyFactory.createLazyObjectInspector(stringTypeInfo, new byte[] { 0 }, 0, serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
LazyMapObjectInspector mapOI = LazyObjectInspectorFactory.getLazySimpleMapObjectInspector(stringOI, stringOI, seps[3], seps[4], serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
LazySimpleStructObjectInspector rowStructOI = (LazySimpleStructObjectInspector) LazyObjectInspectorFactory.getLazySimpleStructObjectInspector(structColNames, Arrays.<ObjectInspector>asList(mapOI, mapOI), (byte) seps[2], serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyObjectInspectorFactory.getLazySimpleStructObjectInspector(columns, Arrays.asList(rowStructOI, stringOI), seps[1], serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), new ColumnVisibility(), accumuloSerDeParams.getRowIdFactory());
Map<String, String> map1 = new HashMap<String, String>(), map2 = new HashMap<String, String>();
map1.put("key10", "value10");
map1.put("key11", "value11");
map2.put("key20", "value20");
map2.put("key21", "value21");
ByteArrayRef byteRef = new ByteArrayRef();
// Default separators are 1-indexed (instead of 0-indexed), thus the separator at offset 1 is
// (byte) 2
// The separator for the hive row is \x02, for the row Id struct, \x03, and the maps \x04 and
// \x05
String accumuloRow = "key10\5value10\4key11\5value11\3key20\5value20\4key21\5value21";
LazyStruct entireStruct = (LazyStruct) LazyFactory.createLazyObject(structOI);
byteRef.setData((accumuloRow + "\2foo").getBytes());
entireStruct.init(byteRef, 0, byteRef.getData().length);
Mutation m = serializer.serialize(entireStruct, structOI);
Assert.assertArrayEquals(accumuloRow.getBytes(), m.getRow());
Assert.assertEquals(1, m.getUpdates().size());
ColumnUpdate update = m.getUpdates().get(0);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq", new String(update.getColumnQualifier()));
Assert.assertEquals("foo", new String(update.getValue()));
AccumuloHiveRow haRow = new AccumuloHiveRow(new String(m.getRow()));
haRow.add("cf", "cq", "foo".getBytes());
LazyAccumuloRow lazyAccumuloRow = new LazyAccumuloRow(structOI);
lazyAccumuloRow.init(haRow, accumuloSerDeParams.getColumnMappings(), accumuloSerDeParams.getRowIdFactory());
List<Object> objects = lazyAccumuloRow.getFieldsAsList();
Assert.assertEquals(2, objects.size());
Assert.assertEquals("foo", objects.get(1).toString());
LazyStruct rowStruct = (LazyStruct) objects.get(0);
List<Object> rowObjects = rowStruct.getFieldsAsList();
Assert.assertEquals(2, rowObjects.size());
LazyMap rowMap = (LazyMap) rowObjects.get(0);
Map<?, ?> actualMap = rowMap.getMap();
System.out.println("Actual map 1: " + actualMap);
Map<String, String> actualStringMap = new HashMap<String, String>();
for (Entry<?, ?> entry : actualMap.entrySet()) {
actualStringMap.put(entry.getKey().toString(), entry.getValue().toString());
}
Assert.assertEquals(map1, actualStringMap);
rowMap = (LazyMap) rowObjects.get(1);
actualMap = rowMap.getMap();
System.out.println("Actual map 2: " + actualMap);
actualStringMap = new HashMap<String, String>();
for (Entry<?, ?> entry : actualMap.entrySet()) {
actualStringMap.put(entry.getKey().toString(), entry.getValue().toString());
}
Assert.assertEquals(map2, actualStringMap);
}
use of org.apache.accumulo.core.data.ColumnUpdate in project hive by apache.
the class TestAccumuloRowSerializer method testMapSerialization.
@Test
public void testMapSerialization() throws IOException, SerDeException {
List<String> columns = Arrays.asList("row", "col");
List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.getMapTypeInfo(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo));
List<String> typeNames = new ArrayList<String>(types.size());
for (TypeInfo type : types) {
typeNames.add(type.getTypeName());
}
Properties tableProperties = new Properties();
tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,cf:*");
tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
tableProperties.setProperty(serdeConstants.COLLECTION_DELIM, ",");
tableProperties.setProperty(serdeConstants.MAPKEY_DELIM, ":");
tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(typeNames));
AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
TypeInfo stringTypeInfo = TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME);
LazyStringObjectInspector stringOI = (LazyStringObjectInspector) LazyFactory.createLazyObjectInspector(stringTypeInfo, new byte[] { 0 }, 0, serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
LazyMapObjectInspector mapOI = LazyObjectInspectorFactory.getLazySimpleMapObjectInspector(stringOI, stringOI, (byte) ',', (byte) ':', serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyObjectInspectorFactory.getLazySimpleStructObjectInspector(columns, Arrays.asList(stringOI, mapOI), (byte) ' ', serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), new ColumnVisibility(), accumuloSerDeParams.getRowIdFactory());
// Create the LazyStruct from the LazyStruct...Inspector
LazyStruct obj = (LazyStruct) LazyFactory.createLazyObject(structOI);
ByteArrayRef byteRef = new ByteArrayRef();
byteRef.setData("row1 cq1:10,cq2:20,cq3:value".getBytes());
obj.init(byteRef, 0, byteRef.getData().length);
Mutation m = (Mutation) serializer.serialize(obj, structOI);
Assert.assertArrayEquals("row1".getBytes(), m.getRow());
List<ColumnUpdate> updates = m.getUpdates();
Assert.assertEquals(3, updates.size());
ColumnUpdate update = updates.get(0);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq1", new String(update.getColumnQualifier()));
Assert.assertEquals("10", new String(update.getValue()));
update = updates.get(1);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq2", new String(update.getColumnQualifier()));
Assert.assertEquals("20", new String(update.getValue()));
update = updates.get(2);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq3", new String(update.getColumnQualifier()));
Assert.assertEquals("value", new String(update.getValue()));
}
use of org.apache.accumulo.core.data.ColumnUpdate in project presto by prestodb.
the class Indexer method index.
/**
* Index the given mutation, adding mutations to the index and metrics table
* <p>
* Like typical use of a BatchWriter, this method does not flush mutations to the underlying index table.
* For higher throughput the modifications to the metrics table are tracked in memory and added to the metrics table when the indexer is flushed or closed.
*
* @param mutation Mutation to index
*/
public void index(Mutation mutation) {
// Increment the cardinality for the number of rows in the table
metrics.get(METRICS_TABLE_ROW_COUNT).incrementAndGet();
// Set the first and last row values of the table based on existing row IDs
if (firstRow == null || byteArrayComparator.compare(mutation.getRow(), firstRow) < 0) {
firstRow = mutation.getRow();
}
if (lastRow == null || byteArrayComparator.compare(mutation.getRow(), lastRow) > 0) {
lastRow = mutation.getRow();
}
// For each column update in this mutation
for (ColumnUpdate columnUpdate : mutation.getUpdates()) {
// Get the column qualifiers we want to index for this column family (if any)
ByteBuffer family = wrap(columnUpdate.getColumnFamily());
Collection<ByteBuffer> indexQualifiers = indexColumns.get(family);
// If we have column qualifiers we want to index for this column family
if (indexQualifiers != null) {
// Check if we want to index this particular qualifier
ByteBuffer qualifier = wrap(columnUpdate.getColumnQualifier());
if (indexQualifiers.contains(qualifier)) {
// If so, create a mutation using the following mapping:
// Row ID = column value
// Column Family = columnqualifier_columnfamily
// Column Qualifier = row ID
// Value = empty
ByteBuffer indexFamily = getIndexColumnFamily(columnUpdate.getColumnFamily(), columnUpdate.getColumnQualifier());
Type type = indexColumnTypes.get(family).get(qualifier);
ColumnVisibility visibility = new ColumnVisibility(columnUpdate.getColumnVisibility());
// If this is an array type, then index each individual element in the array
if (Types.isArrayType(type)) {
Type elementType = Types.getElementType(type);
List<?> elements = serializer.decode(type, columnUpdate.getValue());
for (Object element : elements) {
addIndexMutation(wrap(serializer.encode(elementType, element)), indexFamily, visibility, mutation.getRow());
}
} else {
addIndexMutation(wrap(columnUpdate.getValue()), indexFamily, visibility, mutation.getRow());
}
}
}
}
}
use of org.apache.accumulo.core.data.ColumnUpdate in project hive by apache.
the class TestAccumuloRowSerializer method testVisibilityLabel.
@Test
public void testVisibilityLabel() throws IOException, SerDeException {
List<String> columns = Arrays.asList("row", "cq1", "cq2", "cq3");
List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.stringTypeInfo);
List<String> typeNames = new ArrayList<String>(types.size());
for (TypeInfo type : types) {
typeNames.add(type.getTypeName());
}
Properties tableProperties = new Properties();
tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,cf:cq1#b,cf:cq2#b,cf:cq3");
tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(typeNames));
AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
LazySimpleStructObjectInspector oi = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(columns, types, serDeParams.getSeparators(), serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), new ColumnVisibility("foo"), accumuloSerDeParams.getRowIdFactory());
// Create the LazyStruct from the LazyStruct...Inspector
LazyStruct obj = (LazyStruct) LazyFactory.createLazyObject(oi);
ByteArrayRef byteRef = new ByteArrayRef();
byteRef.setData(new byte[] { 'r', 'o', 'w', '1', ' ', '1', '0', ' ', '2', '0', ' ', 'v', 'a', 'l', 'u', 'e' });
obj.init(byteRef, 0, byteRef.getData().length);
Mutation m = (Mutation) serializer.serialize(obj, oi);
Assert.assertArrayEquals("row1".getBytes(), m.getRow());
List<ColumnUpdate> updates = m.getUpdates();
Assert.assertEquals(3, updates.size());
ByteArrayOutputStream baos = new ByteArrayOutputStream();
DataOutputStream out = new DataOutputStream(baos);
ColumnUpdate update = updates.get(0);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq1", new String(update.getColumnQualifier()));
Assert.assertEquals("foo", new String(update.getColumnVisibility()));
out.writeInt(10);
Assert.assertArrayEquals(baos.toByteArray(), update.getValue());
update = updates.get(1);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq2", new String(update.getColumnQualifier()));
Assert.assertEquals("foo", new String(update.getColumnVisibility()));
baos.reset();
out.writeInt(20);
Assert.assertArrayEquals(baos.toByteArray(), update.getValue());
update = updates.get(2);
Assert.assertEquals("cf", new String(update.getColumnFamily()));
Assert.assertEquals("cq3", new String(update.getColumnQualifier()));
Assert.assertEquals("foo", new String(update.getColumnVisibility()));
Assert.assertEquals("value", new String(update.getValue()));
}
Aggregations