Search in sources :

Example 36 with ColumnVisibility

use of org.apache.accumulo.core.security.ColumnVisibility 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);
}
Also used : ColumnUpdate(org.apache.accumulo.core.data.ColumnUpdate) Configuration(org.apache.hadoop.conf.Configuration) LazySerDeParameters(org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters) HashMap(java.util.HashMap) LazyString(org.apache.hadoop.hive.serde2.lazy.LazyString) Properties(java.util.Properties) AccumuloHiveRow(org.apache.hadoop.hive.accumulo.AccumuloHiveRow) ColumnVisibility(org.apache.accumulo.core.security.ColumnVisibility) LazyStruct(org.apache.hadoop.hive.serde2.lazy.LazyStruct) LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) LazyStringObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyStringObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) LazyMapObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector) LazyStringObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyStringObjectInspector) LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) LazyAccumuloRow(org.apache.hadoop.hive.accumulo.LazyAccumuloRow) LazyMap(org.apache.hadoop.hive.serde2.lazy.LazyMap) LazyMapObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) ByteArrayRef(org.apache.hadoop.hive.serde2.lazy.ByteArrayRef) Mutation(org.apache.accumulo.core.data.Mutation) Test(org.junit.Test)

Example 37 with ColumnVisibility

use of org.apache.accumulo.core.security.ColumnVisibility in project hive by apache.

the class TestAccumuloRowSerializer method testInvalidRowIdOffset.

@Test(expected = IllegalArgumentException.class)
public void testInvalidRowIdOffset() throws SerDeException {
    ArrayList<ColumnMapping> mappings = new ArrayList<ColumnMapping>();
    // Should fail because of the -1
    new AccumuloRowSerializer(-1, null, mappings, new ColumnVisibility(), null);
}
Also used : ArrayList(java.util.ArrayList) ColumnVisibility(org.apache.accumulo.core.security.ColumnVisibility) ColumnMapping(org.apache.hadoop.hive.accumulo.columns.ColumnMapping) Test(org.junit.Test)

Example 38 with ColumnVisibility

use of org.apache.accumulo.core.security.ColumnVisibility 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()));
}
Also used : LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) ColumnUpdate(org.apache.accumulo.core.data.ColumnUpdate) Configuration(org.apache.hadoop.conf.Configuration) LazySerDeParameters(org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters) DataOutputStream(java.io.DataOutputStream) ArrayList(java.util.ArrayList) ByteArrayOutputStream(java.io.ByteArrayOutputStream) Properties(java.util.Properties) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) ByteArrayRef(org.apache.hadoop.hive.serde2.lazy.ByteArrayRef) ColumnVisibility(org.apache.accumulo.core.security.ColumnVisibility) Mutation(org.apache.accumulo.core.data.Mutation) LazyStruct(org.apache.hadoop.hive.serde2.lazy.LazyStruct) Test(org.junit.Test)

Example 39 with ColumnVisibility

use of org.apache.accumulo.core.security.ColumnVisibility 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()));
}
Also used : LazyStringObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyStringObjectInspector) LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) ColumnUpdate(org.apache.accumulo.core.data.ColumnUpdate) Configuration(org.apache.hadoop.conf.Configuration) LazySerDeParameters(org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters) ArrayList(java.util.ArrayList) LazyMapObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector) Properties(java.util.Properties) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) ByteArrayRef(org.apache.hadoop.hive.serde2.lazy.ByteArrayRef) ColumnVisibility(org.apache.accumulo.core.security.ColumnVisibility) Mutation(org.apache.accumulo.core.data.Mutation) LazyStruct(org.apache.hadoop.hive.serde2.lazy.LazyStruct) Test(org.junit.Test)

Example 40 with ColumnVisibility

use of org.apache.accumulo.core.security.ColumnVisibility in project hive by apache.

the class TestHiveAccumuloTableInputFormat method testGetProtectedField.

@Test
public void testGetProtectedField() throws Exception {
    FileInputFormat.addInputPath(conf, new Path("unused"));
    BatchWriterConfig writerConf = new BatchWriterConfig();
    BatchWriter writer = con.createBatchWriter(TEST_TABLE, writerConf);
    Authorizations origAuths = con.securityOperations().getUserAuthorizations(USER);
    con.securityOperations().changeUserAuthorizations(USER, new Authorizations(origAuths.toString() + ",foo"));
    Mutation m = new Mutation("r4");
    m.put(COLUMN_FAMILY, NAME, new ColumnVisibility("foo"), new Value("frank".getBytes()));
    m.put(COLUMN_FAMILY, SID, new ColumnVisibility("foo"), new Value(parseIntBytes("4")));
    m.put(COLUMN_FAMILY, DEGREES, new ColumnVisibility("foo"), new Value(parseDoubleBytes("60.6")));
    m.put(COLUMN_FAMILY, MILLIS, new ColumnVisibility("foo"), new Value(parseLongBytes("777")));
    writer.addMutation(m);
    writer.close();
    conf.set(AccumuloSerDeParameters.AUTHORIZATIONS_KEY, "foo");
    InputSplit[] splits = inputformat.getSplits(conf, 0);
    assertEquals(splits.length, 1);
    RecordReader<Text, AccumuloHiveRow> reader = inputformat.getRecordReader(splits[0], conf, null);
    Text rowId = new Text("r1");
    AccumuloHiveRow row = new AccumuloHiveRow();
    assertTrue(reader.next(rowId, row));
    assertEquals(row.getRowId(), rowId.toString());
    assertTrue(row.hasFamAndQual(COLUMN_FAMILY, NAME));
    assertArrayEquals(row.getValue(COLUMN_FAMILY, NAME), "brian".getBytes());
    rowId = new Text("r2");
    assertTrue(reader.next(rowId, row));
    assertEquals(row.getRowId(), rowId.toString());
    assertTrue(row.hasFamAndQual(COLUMN_FAMILY, NAME));
    assertArrayEquals(row.getValue(COLUMN_FAMILY, NAME), "mark".getBytes());
    rowId = new Text("r3");
    assertTrue(reader.next(rowId, row));
    assertEquals(row.getRowId(), rowId.toString());
    assertTrue(row.hasFamAndQual(COLUMN_FAMILY, NAME));
    assertArrayEquals(row.getValue(COLUMN_FAMILY, NAME), "dennis".getBytes());
    rowId = new Text("r4");
    assertTrue(reader.next(rowId, row));
    assertEquals(row.getRowId(), rowId.toString());
    assertTrue(row.hasFamAndQual(COLUMN_FAMILY, NAME));
    assertArrayEquals(row.getValue(COLUMN_FAMILY, NAME), "frank".getBytes());
    assertFalse(reader.next(rowId, row));
}
Also used : Path(org.apache.hadoop.fs.Path) Authorizations(org.apache.accumulo.core.security.Authorizations) Value(org.apache.accumulo.core.data.Value) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) Text(org.apache.hadoop.io.Text) BatchWriter(org.apache.accumulo.core.client.BatchWriter) Mutation(org.apache.accumulo.core.data.Mutation) ColumnVisibility(org.apache.accumulo.core.security.ColumnVisibility) InputSplit(org.apache.hadoop.mapred.InputSplit) AccumuloHiveRow(org.apache.hadoop.hive.accumulo.AccumuloHiveRow) Test(org.junit.Test)

Aggregations

ColumnVisibility (org.apache.accumulo.core.security.ColumnVisibility)145 Mutation (org.apache.accumulo.core.data.Mutation)62 Text (org.apache.hadoop.io.Text)59 Value (org.apache.accumulo.core.data.Value)57 Key (org.apache.accumulo.core.data.Key)43 BatchWriter (org.apache.accumulo.core.client.BatchWriter)31 Test (org.junit.Test)23 BatchWriterConfig (org.apache.accumulo.core.client.BatchWriterConfig)19 Test (org.junit.jupiter.api.Test)18 ArrayList (java.util.ArrayList)17 Entry (java.util.Map.Entry)17 Authorizations (org.apache.accumulo.core.security.Authorizations)17 TableNotFoundException (org.apache.accumulo.core.client.TableNotFoundException)16 Scanner (org.apache.accumulo.core.client.Scanner)14 MutationsRejectedException (org.apache.accumulo.core.client.MutationsRejectedException)13 Configuration (org.apache.hadoop.conf.Configuration)13 ColumnUpdate (org.apache.accumulo.core.data.ColumnUpdate)12 AccumuloException (org.apache.accumulo.core.client.AccumuloException)11 TMutation (org.apache.accumulo.core.dataImpl.thrift.TMutation)11 IteratorSetting (org.apache.accumulo.core.client.IteratorSetting)10