private List<Tuple> getAllPartitions(Schema outSchema) {
List<TablePartitionProto> partitionList = masterContext.getCatalog().getAllPartitions();
List<Tuple> tuples = new ArrayList<Tuple>(partitionList.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (TablePartitionProto partition : partitionList) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("partition_id".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(partition.getPartitionId()));
} else if ("tid".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(partition.getTid()));
} else if ("partition_name".equalsIgnoreCase(column.getSimpleName())) {
if (partition.hasPartitionName()) {
aTuple.put(fieldId, DatumFactory.createText(partition.getPartitionName()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("path".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(partition.getPath()));
}
}
tuples.add(aTuple);
}
return tuples;
}
/**
* Creates a new TajoRecordConverter.
*
* @param parquetSchema The Parquet schema of the projection.
* @param tajoReadSchema The Tajo schema of the table.
* @param projectionMap An array mapping the projection column to the column index in the table.
*/
public TajoRecordConverter(GroupType parquetSchema, Schema tajoReadSchema, int[] projectionMap) {
this.parquetSchema = parquetSchema;
this.tajoReadSchema = tajoReadSchema;
this.projectionMap = projectionMap;
this.tupleSize = tajoReadSchema.size();
// The projectionMap.length does not match parquetSchema.getFieldCount()
// when the projection contains NULL_TYPE columns. We will skip over the
// NULL_TYPE columns when we construct the converters and populate the
// NULL_TYPE columns with NullDatums in start().
int index = 0;
this.converters = new Converter[parquetSchema.getFieldCount()];
for (int i = 0; i < projectionMap.length; ++i) {
final int projectionIndex = projectionMap[i];
Column column = tajoReadSchema.getColumn(projectionIndex);
if (column.getDataType().getType() == TajoDataTypes.Type.NULL_TYPE) {
continue;
}
Type type = parquetSchema.getType(index);
final int writeIndex = i;
converters[index] =
newConverter(
column,
type,
new ParentValueContainer() {
@Override
void add(Object value) {
TajoRecordConverter.this.set(writeIndex, value);
}
});
++index;
}
}
private List<Tuple> getAllTableOptions(Schema outSchema) {
List<TableOptionProto> optionList = masterContext.getCatalog().getAllTableOptions();
List<Tuple> tuples = new ArrayList<Tuple>(optionList.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (TableOptionProto option : optionList) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("tid".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(option.getTid()));
} else if ("key_".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(option.getKeyval().getKey()));
} else if ("value_".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(option.getKeyval().getValue()));
}
}
tuples.add(aTuple);
}
return tuples;
}
private List<Tuple> getAllTableStats(Schema outSchema) {
List<TableStatsProto> statList = masterContext.getCatalog().getAllTableStats();
List<Tuple> tuples = new ArrayList<Tuple>(statList.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (TableStatsProto stat : statList) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("tid".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(stat.getTid()));
} else if ("num_rows".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(stat.getNumRows()));
} else if ("num_bytes".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(stat.getNumBytes()));
}
}
tuples.add(aTuple);
}
return tuples;
}
private List<Tuple> getTables(Schema outSchema) {
List<TableDescriptorProto> tables = masterContext.getCatalog().getAllTables();
List<Tuple> tuples = new ArrayList<Tuple>(tables.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (TableDescriptorProto table : tables) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("tid".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(table.getTid()));
} else if ("db_id".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(table.getDbId()));
} else if ("table_name".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(table.getName()));
} else if ("table_type".equalsIgnoreCase(column.getSimpleName())) {
if (table.hasTableType()) {
aTuple.put(fieldId, DatumFactory.createText(table.getTableType()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("path".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(table.getPath()));
} else if ("store_type".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(table.getStoreType()));
}
}
tuples.add(aTuple);
}
return tuples;
}
private List<Tuple> getDatabases(Schema outSchema) {
List<DatabaseProto> databases = masterContext.getCatalog().getAllDatabases();
List<Tuple> tuples = new ArrayList<Tuple>(databases.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (DatabaseProto database : databases) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("db_id".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(database.getId()));
} else if ("db_name".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(database.getName()));
} else if ("space_id".equalsIgnoreCase(column.getSimpleName())) {
if (database.hasSpaceId()) {
aTuple.put(fieldId, DatumFactory.createInt4(database.getSpaceId()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
}
}
tuples.add(aTuple);
}
return tuples;
}
private List<Tuple> getTablespaces(Schema outSchema) {
List<TablespaceProto> tablespaces = masterContext.getCatalog().getAllTablespaces();
List<Tuple> tuples = new ArrayList<Tuple>(tablespaces.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (TablespaceProto tablespace : tablespaces) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("space_id".equalsIgnoreCase(column.getSimpleName())) {
if (tablespace.hasId()) {
aTuple.put(fieldId, DatumFactory.createInt4(tablespace.getId()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("space_name".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(tablespace.getSpaceName()));
} else if ("space_handler".equalsIgnoreCase(column.getSimpleName())) {
if (tablespace.hasHandler()) {
aTuple.put(fieldId, DatumFactory.createText(tablespace.getHandler()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("space_uri".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(tablespace.getUri()));
}
}
tuples.add(aTuple);
}
return tuples;
}
private Tuple getWorkerTuple(Schema outSchema, Worker aWorker) {
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple = new VTuple(outSchema.size());
WorkerResource aResource = aWorker.getResource();
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("host".equalsIgnoreCase(column.getSimpleName())) {
if (aWorker.getConnectionInfo() != null && aWorker.getConnectionInfo().getHost() != null) {
aTuple.put(fieldId, DatumFactory.createText(aWorker.getConnectionInfo().getHost()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("port".equalsIgnoreCase(column.getSimpleName())) {
if (aWorker.getConnectionInfo() != null) {
aTuple.put(
fieldId, DatumFactory.createInt4(aWorker.getConnectionInfo().getPeerRpcPort()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("type".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText("Worker"));
} else if ("status".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(aWorker.getState().toString()));
} else if ("RUNNING".equalsIgnoreCase(aWorker.getState().toString())) {
if ("total_cpu".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(aResource.getCpuCoreSlots()));
} else if ("used_mem".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(aResource.getUsedMemoryMB() * 1048576l));
} else if ("total_mem".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(aResource.getMemoryMB() * 1048576l));
} else if ("free_heap".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(aResource.getFreeHeap()));
} else if ("max_heap".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt8(aResource.getMaxHeap()));
} else if ("used_diskslots".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createFloat4(aResource.getUsedDiskSlots()));
} else if ("total_diskslots".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createFloat4(aResource.getDiskSlots()));
} else if ("running_tasks".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(aResource.getNumRunningTasks()));
} else if ("last_heartbeat_ts".equalsIgnoreCase(column.getSimpleName())) {
if (aWorker.getLastHeartbeatTime() > 0) {
aTuple.put(
fieldId,
DatumFactory.createTimestmpDatumWithJavaMillis(aWorker.getLastHeartbeatTime()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
}
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
}
return aTuple;
}
private boolean isClonedSchema(Schema origSchema, Schema newSchema) {
// Check schema of tables
boolean schemaEqual = (origSchema.size() == newSchema.size());
if (schemaEqual == false) {
fail("Number of columns in schema not equal");
return false;
}
for (int col = 0; col < origSchema.size(); col++) {
Column colA = origSchema.getColumn(col);
Column colB = newSchema.getColumn(col);
if (colA.getSimpleName().equals(colB.getSimpleName()) == false) {
fail("Column names at index " + col + " do not match");
return false;
}
if (colA.getDataType().equals(colB.getDataType()) == false) {
fail("Column datatypes at index " + col + " do not match");
return false;
}
}
return true;
}
private List<Tuple> getColumns(Schema outSchema) {
List<ColumnProto> columnsList = masterContext.getCatalog().getAllColumns();
List<Tuple> tuples = new ArrayList<Tuple>(columnsList.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
int columnId = 1, prevtid = -1, tid = 0;
for (ColumnProto column : columnsList) {
aTuple = new VTuple(outSchema.size());
tid = column.getTid();
if (prevtid != tid) {
columnId = 1;
prevtid = tid;
}
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column colObj = columns.get(fieldId);
if ("tid".equalsIgnoreCase(colObj.getSimpleName())) {
if (column.hasTid()) {
aTuple.put(fieldId, DatumFactory.createInt4(tid));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
} else if ("column_name".equalsIgnoreCase(colObj.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(column.getName()));
} else if ("ordinal_position".equalsIgnoreCase(colObj.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(columnId));
} else if ("data_type".equalsIgnoreCase(colObj.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(column.getDataType().getType().toString()));
} else if ("type_length".equalsIgnoreCase(colObj.getSimpleName())) {
DataType dataType = column.getDataType();
if (dataType.hasLength()) {
aTuple.put(fieldId, DatumFactory.createInt4(dataType.getLength()));
} else {
aTuple.put(fieldId, DatumFactory.createNullDatum());
}
}
}
columnId++;
tuples.add(aTuple);
}
return tuples;
}
private List<Tuple> getSessionInfo(Schema outSchema) {
List<Tuple> outputs = Lists.newArrayList();
Tuple eachVariable;
try {
for (Map.Entry<String, String> var :
masterContext.getSessionManager().getAllVariables(sessionId).entrySet()) {
eachVariable = new VTuple(outSchema.size());
eachVariable.put(0, DatumFactory.createText(var.getKey()));
eachVariable.put(1, DatumFactory.createText(var.getValue()));
outputs.add(eachVariable);
}
} catch (InvalidSessionException e) {
LOG.error(e);
}
return outputs;
}
private void writeRecordFields(GroupType schema, Schema tajoSchema, Tuple tuple) {
List<Type> fields = schema.getFields();
// Parquet ignores Tajo NULL_TYPE columns, so the index may differ.
int index = 0;
for (int tajoIndex = 0; tajoIndex < tajoSchema.size(); ++tajoIndex) {
Column column = tajoSchema.getColumn(tajoIndex);
if (column.getDataType().getType() == TajoDataTypes.Type.NULL_TYPE) {
continue;
}
Type fieldType = fields.get(index);
if (!tuple.isBlankOrNull(tajoIndex)) {
recordConsumer.startField(fieldType.getName(), index);
writeValue(column, tuple, tajoIndex);
recordConsumer.endField(fieldType.getName(), index);
} else if (fieldType.isRepetition(Type.Repetition.REQUIRED)) {
throw new RuntimeException("Null-value for required field: " + column.getSimpleName());
}
++index;
}
}
private List<Tuple> getIndexes(Schema outSchema) {
List<IndexProto> indexList = masterContext.getCatalog().getAllIndexes();
List<Tuple> tuples = new ArrayList<Tuple>(indexList.size());
List<Column> columns = outSchema.getRootColumns();
Tuple aTuple;
for (IndexProto index : indexList) {
aTuple = new VTuple(outSchema.size());
for (int fieldId = 0; fieldId < columns.size(); fieldId++) {
Column column = columns.get(fieldId);
if ("db_id".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(index.getDbId()));
} else if ("tid".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createInt4(index.getTId()));
} else if ("index_name".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(index.getIndexName()));
} else if ("column_name".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(index.getColumnName()));
} else if ("data_type".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(index.getDataType()));
} else if ("index_type".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createText(index.getIndexType()));
} else if ("is_unique".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createBool(index.getIsUnique()));
} else if ("is_clustered".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createBool(index.getIsClustered()));
} else if ("is_ascending".equalsIgnoreCase(column.getSimpleName())) {
aTuple.put(fieldId, DatumFactory.createBool(index.getIsAscending()));
}
}
tuples.add(aTuple);
}
return tuples;
}
@Before
public void setUp() throws Exception {
util = new TajoTestingCluster();
util.initTestDir();
util.startCatalogCluster();
catalog = util.getCatalogService();
testDir = CommonTestingUtil.getTestDir(TEST_PATH);
catalog.createTablespace(DEFAULT_TABLESPACE_NAME, testDir.toUri().toString());
catalog.createDatabase(DEFAULT_DATABASE_NAME, DEFAULT_TABLESPACE_NAME);
conf = util.getConfiguration();
// ----------------- dep3 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
Schema dep3Schema = new Schema();
dep3Schema.addColumn("dep_id", Type.INT4);
dep3Schema.addColumn("dep_name", Type.TEXT);
dep3Schema.addColumn("loc_id", Type.INT4);
TableMeta dep3Meta = CatalogUtil.newTableMeta("TEXT");
Path dep3Path = new Path(testDir, "dep3.csv");
Appender appender1 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep3Meta, dep3Schema, dep3Path);
appender1.init();
VTuple tuple = new VTuple(dep3Schema.size());
for (int i = 0; i < 10; i++) {
tuple.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender1.addTuple(tuple);
}
appender1.flush();
appender1.close();
dep3 = CatalogUtil.newTableDesc(DEP3_NAME, dep3Schema, dep3Meta, dep3Path);
catalog.createTable(dep3);
// ----------------- dep4 ------------------------------
// dep_id | dep_name | loc_id
// --------------------------------
// 0 | dep_0 | 1000
// 1 | dep_1 | 1001
// 2 | dep_2 | 1002
// 3 | dep_3 | 1003
// 4 | dep_4 | 1004
// 5 | dep_5 | 1005
// 6 | dep_6 | 1006
// 7 | dep_7 | 1007
// 8 | dep_8 | 1008
// 9 | dep_9 | 1009
// 10 | dep_10 | 1010
Schema dep4Schema = new Schema();
dep4Schema.addColumn("dep_id", Type.INT4);
dep4Schema.addColumn("dep_name", Type.TEXT);
dep4Schema.addColumn("loc_id", Type.INT4);
TableMeta dep4Meta = CatalogUtil.newTableMeta("TEXT");
Path dep4Path = new Path(testDir, "dep4.csv");
Appender appender4 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(dep4Meta, dep4Schema, dep4Path);
appender4.init();
VTuple tuple4 = new VTuple(dep4Schema.size());
for (int i = 0; i < 11; i++) {
tuple4.put(
new Datum[] {
DatumFactory.createInt4(i),
DatumFactory.createText("dept_" + i),
DatumFactory.createInt4(1000 + i)
});
appender4.addTuple(tuple4);
}
appender4.flush();
appender4.close();
dep4 = CatalogUtil.newTableDesc(DEP4_NAME, dep4Schema, dep4Meta, dep4Path);
catalog.createTable(dep4);
// ----------------- job3 ------------------------------
// job_id | job_title
// ----------------------
// 101 | job_101
// 102 | job_102
// 103 | job_103
Schema job3Schema = new Schema();
job3Schema.addColumn("job_id", Type.INT4);
job3Schema.addColumn("job_title", Type.TEXT);
TableMeta job3Meta = CatalogUtil.newTableMeta("TEXT");
Path job3Path = new Path(testDir, "job3.csv");
Appender appender2 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(job3Meta, job3Schema, job3Path);
appender2.init();
VTuple tuple2 = new VTuple(job3Schema.size());
for (int i = 1; i < 4; i++) {
int x = 100 + i;
tuple2.put(
new Datum[] {DatumFactory.createInt4(100 + i), DatumFactory.createText("job_" + x)});
appender2.addTuple(tuple2);
}
appender2.flush();
appender2.close();
job3 = CatalogUtil.newTableDesc(JOB3_NAME, job3Schema, job3Meta, job3Path);
catalog.createTable(job3);
// ---------------------emp3 --------------------
// emp_id | first_name | last_name | dep_id | salary | job_id
// ------------------------------------------------------------
// 11 | fn_11 | ln_11 | 1 | 123 | 101
// 13 | fn_13 | ln_13 | 3 | 369 | 103
// 15 | fn_15 | ln_15 | 5 | 615 | null
// 17 | fn_17 | ln_17 | 7 | 861 | null
// 19 | fn_19 | ln_19 | 9 | 1107 | null
// 21 | fn_21 | ln_21 | 1 | 123 | 101
// 23 | fn_23 | ln_23 | 3 | 369 | 103
Schema emp3Schema = new Schema();
emp3Schema.addColumn("emp_id", Type.INT4);
emp3Schema.addColumn("first_name", Type.TEXT);
emp3Schema.addColumn("last_name", Type.TEXT);
emp3Schema.addColumn("dep_id", Type.INT4);
emp3Schema.addColumn("salary", Type.FLOAT4);
emp3Schema.addColumn("job_id", Type.INT4);
TableMeta emp3Meta = CatalogUtil.newTableMeta("TEXT");
Path emp3Path = new Path(testDir, "emp3.csv");
Appender appender3 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(emp3Meta, emp3Schema, emp3Path);
appender3.init();
VTuple tuple3 = new VTuple(emp3Schema.size());
for (int i = 1; i < 4; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
int y = 20 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(20 + i),
DatumFactory.createText("firstname_" + y),
DatumFactory.createText("lastname_" + y),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createInt4(100 + i)
});
appender3.addTuple(tuple3);
}
for (int i = 5; i < 10; i += 2) {
int x = 10 + i;
tuple3.put(
new Datum[] {
DatumFactory.createInt4(10 + i),
DatumFactory.createText("firstname_" + x),
DatumFactory.createText("lastname_" + x),
DatumFactory.createInt4(i),
DatumFactory.createFloat4(123 * i),
DatumFactory.createNullDatum()
});
appender3.addTuple(tuple3);
}
appender3.flush();
appender3.close();
emp3 = CatalogUtil.newTableDesc(EMP3_NAME, emp3Schema, emp3Meta, emp3Path);
catalog.createTable(emp3);
// ---------------------phone3 --------------------
// emp_id | phone_number
// -----------------------------------------------
// this table is empty, no rows
Schema phone3Schema = new Schema();
phone3Schema.addColumn("emp_id", Type.INT4);
phone3Schema.addColumn("phone_number", Type.TEXT);
TableMeta phone3Meta = CatalogUtil.newTableMeta("TEXT");
Path phone3Path = new Path(testDir, "phone3.csv");
Appender appender5 =
((FileTablespace) TablespaceManager.getLocalFs())
.getAppender(phone3Meta, phone3Schema, phone3Path);
appender5.init();
appender5.flush();
appender5.close();
phone3 = CatalogUtil.newTableDesc(PHONE3_NAME, phone3Schema, phone3Meta, phone3Path);
catalog.createTable(phone3);
analyzer = new SQLAnalyzer();
planner = new LogicalPlanner(catalog, TablespaceManager.getInstance());
defaultContext = LocalTajoTestingUtility.createDummyContext(conf);
}
@Before
public void setup() throws Exception {
this.randomValues = new HashMap<Integer, Integer>();
this.conf = new TajoConf();
util = new TajoTestingCluster();
util.startCatalogCluster();
catalog = util.getMiniCatalogCluster().getCatalog();
Path workDir = CommonTestingUtil.getTestDir();
catalog.createTablespace(DEFAULT_TABLESPACE_NAME, workDir.toUri().toString());
catalog.createDatabase(TajoConstants.DEFAULT_DATABASE_NAME, DEFAULT_TABLESPACE_NAME);
sm = StorageManagerFactory.getStorageManager(conf, workDir);
idxPath = new Path(workDir, "test.idx");
Schema schema = new Schema();
schema.addColumn("managerid", Type.INT4);
schema.addColumn("empid", Type.INT4);
schema.addColumn("deptname", Type.TEXT);
this.idxSchema = new Schema();
idxSchema.addColumn("managerid", Type.INT4);
SortSpec[] sortKeys = new SortSpec[1];
sortKeys[0] = new SortSpec(idxSchema.getColumn("managerid"), true, false);
this.comp = new TupleComparator(idxSchema, sortKeys);
this.writer =
new BSTIndex(conf)
.getIndexWriter(idxPath, BSTIndex.TWO_LEVEL_INDEX, this.idxSchema, this.comp);
writer.setLoadNum(100);
writer.open();
long offset;
meta = CatalogUtil.newTableMeta(StoreType.CSV);
tablePath = StorageUtil.concatPath(workDir, "employee", "table.csv");
fs = tablePath.getFileSystem(conf);
fs.mkdirs(tablePath.getParent());
FileAppender appender =
(FileAppender)
StorageManagerFactory.getStorageManager(conf).getAppender(meta, schema, tablePath);
appender.init();
Tuple tuple = new VTuple(schema.size());
for (int i = 0; i < 10000; i++) {
Tuple key = new VTuple(this.idxSchema.size());
int rndKey = rnd.nextInt(250);
if (this.randomValues.containsKey(rndKey)) {
int t = this.randomValues.remove(rndKey) + 1;
this.randomValues.put(rndKey, t);
} else {
this.randomValues.put(rndKey, 1);
}
key.put(new Datum[] {DatumFactory.createInt4(rndKey)});
tuple.put(
new Datum[] {
DatumFactory.createInt4(rndKey),
DatumFactory.createInt4(rnd.nextInt(10)),
DatumFactory.createText("dept_" + rnd.nextInt(10))
});
offset = appender.getOffset();
appender.addTuple(tuple);
writer.write(key, offset);
}
appender.flush();
appender.close();
writer.close();
TableDesc desc =
new TableDesc(
CatalogUtil.buildFQName(TajoConstants.DEFAULT_DATABASE_NAME, "employee"),
schema,
meta,
sm.getTablePath("employee"));
catalog.createTable(desc);
analyzer = new SQLAnalyzer();
planner = new LogicalPlanner(catalog);
optimizer = new LogicalOptimizer(conf);
}
@Test
public void testColumnKeyValueMapping() throws Exception {
KeyValueSet keyValueSet = new KeyValueSet();
keyValueSet.set(HBaseStorageConstants.META_TABLE_KEY, "test");
keyValueSet.set(HBaseStorageConstants.META_COLUMNS_KEY, ":key,col2:key:,col2:value:#b,col3:");
Schema schema = new Schema();
schema.addColumn("c1", Type.TEXT);
schema.addColumn("c2", Type.TEXT);
schema.addColumn("c3", Type.TEXT);
schema.addColumn("c4", Type.TEXT);
TableMeta tableMeta = new TableMeta("HBASE", keyValueSet);
ColumnMapping columnMapping = new ColumnMapping(schema, tableMeta.getPropertySet());
List<String> cfNames = columnMapping.getColumnFamilyNames();
assertEquals(2, cfNames.size());
assertEquals("col2", cfNames.get(0));
assertEquals("col3", cfNames.get(1));
for (int i = 0; i < columnMapping.getIsBinaryColumns().length; i++) {
if (i == 2) {
assertTrue(columnMapping.getIsBinaryColumns()[i]);
} else {
assertFalse(columnMapping.getIsBinaryColumns()[i]);
}
}
for (int i = 0; i < columnMapping.getIsRowKeyMappings().length; i++) {
if (i == 0) {
assertTrue(columnMapping.getIsRowKeyMappings()[i]);
} else {
assertFalse(columnMapping.getIsRowKeyMappings()[i]);
}
}
String[] expectedColumnNames = {null, null, null, null};
for (int i = 0; i < schema.size(); i++) {
String columnName =
columnMapping.getMappingColumns()[i][1] == null
? null
: new String(columnMapping.getMappingColumns()[i][1]);
assertEquals(expectedColumnNames[i], columnName);
}
for (int i = 0; i < schema.size(); i++) {
if (i == 1) {
assertTrue(columnMapping.getIsColumnKeys()[i]);
} else {
assertFalse(columnMapping.getIsColumnKeys()[i]);
}
}
for (int i = 0; i < schema.size(); i++) {
if (i == 2) {
assertTrue(columnMapping.getIsColumnValues()[i]);
} else {
assertFalse(columnMapping.getIsColumnValues()[i]);
}
}
}
