private boolean isClonedTable(String orignalTable, String newTable) throws Exception {
assertTableExists(newTable);
TableDesc origTableDesc = client.getTableDesc(orignalTable);
TableDesc newTableDesc = client.getTableDesc(newTable);
if (isClonedSchema(origTableDesc.getSchema(), newTableDesc.getSchema()) == false) {
fail("Schema of input tables do not match");
return false;
}
// Check partition information
PartitionMethodDesc origPartMethod = origTableDesc.getPartitionMethod();
PartitionMethodDesc newPartMethod = newTableDesc.getPartitionMethod();
if (origPartMethod != null) {
if (newPartMethod == null) {
fail("New table does not have partition info");
return false;
}
if (isClonedSchema(origPartMethod.getExpressionSchema(), newPartMethod.getExpressionSchema())
== false) {
fail("Partition columns of input tables do not match");
return false;
}
if (origPartMethod.getPartitionType().equals(newPartMethod.getPartitionType()) == false) {
fail("Partition type of input tables do not match");
return false;
}
}
// Check external flag
if (origTableDesc.isExternal() != newTableDesc.isExternal()) {
fail("External table flag on input tables not equal");
return false;
}
if (origTableDesc.getMeta() != null) {
TableMeta origMeta = origTableDesc.getMeta();
TableMeta newMeta = newTableDesc.getMeta();
if (origMeta.getDataFormat().equals(newMeta.getDataFormat()) == false) {
fail("Store type of input tables not equal");
return false;
}
KeyValueSet origOptions = origMeta.getPropertySet();
KeyValueSet newOptions = newMeta.getPropertySet();
if (origOptions.equals(newOptions) == false) {
fail("Meta options of input tables not equal");
return false;
}
}
return true;
}
@Test
public final void testCtasWithOptions() throws Exception {
ResultSet res = executeFile("CtasWithOptions.sql");
res.close();
ResultSet res2 = executeQuery();
resultSetToString(res2);
res2.close();
TableDesc desc =
client.getTableDesc(CatalogUtil.normalizeIdentifier(res2.getMetaData().getTableName(1)));
assertNotNull(desc);
assertTrue("CSV".equalsIgnoreCase(desc.getMeta().getStoreType()));
KeyValueSet options = desc.getMeta().getOptions();
assertNotNull(options);
assertEquals(
StringEscapeUtils.escapeJava("\u0001"), options.get(StorageConstants.TEXT_DELIMITER));
}
@Test
public final void testRightOuter_MergeJoin3() throws IOException, TajoException {
Expr expr = analyzer.parse(QUERIES[3]);
LogicalNode plan = planner.createPlan(defaultContext, expr).getRootBlock().getRoot();
JoinNode joinNode = PlannerUtil.findTopNode(plan, NodeType.JOIN);
Enforcer enforcer = new Enforcer();
enforcer.enforceJoinAlgorithm(joinNode.getPID(), JoinAlgorithm.MERGE_JOIN);
FileFragment[] emp3Frags =
FileTablespace.splitNG(
conf, EMP3_NAME, emp3.getMeta(), new Path(emp3.getUri()), Integer.MAX_VALUE);
FileFragment[] dep4Frags =
FileTablespace.splitNG(
conf, DEP4_NAME, dep4.getMeta(), new Path(dep4.getUri()), Integer.MAX_VALUE);
FileFragment[] merged = TUtil.concat(emp3Frags, dep4Frags);
Path workDir =
CommonTestingUtil.getTestDir(
TajoTestingCluster.DEFAULT_TEST_DIRECTORY + "/testRightOuter_MergeJoin3");
TaskAttemptContext ctx =
new TaskAttemptContext(
new QueryContext(conf), LocalTajoTestingUtility.newTaskAttemptId(), merged, workDir);
ctx.setEnforcer(enforcer);
PhysicalPlanner phyPlanner = new PhysicalPlannerImpl(conf);
PhysicalExec exec = phyPlanner.createPlan(ctx, plan);
ProjectionExec proj = (ProjectionExec) exec;
assertTrue(proj.getChild() instanceof RightOuterMergeJoinExec);
int count = 0;
exec.init();
while (exec.next() != null) {
// TODO check contents
count = count + 1;
}
assertNull(exec.next());
exec.close();
assertEquals(13, count);
}
@Test
public void testAlterTable() throws Exception {
// CREATE_TABLE
TableDesc tableRenameTestDesc = createMockupTable("default", "mycooltable");
catalog.createTable(tableRenameTestDesc);
// RENAME_TABLE
catalog.alterTable(createMockAlterTableName());
assertTrue(catalog.existsTable("default", "mynewcooltable"));
// RENAME_COLUMN
catalog.alterTable(createMockAlterTableRenameColumn());
TableDesc columnRenameDesc = catalog.getTableDesc("default", "mynewcooltable");
assertTrue(columnRenameDesc.getSchema().containsByName("ren" + FieldName1));
// ADD_COLUMN
catalog.alterTable(createMockAlterTableAddColumn());
TableDesc addColumnDesc = catalog.getTableDesc("default", "mynewcooltable");
assertTrue(addColumnDesc.getSchema().containsByName("mynewcol"));
// SET_PROPERTY
TableDesc setPropertyDesc = catalog.getTableDesc("default", "mynewcooltable");
KeyValueSet options = new KeyValueSet();
options.set("timezone", "GMT+9"); // Seoul, Korea
setPropertyDesc.setMeta(new TableMeta("TEXT", options));
String prevTimeZone = setPropertyDesc.getMeta().getProperty("timezone");
String newTimeZone = "GMT-7"; // Silicon Valley, California
catalog.alterTable(createMockAlterTableSetProperty(newTimeZone));
setPropertyDesc = catalog.getTableDesc("default", "mynewcooltable");
assertNotEquals(prevTimeZone, setPropertyDesc.getMeta().getProperty("timezone"));
assertEquals(newTimeZone, setPropertyDesc.getMeta().getProperty("timezone"));
// UNSET_PROPERTY
catalog.alterTable(createMockAlterTableUnsetProperty(Sets.newHashSet("dummy")));
setPropertyDesc = catalog.getTableDesc("default", "mynewcooltable");
assertTrue(setPropertyDesc.getMeta().getPropertySet().containsKey("timezone"));
assertFalse(setPropertyDesc.getMeta().getPropertySet().containsKey("dummy"));
}
@Test
public final void testCtasWithTextFile() throws Exception {
ResultSet res = executeFile("CtasWithTextFile.sql");
res.close();
ResultSet res2 = executeQuery();
resultSetToString(res2);
res2.close();
TableDesc desc =
client.getTableDesc(CatalogUtil.normalizeIdentifier(res2.getMetaData().getTableName(1)));
assertNotNull(desc);
assertTrue("TEXT".equalsIgnoreCase(desc.getMeta().getStoreType()));
}
public void executeDistributedQuery(
QueryContext queryContext,
Session session,
LogicalPlan plan,
String sql,
String jsonExpr,
SubmitQueryResponse.Builder responseBuilder)
throws Exception {
LogicalRootNode rootNode = plan.getRootBlock().getRoot();
TableDesc tableDesc = PlannerUtil.getTableDesc(catalog, plan.getRootBlock().getRoot());
if (tableDesc != null) {
Tablespace space = TablespaceManager.get(tableDesc.getUri()).get();
FormatProperty formatProperty = space.getFormatProperty(tableDesc.getMeta());
if (!formatProperty.isInsertable()) {
throw new UnsupportedException(
String.format("INSERT operation on %s tablespace", tableDesc.getUri().toString()));
}
space.prepareTable(rootNode.getChild());
}
hookManager.doHooks(queryContext, plan);
QueryManager queryManager = this.context.getQueryJobManager();
QueryInfo queryInfo;
queryInfo = queryManager.scheduleQuery(session, queryContext, sql, jsonExpr, rootNode);
responseBuilder.setState(OK);
responseBuilder.setQueryId(queryInfo.getQueryId().getProto());
responseBuilder.setResultType(ResultType.FETCH);
if (queryInfo.getQueryMasterHost() != null) {
responseBuilder.setQueryMasterHost(queryInfo.getQueryMasterHost());
}
responseBuilder.setQueryMasterPort(queryInfo.getQueryMasterClientPort());
LOG.info(
"Query "
+ queryInfo.getQueryId().toString()
+ ","
+ queryInfo.getSql()
+ ","
+ " is forwarded to "
+ queryInfo.getQueryMasterHost()
+ ":"
+ queryInfo.getQueryMasterPort());
}
@Override
public final void createTable(final CatalogProtos.TableDescProto tableDescProto)
throws CatalogException {
HiveCatalogStoreClientPool.HiveCatalogStoreClient client = null;
TableDesc tableDesc = new TableDesc(tableDescProto);
String[] splitted = CatalogUtil.splitFQTableName(tableDesc.getName());
String databaseName = splitted[0];
String tableName = splitted[1];
try {
client = clientPool.getClient();
org.apache.hadoop.hive.metastore.api.Table table =
new org.apache.hadoop.hive.metastore.api.Table();
table.setDbName(databaseName);
table.setTableName(tableName);
table.setParameters(
new HashMap<String, String>(tableDesc.getMeta().getOptions().getAllKeyValus()));
// TODO: set owner
// table.setOwner();
StorageDescriptor sd = new StorageDescriptor();
sd.setSerdeInfo(new SerDeInfo());
sd.getSerdeInfo().setParameters(new HashMap<String, String>());
sd.getSerdeInfo().setName(table.getTableName());
// if tajo set location method, thrift client make exception as follows:
// Caused by: MetaException(message:java.lang.NullPointerException)
// If you want to modify table path, you have to modify on Hive cli.
if (tableDesc.isExternal()) {
table.setTableType(TableType.EXTERNAL_TABLE.name());
table.putToParameters("EXTERNAL", "TRUE");
Path tablePath = new Path(tableDesc.getUri());
FileSystem fs = tablePath.getFileSystem(conf);
if (fs.isFile(tablePath)) {
LOG.warn("A table path is a file, but HiveCatalogStore does not allow a file path.");
sd.setLocation(tablePath.getParent().toString());
} else {
sd.setLocation(tablePath.toString());
}
}
// set column information
List<Column> columns = tableDesc.getSchema().getRootColumns();
ArrayList<FieldSchema> cols = new ArrayList<FieldSchema>(columns.size());
for (Column eachField : columns) {
cols.add(
new FieldSchema(
eachField.getSimpleName(),
HiveCatalogUtil.getHiveFieldType(eachField.getDataType()),
""));
}
sd.setCols(cols);
// set partition keys
if (tableDesc.hasPartition()
&& tableDesc.getPartitionMethod().getPartitionType().equals(PartitionType.COLUMN)) {
List<FieldSchema> partitionKeys = new ArrayList<FieldSchema>();
for (Column eachPartitionKey :
tableDesc.getPartitionMethod().getExpressionSchema().getRootColumns()) {
partitionKeys.add(
new FieldSchema(
eachPartitionKey.getSimpleName(),
HiveCatalogUtil.getHiveFieldType(eachPartitionKey.getDataType()),
""));
}
table.setPartitionKeys(partitionKeys);
}
if (tableDesc.getMeta().getStoreType().equalsIgnoreCase(BuiltinStorages.RCFILE)) {
String serde = tableDesc.getMeta().getOption(StorageConstants.RCFILE_SERDE);
sd.setInputFormat(org.apache.hadoop.hive.ql.io.RCFileInputFormat.class.getName());
sd.setOutputFormat(org.apache.hadoop.hive.ql.io.RCFileOutputFormat.class.getName());
if (StorageConstants.DEFAULT_TEXT_SERDE.equals(serde)) {
sd.getSerdeInfo()
.setSerializationLib(
org.apache.hadoop.hive.serde2.columnar.ColumnarSerDe.class.getName());
} else {
sd.getSerdeInfo()
.setSerializationLib(
org.apache.hadoop.hive.serde2.columnar.LazyBinaryColumnarSerDe.class.getName());
}
if (tableDesc.getMeta().getOptions().containsKey(StorageConstants.RCFILE_NULL)) {
table.putToParameters(
serdeConstants.SERIALIZATION_NULL_FORMAT,
StringEscapeUtils.unescapeJava(
tableDesc.getMeta().getOption(StorageConstants.RCFILE_NULL)));
}
} else if (tableDesc.getMeta().getStoreType().equals(BuiltinStorages.TEXT)) {
sd.getSerdeInfo()
.setSerializationLib(
org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.class.getName());
sd.setInputFormat(org.apache.hadoop.mapred.TextInputFormat.class.getName());
sd.setOutputFormat(
org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat.class.getName());
String fieldDelimiter =
tableDesc
.getMeta()
.getOption(
StorageConstants.TEXT_DELIMITER, StorageConstants.DEFAULT_FIELD_DELIMITER);
// User can use an unicode for filed delimiter such as \u0001, \001.
// In this case, java console will convert this value into "\\u001".
// And hive will un-espace this value again.
// As a result, user can use right field delimiter.
// So, we have to un-escape this value.
sd.getSerdeInfo()
.putToParameters(
serdeConstants.SERIALIZATION_FORMAT,
StringEscapeUtils.unescapeJava(fieldDelimiter));
sd.getSerdeInfo()
.putToParameters(
serdeConstants.FIELD_DELIM, StringEscapeUtils.unescapeJava(fieldDelimiter));
table.getParameters().remove(StorageConstants.TEXT_DELIMITER);
if (tableDesc.getMeta().containsOption(StorageConstants.TEXT_NULL)) {
table.putToParameters(
serdeConstants.SERIALIZATION_NULL_FORMAT,
StringEscapeUtils.unescapeJava(
tableDesc.getMeta().getOption(StorageConstants.TEXT_NULL)));
table.getParameters().remove(StorageConstants.TEXT_NULL);
}
} else if (tableDesc
.getMeta()
.getStoreType()
.equalsIgnoreCase(BuiltinStorages.SEQUENCE_FILE)) {
String serde = tableDesc.getMeta().getOption(StorageConstants.SEQUENCEFILE_SERDE);
sd.setInputFormat(org.apache.hadoop.mapred.SequenceFileInputFormat.class.getName());
sd.setOutputFormat(
org.apache.hadoop.hive.ql.io.HiveSequenceFileOutputFormat.class.getName());
if (StorageConstants.DEFAULT_TEXT_SERDE.equals(serde)) {
sd.getSerdeInfo()
.setSerializationLib(
org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe.class.getName());
String fieldDelimiter =
tableDesc
.getMeta()
.getOption(
StorageConstants.SEQUENCEFILE_DELIMITER,
StorageConstants.DEFAULT_FIELD_DELIMITER);
// User can use an unicode for filed delimiter such as \u0001, \001.
// In this case, java console will convert this value into "\\u001".
// And hive will un-espace this value again.
// As a result, user can use right field delimiter.
// So, we have to un-escape this value.
sd.getSerdeInfo()
.putToParameters(
serdeConstants.SERIALIZATION_FORMAT,
StringEscapeUtils.unescapeJava(fieldDelimiter));
sd.getSerdeInfo()
.putToParameters(
serdeConstants.FIELD_DELIM, StringEscapeUtils.unescapeJava(fieldDelimiter));
table.getParameters().remove(StorageConstants.SEQUENCEFILE_DELIMITER);
} else {
sd.getSerdeInfo()
.setSerializationLib(
org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe.class.getName());
}
if (tableDesc.getMeta().containsOption(StorageConstants.SEQUENCEFILE_NULL)) {
table.putToParameters(
serdeConstants.SERIALIZATION_NULL_FORMAT,
StringEscapeUtils.unescapeJava(
tableDesc.getMeta().getOption(StorageConstants.SEQUENCEFILE_NULL)));
table.getParameters().remove(StorageConstants.SEQUENCEFILE_NULL);
}
} else {
if (tableDesc.getMeta().getStoreType().equalsIgnoreCase(BuiltinStorages.PARQUET)) {
sd.setInputFormat(parquet.hive.DeprecatedParquetInputFormat.class.getName());
sd.setOutputFormat(parquet.hive.DeprecatedParquetOutputFormat.class.getName());
sd.getSerdeInfo()
.setSerializationLib(parquet.hive.serde.ParquetHiveSerDe.class.getName());
} else {
throw new UnsupportedException(
tableDesc.getMeta().getStoreType() + " in HivecatalogStore");
}
}
sd.setSortCols(new ArrayList<Order>());
table.setSd(sd);
client.getHiveClient().createTable(table);
} catch (Throwable t) {
throw new TajoInternalError(t);
} finally {
if (client != null) client.release();
}
}
@Test
public final void testNext() throws IOException, PlanningException {
FileFragment[] frags =
StorageManager.splitNG(
conf, "employee", employee.getMeta(), employee.getPath(), Integer.MAX_VALUE);
Path workDir = new Path(testDir, TestExternalSortExec.class.getName());
TaskAttemptContext ctx =
new TaskAttemptContext(
conf,
LocalTajoTestingUtility.newQueryUnitAttemptId(),
new FileFragment[] {frags[0]},
workDir);
ctx.setEnforcer(new Enforcer());
Expr expr = analyzer.parse(QUERIES[0]);
LogicalPlan plan = planner.createPlan(expr);
LogicalNode rootNode = plan.getRootBlock().getRoot();
PhysicalPlanner phyPlanner = new PhysicalPlannerImpl(conf, sm);
PhysicalExec exec = phyPlanner.createPlan(ctx, rootNode);
ProjectionExec proj = (ProjectionExec) exec;
// TODO - should be planed with user's optimization hint
if (!(proj.getChild() instanceof ExternalSortExec)) {
UnaryPhysicalExec sortExec = proj.getChild();
SeqScanExec scan = sortExec.getChild();
ExternalSortExec extSort =
new ExternalSortExec(ctx, sm, ((MemSortExec) sortExec).getPlan(), scan);
proj.setChild(extSort);
}
Tuple tuple;
Tuple preVal = null;
Tuple curVal;
int cnt = 0;
exec.init();
long start = System.currentTimeMillis();
TupleComparator comparator =
new TupleComparator(
proj.getSchema(),
new SortSpec[] {
new SortSpec(new Column("managerId", Type.INT4)),
new SortSpec(new Column("empId", Type.INT4))
});
while ((tuple = exec.next()) != null) {
curVal = tuple;
if (preVal != null) {
assertTrue(
"prev: " + preVal + ", but cur: " + curVal, comparator.compare(preVal, curVal) <= 0);
}
preVal = curVal;
cnt++;
}
long end = System.currentTimeMillis();
assertEquals(numTuple, cnt);
// for rescan test
preVal = null;
exec.rescan();
cnt = 0;
while ((tuple = exec.next()) != null) {
curVal = tuple;
if (preVal != null) {
assertTrue(
"prev: " + preVal + ", but cur: " + curVal, comparator.compare(preVal, curVal) <= 0);
}
preVal = curVal;
cnt++;
}
assertEquals(numTuple, cnt);
exec.close();
System.out.println("Sort Time: " + (end - start) + " msc");
}