public void execSimpleQuery(
QueryContext queryContext,
Session session,
String query,
LogicalPlan plan,
SubmitQueryResponse.Builder response)
throws Exception {
ScanNode scanNode = plan.getRootBlock().getNode(NodeType.SCAN);
TableDesc desc = scanNode.getTableDesc();
if (desc.hasPartition()) {
scanNode = plan.getRootBlock().getNode(NodeType.PARTITIONS_SCAN);
}
int maxRow = Integer.MAX_VALUE;
if (plan.getRootBlock().hasNode(NodeType.LIMIT)) {
LimitNode limitNode = plan.getRootBlock().getNode(NodeType.LIMIT);
maxRow = (int) limitNode.getFetchFirstNum();
}
if (desc.getStats().getNumRows() == 0) {
desc.getStats().setNumRows(TajoConstants.UNKNOWN_ROW_NUMBER);
}
QueryInfo queryInfo =
context
.getQueryJobManager()
.createNewSimpleQuery(
queryContext, session, query, (LogicalRootNode) plan.getRootBlock().getRoot());
NonForwardQueryResultScanner queryResultScanner =
new NonForwardQueryResultFileScanner(
context.getConf(),
session.getSessionId(),
queryInfo.getQueryId(),
scanNode,
desc,
maxRow);
queryResultScanner.init();
session.addNonForwardQueryResultScanner(queryResultScanner);
response.setState(OK);
response.setQueryId(queryInfo.getQueryId().getProto());
response.setResultType(ResultType.ENCLOSED);
response.setMaxRowNum(maxRow);
response.setTableDesc(desc.getProto());
}
/** Insert row values */
private void insertRowValues(
QueryContext queryContext,
InsertNode insertNode,
SubmitQueryResponse.Builder responseBuilder) {
try {
String nodeUniqName =
insertNode.getTableName() == null
? new Path(insertNode.getUri()).getName()
: insertNode.getTableName();
String queryId = nodeUniqName + "_" + System.currentTimeMillis();
URI finalOutputUri = insertNode.getUri();
Tablespace space = TablespaceManager.get(finalOutputUri).get();
TableMeta tableMeta = new TableMeta(insertNode.getStorageType(), insertNode.getOptions());
tableMeta.putOption(StorageConstants.INSERT_DIRECTLY, Boolean.TRUE.toString());
FormatProperty formatProperty = space.getFormatProperty(tableMeta);
TaskAttemptContext taskAttemptContext;
if (formatProperty
.directInsertSupported()) { // if this format and storage supports direct insertion
taskAttemptContext = new TaskAttemptContext(queryContext, null, null, null, null);
taskAttemptContext.setOutputPath(new Path(finalOutputUri));
EvalExprExec evalExprExec =
new EvalExprExec(taskAttemptContext, (EvalExprNode) insertNode.getChild());
InsertRowsExec exec = new InsertRowsExec(taskAttemptContext, insertNode, evalExprExec);
try {
exec.init();
exec.next();
} finally {
exec.close();
}
} else {
URI stagingSpaceUri =
space.prepareStagingSpace(context.getConf(), queryId, queryContext, tableMeta);
Path stagingDir = new Path(stagingSpaceUri);
Path stagingResultDir = new Path(stagingDir, TajoConstants.RESULT_DIR_NAME);
taskAttemptContext = new TaskAttemptContext(queryContext, null, null, null, stagingDir);
taskAttemptContext.setOutputPath(new Path(stagingResultDir, "part-01-000000"));
insertRowsThroughStaging(
taskAttemptContext, insertNode, new Path(finalOutputUri), stagingDir, stagingResultDir);
}
// set insert stats (how many rows and bytes)
TableStats stats = new TableStats();
stats.setNumBytes(taskAttemptContext.getResultStats().getNumBytes());
stats.setNumRows(taskAttemptContext.getResultStats().getNumRows());
if (insertNode.hasTargetTable()) {
CatalogProtos.UpdateTableStatsProto.Builder builder =
CatalogProtos.UpdateTableStatsProto.newBuilder();
builder.setTableName(insertNode.getTableName());
builder.setStats(stats.getProto());
catalog.updateTableStats(builder.build());
TableDesc desc =
new TableDesc(
insertNode.getTableName(), insertNode.getTargetSchema(), tableMeta, finalOutputUri);
responseBuilder.setTableDesc(desc.getProto());
} else { // If INSERT INTO LOCATION
// Empty TableDesc
List<CatalogProtos.ColumnProto> columns = new ArrayList<CatalogProtos.ColumnProto>();
CatalogProtos.TableDescProto tableDescProto =
CatalogProtos.TableDescProto.newBuilder()
.setTableName(nodeUniqName)
.setMeta(
CatalogProtos.TableProto.newBuilder()
.setStoreType(BuiltinStorages.TEXT)
.build())
.setSchema(CatalogProtos.SchemaProto.newBuilder().addAllFields(columns).build())
.setStats(stats.getProto())
.build();
responseBuilder.setTableDesc(tableDescProto);
}
// If queryId == NULL_QUERY_ID and MaxRowNum == -1, TajoCli prints only number of inserted
// rows.
responseBuilder.setMaxRowNum(-1);
responseBuilder.setQueryId(QueryIdFactory.NULL_QUERY_ID.getProto());
responseBuilder.setResultType(ResultType.NO_RESULT);
responseBuilder.setState(OK);
} catch (Throwable t) {
throw new RuntimeException(t);
}
}
@Override
public final CatalogProtos.TableDescProto getTable(String databaseName, final String tableName)
throws CatalogException {
org.apache.hadoop.hive.ql.metadata.Table table = null;
HiveCatalogStoreClientPool.HiveCatalogStoreClient client = null;
Path path = null;
String storeType = null;
org.apache.tajo.catalog.Schema schema = null;
KeyValueSet options = null;
TableStats stats = null;
PartitionMethodDesc partitions = null;
//////////////////////////////////
// set tajo table schema.
//////////////////////////////////
try {
// get hive table schema
try {
client = clientPool.getClient();
table = HiveCatalogUtil.getTable(client.getHiveClient(), databaseName, tableName);
path = table.getPath();
} catch (NoSuchObjectException nsoe) {
throw new UndefinedTableException(tableName);
} catch (Exception e) {
throw new TajoInternalError(e);
}
// convert HiveCatalogStore field schema into tajo field schema.
schema = new org.apache.tajo.catalog.Schema();
List<FieldSchema> fieldSchemaList = table.getCols();
boolean isPartitionKey = false;
for (FieldSchema eachField : fieldSchemaList) {
isPartitionKey = false;
if (table.getPartitionKeys() != null) {
for (FieldSchema partitionKey : table.getPartitionKeys()) {
if (partitionKey.getName().equals(eachField.getName())) {
isPartitionKey = true;
}
}
}
if (!isPartitionKey) {
String fieldName =
databaseName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ tableName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ eachField.getName();
TajoDataTypes.Type dataType =
HiveCatalogUtil.getTajoFieldType(eachField.getType().toString());
schema.addColumn(fieldName, dataType);
}
}
// validate field schema.
HiveCatalogUtil.validateSchema(table);
stats = new TableStats();
options = new KeyValueSet();
options.putAll(table.getParameters());
options.remove("EXTERNAL");
Properties properties = table.getMetadata();
if (properties != null) {
// set field delimiter
String fieldDelimiter = "", nullFormat = "";
if (properties.getProperty(serdeConstants.FIELD_DELIM) != null) {
fieldDelimiter = properties.getProperty(serdeConstants.FIELD_DELIM);
} else {
// if hive table used default row format delimiter, Properties doesn't have it.
// So, Tajo must set as follows:
fieldDelimiter = "\u0001";
}
// set null format
if (properties.getProperty(serdeConstants.SERIALIZATION_NULL_FORMAT) != null) {
nullFormat = properties.getProperty(serdeConstants.SERIALIZATION_NULL_FORMAT);
} else {
nullFormat = "\\N";
}
options.remove(serdeConstants.SERIALIZATION_NULL_FORMAT);
// set file output format
String fileOutputformat =
properties.getProperty(hive_metastoreConstants.FILE_OUTPUT_FORMAT);
storeType = HiveCatalogUtil.getStoreType(fileOutputformat);
if (storeType.equalsIgnoreCase("TEXT")) {
options.set(
StorageConstants.TEXT_DELIMITER, StringEscapeUtils.escapeJava(fieldDelimiter));
options.set(StorageConstants.TEXT_NULL, StringEscapeUtils.escapeJava(nullFormat));
} else if (storeType.equals("RCFILE")) {
options.set(StorageConstants.RCFILE_NULL, StringEscapeUtils.escapeJava(nullFormat));
String serde = properties.getProperty(serdeConstants.SERIALIZATION_LIB);
if (LazyBinaryColumnarSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.RCFILE_SERDE, StorageConstants.DEFAULT_BINARY_SERDE);
} else if (ColumnarSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.RCFILE_SERDE, StorageConstants.DEFAULT_TEXT_SERDE);
}
} else if (storeType.equals("SEQUENCEFILE")) {
options.set(
StorageConstants.SEQUENCEFILE_DELIMITER,
StringEscapeUtils.escapeJava(fieldDelimiter));
options.set(StorageConstants.SEQUENCEFILE_NULL, StringEscapeUtils.escapeJava(nullFormat));
String serde = properties.getProperty(serdeConstants.SERIALIZATION_LIB);
if (LazyBinarySerDe.class.getName().equals(serde)) {
options.set(StorageConstants.SEQUENCEFILE_SERDE, StorageConstants.DEFAULT_BINARY_SERDE);
} else if (LazySimpleSerDe.class.getName().equals(serde)) {
options.set(StorageConstants.SEQUENCEFILE_SERDE, StorageConstants.DEFAULT_TEXT_SERDE);
}
}
// set data size
long totalSize = 0;
if (properties.getProperty("totalSize") != null) {
totalSize = Long.parseLong(properties.getProperty("totalSize"));
} else {
try {
FileSystem fs = path.getFileSystem(conf);
if (fs.exists(path)) {
totalSize = fs.getContentSummary(path).getLength();
}
} catch (IOException ioe) {
throw new TajoInternalError(ioe);
}
}
stats.setNumBytes(totalSize);
}
// set partition keys
List<FieldSchema> partitionKeys = table.getPartitionKeys();
if (null != partitionKeys) {
org.apache.tajo.catalog.Schema expressionSchema = new org.apache.tajo.catalog.Schema();
StringBuilder sb = new StringBuilder();
if (partitionKeys.size() > 0) {
for (int i = 0; i < partitionKeys.size(); i++) {
FieldSchema fieldSchema = partitionKeys.get(i);
TajoDataTypes.Type dataType =
HiveCatalogUtil.getTajoFieldType(fieldSchema.getType().toString());
String fieldName =
databaseName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ tableName
+ CatalogConstants.IDENTIFIER_DELIMITER
+ fieldSchema.getName();
expressionSchema.addColumn(new Column(fieldName, dataType));
if (i > 0) {
sb.append(",");
}
sb.append(fieldSchema.getName());
}
partitions =
new PartitionMethodDesc(
databaseName, tableName, PartitionType.COLUMN, sb.toString(), expressionSchema);
}
}
} finally {
if (client != null) client.release();
}
TableMeta meta = new TableMeta(storeType, options);
TableDesc tableDesc = new TableDesc(databaseName + "." + tableName, schema, meta, path.toUri());
if (table.getTableType().equals(TableType.EXTERNAL_TABLE)) {
tableDesc.setExternal(true);
}
if (stats != null) {
tableDesc.setStats(stats);
}
if (partitions != null) {
tableDesc.setPartitionMethod(partitions);
}
return tableDesc.getProto();
}