org.apache.hadoop.hive.ql.ddl

Class DDLUtils

java.lang.Object

org.apache.hadoop.hive.ql.ddl.DDLUtils

public final class DDLUtils
extends Object

Utilities used by some DDLOperations.

Method Summary

Modifier and Type	Method and Description
static void	addDbAndTableToOutputs(org.apache.hadoop.hive.metastore.api.Database database, TableName tableName, TableType type, boolean isTemporary, Map<String,String> properties, Set<WriteEntity> outputs)
static boolean	addIfAbsentByName(WriteEntity newWriteEntity, DDLOperationContext context)
static boolean	addIfAbsentByName(WriteEntity newWriteEntity, Set<WriteEntity> outputs) There are many places where "duplicate" Read/WriteEnity objects are added.
static void	addServiceOutput(HiveConf conf, Set<WriteEntity> outputs)
static boolean	allowOperationInReplicationScope(Hive db, String tableName, Map<String,String> partSpec, ReplicationSpec replicationSpec) Validate if the given table/partition is eligible for update.
static String	getFQName(ASTNode node) Get the fully qualified name in the node.
static void	validateSerDe(String serdeName, DDLOperationContext context) Check if the given serde is valid.
static void	validateSerDe(String serdeName, HiveConf conf)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

addIfAbsentByName

public static boolean addIfAbsentByName(WriteEntity newWriteEntity,
                                        Set<WriteEntity> outputs)

There are many places where "duplicate" Read/WriteEnity objects are added. The way this was initially implemented, the duplicate just replaced the previous object. (work.getOutputs() is a Set and WriteEntity#equals() relies on name) This may be benign for ReadEntity and perhaps was benign for WriteEntity before WriteType was added. Now that WriteEntity has a WriteType it replaces it with one with possibly different WriteEntity.WriteType. It's hard to imagine how this is desirable. As of HIVE-14993, WriteEntity with different WriteType must be considered different. So WriteEntity created in DDLTask cause extra output in golden files, but only because DDLTask sets a different WriteType for the same Entity. In the spirit of bug-for-bug compatibility, this method ensures we only add new WriteEntity if it's really new.

Returns:

true if item was added

addIfAbsentByName

public static boolean addIfAbsentByName(WriteEntity newWriteEntity,
                                        DDLOperationContext context)

validateSerDe

public static void validateSerDe(String serdeName,
                                 DDLOperationContext context)
                          throws HiveException

Check if the given serde is valid.

Throws:

HiveException

validateSerDe

public static void validateSerDe(String serdeName,
                                 HiveConf conf)
                          throws HiveException

Throws:

HiveException

allowOperationInReplicationScope

public static boolean allowOperationInReplicationScope(Hive db,
                                                       String tableName,
                                                       Map<String,String> partSpec,
                                                       ReplicationSpec replicationSpec)
                                                throws HiveException

Validate if the given table/partition is eligible for update.

Parameters:

db - Database.

tableName - Table name of format db.table

partSpec - Partition spec for the partition

replicationSpec - Replications specification

Returns:

boolean true if allow the operation

Throws:

HiveException

addServiceOutput

public static void addServiceOutput(HiveConf conf,
                                    Set<WriteEntity> outputs)
                             throws SemanticException

Throws:

SemanticException

getFQName

public static String getFQName(ASTNode node)

Get the fully qualified name in the node. E.g. the node of the form ^(DOT ^(DOT a b) c) will generate a name of the form "a.b.c".

addDbAndTableToOutputs

public static void addDbAndTableToOutputs(org.apache.hadoop.hive.metastore.api.Database database,
                                          TableName tableName,
                                          TableType type,
                                          boolean isTemporary,
                                          Map<String,String> properties,
                                          Set<WriteEntity> outputs)