All Downloads are FREE. Search and download functionalities are using the official Maven repository.

org.apache.hadoop.hive.ql.parse.DDLSemanticAnalyzer Maven / Gradle / Ivy

There is a newer version: 4.0.1
Show newest version
/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package org.apache.hadoop.hive.ql.parse;

import static org.apache.hadoop.hive.ql.parse.HiveParser.TOK_DATABASELOCATION;
import static org.apache.hadoop.hive.ql.parse.HiveParser.TOK_DATABASEPROPERTIES;

import java.io.FileNotFoundException;
import java.io.Serializable;
import java.lang.reflect.Constructor;
import java.lang.reflect.InvocationTargetException;
import java.net.URI;
import java.net.URISyntaxException;
import java.util.ArrayList;
import java.util.Collection;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Iterator;
import java.util.LinkedHashMap;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Set;

import org.antlr.runtime.tree.CommonTree;
import org.antlr.runtime.tree.Tree;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hive.common.StatsSetupConst;
import org.apache.hadoop.hive.conf.HiveConf;
import org.apache.hadoop.hive.conf.HiveConf.ConfVars;
import org.apache.hadoop.hive.metastore.TableType;
import org.apache.hadoop.hive.metastore.Warehouse;
import org.apache.hadoop.hive.metastore.api.Database;
import org.apache.hadoop.hive.metastore.api.EnvironmentContext;
import org.apache.hadoop.hive.metastore.api.FieldSchema;
import org.apache.hadoop.hive.metastore.api.MetaException;
import org.apache.hadoop.hive.metastore.api.Order;
import org.apache.hadoop.hive.metastore.api.SQLCheckConstraint;
import org.apache.hadoop.hive.metastore.api.SQLDefaultConstraint;
import org.apache.hadoop.hive.metastore.api.SQLForeignKey;
import org.apache.hadoop.hive.metastore.api.SQLNotNullConstraint;
import org.apache.hadoop.hive.metastore.api.SQLPrimaryKey;
import org.apache.hadoop.hive.metastore.api.SQLUniqueConstraint;
import org.apache.hadoop.hive.metastore.api.SkewedInfo;
import org.apache.hadoop.hive.metastore.api.WMMapping;
import org.apache.hadoop.hive.metastore.api.WMNullablePool;
import org.apache.hadoop.hive.metastore.api.WMNullableResourcePlan;
import org.apache.hadoop.hive.metastore.api.WMPool;
import org.apache.hadoop.hive.metastore.api.WMResourcePlanStatus;
import org.apache.hadoop.hive.metastore.api.WMTrigger;
import org.apache.hadoop.hive.metastore.api.hive_metastoreConstants;
import org.apache.hadoop.hive.metastore.utils.MetaStoreUtils;
import org.apache.hadoop.hive.ql.Driver;
import org.apache.hadoop.hive.ql.ErrorMsg;
import org.apache.hadoop.hive.ql.QueryState;
import org.apache.hadoop.hive.ql.exec.ArchiveUtils;
import org.apache.hadoop.hive.ql.exec.ColumnStatsUpdateTask;
import org.apache.hadoop.hive.ql.exec.DDLTask;
import org.apache.hadoop.hive.ql.exec.FunctionRegistry;
import org.apache.hadoop.hive.ql.exec.Task;
import org.apache.hadoop.hive.ql.exec.TaskFactory;
import org.apache.hadoop.hive.ql.exec.Utilities;
import org.apache.hadoop.hive.ql.hooks.Entity.Type;
import org.apache.hadoop.hive.ql.hooks.ReadEntity;
import org.apache.hadoop.hive.ql.hooks.WriteEntity;
import org.apache.hadoop.hive.ql.hooks.WriteEntity.WriteType;
import org.apache.hadoop.hive.ql.io.AcidUtils;
import org.apache.hadoop.hive.ql.io.RCFileInputFormat;
import org.apache.hadoop.hive.ql.io.orc.OrcInputFormat;
import org.apache.hadoop.hive.ql.lib.Node;
import org.apache.hadoop.hive.ql.lockmgr.HiveTxnManager;
import org.apache.hadoop.hive.ql.lockmgr.LockException;
import org.apache.hadoop.hive.ql.lockmgr.TxnManagerFactory;
import org.apache.hadoop.hive.ql.metadata.DefaultConstraint;
import org.apache.hadoop.hive.ql.metadata.Hive;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.metadata.HiveUtils;
import org.apache.hadoop.hive.ql.metadata.InvalidTableException;
import org.apache.hadoop.hive.ql.metadata.NotNullConstraint;
import org.apache.hadoop.hive.ql.metadata.Partition;
import org.apache.hadoop.hive.ql.metadata.Table;
import org.apache.hadoop.hive.ql.parse.authorization.AuthorizationParseUtils;
import org.apache.hadoop.hive.ql.parse.authorization.HiveAuthorizationTaskFactory;
import org.apache.hadoop.hive.ql.parse.authorization.HiveAuthorizationTaskFactoryImpl;
import org.apache.hadoop.hive.ql.plan.AbortTxnsDesc;
import org.apache.hadoop.hive.ql.plan.AddPartitionDesc;
import org.apache.hadoop.hive.ql.plan.AddPartitionDesc.OnePartitionDesc;
import org.apache.hadoop.hive.ql.plan.AlterDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.AlterMaterializedViewDesc;
import org.apache.hadoop.hive.ql.plan.AlterMaterializedViewDesc.AlterMaterializedViewTypes;
import org.apache.hadoop.hive.ql.plan.AlterResourcePlanDesc;
import org.apache.hadoop.hive.ql.plan.AlterTableAlterPartDesc;
import org.apache.hadoop.hive.ql.plan.AlterTableDesc;
import org.apache.hadoop.hive.ql.plan.AlterTableDesc.AlterTableTypes;
import org.apache.hadoop.hive.ql.plan.DDLDesc.DDLDescWithWriteId;
import org.apache.hadoop.hive.ql.plan.AlterTableExchangePartition;
import org.apache.hadoop.hive.ql.plan.AlterTableSimpleDesc;
import org.apache.hadoop.hive.ql.plan.AlterWMTriggerDesc;
import org.apache.hadoop.hive.ql.plan.BasicStatsWork;
import org.apache.hadoop.hive.ql.plan.CacheMetadataDesc;
import org.apache.hadoop.hive.ql.plan.ColumnStatsUpdateWork;
import org.apache.hadoop.hive.ql.plan.CreateDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.CreateOrAlterWMMappingDesc;
import org.apache.hadoop.hive.ql.plan.CreateOrAlterWMPoolDesc;
import org.apache.hadoop.hive.ql.plan.CreateOrDropTriggerToPoolMappingDesc;
import org.apache.hadoop.hive.ql.plan.CreateResourcePlanDesc;
import org.apache.hadoop.hive.ql.plan.CreateWMTriggerDesc;
import org.apache.hadoop.hive.ql.plan.DDLDesc;
import org.apache.hadoop.hive.ql.plan.DDLWork;
import org.apache.hadoop.hive.ql.plan.DescDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.DescFunctionDesc;
import org.apache.hadoop.hive.ql.plan.DescTableDesc;
import org.apache.hadoop.hive.ql.plan.DropDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.DropResourcePlanDesc;
import org.apache.hadoop.hive.ql.plan.DropTableDesc;
import org.apache.hadoop.hive.ql.plan.DropWMMappingDesc;
import org.apache.hadoop.hive.ql.plan.DropWMPoolDesc;
import org.apache.hadoop.hive.ql.plan.DropWMTriggerDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc;
import org.apache.hadoop.hive.ql.plan.HiveOperation;
import org.apache.hadoop.hive.ql.plan.KillQueryDesc;
import org.apache.hadoop.hive.ql.plan.ListBucketingCtx;
import org.apache.hadoop.hive.ql.plan.LoadTableDesc;
import org.apache.hadoop.hive.ql.plan.LockDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.LockTableDesc;
import org.apache.hadoop.hive.ql.plan.MoveWork;
import org.apache.hadoop.hive.ql.plan.MsckDesc;
import org.apache.hadoop.hive.ql.plan.PlanUtils;
import org.apache.hadoop.hive.ql.plan.PrincipalDesc;
import org.apache.hadoop.hive.ql.plan.RenamePartitionDesc;
import org.apache.hadoop.hive.ql.plan.RoleDDLDesc;
import org.apache.hadoop.hive.ql.plan.ShowColumnsDesc;
import org.apache.hadoop.hive.ql.plan.ShowCompactionsDesc;
import org.apache.hadoop.hive.ql.plan.ShowConfDesc;
import org.apache.hadoop.hive.ql.plan.ShowCreateDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.ShowCreateTableDesc;
import org.apache.hadoop.hive.ql.plan.ShowDatabasesDesc;
import org.apache.hadoop.hive.ql.plan.ShowFunctionsDesc;
import org.apache.hadoop.hive.ql.plan.ShowGrantDesc;
import org.apache.hadoop.hive.ql.plan.ShowLocksDesc;
import org.apache.hadoop.hive.ql.plan.ShowPartitionsDesc;
import org.apache.hadoop.hive.ql.plan.ShowResourcePlanDesc;
import org.apache.hadoop.hive.ql.plan.ShowTableStatusDesc;
import org.apache.hadoop.hive.ql.plan.ShowTablesDesc;
import org.apache.hadoop.hive.ql.plan.ShowTblPropertiesDesc;
import org.apache.hadoop.hive.ql.plan.ShowTxnsDesc;
import org.apache.hadoop.hive.ql.plan.StatsWork;
import org.apache.hadoop.hive.ql.plan.SwitchDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.TableDesc;
import org.apache.hadoop.hive.ql.plan.TruncateTableDesc;
import org.apache.hadoop.hive.ql.plan.UnlockDatabaseDesc;
import org.apache.hadoop.hive.ql.plan.UnlockTableDesc;
import org.apache.hadoop.hive.ql.session.SessionState;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDF;
import org.apache.hadoop.hive.serde.serdeConstants;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorConverters;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorConverters.Converter;
import org.apache.hadoop.hive.serde2.typeinfo.CharTypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.TimestampLocalTZTypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfoFactory;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils;
import org.apache.hadoop.hive.serde2.typeinfo.VarcharTypeInfo;
import org.apache.hadoop.mapred.InputFormat;
import org.apache.hadoop.util.StringUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;

import com.google.common.collect.ImmutableList;
import com.google.common.collect.Lists;

/**
 * DDLSemanticAnalyzer.
 *
 */
public class DDLSemanticAnalyzer extends BaseSemanticAnalyzer {
  private static final Logger LOG = LoggerFactory.getLogger(DDLSemanticAnalyzer.class);
  private static final Map TokenToTypeName = new HashMap();

  private final Set reservedPartitionValues;
  private final HiveAuthorizationTaskFactory hiveAuthorizationTaskFactory;
  private WriteEntity alterTableOutput;
  // Equivalent to acidSinks, but for DDL operations that change data.
  private DDLDesc.DDLDescWithWriteId ddlDescWithWriteId;

  static {
    TokenToTypeName.put(HiveParser.TOK_BOOLEAN, serdeConstants.BOOLEAN_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_TINYINT, serdeConstants.TINYINT_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_SMALLINT, serdeConstants.SMALLINT_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_INT, serdeConstants.INT_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_BIGINT, serdeConstants.BIGINT_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_FLOAT, serdeConstants.FLOAT_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_DOUBLE, serdeConstants.DOUBLE_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_STRING, serdeConstants.STRING_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_CHAR, serdeConstants.CHAR_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_VARCHAR, serdeConstants.VARCHAR_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_BINARY, serdeConstants.BINARY_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_DATE, serdeConstants.DATE_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_DATETIME, serdeConstants.DATETIME_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_TIMESTAMP, serdeConstants.TIMESTAMP_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_TIMESTAMPLOCALTZ, serdeConstants.TIMESTAMPLOCALTZ_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_INTERVAL_YEAR_MONTH, serdeConstants.INTERVAL_YEAR_MONTH_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_INTERVAL_DAY_TIME, serdeConstants.INTERVAL_DAY_TIME_TYPE_NAME);
    TokenToTypeName.put(HiveParser.TOK_DECIMAL, serdeConstants.DECIMAL_TYPE_NAME);
  }

  public static String getTypeName(ASTNode node) throws SemanticException {
    int token = node.getType();
    String typeName;

    // datetime type isn't currently supported
    if (token == HiveParser.TOK_DATETIME) {
      throw new SemanticException(ErrorMsg.UNSUPPORTED_TYPE.getMsg());
    }

    switch (token) {
    case HiveParser.TOK_CHAR:
      CharTypeInfo charTypeInfo = ParseUtils.getCharTypeInfo(node);
      typeName = charTypeInfo.getQualifiedName();
      break;
    case HiveParser.TOK_VARCHAR:
      VarcharTypeInfo varcharTypeInfo = ParseUtils.getVarcharTypeInfo(node);
      typeName = varcharTypeInfo.getQualifiedName();
      break;
    case HiveParser.TOK_TIMESTAMPLOCALTZ:
      HiveConf conf;
      try {
        conf = Hive.get().getConf();
      } catch (HiveException e) {
        throw new SemanticException(e);
      }
      TimestampLocalTZTypeInfo timestampLocalTZTypeInfo = TypeInfoFactory.getTimestampTZTypeInfo(
          conf.getLocalTimeZone());
      typeName = timestampLocalTZTypeInfo.getQualifiedName();
      break;
    case HiveParser.TOK_DECIMAL:
      DecimalTypeInfo decTypeInfo = ParseUtils.getDecimalTypeTypeInfo(node);
      typeName = decTypeInfo.getQualifiedName();
      break;
    default:
      typeName = TokenToTypeName.get(token);
    }
    return typeName;
  }

  public DDLSemanticAnalyzer(QueryState queryState) throws SemanticException {
    this(queryState, createHiveDB(queryState.getConf()));
  }

  public DDLSemanticAnalyzer(QueryState queryState, Hive db) throws SemanticException {
    super(queryState, db);
    reservedPartitionValues = new HashSet();
    // Partition can't have this name
    reservedPartitionValues.add(HiveConf.getVar(conf, ConfVars.DEFAULTPARTITIONNAME));
    reservedPartitionValues.add(HiveConf.getVar(conf, ConfVars.DEFAULT_ZOOKEEPER_PARTITION_NAME));
    // Partition value can't end in this suffix
    reservedPartitionValues.add(HiveConf.getVar(conf, ConfVars.METASTORE_INT_ORIGINAL));
    reservedPartitionValues.add(HiveConf.getVar(conf, ConfVars.METASTORE_INT_ARCHIVED));
    reservedPartitionValues.add(HiveConf.getVar(conf, ConfVars.METASTORE_INT_EXTRACTED));
    hiveAuthorizationTaskFactory = createAuthorizationTaskFactory(conf, db);
  }

  @Override
  public void analyzeInternal(ASTNode input) throws SemanticException {

    ASTNode ast = input;
    switch (ast.getType()) {
    case HiveParser.TOK_ALTERTABLE: {
      ast = (ASTNode) input.getChild(1);
      String[] qualified = getQualifiedTableName((ASTNode) input.getChild(0));
      // TODO CAT - for now always use the default catalog.  Eventually will want to see if
      // the user specified a catalog
      String catName = MetaStoreUtils.getDefaultCatalog(conf);
      String tableName = getDotName(qualified);
      HashMap partSpec = null;
      ASTNode partSpecNode = (ASTNode)input.getChild(2);
      if (partSpecNode != null) {
        //  We can use alter table partition rename to convert/normalize the legacy partition
        //  column values. In so, we should not enable the validation to the old partition spec
        //  passed in this command.
        if (ast.getType() == HiveParser.TOK_ALTERTABLE_RENAMEPART) {
          partSpec = getPartSpec(partSpecNode);
        } else {
          partSpec = getValidatedPartSpec(getTable(tableName), partSpecNode, conf, false);
        }
      }

      if (ast.getType() == HiveParser.TOK_ALTERTABLE_RENAME) {
        analyzeAlterTableRename(qualified, ast, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_TOUCH) {
        analyzeAlterTableTouch(qualified, ast);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_ARCHIVE) {
        analyzeAlterTableArchive(qualified, ast, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_UNARCHIVE) {
        analyzeAlterTableArchive(qualified, ast, true);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_ADDCOLS) {
        analyzeAlterTableModifyCols(qualified, ast, partSpec, AlterTableTypes.ADDCOLS);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_REPLACECOLS) {
        analyzeAlterTableModifyCols(qualified, ast, partSpec, AlterTableTypes.REPLACECOLS);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_RENAMECOL) {
        analyzeAlterTableRenameCol(catName, qualified, ast, partSpec);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_ADDPARTS) {
        analyzeAlterTableAddParts(qualified, ast, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_DROPPARTS) {
        analyzeAlterTableDropParts(qualified, ast, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_PARTCOLTYPE) {
        analyzeAlterTablePartColType(qualified, ast);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_PROPERTIES) {
        analyzeAlterTableProps(qualified, null, ast, false, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_DROPPROPERTIES) {
        analyzeAlterTableProps(qualified, null, ast, false, true);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_UPDATESTATS) {
        analyzeAlterTableProps(qualified, partSpec, ast, false, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_SKEWED) {
        analyzeAltertableSkewedby(qualified, ast);
      } else if (ast.getType() == HiveParser.TOK_ALTERTABLE_EXCHANGEPARTITION) {
        analyzeExchangePartition(qualified, ast);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_FILEFORMAT) {
        analyzeAlterTableFileFormat(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_LOCATION) {
        analyzeAlterTableLocation(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_MERGEFILES) {
        analyzeAlterTablePartMergeFiles(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_SERIALIZER) {
        analyzeAlterTableSerde(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_SERDEPROPERTIES) {
        analyzeAlterTableSerdeProps(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_RENAMEPART) {
        analyzeAlterTableRenamePart(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_SKEWED_LOCATION) {
        analyzeAlterTableSkewedLocation(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_BUCKETS) {
        analyzeAlterTableBucketNum(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_CLUSTER_SORT) {
        analyzeAlterTableClusterSort(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_COMPACT) {
        analyzeAlterTableCompact(ast, tableName, partSpec);
      } else if(ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_UPDATECOLSTATS){
        analyzeAlterTableUpdateStats(ast, tableName, partSpec);
      } else if(ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_DROPCONSTRAINT) {
        analyzeAlterTableDropConstraint(ast, tableName);
      } else if(ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_ADDCONSTRAINT) {
          analyzeAlterTableAddConstraint(ast, tableName);
      } else if(ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_UPDATECOLUMNS) {
        analyzeAlterTableUpdateColumns(ast, tableName, partSpec);
      } else if (ast.getToken().getType() == HiveParser.TOK_ALTERTABLE_OWNER) {
        analyzeAlterTableOwner(ast, tableName);
      }
      break;
    }
    case HiveParser.TOK_DROPTABLE:
      analyzeDropTable(ast, null);
      break;
    case HiveParser.TOK_TRUNCATETABLE:
      analyzeTruncateTable(ast);
      break;
    case HiveParser.TOK_DESCTABLE:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeDescribeTable(ast);
      break;
    case HiveParser.TOK_SHOWDATABASES:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowDatabases(ast);
      break;
    case HiveParser.TOK_SHOWTABLES:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowTables(ast);
      break;
    case HiveParser.TOK_SHOWCOLUMNS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowColumns(ast);
      break;
    case HiveParser.TOK_SHOW_TABLESTATUS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowTableStatus(ast);
      break;
    case HiveParser.TOK_SHOW_TBLPROPERTIES:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowTableProperties(ast);
      break;
    case HiveParser.TOK_SHOWFUNCTIONS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowFunctions(ast);
      break;
    case HiveParser.TOK_SHOWLOCKS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowLocks(ast);
      break;
    case HiveParser.TOK_SHOWDBLOCKS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowDbLocks(ast);
      break;
    case HiveParser.TOK_SHOW_COMPACTIONS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowCompactions(ast);
      break;
    case HiveParser.TOK_SHOW_TRANSACTIONS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowTxns(ast);
      break;
    case HiveParser.TOK_ABORT_TRANSACTIONS:
      analyzeAbortTxns(ast);
      break;
    case HiveParser.TOK_KILL_QUERY:
      analyzeKillQuery(ast);
      break;
    case HiveParser.TOK_SHOWCONF:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowConf(ast);
      break;
    case HiveParser.TOK_SHOWVIEWS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowViews(ast);
      break;
    case HiveParser.TOK_SHOWMATERIALIZEDVIEWS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowMaterializedViews(ast);
      break;
    case HiveParser.TOK_DESCFUNCTION:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeDescFunction(ast);
      break;
    case HiveParser.TOK_DESCDATABASE:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeDescDatabase(ast);
      break;
    case HiveParser.TOK_MSCK:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeMetastoreCheck(ast);
      break;
    case HiveParser.TOK_DROPVIEW:
      analyzeDropTable(ast, TableType.VIRTUAL_VIEW);
      break;
    case HiveParser.TOK_DROP_MATERIALIZED_VIEW:
      analyzeDropTable(ast, TableType.MATERIALIZED_VIEW);
      break;
    case HiveParser.TOK_ALTERVIEW: {
      String[] qualified = getQualifiedTableName((ASTNode) ast.getChild(0));
      ast = (ASTNode) ast.getChild(1);
      if (ast.getType() == HiveParser.TOK_ALTERVIEW_PROPERTIES) {
        analyzeAlterTableProps(qualified, null, ast, true, false);
      } else if (ast.getType() == HiveParser.TOK_ALTERVIEW_DROPPROPERTIES) {
        analyzeAlterTableProps(qualified, null, ast, true, true);
      } else if (ast.getType() == HiveParser.TOK_ALTERVIEW_ADDPARTS) {
        analyzeAlterTableAddParts(qualified, ast, true);
      } else if (ast.getType() == HiveParser.TOK_ALTERVIEW_DROPPARTS) {
        analyzeAlterTableDropParts(qualified, ast, true);
      } else if (ast.getType() == HiveParser.TOK_ALTERVIEW_RENAME) {
        analyzeAlterTableRename(qualified, ast, true);
      }
      break;
    }
    case HiveParser.TOK_ALTER_MATERIALIZED_VIEW: {
      ast = (ASTNode) input.getChild(1);
      String[] qualified = getQualifiedTableName((ASTNode) input.getChild(0));
      String tableName = getDotName(qualified);

      if (ast.getType() == HiveParser.TOK_ALTER_MATERIALIZED_VIEW_REWRITE) {
        analyzeAlterMaterializedViewRewrite(tableName, ast);
      }
      break;
    }
    case HiveParser.TOK_SHOWPARTITIONS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowPartitions(ast);
      break;
    case HiveParser.TOK_SHOW_CREATEDATABASE:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowCreateDatabase(ast);
      break;
    case HiveParser.TOK_SHOW_CREATETABLE:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowCreateTable(ast);
      break;
    case HiveParser.TOK_LOCKTABLE:
      analyzeLockTable(ast);
      break;
    case HiveParser.TOK_UNLOCKTABLE:
      analyzeUnlockTable(ast);
      break;
    case HiveParser.TOK_LOCKDB:
      analyzeLockDatabase(ast);
      break;
    case HiveParser.TOK_UNLOCKDB:
      analyzeUnlockDatabase(ast);
      break;
    case HiveParser.TOK_CREATEDATABASE:
      analyzeCreateDatabase(ast);
      break;
    case HiveParser.TOK_DROPDATABASE:
      analyzeDropDatabase(ast);
      break;
    case HiveParser.TOK_SWITCHDATABASE:
      analyzeSwitchDatabase(ast);
      break;
    case HiveParser.TOK_ALTERDATABASE_PROPERTIES:
      analyzeAlterDatabaseProperties(ast);
      break;
    case HiveParser.TOK_ALTERDATABASE_OWNER:
      analyzeAlterDatabaseOwner(ast);
      break;
    case HiveParser.TOK_ALTERDATABASE_LOCATION:
      analyzeAlterDatabaseLocation(ast);
      break;
    case HiveParser.TOK_CREATEROLE:
      analyzeCreateRole(ast);
      break;
    case HiveParser.TOK_DROPROLE:
      analyzeDropRole(ast);
      break;
    case HiveParser.TOK_SHOW_ROLE_GRANT:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowRoleGrant(ast);
      break;
    case HiveParser.TOK_SHOW_ROLE_PRINCIPALS:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowRolePrincipals(ast);
      break;
    case HiveParser.TOK_SHOW_ROLES:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowRoles(ast);
      break;
    case HiveParser.TOK_GRANT_ROLE:
      analyzeGrantRevokeRole(true, ast);
      break;
    case HiveParser.TOK_REVOKE_ROLE:
      analyzeGrantRevokeRole(false, ast);
      break;
    case HiveParser.TOK_GRANT:
      analyzeGrant(ast);
      break;
    case HiveParser.TOK_SHOW_GRANT:
      ctx.setResFile(ctx.getLocalTmpPath());
      analyzeShowGrant(ast);
      break;
    case HiveParser.TOK_REVOKE:
      analyzeRevoke(ast);
      break;
   case HiveParser.TOK_SHOW_SET_ROLE:
     analyzeSetShowRole(ast);
     break;
   case HiveParser.TOK_CACHE_METADATA:
     analyzeCacheMetadata(ast);
     break;
   case HiveParser.TOK_CREATE_RP:
     analyzeCreateResourcePlan(ast);
     break;
   case HiveParser.TOK_SHOW_RP:
     ctx.setResFile(ctx.getLocalTmpPath());
     analyzeShowResourcePlan(ast);
     break;
   case HiveParser.TOK_ALTER_RP:
     analyzeAlterResourcePlan(ast);
     break;
   case HiveParser.TOK_DROP_RP:
     analyzeDropResourcePlan(ast);
     break;
   case HiveParser.TOK_CREATE_TRIGGER:
     analyzeCreateTrigger(ast);
     break;
   case HiveParser.TOK_ALTER_TRIGGER:
     analyzeAlterTrigger(ast);
     break;
   case HiveParser.TOK_DROP_TRIGGER:
     analyzeDropTrigger(ast);
     break;
   case HiveParser.TOK_CREATE_POOL:
     analyzeCreatePool(ast);
     break;
   case HiveParser.TOK_ALTER_POOL:
     analyzeAlterPool(ast);
     break;
   case HiveParser.TOK_DROP_POOL:
     analyzeDropPool(ast);
     break;
   case HiveParser.TOK_CREATE_MAPPING:
     analyzeCreateOrAlterMapping(ast, false);
     break;
   case HiveParser.TOK_ALTER_MAPPING:
     analyzeCreateOrAlterMapping(ast, true);
     break;
   case HiveParser.TOK_DROP_MAPPING:
     analyzeDropMapping(ast);
     break;
   default:
      throw new SemanticException("Unsupported command: " + ast);
    }
    if (fetchTask != null && !rootTasks.isEmpty()) {
      rootTasks.get(rootTasks.size() - 1).setFetchSource(true);
    }
  }

  private void analyzeCacheMetadata(ASTNode ast) throws SemanticException {
    Table tbl = AnalyzeCommandUtils.getTable(ast, this);
    Map partSpec = null;
    CacheMetadataDesc desc;
    // In 2 cases out of 3, we could pass the path and type directly to metastore...
    if (AnalyzeCommandUtils.isPartitionLevelStats(ast)) {
      partSpec = AnalyzeCommandUtils.getPartKeyValuePairsFromAST(tbl, ast, conf);
      Partition part = getPartition(tbl, partSpec, true);
      desc = new CacheMetadataDesc(tbl.getDbName(), tbl.getTableName(), part.getName());
      inputs.add(new ReadEntity(part));
    } else {
      // Should we get all partitions for a partitioned table?
      desc = new CacheMetadataDesc(tbl.getDbName(), tbl.getTableName(), tbl.isPartitioned());
      inputs.add(new ReadEntity(tbl));
    }
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeAlterTableUpdateStats(ASTNode ast, String tblName, Map partSpec)
      throws SemanticException {
    String colName = getUnescapedName((ASTNode) ast.getChild(0));
    Map mapProp = getProps((ASTNode) (ast.getChild(1)).getChild(0));

    Table tbl = getTable(tblName);
    String partName = null;
    if (partSpec != null) {
      try {
        partName = Warehouse.makePartName(partSpec, false);
      } catch (MetaException e) {
        throw new SemanticException("partition " + partSpec.toString()
            + " not found");
      }
    }

    String colType = null;
    List cols = tbl.getCols();
    for (FieldSchema col : cols) {
      if (colName.equalsIgnoreCase(col.getName())) {
        colType = col.getType();
        break;
      }
    }

    if (colType == null) {
      throw new SemanticException("column type not found");
    }

    ColumnStatsUpdateWork columnStatsUpdateWork =
        new ColumnStatsUpdateWork(partName, mapProp, tbl.getDbName(), tbl.getTableName(), colName, colType);
    ColumnStatsUpdateTask cStatsUpdateTask = (ColumnStatsUpdateTask) TaskFactory
        .get(columnStatsUpdateWork);
    rootTasks.add(cStatsUpdateTask);
  }

  private void analyzeSetShowRole(ASTNode ast) throws SemanticException {
    switch (ast.getChildCount()) {
      case 0:
        ctx.setResFile(ctx.getLocalTmpPath());
        rootTasks.add(hiveAuthorizationTaskFactory.createShowCurrentRoleTask(
        getInputs(), getOutputs(), ctx.getResFile()));
        setFetchTask(createFetchTask(RoleDDLDesc.getRoleNameSchema()));
        break;
      case 1:
        rootTasks.add(hiveAuthorizationTaskFactory.createSetRoleTask(
        BaseSemanticAnalyzer.unescapeIdentifier(ast.getChild(0).getText()),
        getInputs(), getOutputs()));
        break;
      default:
        throw new SemanticException("Internal error. ASTNode expected to have 0 or 1 child. "
        + ast.dump());
    }
  }

  private void analyzeGrantRevokeRole(boolean grant, ASTNode ast) throws SemanticException {
    Task task;
    if(grant) {
      task = hiveAuthorizationTaskFactory.createGrantRoleTask(ast, getInputs(), getOutputs());
    } else {
      task = hiveAuthorizationTaskFactory.createRevokeRoleTask(ast, getInputs(), getOutputs());
    }
    if(task != null) {
      rootTasks.add(task);
    }
  }

  private void analyzeShowGrant(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createShowGrantTask(ast, ctx.getResFile(), getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
      setFetchTask(createFetchTask(ShowGrantDesc.getSchema()));
    }
  }

  private void analyzeGrant(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createGrantTask(ast, getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
    }
  }

  private void analyzeRevoke(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createRevokeTask(ast, getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
    }
  }

  private void analyzeCreateRole(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createCreateRoleTask(ast, getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
    }
  }

  private void analyzeDropRole(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createDropRoleTask(ast, getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
    }
  }

  private void analyzeShowRoleGrant(ASTNode ast) throws SemanticException {
    Task task = hiveAuthorizationTaskFactory.
        createShowRoleGrantTask(ast, ctx.getResFile(), getInputs(), getOutputs());
    if(task != null) {
      rootTasks.add(task);
      setFetchTask(createFetchTask(RoleDDLDesc.getRoleShowGrantSchema()));
    }
  }

  private void analyzeShowRolePrincipals(ASTNode ast) throws SemanticException {
    Task roleDDLTask = (Task) hiveAuthorizationTaskFactory
        .createShowRolePrincipalsTask(ast, ctx.getResFile(), getInputs(), getOutputs());

    if (roleDDLTask != null) {
      rootTasks.add(roleDDLTask);
      setFetchTask(createFetchTask(RoleDDLDesc.getShowRolePrincipalsSchema()));
    }
  }

  private void analyzeShowRoles(ASTNode ast) throws SemanticException {
    @SuppressWarnings("unchecked")
    Task roleDDLTask = (Task) hiveAuthorizationTaskFactory
        .createShowRolesTask(ast, ctx.getResFile(), getInputs(), getOutputs());

    if (roleDDLTask != null) {
      rootTasks.add(roleDDLTask);
      setFetchTask(createFetchTask(RoleDDLDesc.getRoleNameSchema()));
    }
  }

  private void analyzeAlterDatabaseProperties(ASTNode ast) throws SemanticException {

    String dbName = unescapeIdentifier(ast.getChild(0).getText());
    Map dbProps = null;

    for (int i = 1; i < ast.getChildCount(); i++) {
      ASTNode childNode = (ASTNode) ast.getChild(i);
      switch (childNode.getToken().getType()) {
      case HiveParser.TOK_DATABASEPROPERTIES:
        dbProps = DDLSemanticAnalyzer.getProps((ASTNode) childNode.getChild(0));
        break;
      default:
        throw new SemanticException("Unrecognized token in CREATE DATABASE statement");
      }
    }
    AlterDatabaseDesc alterDesc = new AlterDatabaseDesc(dbName, dbProps, null);
    addAlterDbDesc(alterDesc);
  }

  private void addAlterDbDesc(AlterDatabaseDesc alterDesc) throws SemanticException {
    Database database = getDatabase(alterDesc.getDatabaseName());
    outputs.add(new WriteEntity(database, WriteEntity.WriteType.DDL_NO_LOCK));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), alterDesc)));
  }

  private void analyzeAlterDatabaseOwner(ASTNode ast) throws SemanticException {
    String dbName = getUnescapedName((ASTNode) ast.getChild(0));
    PrincipalDesc principalDesc = AuthorizationParseUtils.getPrincipalDesc((ASTNode) ast
        .getChild(1));

    // The syntax should not allow these fields to be null, but lets verify
    String nullCmdMsg = "can't be null in alter database set owner command";
    if(principalDesc.getName() == null){
      throw new SemanticException("Owner name " + nullCmdMsg);
    }
    if(principalDesc.getType() == null){
      throw new SemanticException("Owner type " + nullCmdMsg);
    }

    AlterDatabaseDesc alterDesc = new AlterDatabaseDesc(dbName, principalDesc, null);
    addAlterDbDesc(alterDesc);
  }

  private void analyzeAlterDatabaseLocation(ASTNode ast) throws SemanticException {
    String dbName = getUnescapedName((ASTNode) ast.getChild(0));
    String newLocation = unescapeSQLString(ast.getChild(1).getText());
    addLocationToOutputs(newLocation);
    AlterDatabaseDesc alterDesc = new AlterDatabaseDesc(dbName, newLocation);
    addAlterDbDesc(alterDesc);
  }

  private void analyzeExchangePartition(String[] qualified, ASTNode ast) throws SemanticException {
    Table destTable = getTable(qualified);
    Table sourceTable = getTable(getUnescapedName((ASTNode)ast.getChild(1)));

    // Get the partition specs
    Map partSpecs = getValidatedPartSpec(sourceTable, (ASTNode)ast.getChild(0), conf, false);
    validatePartitionValues(partSpecs);
    boolean sameColumns = MetaStoreUtils.compareFieldColumns(
        destTable.getAllCols(), sourceTable.getAllCols());
    boolean samePartitions = MetaStoreUtils.compareFieldColumns(
        destTable.getPartitionKeys(), sourceTable.getPartitionKeys());
    if (!sameColumns || !samePartitions) {
      throw new SemanticException(ErrorMsg.TABLES_INCOMPATIBLE_SCHEMAS.getMsg());
    }

    // Exchange partition is not allowed with transactional tables.
    // If only source is transactional table, then target will see deleted rows too as no snapshot
    // isolation applicable for non-acid tables.
    // If only target is transactional table, then data would be visible to all ongoing transactions
    // affecting the snapshot isolation.
    // If both source and targets are transactional tables, then target partition may have delta/base
    // files with write IDs may not be valid. It may affect snapshot isolation for on-going txns as well.
    if (AcidUtils.isTransactionalTable(sourceTable) || AcidUtils.isTransactionalTable(destTable)) {
      throw new SemanticException(ErrorMsg.EXCHANGE_PARTITION_NOT_ALLOWED_WITH_TRANSACTIONAL_TABLES.getMsg());
    }

    // check if source partition exists
    getPartitions(sourceTable, partSpecs, true);

    // Verify that the partitions specified are continuous
    // If a subpartition value is specified without specifying a partition's value
    // then we throw an exception
    int counter = isPartitionValueContinuous(sourceTable.getPartitionKeys(), partSpecs);
    if (counter < 0) {
      throw new SemanticException(
          ErrorMsg.PARTITION_VALUE_NOT_CONTINUOUS.getMsg(partSpecs.toString()));
    }
    List destPartitions = null;
    try {
      destPartitions = getPartitions(destTable, partSpecs, true);
    } catch (SemanticException ex) {
      // We should expect a semantic exception being throw as this partition
      // should not be present.
    }
    if (destPartitions != null) {
      // If any destination partition is present then throw a Semantic Exception.
      throw new SemanticException(ErrorMsg.PARTITION_EXISTS.getMsg(destPartitions.toString()));
    }
    AlterTableExchangePartition alterTableExchangePartition =
      new AlterTableExchangePartition(sourceTable, destTable, partSpecs);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTableExchangePartition)));

    inputs.add(new ReadEntity(sourceTable));
    outputs.add(new WriteEntity(destTable, WriteType.DDL_SHARED));
  }

  /**
   * @param partitionKeys the list of partition keys of the table
   * @param partSpecs the partition specs given by the user
   * @return >=0 if no subpartition value is specified without a partition's
   *         value being specified else it returns -1
   */
  private int isPartitionValueContinuous(List partitionKeys,
      Map partSpecs) {
    int counter = 0;
    for (FieldSchema partitionKey : partitionKeys) {
      if (partSpecs.containsKey(partitionKey.getName())) {
        counter++;
        continue;
      }
      return partSpecs.size() == counter ? counter : -1;
    }
    return counter;
  }

  private void analyzeCreateResourcePlan(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() == 0) {
      throw new SemanticException("Expected name in CREATE RESOURCE PLAN statement");
    }
    String resourcePlanName = unescapeIdentifier(ast.getChild(0).getText());
    Integer queryParallelism = null;
    String likeName = null;
    for (int i = 1; i < ast.getChildCount(); ++i) {
      Tree child = ast.getChild(i);
      switch (child.getType()) {
      case HiveParser.TOK_QUERY_PARALLELISM:
        // Note: later we may be able to set multiple things together (except LIKE).
        if (queryParallelism == null && likeName == null) {
          queryParallelism = Integer.parseInt(child.getChild(0).getText());
        } else {
          throw new SemanticException("Conflicting create arguments " + ast.toStringTree());
        }
        break;
      case HiveParser.TOK_LIKERP:
        if (queryParallelism == null && likeName == null) {
          likeName = unescapeIdentifier(child.getChild(0).getText());
        } else {
          throw new SemanticException("Conflicting create arguments " + ast.toStringTree());
        }
        break;
      default: throw new SemanticException("Invalid create arguments " + ast.toStringTree());
      }
    }
    CreateResourcePlanDesc desc = new CreateResourcePlanDesc(
        resourcePlanName, queryParallelism, likeName);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeShowResourcePlan(ASTNode ast) throws SemanticException {
    String rpName = null;
    if (ast.getChildCount() > 0) {
      rpName = unescapeIdentifier(ast.getChild(0).getText());
    }
    if (ast.getChildCount() > 1) {
      throw new SemanticException("Invalid syntax for SHOW RESOURCE PLAN statement");
    }
    ShowResourcePlanDesc showResourcePlanDesc = new ShowResourcePlanDesc(rpName, ctx.getResFile());
    addServiceOutput();
    rootTasks.add(TaskFactory.get(
        new DDLWork(getInputs(), getOutputs(), showResourcePlanDesc)));
    setFetchTask(createFetchTask(showResourcePlanDesc.getSchema(rpName)));
  }

  private void analyzeAlterResourcePlan(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() < 1) {
      throw new SemanticException("Incorrect syntax");
    }
    Tree nameOrGlobal = ast.getChild(0);
    switch (nameOrGlobal.getType()) {
    case HiveParser.TOK_ENABLE:
      // This command exists solely to output this message. TODO: can we do it w/o an error?
      throw new SemanticException("Activate a resource plan to enable workload management");
    case HiveParser.TOK_DISABLE:
      WMNullableResourcePlan anyRp = new WMNullableResourcePlan();
      anyRp.setStatus(WMResourcePlanStatus.ENABLED);
      AlterResourcePlanDesc desc = new AlterResourcePlanDesc(
          anyRp, null, false, false, true, false);
      addServiceOutput();
        rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
      return;
    default: // Continue to handle changes to a specific plan.
    }
    if (ast.getChildCount() < 2) {
      throw new SemanticException("Invalid syntax for ALTER RESOURCE PLAN statement");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    WMNullableResourcePlan resourcePlan = new WMNullableResourcePlan();
    boolean isEnableActivate = false, isReplace = false;
    boolean validate = false;
    for (int i = 1; i < ast.getChildCount(); ++i) {
      Tree child = ast.getChild(i);
      switch (child.getType()) {
      case HiveParser.TOK_VALIDATE:
        validate = true;
        break;
      case HiveParser.TOK_ACTIVATE:
        if (resourcePlan.getStatus() == WMResourcePlanStatus.ENABLED) {
          isEnableActivate = true;
        }
        if (child.getChildCount() > 1) {
          throw new SemanticException("Expected 0 or 1 arguments " + ast.toStringTree());
        } else if (child.getChildCount() == 1) {
          if (child.getChild(0).getType() != HiveParser.TOK_REPLACE) {
            throw new SemanticException("Incorrect syntax " + ast.toStringTree());
          }
          isReplace = true;
          isEnableActivate = false; // Implied.
        }
        resourcePlan.setStatus(WMResourcePlanStatus.ACTIVE);
        break;
      case HiveParser.TOK_ENABLE:
        if (resourcePlan.getStatus() == WMResourcePlanStatus.ACTIVE) {
          isEnableActivate = !isReplace;
        } else {
          resourcePlan.setStatus(WMResourcePlanStatus.ENABLED);
        }
        break;
      case HiveParser.TOK_DISABLE:
        resourcePlan.setStatus(WMResourcePlanStatus.DISABLED);
        break;
      case HiveParser.TOK_REPLACE:
        isReplace = true;
        if (child.getChildCount() > 1) {
          throw new SemanticException("Expected 0 or 1 arguments " + ast.toStringTree());
        } else if (child.getChildCount() == 1) {
          // Replace is essentially renaming a plan to the name of an existing plan, with backup.
          resourcePlan.setName(unescapeIdentifier(child.getChild(0).getText()));
        } else {
          resourcePlan.setStatus(WMResourcePlanStatus.ACTIVE);
        }
        break;
      case HiveParser.TOK_QUERY_PARALLELISM: {
        if (child.getChildCount() != 1) {
          throw new SemanticException("Expected one argument");
        }
        Tree val = child.getChild(0);
        resourcePlan.setIsSetQueryParallelism(true);
        if (val.getType() == HiveParser.TOK_NULL) {
          resourcePlan.unsetQueryParallelism();
        } else {
          resourcePlan.setQueryParallelism(Integer.parseInt(val.getText()));
        }
        break;
      }
      case HiveParser.TOK_DEFAULT_POOL: {
        if (child.getChildCount() != 1) {
          throw new SemanticException("Expected one argument");
        }
        Tree val = child.getChild(0);
        resourcePlan.setIsSetDefaultPoolPath(true);
        if (val.getType() == HiveParser.TOK_NULL) {
          resourcePlan.unsetDefaultPoolPath();
        } else {
          resourcePlan.setDefaultPoolPath(poolPath(child.getChild(0)));
        }
        break;
      }
      case HiveParser.TOK_RENAME:
        if (child.getChildCount() != 1) {
          throw new SemanticException("Expected one argument");
        }
        resourcePlan.setName(unescapeIdentifier(child.getChild(0).getText()));
        break;
      default:
        throw new SemanticException(
          "Unexpected token in alter resource plan statement: " + child.getType());
      }
    }
    AlterResourcePlanDesc desc = new AlterResourcePlanDesc(
        resourcePlan, rpName, validate, isEnableActivate, false, isReplace);
    if (validate) {
      ctx.setResFile(ctx.getLocalTmpPath());
      desc.setResFile(ctx.getResFile().toString());
    }
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
    if (validate) {
      setFetchTask(createFetchTask(AlterResourcePlanDesc.getSchema()));
    }
  }

  private void analyzeDropResourcePlan(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() == 0) {
      throw new SemanticException("Expected name in DROP RESOURCE PLAN statement");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    DropResourcePlanDesc desc = new DropResourcePlanDesc(rpName);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(
        new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeCreateTrigger(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() != 4) {
      throw new SemanticException("Invalid syntax for create trigger statement");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String triggerName = unescapeIdentifier(ast.getChild(1).getText());
    String triggerExpression = buildTriggerExpression((ASTNode)ast.getChild(2));
    String actionExpression = buildTriggerActionExpression((ASTNode)ast.getChild(3));

    WMTrigger trigger = new WMTrigger(rpName, triggerName);
    trigger.setTriggerExpression(triggerExpression);
    trigger.setActionExpression(actionExpression);

    CreateWMTriggerDesc desc = new CreateWMTriggerDesc(trigger);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private String buildTriggerExpression(ASTNode ast) throws SemanticException {
    if (ast.getType() != HiveParser.TOK_TRIGGER_EXPRESSION || ast.getChildCount() == 0) {
      throw new SemanticException("Invalid trigger expression.");
    }
    StringBuilder builder = new StringBuilder();
    for (int i = 0; i < ast.getChildCount(); ++i) {
      builder.append(ast.getChild(i).getText()); // Don't strip quotes.
      builder.append(' ');
    }
    builder.deleteCharAt(builder.length() - 1);
    return builder.toString();
  }

  private String poolPath(Tree ast) {
    StringBuilder builder = new StringBuilder();
    builder.append(unescapeIdentifier(ast.getText()));
    for (int i = 0; i < ast.getChildCount(); ++i) {
      // DOT is not affected
      builder.append(unescapeIdentifier(ast.getChild(i).getText()));
    }
    return builder.toString();
  }

  private String buildTriggerActionExpression(ASTNode ast) throws SemanticException {
    switch (ast.getType()) {
    case HiveParser.KW_KILL:
      return "KILL";
    case HiveParser.KW_MOVE:
      if (ast.getChildCount() != 1) {
        throw new SemanticException("Invalid move to clause in trigger action.");
      }
      String poolPath = poolPath(ast.getChild(0));
      return "MOVE TO " + poolPath;
    default:
      throw new SemanticException("Unknown token in action clause: " + ast.getType());
    }
  }

  private void analyzeAlterTrigger(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() != 4) {
      throw new SemanticException("Invalid syntax for alter trigger statement");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String triggerName = unescapeIdentifier(ast.getChild(1).getText());
    String triggerExpression = buildTriggerExpression((ASTNode)ast.getChild(2));
    String actionExpression = buildTriggerActionExpression((ASTNode)ast.getChild(3));

    WMTrigger trigger = new WMTrigger(rpName, triggerName);
    trigger.setTriggerExpression(triggerExpression);
    trigger.setActionExpression(actionExpression);

    AlterWMTriggerDesc desc = new AlterWMTriggerDesc(trigger);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeDropTrigger(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() != 2) {
      throw new SemanticException("Invalid syntax for drop trigger.");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String triggerName = unescapeIdentifier(ast.getChild(1).getText());

    DropWMTriggerDesc desc = new DropWMTriggerDesc(rpName, triggerName);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(
        new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeCreatePool(ASTNode ast) throws SemanticException {
    // TODO: allow defaults for e.g. scheduling policy.
    if (ast.getChildCount() < 3) {
      throw new SemanticException("Expected more arguments: " + ast.toStringTree());
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String poolPath = poolPath(ast.getChild(1));
    WMPool pool = new WMPool(rpName, poolPath);
    for (int i = 2; i < ast.getChildCount(); ++i) {
      Tree child = ast.getChild(i);
      if (child.getChildCount() != 1) {
        throw new SemanticException("Expected 1 paramter for: " + child.getText());
      }
      String param = child.getChild(0).getText();
      switch (child.getType()) {
      case HiveParser.TOK_ALLOC_FRACTION:
        pool.setAllocFraction(Double.parseDouble(param));
        break;
      case HiveParser.TOK_QUERY_PARALLELISM:
        pool.setQueryParallelism(Integer.parseInt(param));
        break;
      case HiveParser.TOK_SCHEDULING_POLICY:
        String schedulingPolicyStr = PlanUtils.stripQuotes(param);
        if (!MetaStoreUtils.isValidSchedulingPolicy(schedulingPolicyStr)) {
          throw new SemanticException("Invalid scheduling policy " + schedulingPolicyStr);
        }
        pool.setSchedulingPolicy(schedulingPolicyStr);
        break;
      case HiveParser.TOK_PATH:
        throw new SemanticException("Invalid parameter path in create pool");
      }
    }
    if (!pool.isSetAllocFraction()) {
      throw new SemanticException("alloc_fraction should be specified for a pool");
    }
    if (!pool.isSetQueryParallelism()) {
      throw new SemanticException("query_parallelism should be specified for a pool");
    }
    CreateOrAlterWMPoolDesc desc = new CreateOrAlterWMPoolDesc(pool, poolPath, false);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(
        new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeAlterPool(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() < 3) {
      throw new SemanticException("Invalid syntax for alter pool: " + ast.toStringTree());
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    Tree poolTarget = ast.getChild(1);

    boolean isUnmanagedPool = false;
    String poolPath = null;
    if (poolTarget.getType() == HiveParser.TOK_UNMANAGED) {
      isUnmanagedPool = true;
    } else {
      poolPath = poolPath(ast.getChild(1));
    }

    WMNullablePool poolChanges = null;
    boolean hasTrigger = false;
    for (int i = 2; i < ast.getChildCount(); ++i) {
      Tree child = ast.getChild(i);
      if (child.getChildCount() != 1) {
        throw new SemanticException("Invalid syntax in alter pool expected parameter.");
      }
      Tree param = child.getChild(0);
      if (child.getType() == HiveParser.TOK_ADD_TRIGGER
          || child.getType() == HiveParser.TOK_DROP_TRIGGER) {
        hasTrigger = true;
        boolean drop = child.getType() == HiveParser.TOK_DROP_TRIGGER;
        String triggerName = unescapeIdentifier(param.getText());
        rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
            new CreateOrDropTriggerToPoolMappingDesc(
                rpName, triggerName, poolPath, drop, isUnmanagedPool))));
      } else {
        if (isUnmanagedPool) {
          throw new SemanticException("Cannot alter the unmanaged pool");
        }
        if (poolChanges == null) {
          poolChanges = new WMNullablePool(rpName, null);
        }
        switch (child.getType()) {
        case HiveParser.TOK_ALLOC_FRACTION:
          poolChanges.setAllocFraction(Double.parseDouble(param.getText()));
          break;
        case HiveParser.TOK_QUERY_PARALLELISM:
          poolChanges.setQueryParallelism(Integer.parseInt(param.getText()));
          break;
        case HiveParser.TOK_SCHEDULING_POLICY:
          poolChanges.setIsSetSchedulingPolicy(true);
          if (param.getType() != HiveParser.TOK_NULL) {
            poolChanges.setSchedulingPolicy(PlanUtils.stripQuotes(param.getText()));
          }
          break;
        case HiveParser.TOK_PATH:
          poolChanges.setPoolPath(poolPath(param));
          break;
        default: throw new SemanticException("Incorrect alter syntax: " + child.toStringTree());
        }
      }
    }

    if (poolChanges != null || hasTrigger) {
      addServiceOutput();
    }
    if (poolChanges != null) {
      if (!poolChanges.isSetPoolPath()) {
        poolChanges.setPoolPath(poolPath);
      }
      rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
          new CreateOrAlterWMPoolDesc(poolChanges, poolPath, true))));
    }
  }

  private void analyzeDropPool(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() != 2) {
      throw new SemanticException("Invalid syntax for drop pool.");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String poolPath = poolPath(ast.getChild(1));

    DropWMPoolDesc desc = new DropWMPoolDesc(rpName, poolPath);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(
        new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeCreateOrAlterMapping(ASTNode ast, boolean update) throws SemanticException {
    if (ast.getChildCount() < 4) {
      throw new SemanticException("Invalid syntax for create or alter mapping.");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String entityType = ast.getChild(1).getText();
    String entityName = PlanUtils.stripQuotes(ast.getChild(2).getText());
    WMMapping mapping = new WMMapping(rpName, entityType, entityName);
    Tree dest = ast.getChild(3);
    if (dest.getType() != HiveParser.TOK_UNMANAGED) {
      mapping.setPoolPath(poolPath(dest));
    } // Null path => unmanaged
    if (ast.getChildCount() == 5) {
      mapping.setOrdering(Integer.valueOf(ast.getChild(4).getText()));
    }

    CreateOrAlterWMMappingDesc desc = new CreateOrAlterWMMappingDesc(mapping, update);
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeDropMapping(ASTNode ast) throws SemanticException {
    if (ast.getChildCount() != 3) {
      throw new SemanticException("Invalid syntax for drop mapping.");
    }
    String rpName = unescapeIdentifier(ast.getChild(0).getText());
    String entityType = ast.getChild(1).getText();
    String entityName = PlanUtils.stripQuotes(ast.getChild(2).getText());

    DropWMMappingDesc desc = new DropWMMappingDesc(new WMMapping(rpName, entityType, entityName));
    addServiceOutput();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeCreateDatabase(ASTNode ast) throws SemanticException {
    String dbName = unescapeIdentifier(ast.getChild(0).getText());
    boolean ifNotExists = false;
    String dbComment = null;
    String dbLocation = null;
    Map dbProps = null;

    for (int i = 1; i < ast.getChildCount(); i++) {
      ASTNode childNode = (ASTNode) ast.getChild(i);
      switch (childNode.getToken().getType()) {
      case HiveParser.TOK_IFNOTEXISTS:
        ifNotExists = true;
        break;
      case HiveParser.TOK_DATABASECOMMENT:
        dbComment = unescapeSQLString(childNode.getChild(0).getText());
        break;
      case TOK_DATABASEPROPERTIES:
        dbProps = DDLSemanticAnalyzer.getProps((ASTNode) childNode.getChild(0));
        break;
      case TOK_DATABASELOCATION:
        dbLocation = unescapeSQLString(childNode.getChild(0).getText());
        addLocationToOutputs(dbLocation);
        break;
      default:
        throw new SemanticException("Unrecognized token in CREATE DATABASE statement");
      }
    }

    CreateDatabaseDesc createDatabaseDesc =
        new CreateDatabaseDesc(dbName, dbComment, dbLocation, ifNotExists);
    if (dbProps != null) {
      createDatabaseDesc.setDatabaseProperties(dbProps);
    }
    Database database = new Database(dbName, dbComment, dbLocation, dbProps);
    outputs.add(new WriteEntity(database, WriteEntity.WriteType.DDL_NO_LOCK));

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        createDatabaseDesc)));
  }

  private void analyzeDropDatabase(ASTNode ast) throws SemanticException {
    String dbName = unescapeIdentifier(ast.getChild(0).getText());
    boolean ifExists = false;
    boolean ifCascade = false;

    if (null != ast.getFirstChildWithType(HiveParser.TOK_IFEXISTS)) {
      ifExists = true;
    }

    if (null != ast.getFirstChildWithType(HiveParser.TOK_CASCADE)) {
      ifCascade = true;
    }

    Database database = getDatabase(dbName, !ifExists);
    if (database == null) {
      return;
    }

    // if cascade=true, then we need to authorize the drop table action as well
    if (ifCascade) {
      // add the tables as well to outputs
      List tableNames;
      // get names of all tables under this dbName
      try {
        tableNames = db.getAllTables(dbName);
      } catch (HiveException e) {
        throw new SemanticException(e);
      }
      // add tables to outputs
      if (tableNames != null) {
        for (String tableName : tableNames) {
          Table table = getTable(dbName, tableName, true);
          // We want no lock here, as the database lock will cover the tables,
          // and putting a lock will actually cause us to deadlock on ourselves.
          outputs.add(new WriteEntity(table, WriteEntity.WriteType.DDL_NO_LOCK));
        }
      }
    }
    inputs.add(new ReadEntity(database));
    outputs.add(new WriteEntity(database, WriteEntity.WriteType.DDL_EXCLUSIVE));

    DropDatabaseDesc dropDatabaseDesc = new DropDatabaseDesc(dbName, ifExists, ifCascade,
        new ReplicationSpec());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), dropDatabaseDesc)));
  }

  private void analyzeSwitchDatabase(ASTNode ast) throws SemanticException {
    String dbName = unescapeIdentifier(ast.getChild(0).getText());
    Database database = getDatabase(dbName, true);
    ReadEntity dbReadEntity = new ReadEntity(database);
    dbReadEntity.noLockNeeded();
    inputs.add(dbReadEntity);
    SwitchDatabaseDesc switchDatabaseDesc = new SwitchDatabaseDesc(dbName);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        switchDatabaseDesc)));
  }



  private void analyzeDropTable(ASTNode ast, TableType expectedType)
      throws SemanticException {
    String tableName = getUnescapedName((ASTNode) ast.getChild(0));
    boolean ifExists = (ast.getFirstChildWithType(HiveParser.TOK_IFEXISTS) != null);
    // we want to signal an error if the table/view doesn't exist and we're
    // configured not to fail silently
    boolean throwException =
        !ifExists && !HiveConf.getBoolVar(conf, ConfVars.DROPIGNORESNONEXISTENT);

    ReplicationSpec replicationSpec = new ReplicationSpec(ast);

    Table tab = getTable(tableName, throwException);
    if (tab != null) {
      inputs.add(new ReadEntity(tab));
      outputs.add(new WriteEntity(tab, WriteEntity.WriteType.DDL_EXCLUSIVE));
    }

    boolean ifPurge = (ast.getFirstChildWithType(HiveParser.KW_PURGE) != null);
    DropTableDesc dropTblDesc = new DropTableDesc(tableName, expectedType, ifExists, ifPurge, replicationSpec);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        dropTblDesc)));
  }

  private void analyzeTruncateTable(ASTNode ast) throws SemanticException {
    ASTNode root = (ASTNode) ast.getChild(0); // TOK_TABLE_PARTITION
    String tableName = getUnescapedName((ASTNode) root.getChild(0));

    Table table = getTable(tableName, true);
    if (table.getTableType() != TableType.MANAGED_TABLE) {
      throw new SemanticException(ErrorMsg.TRUNCATE_FOR_NON_MANAGED_TABLE.format(tableName));
    }
    if (table.isNonNative()) {
      throw new SemanticException(ErrorMsg.TRUNCATE_FOR_NON_NATIVE_TABLE.format(tableName)); //TODO
    }
    if (!table.isPartitioned() && root.getChildCount() > 1) {
      throw new SemanticException(ErrorMsg.PARTSPEC_FOR_NON_PARTITIONED_TABLE.format(tableName));
    }
    Map partSpec = getPartSpec((ASTNode) root.getChild(1));
    if (partSpec == null) {
      if (!table.isPartitioned()) {
        outputs.add(new WriteEntity(table, WriteEntity.WriteType.DDL_EXCLUSIVE));
      } else {
        for (Partition partition : getPartitions(table, null, false)) {
          outputs.add(new WriteEntity(partition, WriteEntity.WriteType.DDL_EXCLUSIVE));
        }
      }
    } else {
      if (isFullSpec(table, partSpec)) {
        validatePartSpec(table, partSpec, (ASTNode) root.getChild(1), conf, true);
        Partition partition = getPartition(table, partSpec, true);
        outputs.add(new WriteEntity(partition, WriteEntity.WriteType.DDL_EXCLUSIVE));
      } else {
        validatePartSpec(table, partSpec, (ASTNode) root.getChild(1), conf, false);
        for (Partition partition : getPartitions(table, partSpec, false)) {
          outputs.add(new WriteEntity(partition, WriteEntity.WriteType.DDL_EXCLUSIVE));
        }
      }
    }

    TruncateTableDesc truncateTblDesc = new TruncateTableDesc(tableName, partSpec, null);

    DDLWork ddlWork = new DDLWork(getInputs(), getOutputs(), truncateTblDesc);
    Task truncateTask = TaskFactory.get(ddlWork);

    // Is this a truncate column command
    List columnNames = null;
    if (ast.getChildCount() == 2) {
      try {
        columnNames = getColumnNames((ASTNode)ast.getChild(1));

        // It would be possible to support this, but this is such a pointless command.
        if (AcidUtils.isInsertOnlyTable(table.getParameters())) {
          throw new SemanticException("Truncating MM table columns not presently supported");
        }

        List bucketCols = null;
        Class inputFormatClass = null;
        boolean isArchived = false;
        Path newTblPartLoc = null;
        Path oldTblPartLoc = null;
        List cols = null;
        ListBucketingCtx lbCtx = null;
        boolean isListBucketed = false;
        List listBucketColNames = null;

        if (table.isPartitioned()) {
          Partition part = db.getPartition(table, partSpec, false);

          Path tabPath = table.getPath();
          Path partPath = part.getDataLocation();

          // if the table is in a different dfs than the partition,
          // replace the partition's dfs with the table's dfs.
          newTblPartLoc = new Path(tabPath.toUri().getScheme(), tabPath.toUri()
              .getAuthority(), partPath.toUri().getPath());

          oldTblPartLoc = partPath;

          cols = part.getCols();
          bucketCols = part.getBucketCols();
          inputFormatClass = part.getInputFormatClass();
          isArchived = ArchiveUtils.isArchived(part);
          lbCtx = constructListBucketingCtx(part.getSkewedColNames(), part.getSkewedColValues(),
              part.getSkewedColValueLocationMaps(), part.isStoredAsSubDirectories(), conf);
          isListBucketed = part.isStoredAsSubDirectories();
          listBucketColNames = part.getSkewedColNames();
        } else {
          // input and output are the same
          oldTblPartLoc = table.getPath();
          newTblPartLoc = table.getPath();
          cols  = table.getCols();
          bucketCols = table.getBucketCols();
          inputFormatClass = table.getInputFormatClass();
          lbCtx = constructListBucketingCtx(table.getSkewedColNames(), table.getSkewedColValues(),
              table.getSkewedColValueLocationMaps(), table.isStoredAsSubDirectories(), conf);
          isListBucketed = table.isStoredAsSubDirectories();
          listBucketColNames = table.getSkewedColNames();
        }

        // throw a HiveException for non-rcfile.
        if (!inputFormatClass.equals(RCFileInputFormat.class)) {
          throw new SemanticException(ErrorMsg.TRUNCATE_COLUMN_NOT_RC.getMsg());
        }

        // throw a HiveException if the table/partition is archived
        if (isArchived) {
          throw new SemanticException(ErrorMsg.TRUNCATE_COLUMN_ARCHIVED.getMsg());
        }

        Set columnIndexes = new HashSet();
        for (String columnName : columnNames) {
          boolean found = false;
          for (int columnIndex = 0; columnIndex < cols.size(); columnIndex++) {
            if (columnName.equalsIgnoreCase(cols.get(columnIndex).getName())) {
              columnIndexes.add(columnIndex);
              found = true;
              break;
            }
          }
          // Throw an exception if the user is trying to truncate a column which doesn't exist
          if (!found) {
            throw new SemanticException(ErrorMsg.INVALID_COLUMN.getMsg(columnName));
          }
          // Throw an exception if the table/partition is bucketed on one of the columns
          for (String bucketCol : bucketCols) {
            if (bucketCol.equalsIgnoreCase(columnName)) {
              throw new SemanticException(ErrorMsg.TRUNCATE_BUCKETED_COLUMN.getMsg(columnName));
            }
          }
          if (isListBucketed) {
            for (String listBucketCol : listBucketColNames) {
              if (listBucketCol.equalsIgnoreCase(columnName)) {
                throw new SemanticException(
                    ErrorMsg.TRUNCATE_LIST_BUCKETED_COLUMN.getMsg(columnName));
              }
            }
          }
        }

        truncateTblDesc.setColumnIndexes(new ArrayList(columnIndexes));
        truncateTblDesc.setInputDir(oldTblPartLoc);
        truncateTblDesc.setLbCtx(lbCtx);

        addInputsOutputsAlterTable(tableName, partSpec, AlterTableTypes.TRUNCATE);
        ddlWork.setNeedLock(true);
        TableDesc tblDesc = Utilities.getTableDesc(table);
        // Write the output to temporary directory and move it to the final location at the end
        // so the operation is atomic.
        Path queryTmpdir = ctx.getExternalTmpPath(newTblPartLoc);
        truncateTblDesc.setOutputDir(queryTmpdir);
        LoadTableDesc ltd = new LoadTableDesc(queryTmpdir, tblDesc,
            partSpec == null ? new HashMap<>() : partSpec);
        ltd.setLbCtx(lbCtx);
        @SuppressWarnings("unchecked")
        Task moveTsk =
            TaskFactory.get(new MoveWork(null, null, ltd, null, false));
        truncateTask.addDependentTask(moveTsk);

        // Recalculate the HDFS stats if auto gather stats is set
        if (conf.getBoolVar(HiveConf.ConfVars.HIVESTATSAUTOGATHER)) {
          BasicStatsWork basicStatsWork;
          if (oldTblPartLoc.equals(newTblPartLoc)) {
            // If we're merging to the same location, we can avoid some metastore calls
            TableSpec tablepart = new TableSpec(this.db, conf, root);
            basicStatsWork = new BasicStatsWork(tablepart);
          } else {
            basicStatsWork = new BasicStatsWork(ltd);
          }
          basicStatsWork.setNoStatsAggregator(true);
          basicStatsWork.setClearAggregatorStats(true);
          StatsWork columnStatsWork = new StatsWork(table, basicStatsWork, conf);

          Task statTask = TaskFactory.get(columnStatsWork);
          moveTsk.addDependentTask(statTask);
        }
      } catch (HiveException e) {
        throw new SemanticException(e);
      }
    }

    rootTasks.add(truncateTask);
  }

  public static boolean isFullSpec(Table table, Map partSpec) {
    for (FieldSchema partCol : table.getPartCols()) {
      if (partSpec.get(partCol.getName()) == null) {
        return false;
      }
    }
    return true;
  }

  private void validateAlterTableType(Table tbl, AlterTableTypes op) throws SemanticException {
    validateAlterTableType(tbl, op, false);
  }

  private void validateAlterTableType(Table tbl, AlterTableTypes op, boolean expectView)
      throws SemanticException {
    if (tbl.isView()) {
      if (!expectView) {
        throw new SemanticException(ErrorMsg.ALTER_COMMAND_FOR_VIEWS.getMsg());
      }

      switch (op) {
      case ADDPARTITION:
      case DROPPARTITION:
      case RENAMEPARTITION:
      case ADDPROPS:
      case DROPPROPS:
      case RENAME:
        // allow this form
        break;
      default:
        throw new SemanticException(ErrorMsg.ALTER_VIEW_DISALLOWED_OP.getMsg(op.toString()));
      }
    } else {
      if (expectView) {
        throw new SemanticException(ErrorMsg.ALTER_COMMAND_FOR_TABLES.getMsg());
      }
    }
    if (tbl.isNonNative() && !AlterTableTypes.nonNativeTableAllowedTypes.contains(op)) {
      throw new SemanticException(ErrorMsg.ALTER_TABLE_NON_NATIVE.getMsg(tbl.getTableName()));
    }
  }

  private boolean hasConstraintsEnabled(final String tblName) throws SemanticException{

    NotNullConstraint nnc = null;
    DefaultConstraint dc = null;
    try {
      // retrieve enabled NOT NULL constraint from metastore
      nnc = Hive.get().getEnabledNotNullConstraints(
          db.getDatabaseCurrent().getName(), tblName);
      dc = Hive.get().getEnabledDefaultConstraints(
          db.getDatabaseCurrent().getName(), tblName);
    } catch (Exception e) {
      if (e instanceof SemanticException) {
        throw (SemanticException) e;
      } else {
        throw (new RuntimeException(e));
      }
    }
    if((nnc != null  && !nnc.getNotNullConstraints().isEmpty())
        || (dc != null && !dc.getDefaultConstraints().isEmpty())) {
      return true;
    }
    return false;
  }

  private void analyzeAlterTableProps(String[] qualified, HashMap partSpec,
      ASTNode ast, boolean expectView, boolean isUnset) throws SemanticException {

    String tableName = getDotName(qualified);
    HashMap mapProp = getProps((ASTNode) (ast.getChild(0))
        .getChild(0));
    EnvironmentContext environmentContext = null;
    // we need to check if the properties are valid, especially for stats.
    // they might be changed via alter table .. update statistics or
    // alter table .. set tblproperties. If the property is not row_count
    // or raw_data_size, it could not be changed through update statistics
    boolean changeStatsSucceeded = false;
    for (Entry entry : mapProp.entrySet()) {
      // we make sure that we do not change anything if there is anything
      // wrong.
      if (entry.getKey().equals(StatsSetupConst.ROW_COUNT)
          || entry.getKey().equals(StatsSetupConst.RAW_DATA_SIZE)) {
        try {
          Long.parseLong(entry.getValue());
          changeStatsSucceeded = true;
        } catch (Exception e) {
          throw new SemanticException("AlterTable " + entry.getKey() + " failed with value "
              + entry.getValue());
        }
      }
      // if table is being modified to be external we need to make sure existing table
      // doesn't have enabled constraint since constraints are disallowed with such tables
      else if(entry.getKey().equals("external") && entry.getValue().equals("true")){
        if(hasConstraintsEnabled(qualified[1])){
          throw new SemanticException(
              ErrorMsg.INVALID_CSTR_SYNTAX.getMsg("Table: " + tableName + " has constraints enabled."
                  + "Please remove those constraints to change this property."));
        }
      }
      else {
        if (queryState.getCommandType()
            .equals(HiveOperation.ALTERTABLE_UPDATETABLESTATS.getOperationName())
            || queryState.getCommandType()
                .equals(HiveOperation.ALTERTABLE_UPDATEPARTSTATS.getOperationName())) {
          throw new SemanticException("AlterTable UpdateStats " + entry.getKey()
              + " failed because the only valid keys are " + StatsSetupConst.ROW_COUNT + " and "
              + StatsSetupConst.RAW_DATA_SIZE);
        }
      }

      if (changeStatsSucceeded) {
        environmentContext = new EnvironmentContext();
        environmentContext.putToProperties(StatsSetupConst.STATS_GENERATED, StatsSetupConst.USER);
      }
    }
    AlterTableDesc alterTblDesc = null;
    if (isUnset == true) {
      alterTblDesc = new AlterTableDesc(AlterTableTypes.DROPPROPS, partSpec, expectView);
      if (ast.getChild(1) != null) {
        alterTblDesc.setDropIfExists(true);
      }
    } else {
      alterTblDesc = new AlterTableDesc(AlterTableTypes.ADDPROPS, partSpec, expectView);
    }
    alterTblDesc.setProps(mapProp);
    alterTblDesc.setEnvironmentContext(environmentContext);
    alterTblDesc.setOldName(tableName);

    boolean isPotentialMmSwitch = AcidUtils.isTablePropertyTransactional(mapProp)
        || mapProp.containsKey(hive_metastoreConstants.TABLE_TRANSACTIONAL_PROPERTIES);
    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc, isPotentialMmSwitch);

    DDLWork ddlWork = new DDLWork(getInputs(), getOutputs(), alterTblDesc);
    if (isPotentialMmSwitch) {
      this.ddlDescWithWriteId = alterTblDesc;
      ddlWork.setNeedLock(true); // Hmm... why don't many other operations here need locks?
    }

    rootTasks.add(TaskFactory.get(ddlWork));
  }

  @Override
  public DDLDescWithWriteId getAcidDdlDesc() {
    return ddlDescWithWriteId;
  }

  private void analyzeAlterTableSerdeProps(ASTNode ast, String tableName,
      HashMap partSpec)
      throws SemanticException {
    HashMap mapProp = getProps((ASTNode) (ast.getChild(0))
        .getChild(0));
    AlterTableDesc alterTblDesc = new AlterTableDesc(
        AlterTableTypes.ADDSERDEPROPS);
    alterTblDesc.setProps(mapProp);
    alterTblDesc.setOldName(tableName);
    alterTblDesc.setPartSpec(partSpec);

    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableSerde(ASTNode ast, String tableName,
      HashMap partSpec)
      throws SemanticException {

    String serdeName = unescapeSQLString(ast.getChild(0).getText());
    AlterTableDesc alterTblDesc = new AlterTableDesc(AlterTableTypes.ADDSERDE);
    if (ast.getChildCount() > 1) {
      HashMap mapProp = getProps((ASTNode) (ast.getChild(1))
          .getChild(0));
      alterTblDesc.setProps(mapProp);
    }
    alterTblDesc.setOldName(tableName);
    alterTblDesc.setSerdeName(serdeName);
    alterTblDesc.setPartSpec(partSpec);

    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableFileFormat(ASTNode ast, String tableName,
      HashMap partSpec)
      throws SemanticException {

    StorageFormat format = new StorageFormat(conf);
    ASTNode child = (ASTNode) ast.getChild(0);

    if (!format.fillStorageFormat(child)) {
      throw new AssertionError("Unknown token " + child.getText());
    }

    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, format.getInputFormat(),
        format.getOutputFormat(), format.getSerde(), format.getStorageHandler(), partSpec);

    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private WriteType determineAlterTableWriteType(Table tab, AlterTableDesc desc, AlterTableTypes op) {
    boolean convertingToAcid = false;
    if(desc != null && desc.getProps() != null && Boolean.parseBoolean(desc.getProps().get(hive_metastoreConstants.TABLE_IS_TRANSACTIONAL))) {
      convertingToAcid = true;
    }
    if(!AcidUtils.isTransactionalTable(tab) && convertingToAcid) {
      //non-acid to transactional conversion (property itself) must be mutexed to prevent concurrent writes.
      // See HIVE-16688 for use cases.
      return WriteType.DDL_EXCLUSIVE;
    }
    return WriteEntity.determineAlterTableWriteType(op);
  }
  private void addInputsOutputsAlterTable(String tableName, Map partSpec,
      AlterTableTypes op) throws SemanticException {
    addInputsOutputsAlterTable(tableName, partSpec, null, op, false);
  }

  private void addInputsOutputsAlterTable(String tableName, Map partSpec,
      AlterTableDesc desc, boolean doForceExclusive) throws SemanticException {
     addInputsOutputsAlterTable(tableName, partSpec, desc, desc.getOp(), doForceExclusive);
  }

  private void addInputsOutputsAlterTable(String tableName, Map partSpec,
     AlterTableDesc desc) throws SemanticException {
    addInputsOutputsAlterTable(tableName, partSpec, desc, desc.getOp(), false);
  }

  private void addInputsOutputsAlterTable(String tableName, Map partSpec,
      AlterTableDesc desc, AlterTableTypes op, boolean doForceExclusive) throws SemanticException {
    boolean isCascade = desc != null && desc.getIsCascade();
    boolean alterPartitions = partSpec != null && !partSpec.isEmpty();
    //cascade only occurs at table level then cascade to partition level
    if (isCascade && alterPartitions) {
      throw new SemanticException(
          ErrorMsg.ALTER_TABLE_PARTITION_CASCADE_NOT_SUPPORTED, op.getName());
    }

    Table tab = getTable(tableName, true);
    // cascade only occurs with partitioned table
    if (isCascade && !tab.isPartitioned()) {
      throw new SemanticException(
          ErrorMsg.ALTER_TABLE_NON_PARTITIONED_TABLE_CASCADE_NOT_SUPPORTED);
    }

    // Determine the lock type to acquire
    WriteEntity.WriteType writeType = doForceExclusive
        ? WriteType.DDL_EXCLUSIVE : determineAlterTableWriteType(tab, desc, op);

    if (!alterPartitions) {
      inputs.add(new ReadEntity(tab));
      alterTableOutput = new WriteEntity(tab, writeType);
      outputs.add(alterTableOutput);
      //do not need the lock for partitions since they are covered by the table lock
      if (isCascade) {
        for (Partition part : getPartitions(tab, partSpec, false)) {
          outputs.add(new WriteEntity(part, WriteEntity.WriteType.DDL_NO_LOCK));
        }
      }
    } else {
      ReadEntity re = new ReadEntity(tab);
      // In the case of altering a table for its partitions we don't need to lock the table
      // itself, just the partitions.  But the table will have a ReadEntity.  So mark that
      // ReadEntity as no lock.
      re.noLockNeeded();
      inputs.add(re);

      if (isFullSpec(tab, partSpec)) {
        // Fully specified partition spec
        Partition part = getPartition(tab, partSpec, true);
        outputs.add(new WriteEntity(part, writeType));
      } else {
        // Partial partition spec supplied. Make sure this is allowed.
        if (!AlterTableDesc.doesAlterTableTypeSupportPartialPartitionSpec(op)) {
          throw new SemanticException(
              ErrorMsg.ALTER_TABLE_TYPE_PARTIAL_PARTITION_SPEC_NO_SUPPORTED, op.getName());
        } else if (!conf.getBoolVar(HiveConf.ConfVars.DYNAMICPARTITIONING)) {
          throw new SemanticException(ErrorMsg.DYNAMIC_PARTITION_DISABLED);
        }

        for (Partition part : getPartitions(tab, partSpec, true)) {
          outputs.add(new WriteEntity(part, writeType));
        }
      }
    }

    if (desc != null) {
      validateAlterTableType(tab, op, desc.getExpectView());

      // validate Unset Non Existed Table Properties
      if (op == AlterTableDesc.AlterTableTypes.DROPPROPS && !desc.getIsDropIfExists()) {
        Map tableParams = tab.getTTable().getParameters();
        for (String currKey : desc.getProps().keySet()) {
          if (!tableParams.containsKey(currKey)) {
            String errorMsg =
                "The following property " + currKey +
                " does not exist in " + tab.getTableName();
            throw new SemanticException(
              ErrorMsg.ALTER_TBL_UNSET_NON_EXIST_PROPERTY.getMsg(errorMsg));
          }
        }
      }
    }
  }

  private void analyzeAlterTableOwner(ASTNode ast, String tableName) throws SemanticException {
    PrincipalDesc ownerPrincipal = AuthorizationParseUtils.getPrincipalDesc((ASTNode) ast.getChild(0));

    if (ownerPrincipal.getType() == null) {
      throw new SemanticException("Owner type can't be null in alter table set owner command");
    }

    if (ownerPrincipal.getName() == null) {
      throw new SemanticException("Owner name can't be null in alter table set owner command");
    }

    AlterTableDesc alterTblDesc  = new AlterTableDesc(tableName, ownerPrincipal);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), alterTblDesc), conf));
  }

  private void analyzeAlterTableLocation(ASTNode ast, String tableName,
      HashMap partSpec) throws SemanticException {

    String newLocation = unescapeSQLString(ast.getChild(0).getText());
    try {
      // To make sure host/port pair is valid, the status of the location
      // does not matter
      FileSystem.get(new URI(newLocation), conf).getFileStatus(new Path(newLocation));
    } catch (FileNotFoundException e) {
      // Only check host/port pair is valid, wheter the file exist or not does not matter
    } catch (Exception e) {
      throw new SemanticException("Cannot connect to namenode, please check if host/port pair for " + newLocation + " is valid", e);
    }
    addLocationToOutputs(newLocation);
    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, newLocation, partSpec);

    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));

  }

  private void analyzeAlterTablePartMergeFiles(ASTNode ast,
      String tableName, HashMap partSpec)
      throws SemanticException {
    AlterTablePartMergeFilesDesc mergeDesc = new AlterTablePartMergeFilesDesc(
        tableName, partSpec);

    List inputDir = new ArrayList();
    Path oldTblPartLoc = null;
    Path newTblPartLoc = null;
    Table tblObj = null;
    ListBucketingCtx lbCtx = null;

    try {
      tblObj = getTable(tableName);
      if(AcidUtils.isTransactionalTable(tblObj)) {
        LinkedHashMap newPartSpec = null;
        if (partSpec != null) {
          newPartSpec = new LinkedHashMap<>(partSpec);
        }

        boolean isBlocking = !HiveConf.getBoolVar(conf,
            ConfVars.TRANSACTIONAL_CONCATENATE_NOBLOCK, false);
        AlterTableSimpleDesc desc = new AlterTableSimpleDesc(
            tableName, newPartSpec, "MAJOR", isBlocking);

        rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
        return;
      }
      mergeDesc.setTableDesc(Utilities.getTableDesc(tblObj));

      List bucketCols = null;
      Class inputFormatClass = null;
      boolean isArchived = false;
      if (tblObj.isPartitioned()) {
        if (partSpec == null) {
          throw new SemanticException("source table " + tableName
              + " is partitioned but no partition desc found.");
        } else {
          Partition part = getPartition(tblObj, partSpec, false);
          if (part == null) {
            throw new SemanticException("source table " + tableName
                + " is partitioned but partition not found.");
          }
          bucketCols = part.getBucketCols();
          inputFormatClass = part.getInputFormatClass();
          isArchived = ArchiveUtils.isArchived(part);

          Path tabPath = tblObj.getPath();
          Path partPath = part.getDataLocation();

          // if the table is in a different dfs than the partition,
          // replace the partition's dfs with the table's dfs.
          newTblPartLoc = new Path(tabPath.toUri().getScheme(), tabPath.toUri()
              .getAuthority(), partPath.toUri().getPath());

          oldTblPartLoc = partPath;

          lbCtx = constructListBucketingCtx(part.getSkewedColNames(), part.getSkewedColValues(),
              part.getSkewedColValueLocationMaps(), part.isStoredAsSubDirectories(), conf);
        }
      } else {
        inputFormatClass = tblObj.getInputFormatClass();
        bucketCols = tblObj.getBucketCols();

        // input and output are the same
        oldTblPartLoc = tblObj.getPath();
        newTblPartLoc = tblObj.getPath();

        lbCtx = constructListBucketingCtx(tblObj.getSkewedColNames(), tblObj.getSkewedColValues(),
            tblObj.getSkewedColValueLocationMaps(), tblObj.isStoredAsSubDirectories(), conf);
      }

      // throw a HiveException for other than rcfile and orcfile.
      if (!((inputFormatClass.equals(RCFileInputFormat.class) ||
          (inputFormatClass.equals(OrcInputFormat.class))))) {
        throw new SemanticException(ErrorMsg.CONCATENATE_UNSUPPORTED_FILE_FORMAT.getMsg());
      }
      mergeDesc.setInputFormatClass(inputFormatClass);

      // throw a HiveException if the table/partition is bucketized
      if (bucketCols != null && bucketCols.size() > 0) {
        throw new SemanticException(ErrorMsg.CONCATENATE_UNSUPPORTED_TABLE_BUCKETED.getMsg());
      }

      // throw a HiveException if the table/partition is archived
      if (isArchived) {
        throw new SemanticException(ErrorMsg.CONCATENATE_UNSUPPORTED_PARTITION_ARCHIVED.getMsg());
      }

      // non-native and non-managed tables are not supported as MoveTask requires filenames to be in specific format,
      // violating which can cause data loss
      if (tblObj.isNonNative()) {
        throw new SemanticException(ErrorMsg.CONCATENATE_UNSUPPORTED_TABLE_NON_NATIVE.getMsg());
      }

      if (tblObj.getTableType() != TableType.MANAGED_TABLE) {
        throw new SemanticException(ErrorMsg.CONCATENATE_UNSUPPORTED_TABLE_NOT_MANAGED.getMsg());
      }

      inputDir.add(oldTblPartLoc);

      mergeDesc.setInputDir(inputDir);

      mergeDesc.setLbCtx(lbCtx);

      addInputsOutputsAlterTable(tableName, partSpec, AlterTableTypes.MERGEFILES);
      DDLWork ddlWork = new DDLWork(getInputs(), getOutputs(), mergeDesc);
      ddlWork.setNeedLock(true);
      Task mergeTask = TaskFactory.get(ddlWork);
      TableDesc tblDesc = Utilities.getTableDesc(tblObj);
      Path queryTmpdir = ctx.getExternalTmpPath(newTblPartLoc);
      mergeDesc.setOutputDir(queryTmpdir);
      // No need to handle MM tables - unsupported path.
      LoadTableDesc ltd = new LoadTableDesc(queryTmpdir, tblDesc,
          partSpec == null ? new HashMap<>() : partSpec);
      ltd.setLbCtx(lbCtx);
      Task moveTsk =
          TaskFactory.get(new MoveWork(null, null, ltd, null, false));
      mergeTask.addDependentTask(moveTsk);

      if (conf.getBoolVar(HiveConf.ConfVars.HIVESTATSAUTOGATHER)) {
        BasicStatsWork basicStatsWork;
        if (oldTblPartLoc.equals(newTblPartLoc)) {
          // If we're merging to the same location, we can avoid some metastore calls
          TableSpec tableSpec = new TableSpec(db, tableName, partSpec);
          basicStatsWork = new BasicStatsWork(tableSpec);
        } else {
          basicStatsWork = new BasicStatsWork(ltd);
        }
        basicStatsWork.setNoStatsAggregator(true);
        basicStatsWork.setClearAggregatorStats(true);
        StatsWork columnStatsWork = new StatsWork(tblObj, basicStatsWork, conf);

        Task statTask = TaskFactory.get(columnStatsWork);
        moveTsk.addDependentTask(statTask);
      }

      rootTasks.add(mergeTask);
    } catch (Exception e) {
      throw new SemanticException(e);
    }
  }

  private void analyzeAlterTableClusterSort(ASTNode ast, String tableName,
      HashMap partSpec) throws SemanticException {

    AlterTableDesc alterTblDesc;
    switch (ast.getChild(0).getType()) {
    case HiveParser.TOK_NOT_CLUSTERED:
      alterTblDesc = new AlterTableDesc(tableName, -1, new ArrayList(),
          new ArrayList(), partSpec);
      break;
    case HiveParser.TOK_NOT_SORTED:
      alterTblDesc = new AlterTableDesc(tableName, true, partSpec);
      break;
    case HiveParser.TOK_ALTERTABLE_BUCKETS:
      ASTNode buckets = (ASTNode) ast.getChild(0);
      List bucketCols = getColumnNames((ASTNode) buckets.getChild(0));
      List sortCols = new ArrayList();
      int numBuckets = -1;
      if (buckets.getChildCount() == 2) {
        numBuckets = Integer.parseInt(buckets.getChild(1).getText());
      } else {
        sortCols = getColumnNamesOrder((ASTNode) buckets.getChild(1));
        numBuckets = Integer.parseInt(buckets.getChild(2).getText());
      }
      if (numBuckets <= 0) {
        throw new SemanticException(ErrorMsg.INVALID_BUCKET_NUMBER.getMsg());
      }

      alterTblDesc = new AlterTableDesc(tableName, numBuckets,
          bucketCols, sortCols, partSpec);
      break;
    default:
      throw new SemanticException("Invalid operation " + ast.getChild(0).getType());
    }
    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), alterTblDesc)));
  }

  private void analyzeAlterTableCompact(ASTNode ast, String tableName,
      HashMap partSpec) throws SemanticException {

    String type = unescapeSQLString(ast.getChild(0).getText()).toLowerCase();

    if (!type.equals("minor") && !type.equals("major")) {
      throw new SemanticException(ErrorMsg.INVALID_COMPACTION_TYPE.getMsg());
    }

    LinkedHashMap newPartSpec = null;
    if (partSpec != null) {
      newPartSpec = new LinkedHashMap(partSpec);
    }

    HashMap mapProp = null;
    boolean isBlocking = false;

    for(int i = 0; i < ast.getChildCount(); i++) {
      switch(ast.getChild(i).getType()) {
        case HiveParser.TOK_TABLEPROPERTIES:
          mapProp = getProps((ASTNode) (ast.getChild(i)).getChild(0));
          break;
        case HiveParser.TOK_BLOCKING:
          isBlocking = true;
          break;
      }
    }
    AlterTableSimpleDesc desc = new AlterTableSimpleDesc(
      tableName, newPartSpec, type, isBlocking);
    desc.setProps(mapProp);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void analyzeAlterTableDropConstraint(ASTNode ast, String tableName)
    throws SemanticException {
    String dropConstraintName = unescapeIdentifier(ast.getChild(0).getText());
    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, dropConstraintName, (ReplicationSpec)null);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableAddConstraint(ASTNode ast, String tableName)
      throws SemanticException {
    ASTNode parent = (ASTNode) ast.getParent();
    String[] qualifiedTabName = getQualifiedTableName((ASTNode) parent.getChild(0));
    // TODO CAT - for now always use the default catalog.  Eventually will want to see if
    // the user specified a catalog
    String catName = MetaStoreUtils.getDefaultCatalog(conf);
    ASTNode child = (ASTNode) ast.getChild(0);
    List primaryKeys = new ArrayList<>();
    List foreignKeys = new ArrayList<>();
    List uniqueConstraints = new ArrayList<>();
    List checkConstraints = new ArrayList<>();

    switch (child.getToken().getType()) {
    case HiveParser.TOK_UNIQUE:
      BaseSemanticAnalyzer.processUniqueConstraints(catName, qualifiedTabName[0], qualifiedTabName[1],
          child, uniqueConstraints);
      break;
    case HiveParser.TOK_PRIMARY_KEY:
      BaseSemanticAnalyzer.processPrimaryKeys(qualifiedTabName[0], qualifiedTabName[1],
          child, primaryKeys);
      break;
    case HiveParser.TOK_FOREIGN_KEY:
      BaseSemanticAnalyzer.processForeignKeys(qualifiedTabName[0], qualifiedTabName[1],
          child, foreignKeys);
      break;
    case HiveParser.TOK_CHECK_CONSTRAINT:
      BaseSemanticAnalyzer.processCheckConstraints(catName, qualifiedTabName[0], qualifiedTabName[1],
          child, null, checkConstraints, child,
          this.ctx.getTokenRewriteStream());
      break;
    default:
      throw new SemanticException(ErrorMsg.NOT_RECOGNIZED_CONSTRAINT.getMsg(
          child.getToken().getText()));
    }
    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, primaryKeys, foreignKeys,
        uniqueConstraints, null, null, checkConstraints, null);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableUpdateColumns(ASTNode ast, String tableName,
      HashMap partSpec) throws SemanticException {

    boolean isCascade = false;
    if (null != ast.getFirstChildWithType(HiveParser.TOK_CASCADE)) {
      isCascade = true;
    }

    AlterTableDesc alterTblDesc = new AlterTableDesc(AlterTableTypes.UPDATECOLUMNS);
    alterTblDesc.setOldName(tableName);
    alterTblDesc.setIsCascade(isCascade);
    alterTblDesc.setPartSpec(partSpec);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
            alterTblDesc), conf));
  }

  static HashMap getProps(ASTNode prop) {
    // Must be deterministic order map for consistent q-test output across Java versions
    HashMap mapProp = new LinkedHashMap();
    readProps(prop, mapProp);
    return mapProp;
  }

  /**
   * Utility class to resolve QualifiedName
   */
  static class QualifiedNameUtil {

    // delimiter to check DOT delimited qualified names
    static final String delimiter = "\\.";

    /**
     * Get the fully qualified name in the ast. e.g. the ast of the form ^(DOT
     * ^(DOT a b) c) will generate a name of the form a.b.c
     *
     * @param ast
     *          The AST from which the qualified name has to be extracted
     * @return String
     */
    static public String getFullyQualifiedName(ASTNode ast) {
      if (ast.getChildCount() == 0) {
        return ast.getText();
      } else if (ast.getChildCount() == 2) {
        return getFullyQualifiedName((ASTNode) ast.getChild(0)) + "."
        + getFullyQualifiedName((ASTNode) ast.getChild(1));
      } else if (ast.getChildCount() == 3) {
        return getFullyQualifiedName((ASTNode) ast.getChild(0)) + "."
        + getFullyQualifiedName((ASTNode) ast.getChild(1)) + "."
        + getFullyQualifiedName((ASTNode) ast.getChild(2));
      } else {
        return null;
      }
    }

    // get the column path
    // return column name if exists, column could be DOT separated.
    // example: lintString.$elem$.myint
    // return table name for column name if no column has been specified.
    static public String getColPath(
      Hive db,
      ASTNode node,
      String dbName,
      String tableName,
      Map partSpec) throws SemanticException {

      // if this ast has only one child, then no column name specified.
      if (node.getChildCount() == 1) {
        return tableName;
      }

      ASTNode columnNode = null;
      // Second child node could be partitionspec or column
      if (node.getChildCount() > 1) {
        if (partSpec == null) {
          columnNode = (ASTNode) node.getChild(1);
        } else {
          columnNode = (ASTNode) node.getChild(2);
        }
      }

      if (columnNode != null) {
        if (dbName == null) {
          return tableName + "." + QualifiedNameUtil.getFullyQualifiedName(columnNode);
        } else {
          return tableName.substring(dbName.length() + 1, tableName.length()) + "." +
              QualifiedNameUtil.getFullyQualifiedName(columnNode);
        }
      } else {
        return tableName;
      }
    }

    // get partition metadata
    static public Map getPartitionSpec(Hive db, ASTNode ast, String tableName)
      throws SemanticException {
      ASTNode partNode = null;
      // if this ast has only one child, then no partition spec specified.
      if (ast.getChildCount() == 1) {
        return null;
      }

      // if ast has two children
      // the 2nd child could be partition spec or columnName
      // if the ast has 3 children, the second *has to* be partition spec
      if (ast.getChildCount() > 2 && (((ASTNode) ast.getChild(1)).getType() != HiveParser.TOK_PARTSPEC)) {
        throw new SemanticException(((ASTNode) ast.getChild(1)).getType() + " is not a partition specification");
      }

      if (((ASTNode) ast.getChild(1)).getType() == HiveParser.TOK_PARTSPEC) {
        partNode = (ASTNode) ast.getChild(1);
      }

      if (partNode != null) {
        Table tab = null;
        try {
          tab = db.getTable(tableName);
        }
        catch (InvalidTableException e) {
          throw new SemanticException(ErrorMsg.INVALID_TABLE.getMsg(tableName), e);
        }
        catch (HiveException e) {
          throw new SemanticException(e.getMessage(), e);
        }

        HashMap partSpec = null;
        try {
          partSpec = getValidatedPartSpec(tab, partNode, db.getConf(), false);
        } catch (SemanticException e) {
          // get exception in resolving partition
          // it could be DESCRIBE table key
          // return null
          // continue processing for DESCRIBE table key
          return null;
        }

        if (partSpec != null) {
          Partition part = null;
          try {
            part = db.getPartition(tab, partSpec, false);
          } catch (HiveException e) {
            // if get exception in finding partition
            // it could be DESCRIBE table key
            // return null
            // continue processing for DESCRIBE table key
            return null;
          }

          // if partition is not found
          // it is DESCRIBE table partition
          // invalid partition exception
          if (part == null) {
            throw new SemanticException(ErrorMsg.INVALID_PARTITION.getMsg(partSpec.toString()));
          }

          // it is DESCRIBE table partition
          // return partition metadata
          return partSpec;
        }
      }

      return null;
    }

  }

  private void validateDatabase(String databaseName) throws SemanticException {
    try {
      if (!db.databaseExists(databaseName)) {
        throw new SemanticException(ErrorMsg.DATABASE_NOT_EXISTS.getMsg(databaseName));
      }
    } catch (HiveException e) {
      throw new SemanticException(ErrorMsg.DATABASE_NOT_EXISTS.getMsg(databaseName), e);
    }
  }

  private void validateTable(String tableName, Map partSpec)
      throws SemanticException {
    Table tab = getTable(tableName);
    if (partSpec != null) {
      getPartition(tab, partSpec, true);
    }
  }

  /**
   * A query like this will generate a tree as follows
   *   "describe formatted default.maptable partition (b=100) id;"
   * TOK_TABTYPE
   *   TOK_TABNAME --> root for tablename, 2 child nodes mean DB specified
   *     default
   *     maptable
   *   TOK_PARTSPEC  --> root node for partition spec. else columnName
   *     TOK_PARTVAL
   *       b
   *       100
   *   id           --> root node for columnName
   * formatted
   */
  private void analyzeDescribeTable(ASTNode ast) throws SemanticException {
    ASTNode tableTypeExpr = (ASTNode) ast.getChild(0);

    String dbName    = null;
    String tableName = null;
    String colPath   = null;
    Map partSpec = null;

    ASTNode tableNode = null;

    // process the first node to extract tablename
    // tablename is either TABLENAME or DBNAME.TABLENAME if db is given
    if (((ASTNode) tableTypeExpr.getChild(0)).getType() == HiveParser.TOK_TABNAME) {
      tableNode = (ASTNode) tableTypeExpr.getChild(0);
      if (tableNode.getChildCount() == 1) {
        tableName = ((ASTNode) tableNode.getChild(0)).getText();
      } else {
        dbName    = ((ASTNode) tableNode.getChild(0)).getText();
        tableName = dbName + "." + ((ASTNode) tableNode.getChild(1)).getText();
      }
    } else {
      throw new SemanticException(((ASTNode) tableTypeExpr.getChild(0)).getText() + " is not an expected token type");
    }

    // process the second child,if exists, node to get partition spec(s)
    partSpec = QualifiedNameUtil.getPartitionSpec(db, tableTypeExpr, tableName);

    // process the third child node,if exists, to get partition spec(s)
    colPath  = QualifiedNameUtil.getColPath(db, tableTypeExpr, dbName, tableName, partSpec);

    // if database is not the one currently using
    // validate database
    if (dbName != null) {
      validateDatabase(dbName);
    }
    if (partSpec != null) {
      validateTable(tableName, partSpec);
    }

    DescTableDesc descTblDesc = new DescTableDesc(
      ctx.getResFile(), tableName, partSpec, colPath);

    boolean showColStats = false;
    if (ast.getChildCount() == 2) {
      int descOptions = ast.getChild(1).getType();
      descTblDesc.setFormatted(descOptions == HiveParser.KW_FORMATTED);
      descTblDesc.setExt(descOptions == HiveParser.KW_EXTENDED);
      // in case of "DESCRIBE FORMATTED tablename column_name" statement, colPath
      // will contain tablename.column_name. If column_name is not specified
      // colPath will be equal to tableName. This is how we can differentiate
      // if we are describing a table or column
      if (!colPath.equalsIgnoreCase(tableName) && descTblDesc.isFormatted()) {
        showColStats = true;
      }
    }

    inputs.add(new ReadEntity(getTable(tableName)));
    Task ddlTask = TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        descTblDesc));
    rootTasks.add(ddlTask);
    String schema = DescTableDesc.getSchema(showColStats);
    setFetchTask(createFetchTask(schema));
    LOG.info("analyzeDescribeTable done");
  }

  /**
   * Describe database.
   *
   * @param ast
   * @throws SemanticException
   */
  private void analyzeDescDatabase(ASTNode ast) throws SemanticException {

    boolean isExtended;
    String dbName;

    if (ast.getChildCount() == 1) {
      dbName = stripQuotes(ast.getChild(0).getText());
      isExtended = false;
    } else if (ast.getChildCount() == 2) {
      dbName = stripQuotes(ast.getChild(0).getText());
      isExtended = true;
    } else {
      throw new SemanticException("Unexpected Tokens at DESCRIBE DATABASE");
    }

    DescDatabaseDesc descDbDesc = new DescDatabaseDesc(ctx.getResFile(),
        dbName, isExtended);
    inputs.add(new ReadEntity(getDatabase(dbName)));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), descDbDesc)));
    setFetchTask(createFetchTask(descDbDesc.getSchema()));
  }

  public static HashMap getPartSpec(ASTNode partspec)
      throws SemanticException {
    if (partspec == null) {
      return null;
    }
    HashMap partSpec = new LinkedHashMap();
    for (int i = 0; i < partspec.getChildCount(); ++i) {
      ASTNode partspec_val = (ASTNode) partspec.getChild(i);
      String key = partspec_val.getChild(0).getText();
      String val = null;
      if (partspec_val.getChildCount() > 1) {
        val = stripQuotes(partspec_val.getChild(1).getText());
      }
      partSpec.put(key.toLowerCase(), val);
    }
    return partSpec;
  }

  public static HashMap getValidatedPartSpec(Table table, ASTNode astNode,
      HiveConf conf, boolean shouldBeFull) throws SemanticException {
    HashMap partSpec = getPartSpec(astNode);
    if (partSpec != null && !partSpec.isEmpty()) {
      validatePartSpec(table, partSpec, astNode, conf, shouldBeFull);
    }
    return partSpec;
  }

  private void analyzeShowPartitions(ASTNode ast) throws SemanticException {
    ShowPartitionsDesc showPartsDesc;
    String tableName = getUnescapedName((ASTNode) ast.getChild(0));
    List> partSpecs = getPartitionSpecs(getTable(tableName), ast);
    // We only can have a single partition spec
    assert (partSpecs.size() <= 1);
    Map partSpec = null;
    if (partSpecs.size() > 0) {
      partSpec = partSpecs.get(0);
    }

    validateTable(tableName, null);

    showPartsDesc = new ShowPartitionsDesc(tableName, ctx.getResFile(), partSpec);
    inputs.add(new ReadEntity(getTable(tableName)));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showPartsDesc)));
    setFetchTask(createFetchTask(showPartsDesc.getSchema()));
  }

  private void analyzeShowCreateDatabase(ASTNode ast) throws SemanticException {
    String dbName = getUnescapedName((ASTNode)ast.getChild(0));
    ShowCreateDatabaseDesc showCreateDbDesc =
        new ShowCreateDatabaseDesc(dbName, ctx.getResFile().toString());

    Database database = getDatabase(dbName);
    inputs.add(new ReadEntity(database));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showCreateDbDesc)));
    setFetchTask(createFetchTask(showCreateDbDesc.getSchema()));
  }


  private void analyzeShowCreateTable(ASTNode ast) throws SemanticException {
    ShowCreateTableDesc showCreateTblDesc;
    String tableName = getUnescapedName((ASTNode)ast.getChild(0));
    showCreateTblDesc = new ShowCreateTableDesc(tableName, ctx.getResFile().toString());

    Table tab = getTable(tableName);
    inputs.add(new ReadEntity(tab));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showCreateTblDesc)));
    setFetchTask(createFetchTask(showCreateTblDesc.getSchema()));
  }

  private void analyzeShowDatabases(ASTNode ast) throws SemanticException {
    ShowDatabasesDesc showDatabasesDesc;
    if (ast.getChildCount() == 1) {
      String databasePattern = unescapeSQLString(ast.getChild(0).getText());
      showDatabasesDesc = new ShowDatabasesDesc(ctx.getResFile(), databasePattern);
    } else {
      showDatabasesDesc = new ShowDatabasesDesc(ctx.getResFile());
    }
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), showDatabasesDesc)));
    setFetchTask(createFetchTask(showDatabasesDesc.getSchema()));
  }

  private void analyzeShowTables(ASTNode ast) throws SemanticException {
    ShowTablesDesc showTblsDesc;
    String dbName = SessionState.get().getCurrentDatabase();
    String tableNames = null;

    if (ast.getChildCount() > 3) {
      throw new SemanticException(ErrorMsg.INVALID_AST_TREE.getMsg(ast.toStringTree()));
    }

    switch (ast.getChildCount()) {
    case 1: // Uses a pattern
      tableNames = unescapeSQLString(ast.getChild(0).getText());
      showTblsDesc = new ShowTablesDesc(ctx.getResFile(), dbName, tableNames);
      break;
    case 2: // Specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      validateDatabase(dbName);
      showTblsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      break;
    case 3: // Uses a pattern and specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      tableNames = unescapeSQLString(ast.getChild(2).getText());
      validateDatabase(dbName);
      showTblsDesc = new ShowTablesDesc(ctx.getResFile(), dbName, tableNames);
      break;
    default: // No pattern or DB
      showTblsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      break;
    }
    inputs.add(new ReadEntity(getDatabase(dbName)));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showTblsDesc)));
    setFetchTask(createFetchTask(showTblsDesc.getSchema()));
  }

  private void analyzeShowColumns(ASTNode ast) throws SemanticException {

  // table name has to be present so min child 1 and max child 4
    if (ast.getChildCount() > 4 || ast.getChildCount()<1) {
      throw new SemanticException(ErrorMsg.INVALID_AST_TREE.getMsg(ast.toStringTree()));
    }

    String tableName = getUnescapedName((ASTNode) ast.getChild(0));

    ShowColumnsDesc showColumnsDesc = null;
    String pattern = null;
    switch(ast.getChildCount()) {
      case 1: //  only tablename no pattern and db
        showColumnsDesc = new ShowColumnsDesc(ctx.getResFile(), tableName);
        break;
      case 2: // tablename and pattern
        pattern = unescapeSQLString(ast.getChild(1).getText());
        showColumnsDesc = new ShowColumnsDesc(ctx.getResFile(), tableName, pattern);
        break;
      case 3: // specifies db
        if (tableName.contains(".")) {
          throw new SemanticException("Duplicates declaration for database name");
        }
        tableName = getUnescapedName((ASTNode) ast.getChild(2)) + "." + tableName;
        showColumnsDesc = new ShowColumnsDesc(ctx.getResFile(), tableName);
        break;
      case 4: // specifies db and pattern
        if (tableName.contains(".")) {
          throw new SemanticException("Duplicates declaration for database name");
        }
        tableName = getUnescapedName((ASTNode) ast.getChild(2)) + "." + tableName;
        pattern = unescapeSQLString(ast.getChild(3).getText());
        showColumnsDesc = new ShowColumnsDesc(ctx.getResFile(), tableName, pattern);
        break;
      default:
        break;
    }

    Table tab = getTable(tableName);
    inputs.add(new ReadEntity(tab));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showColumnsDesc)));
    setFetchTask(createFetchTask(showColumnsDesc.getSchema()));
  }

  private void analyzeShowTableStatus(ASTNode ast) throws SemanticException {
    ShowTableStatusDesc showTblStatusDesc;
    String tableNames = getUnescapedName((ASTNode) ast.getChild(0));
    String dbName = SessionState.get().getCurrentDatabase();
    int children = ast.getChildCount();
    HashMap partSpec = null;
    if (children >= 2) {
      if (children > 3) {
        throw new SemanticException(ErrorMsg.INVALID_AST_TREE.getMsg());
      }
      for (int i = 1; i < children; i++) {
        ASTNode child = (ASTNode) ast.getChild(i);
        if (child.getToken().getType() == HiveParser.Identifier) {
          dbName = unescapeIdentifier(child.getText());
        } else if (child.getToken().getType() == HiveParser.TOK_PARTSPEC) {
          partSpec = getValidatedPartSpec(getTable(tableNames), child, conf, false);
        } else {
          throw new SemanticException(ErrorMsg.INVALID_AST_TREE.getMsg(child.toStringTree() +
            " , Invalid token " + child.getToken().getType()));
        }
      }
    }

    if (partSpec != null) {
      validateTable(tableNames, partSpec);
    }

    showTblStatusDesc = new ShowTableStatusDesc(ctx.getResFile().toString(), dbName,
        tableNames, partSpec);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showTblStatusDesc)));
    setFetchTask(createFetchTask(showTblStatusDesc.getSchema()));
  }

  private void analyzeShowTableProperties(ASTNode ast) throws SemanticException {
    ShowTblPropertiesDesc showTblPropertiesDesc;
    String[] qualified = getQualifiedTableName((ASTNode) ast.getChild(0));
    String propertyName = null;
    if (ast.getChildCount() > 1) {
      propertyName = unescapeSQLString(ast.getChild(1).getText());
    }

    String tableNames = getDotName(qualified);
    validateTable(tableNames, null);

    showTblPropertiesDesc = new ShowTblPropertiesDesc(ctx.getResFile().toString(), tableNames,
        propertyName);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showTblPropertiesDesc)));
    setFetchTask(createFetchTask(showTblPropertiesDesc.getSchema()));
  }

  /**
   * Add the task according to the parsed command tree. This is used for the CLI
   * command "SHOW FUNCTIONS;".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsin failed
   */
  private void analyzeShowFunctions(ASTNode ast) throws SemanticException {
    ShowFunctionsDesc showFuncsDesc;
    if (ast.getChildCount() == 1) {
      String funcNames = stripQuotes(ast.getChild(0).getText());
      showFuncsDesc = new ShowFunctionsDesc(ctx.getResFile(), funcNames);
    } else if (ast.getChildCount() == 2) {
      assert (ast.getChild(0).getType() == HiveParser.KW_LIKE);
      String funcNames = stripQuotes(ast.getChild(1).getText());
      showFuncsDesc = new ShowFunctionsDesc(ctx.getResFile(), funcNames, true);
    } else {
      showFuncsDesc = new ShowFunctionsDesc(ctx.getResFile());
    }
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showFuncsDesc)));
    setFetchTask(createFetchTask(showFuncsDesc.getSchema()));
  }

  /**
   * Add the task according to the parsed command tree. This is used for the CLI
   * command "SHOW LOCKS;".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeShowLocks(ASTNode ast) throws SemanticException {
    String tableName = null;
    HashMap partSpec = null;
    boolean isExtended = false;

    if (ast.getChildCount() >= 1) {
      // table for which show locks is being executed
      for (int i = 0; i < ast.getChildCount(); i++) {
        ASTNode child = (ASTNode) ast.getChild(i);
        if (child.getType() == HiveParser.TOK_TABTYPE) {
          ASTNode tableTypeExpr = child;
          tableName =
            QualifiedNameUtil.getFullyQualifiedName((ASTNode) tableTypeExpr.getChild(0));
          // get partition metadata if partition specified
          if (tableTypeExpr.getChildCount() == 2) {
            ASTNode partSpecNode = (ASTNode) tableTypeExpr.getChild(1);
            partSpec = getValidatedPartSpec(getTable(tableName), partSpecNode, conf, false);
          }
        } else if (child.getType() == HiveParser.KW_EXTENDED) {
          isExtended = true;
        }
      }
    }

    HiveTxnManager txnManager = null;
    try {
      txnManager = TxnManagerFactory.getTxnManagerFactory().getTxnManager(conf);
    } catch (LockException e) {
      throw new SemanticException(e.getMessage());
    }

    ShowLocksDesc showLocksDesc = new ShowLocksDesc(ctx.getResFile(), tableName,
        partSpec, isExtended, txnManager.useNewShowLocksFormat());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showLocksDesc)));
    setFetchTask(createFetchTask(showLocksDesc.getSchema()));

    // Need to initialize the lock manager
    ctx.setNeedLockMgr(true);
  }

   /**
    * Add the task according to the parsed command tree. This is used for the CLI
   * command "SHOW LOCKS DATABASE database [extended];".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeShowDbLocks(ASTNode ast) throws SemanticException {
    boolean isExtended = (ast.getChildCount() > 1);
    String dbName = stripQuotes(ast.getChild(0).getText());

    HiveTxnManager txnManager = null;
    try {
      txnManager = TxnManagerFactory.getTxnManagerFactory().getTxnManager(conf);
    } catch (LockException e) {
      throw new SemanticException(e.getMessage());
    }

    ShowLocksDesc showLocksDesc = new ShowLocksDesc(ctx.getResFile(), dbName,
                                                    isExtended, txnManager.useNewShowLocksFormat());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showLocksDesc)));
    setFetchTask(createFetchTask(showLocksDesc.getSchema()));

    // Need to initialize the lock manager
    ctx.setNeedLockMgr(true);
  }

  private void analyzeShowConf(ASTNode ast) throws SemanticException {
    String confName = stripQuotes(ast.getChild(0).getText());
    ShowConfDesc showConfDesc = new ShowConfDesc(ctx.getResFile(), confName);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showConfDesc)));
    setFetchTask(createFetchTask(showConfDesc.getSchema()));
  }

  private void analyzeShowViews(ASTNode ast) throws SemanticException {
    ShowTablesDesc showViewsDesc;
    String dbName = SessionState.get().getCurrentDatabase();
    String viewNames = null;

    if (ast.getChildCount() > 3) {
      throw new SemanticException(ErrorMsg.GENERIC_ERROR.getMsg());
    }

    switch (ast.getChildCount()) {
    case 1: // Uses a pattern
      viewNames = unescapeSQLString(ast.getChild(0).getText());
      showViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName, viewNames, TableType.VIRTUAL_VIEW);
      break;
    case 2: // Specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      validateDatabase(dbName);
      showViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      showViewsDesc.setType(TableType.VIRTUAL_VIEW);
      break;
    case 3: // Uses a pattern and specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      viewNames = unescapeSQLString(ast.getChild(2).getText());
      validateDatabase(dbName);
      showViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName, viewNames, TableType.VIRTUAL_VIEW);
      break;
    default: // No pattern or DB
      showViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      showViewsDesc.setType(TableType.VIRTUAL_VIEW);
      break;
    }

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showViewsDesc)));
    setFetchTask(createFetchTask(showViewsDesc.getSchema()));
  }

  private void analyzeShowMaterializedViews(ASTNode ast) throws SemanticException {
    ShowTablesDesc showMaterializedViewsDesc;
    String dbName = SessionState.get().getCurrentDatabase();
    String materializedViewNames = null;

    if (ast.getChildCount() > 3) {
      throw new SemanticException(ErrorMsg.GENERIC_ERROR.getMsg());
    }

    switch (ast.getChildCount()) {
    case 1: // Uses a pattern
      materializedViewNames = unescapeSQLString(ast.getChild(0).getText());
      showMaterializedViewsDesc = new ShowTablesDesc(
          ctx.getResFile(), dbName, materializedViewNames, TableType.MATERIALIZED_VIEW);
      break;
    case 2: // Specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      validateDatabase(dbName);
      showMaterializedViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      showMaterializedViewsDesc.setType(TableType.MATERIALIZED_VIEW);
      break;
    case 3: // Uses a pattern and specifies a DB
      assert (ast.getChild(0).getType() == HiveParser.TOK_FROM);
      dbName = unescapeIdentifier(ast.getChild(1).getText());
      materializedViewNames = unescapeSQLString(ast.getChild(2).getText());
      validateDatabase(dbName);
      showMaterializedViewsDesc = new ShowTablesDesc(
          ctx.getResFile(), dbName, materializedViewNames, TableType.MATERIALIZED_VIEW);
      break;
    default: // No pattern or DB
      showMaterializedViewsDesc = new ShowTablesDesc(ctx.getResFile(), dbName);
      showMaterializedViewsDesc.setType(TableType.MATERIALIZED_VIEW);
      break;
    }

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        showMaterializedViewsDesc)));
    setFetchTask(createFetchTask(showMaterializedViewsDesc.getSchema()));
  }

  /**
   * Add the task according to the parsed command tree. This is used for the CLI
   * command "LOCK TABLE ..;".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeLockTable(ASTNode ast)
      throws SemanticException {
    String tableName = getUnescapedName((ASTNode) ast.getChild(0)).toLowerCase();
    String mode = unescapeIdentifier(ast.getChild(1).getText().toUpperCase());
    List> partSpecs = getPartitionSpecs(getTable(tableName), ast);

    // We only can have a single partition spec
    assert (partSpecs.size() <= 1);
    Map partSpec = null;
    if (partSpecs.size() > 0) {
      partSpec = partSpecs.get(0);
    }

    LockTableDesc lockTblDesc = new LockTableDesc(tableName, mode, partSpec,
        HiveConf.getVar(conf, ConfVars.HIVEQUERYID));
    lockTblDesc.setQueryStr(this.ctx.getCmd());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        lockTblDesc)));

    // Need to initialize the lock manager
    ctx.setNeedLockMgr(true);
  }

  /**
   * Add a task to execute "SHOW COMPACTIONS"
   * @param ast The parsed command tree.
   * @throws SemanticException Parsing failed.
   */
  private void analyzeShowCompactions(ASTNode ast) throws SemanticException {
    ShowCompactionsDesc desc = new ShowCompactionsDesc(ctx.getResFile());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
    setFetchTask(createFetchTask(desc.getSchema()));
  }

  /**
   * Add a task to execute "SHOW COMPACTIONS"
   * @param ast The parsed command tree.
   * @throws SemanticException Parsing failed.
   */
  private void analyzeShowTxns(ASTNode ast) throws SemanticException {
    ShowTxnsDesc desc = new ShowTxnsDesc(ctx.getResFile());
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
    setFetchTask(createFetchTask(desc.getSchema()));
  }

  /**
   * Add a task to execute "ABORT TRANSACTIONS"
   * @param ast The parsed command tree
   * @throws SemanticException Parsing failed
   */
  private void analyzeAbortTxns(ASTNode ast) throws SemanticException {
    List txnids = new ArrayList();
    int numChildren = ast.getChildCount();
    for (int i = 0; i < numChildren; i++) {
      txnids.add(Long.parseLong(ast.getChild(i).getText()));
    }
    AbortTxnsDesc desc = new AbortTxnsDesc(txnids);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

   /**
   * Add a task to execute "Kill query"
   * @param ast The parsed command tree
   * @throws SemanticException Parsing failed
   */
  private void analyzeKillQuery(ASTNode ast) throws SemanticException {
    List queryIds = new ArrayList();
    int numChildren = ast.getChildCount();
    for (int i = 0; i < numChildren; i++) {
      queryIds.add(stripQuotes(ast.getChild(i).getText()));
    }
    addServiceOutput();
    KillQueryDesc desc = new KillQueryDesc(queryIds);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), desc)));
  }

  private void addServiceOutput() throws SemanticException {
    String hs2Hostname = getHS2Host();
    if (hs2Hostname != null) {
      outputs.add(new WriteEntity(hs2Hostname, Type.SERVICE_NAME));
    }
  }

  private String getHS2Host() throws SemanticException {
    if (SessionState.get().isHiveServerQuery()) {
      return SessionState.get().getHiveServer2Host();
    }
    if (conf.getBoolVar(ConfVars.HIVE_TEST_AUTHORIZATION_SQLSTD_HS2_MODE)) {
      // dummy value for use in tests
      return "dummyHostnameForTest";
    }
    throw new SemanticException("Kill query is only supported in HiveServer2 (not hive cli)");
  }

  /**
   * Add the task according to the parsed command tree. This is used for the CLI
   * command "UNLOCK TABLE ..;".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeUnlockTable(ASTNode ast)
      throws SemanticException {
    String tableName = getUnescapedName((ASTNode) ast.getChild(0));
    List> partSpecs = getPartitionSpecs(getTable(tableName), ast);

    // We only can have a single partition spec
    assert (partSpecs.size() <= 1);
    Map partSpec = null;
    if (partSpecs.size() > 0) {
      partSpec = partSpecs.get(0);
    }

    UnlockTableDesc unlockTblDesc = new UnlockTableDesc(tableName, partSpec);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        unlockTblDesc)));

    // Need to initialize the lock manager
    ctx.setNeedLockMgr(true);
  }

  private void analyzeLockDatabase(ASTNode ast) throws SemanticException {
    String dbName = unescapeIdentifier(ast.getChild(0).getText());
    String mode  = unescapeIdentifier(ast.getChild(1).getText().toUpperCase());

    inputs.add(new ReadEntity(getDatabase(dbName)));
    // Lock database operation is to acquire the lock explicitly, the operation
    // itself doesn't need to be locked. Set the WriteEntity as WriteType:
    // DDL_NO_LOCK here, otherwise it will conflict with Hive's transaction.
    outputs.add(new WriteEntity(getDatabase(dbName), WriteType.DDL_NO_LOCK));

    LockDatabaseDesc lockDatabaseDesc = new LockDatabaseDesc(dbName, mode,
                        HiveConf.getVar(conf, ConfVars.HIVEQUERYID));
    lockDatabaseDesc.setQueryStr(ctx.getCmd());
    DDLWork work = new DDLWork(getInputs(), getOutputs(), lockDatabaseDesc);
    rootTasks.add(TaskFactory.get(work));
    ctx.setNeedLockMgr(true);
  }

  private void analyzeUnlockDatabase(ASTNode ast) throws SemanticException {
    String dbName = unescapeIdentifier(ast.getChild(0).getText());

    inputs.add(new ReadEntity(getDatabase(dbName)));
    // Unlock database operation is to release the lock explicitly, the
    // operation itself don't need to be locked. Set the WriteEntity as
    // WriteType: DDL_NO_LOCK here, otherwise it will conflict with
    // Hive's transaction.
    outputs.add(new WriteEntity(getDatabase(dbName), WriteType.DDL_NO_LOCK));

    UnlockDatabaseDesc unlockDatabaseDesc = new UnlockDatabaseDesc(dbName);
    DDLWork work = new DDLWork(getInputs(), getOutputs(), unlockDatabaseDesc);
    rootTasks.add(TaskFactory.get(work));
    // Need to initialize the lock manager
    ctx.setNeedLockMgr(true);
  }

  /**
   * Add the task according to the parsed command tree. This is used for the CLI
   * command "DESCRIBE FUNCTION;".
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeDescFunction(ASTNode ast) throws SemanticException {
    String funcName;
    boolean isExtended;

    if (ast.getChildCount() == 1) {
      funcName = stripQuotes(ast.getChild(0).getText());
      isExtended = false;
    } else if (ast.getChildCount() == 2) {
      funcName = stripQuotes(ast.getChild(0).getText());
      isExtended = true;
    } else {
      throw new SemanticException("Unexpected Tokens at DESCRIBE FUNCTION");
    }

    DescFunctionDesc descFuncDesc = new DescFunctionDesc(ctx.getResFile(),
        funcName, isExtended);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        descFuncDesc)));
    setFetchTask(createFetchTask(descFuncDesc.getSchema()));
  }


  private void analyzeAlterTableRename(String[] source, ASTNode ast, boolean expectView)
      throws SemanticException {
    String[] target = getQualifiedTableName((ASTNode) ast.getChild(0));

    String sourceName = getDotName(source);
    String targetName = getDotName(target);

    AlterTableDesc alterTblDesc = new AlterTableDesc(sourceName, targetName, expectView, null);
    addInputsOutputsAlterTable(sourceName, null, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableRenameCol(String catName, String[] qualified, ASTNode ast,
      HashMap partSpec) throws SemanticException {
    String newComment = null;
    boolean first = false;
    String flagCol = null;
    boolean isCascade = false;
    //col_old_name col_new_name column_type [COMMENT col_comment] [FIRST|AFTER column_name] [CASCADE|RESTRICT]
    String oldColName = ast.getChild(0).getText();
    String newColName = ast.getChild(1).getText();
    String newType = getTypeStringFromAST((ASTNode) ast.getChild(2));
    ASTNode constraintChild = null;
    int childCount = ast.getChildCount();
    for (int i = 3; i < childCount; i++) {
      ASTNode child = (ASTNode)ast.getChild(i);
      switch (child.getToken().getType()) {
        case HiveParser.StringLiteral:
          newComment = unescapeSQLString(child.getText());
          break;
        case HiveParser.TOK_ALTERTABLE_CHANGECOL_AFTER_POSITION:
          flagCol = unescapeIdentifier(child.getChild(0).getText());
          break;
        case HiveParser.KW_FIRST:
          first = true;
          break;
        case HiveParser.TOK_CASCADE:
          isCascade = true;
          break;
        case HiveParser.TOK_RESTRICT:
          break;
        default:
          constraintChild = (ASTNode) child;
      }
    }
    List primaryKeys = null;
    List foreignKeys = null;
    List uniqueConstraints = null;
    List notNullConstraints = null;
    List defaultConstraints= null;
    List checkConstraints= null;
    if (constraintChild != null) {
      // Process column constraint
      switch (constraintChild.getToken().getType()) {
      case HiveParser.TOK_CHECK_CONSTRAINT:
        checkConstraints = new ArrayList<>();
        processCheckConstraints(catName, qualified[0], qualified[1], constraintChild,
                                  ImmutableList.of(newColName), checkConstraints, (ASTNode)ast.getChild(2),
                                this.ctx.getTokenRewriteStream());
        break;
      case HiveParser.TOK_DEFAULT_VALUE:
        defaultConstraints = new ArrayList<>();
        processDefaultConstraints(catName, qualified[0], qualified[1], constraintChild,
                                  ImmutableList.of(newColName), defaultConstraints, (ASTNode)ast.getChild(2),
                                  this.ctx.getTokenRewriteStream());
        break;
      case HiveParser.TOK_NOT_NULL:
        notNullConstraints = new ArrayList<>();
        processNotNullConstraints(catName, qualified[0], qualified[1], constraintChild,
                                  ImmutableList.of(newColName), notNullConstraints);
        break;
      case HiveParser.TOK_UNIQUE:
        uniqueConstraints = new ArrayList<>();
        processUniqueConstraints(catName, qualified[0], qualified[1], constraintChild,
                                 ImmutableList.of(newColName), uniqueConstraints);
        break;
      case HiveParser.TOK_PRIMARY_KEY:
        primaryKeys = new ArrayList<>();
        processPrimaryKeys(qualified[0], qualified[1], constraintChild,
                           ImmutableList.of(newColName), primaryKeys);
        break;
      case HiveParser.TOK_FOREIGN_KEY:
        foreignKeys = new ArrayList<>();
        processForeignKeys(qualified[0], qualified[1], constraintChild,
                           foreignKeys);
        break;
      default:
        throw new SemanticException(ErrorMsg.NOT_RECOGNIZED_CONSTRAINT.getMsg(
            constraintChild.getToken().getText()));
      }
    }

    /* Validate the operation of renaming a column name. */
    Table tab = getTable(qualified);

    if(checkConstraints != null && !checkConstraints.isEmpty()) {
      validateCheckConstraint(tab.getCols(), checkConstraints, ctx.getConf());
    }

    if(tab.getTableType() == TableType.EXTERNAL_TABLE
        && hasEnabledOrValidatedConstraints(notNullConstraints, defaultConstraints, checkConstraints)){
      throw new SemanticException(
          ErrorMsg.INVALID_CSTR_SYNTAX.getMsg("Constraints are disallowed with External tables. "
              + "Only RELY is allowed."));
    }

    SkewedInfo skewInfo = tab.getTTable().getSd().getSkewedInfo();
    if ((null != skewInfo)
        && (null != skewInfo.getSkewedColNames())
        && skewInfo.getSkewedColNames().contains(oldColName)) {
      throw new SemanticException(oldColName
          + ErrorMsg.ALTER_TABLE_NOT_ALLOWED_RENAME_SKEWED_COLUMN.getMsg());
    }

    String tblName = getDotName(qualified);
    AlterTableDesc alterTblDesc;
    if (primaryKeys == null && foreignKeys == null
            && uniqueConstraints == null && notNullConstraints == null && defaultConstraints == null
        && checkConstraints == null) {
      alterTblDesc = new AlterTableDesc(tblName, partSpec,
          unescapeIdentifier(oldColName), unescapeIdentifier(newColName),
          newType, newComment, first, flagCol, isCascade);
    } else {
      alterTblDesc = new AlterTableDesc(tblName, partSpec,
          unescapeIdentifier(oldColName), unescapeIdentifier(newColName),
          newType, newComment, first, flagCol, isCascade,
          primaryKeys, foreignKeys, uniqueConstraints, notNullConstraints, defaultConstraints, checkConstraints);
    }
    addInputsOutputsAlterTable(tblName, partSpec, alterTblDesc);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableRenamePart(ASTNode ast, String tblName,
      HashMap oldPartSpec) throws SemanticException {
    Table tab = getTable(tblName, true);
    validateAlterTableType(tab, AlterTableTypes.RENAMEPARTITION);
    Map newPartSpec =
        getValidatedPartSpec(tab, (ASTNode)ast.getChild(0), conf, false);
    if (newPartSpec == null) {
      throw new SemanticException("RENAME PARTITION Missing Destination" + ast);
    }
    ReadEntity re = new ReadEntity(tab);
    re.noLockNeeded();
    inputs.add(re);

    List> partSpecs = new ArrayList>();
    partSpecs.add(oldPartSpec);
    partSpecs.add(newPartSpec);
    addTablePartsOutputs(tab, partSpecs, WriteEntity.WriteType.DDL_EXCLUSIVE);
    RenamePartitionDesc renamePartitionDesc = new RenamePartitionDesc(tblName, oldPartSpec, newPartSpec, null);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        renamePartitionDesc)));
  }

  private void analyzeAlterTableBucketNum(ASTNode ast, String tblName,
      HashMap partSpec) throws SemanticException {
    Table tab = getTable(tblName, true);
    if (tab.getBucketCols() == null || tab.getBucketCols().isEmpty()) {
      throw new SemanticException(ErrorMsg.ALTER_BUCKETNUM_NONBUCKETIZED_TBL.getMsg());
    }
    validateAlterTableType(tab, AlterTableTypes.ALTERBUCKETNUM);
    inputs.add(new ReadEntity(tab));

    int bucketNum = Integer.parseInt(ast.getChild(0).getText());
    AlterTableDesc alterBucketNum = new AlterTableDesc(tblName, partSpec, bucketNum);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterBucketNum)));
  }

  private void analyzeAlterTableModifyCols(String[] qualified, ASTNode ast,
      HashMap partSpec, AlterTableTypes alterType) throws SemanticException {

    String tblName = getDotName(qualified);
    List newCols = getColumns((ASTNode) ast.getChild(0));
    boolean isCascade = false;
    if (null != ast.getFirstChildWithType(HiveParser.TOK_CASCADE)) {
      isCascade = true;
    }

    AlterTableDesc alterTblDesc = new AlterTableDesc(tblName, partSpec, newCols,
        alterType, isCascade);

    addInputsOutputsAlterTable(tblName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void analyzeAlterTableDropParts(String[] qualified, ASTNode ast, boolean expectView)
      throws SemanticException {

    boolean ifExists = (ast.getFirstChildWithType(HiveParser.TOK_IFEXISTS) != null)
        || HiveConf.getBoolVar(conf, ConfVars.DROPIGNORESNONEXISTENT);
    // If the drop has to fail on non-existent partitions, we cannot batch expressions.
    // That is because we actually have to check each separate expression for existence.
    // We could do a small optimization for the case where expr has all columns and all
    // operators are equality, if we assume those would always match one partition (which
    // may not be true with legacy, non-normalized column values). This is probably a
    // popular case but that's kinda hacky. Let's not do it for now.
    boolean canGroupExprs = ifExists;

    boolean mustPurge = (ast.getFirstChildWithType(HiveParser.KW_PURGE) != null);
    ReplicationSpec replicationSpec = new ReplicationSpec(ast);

    Table tab = null;
    try {
      tab = getTable(qualified);
    } catch (SemanticException se){
      if (replicationSpec.isInReplicationScope() &&
            (
                (se.getCause() instanceof InvalidTableException)
                ||  (se.getMessage().contains(ErrorMsg.INVALID_TABLE.getMsg()))
            )){
        // If we're inside a replication scope, then the table not existing is not an error.
        // We just return in that case, no drop needed.
        return;
        // TODO : the contains message check is fragile, we should refactor SemanticException to be
        // queriable for error code, and not simply have a message
        // NOTE : IF_EXISTS might also want to invoke this, but there's a good possibility
        // that IF_EXISTS is stricter about table existence, and applies only to the ptn.
        // Therefore, ignoring IF_EXISTS here.
      } else {
        throw se;
      }
    }
    Map> partSpecs =
        getFullPartitionSpecs(ast, tab, canGroupExprs);
    if (partSpecs.isEmpty())
     {
      return; // nothing to do
    }

    validateAlterTableType(tab, AlterTableTypes.DROPPARTITION, expectView);
    ReadEntity re = new ReadEntity(tab);
    re.noLockNeeded();
    inputs.add(re);

    addTableDropPartsOutputs(tab, partSpecs.values(), !ifExists);

    DropTableDesc dropTblDesc =
        new DropTableDesc(getDotName(qualified), partSpecs, expectView ? TableType.VIRTUAL_VIEW : null,
                mustPurge, replicationSpec);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), dropTblDesc)));
  }

  private void analyzeAlterTablePartColType(String[] qualified, ASTNode ast)
      throws SemanticException {


    // check if table exists.
    Table tab = getTable(qualified);
    inputs.add(new ReadEntity(tab));

    // validate the DDL is a valid operation on the table.
    validateAlterTableType(tab, AlterTableTypes.ALTERPARTITION, false);

    // Alter table ... partition column ( column newtype) only takes one column at a time.
    // It must have a column name followed with type.
    ASTNode colAst = (ASTNode) ast.getChild(0);

    FieldSchema newCol = new FieldSchema();

    // get column name
    String name = colAst.getChild(0).getText().toLowerCase();
    newCol.setName(unescapeIdentifier(name));

    // get column type
    ASTNode typeChild = (ASTNode) (colAst.getChild(1));
    newCol.setType(getTypeStringFromAST(typeChild));

    if (colAst.getChildCount() == 3) {
      newCol.setComment(unescapeSQLString(colAst.getChild(2).getText()));
    }

    // check if column is defined or not
    boolean fFoundColumn = false;
    for( FieldSchema col : tab.getTTable().getPartitionKeys()) {
      if (col.getName().compareTo(newCol.getName()) == 0) {
        fFoundColumn = true;
      }
    }

    // raise error if we could not find the column
    if (!fFoundColumn) {
      throw new SemanticException(ErrorMsg.INVALID_COLUMN.getMsg(newCol.getName()));
    }

    AlterTableAlterPartDesc alterTblAlterPartDesc =
            new AlterTableAlterPartDesc(getDotName(qualified), newCol);

    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblAlterPartDesc)));
  }

    /**
   * Add one or more partitions to a table. Useful when the data has been copied
   * to the right location by some other process.
   *
   * @param ast
   *          The parsed command tree.
   *
   * @param expectView
   *          True for ALTER VIEW, false for ALTER TABLE.
   *
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeAlterTableAddParts(String[] qualified, CommonTree ast, boolean expectView)
      throws SemanticException {

    // ^(TOK_ALTERTABLE_ADDPARTS identifier ifNotExists? alterStatementSuffixAddPartitionsElement+)
    boolean ifNotExists = ast.getChild(0).getType() == HiveParser.TOK_IFNOTEXISTS;

    Table tab = getTable(qualified);
    boolean isView = tab.isView();
    validateAlterTableType(tab, AlterTableTypes.ADDPARTITION, expectView);
    outputs.add(new WriteEntity(tab,
        /*use DDL_EXCLUSIVE to cause X lock to prevent races between concurrent add partition calls
        with IF NOT EXISTS.  w/o this 2 concurrent calls to add the same partition may both add
        data since for transactional tables creating partition metadata and moving data there are
        2 separate actions. */
        ifNotExists && AcidUtils.isTransactionalTable(tab) ? WriteType.DDL_EXCLUSIVE
        : WriteEntity.WriteType.DDL_SHARED));

    int numCh = ast.getChildCount();
    int start = ifNotExists ? 1 : 0;

    String currentLocation = null;
    Map currentPart = null;
    // Parser has done some verification, so the order of tokens doesn't need to be verified here.
    AddPartitionDesc addPartitionDesc =
        new AddPartitionDesc(tab.getDbName(), tab.getTableName(), ifNotExists);
    for (int num = start; num < numCh; num++) {
      ASTNode child = (ASTNode) ast.getChild(num);
      switch (child.getToken().getType()) {
      case HiveParser.TOK_PARTSPEC:
        if (currentPart != null) {
          addPartitionDesc.addPartition(currentPart, currentLocation);
          currentLocation = null;
        }
        currentPart = getValidatedPartSpec(tab, child, conf, true);
        validatePartitionValues(currentPart); // validate reserved values
        break;
      case HiveParser.TOK_PARTITIONLOCATION:
        // if location specified, set in partition
        if (isView) {
          throw new SemanticException("LOCATION clause illegal for view partition");
        }
        currentLocation = unescapeSQLString(child.getChild(0).getText());
        inputs.add(toReadEntity(currentLocation));
        break;
      default:
        throw new SemanticException("Unknown child: " + child);
      }
    }

    // add the last one
    if (currentPart != null) {
      addPartitionDesc.addPartition(currentPart, currentLocation);
    }

    if (this.conf.getBoolVar(HiveConf.ConfVars.HIVESTATSAUTOGATHER)) {
      for (int index = 0; index < addPartitionDesc.getPartitionCount(); index++) {
        OnePartitionDesc desc = addPartitionDesc.getPartition(index);
        if (desc.getLocation() == null) {
          if (desc.getPartParams() == null) {
            desc.setPartParams(new HashMap());
          }
          StatsSetupConst.setStatsStateForCreateTable(desc.getPartParams(),
              MetaStoreUtils.getColumnNames(tab.getCols()), StatsSetupConst.TRUE);
        }
      }
    }

    if (addPartitionDesc.getPartitionCount() == 0) {
      // nothing to do
      return;
    }

    Task ddlTask =
        TaskFactory.get(new DDLWork(getInputs(), getOutputs(), addPartitionDesc));
    rootTasks.add(ddlTask);
    handleTransactionalTable(tab, addPartitionDesc, ddlTask);

    if (isView) {
      // Compile internal query to capture underlying table partition dependencies
      StringBuilder cmd = new StringBuilder();
      cmd.append("SELECT * FROM ");
      cmd.append(HiveUtils.unparseIdentifier(getDotName(qualified)));
      cmd.append(" WHERE ");
      boolean firstOr = true;
      for (int i = 0; i < addPartitionDesc.getPartitionCount(); ++i) {
        AddPartitionDesc.OnePartitionDesc partitionDesc = addPartitionDesc.getPartition(i);
        if (firstOr) {
          firstOr = false;
        } else {
          cmd.append(" OR ");
        }
        boolean firstAnd = true;
        cmd.append("(");
        for (Map.Entry entry : partitionDesc.getPartSpec().entrySet()) {
          if (firstAnd) {
            firstAnd = false;
          } else {
            cmd.append(" AND ");
          }
          cmd.append(HiveUtils.unparseIdentifier(entry.getKey()));
          cmd.append(" = '");
          cmd.append(HiveUtils.escapeString(entry.getValue()));
          cmd.append("'");
        }
        cmd.append(")");
      }
      SessionState ss = SessionState.get();
      // TODO: should this use getUserFromAuthenticator?
      String uName = (ss == null? null: ss.getUserName());
      Driver driver = new Driver(conf, uName, queryState.getLineageState());
      int rc = driver.compile(cmd.toString(), false);
      if (rc != 0) {
        throw new SemanticException(ErrorMsg.NO_VALID_PARTN.getMsg());
      }
      inputs.addAll(driver.getPlan().getInputs());
    }
  }

  /**
   * Add partition for Transactional tables needs to add (copy/rename) the data so that it lands
   * in a delta_x_x/ folder in the partition dir.
   */
  private void handleTransactionalTable(Table tab, AddPartitionDesc addPartitionDesc,
      Task ddlTask) throws SemanticException {
    if(!AcidUtils.isTransactionalTable(tab)) {
      return;
    }
    Long writeId = null;
    int stmtId = 0;

    for (int index = 0; index < addPartitionDesc.getPartitionCount(); index++) {
      OnePartitionDesc desc = addPartitionDesc.getPartition(index);
      if (desc.getLocation() != null) {
        if(addPartitionDesc.isIfNotExists()) {
          //Don't add partition data if it already exists
          Partition oldPart = getPartition(tab, desc.getPartSpec(), false);
          if(oldPart != null) {
            continue;
          }
        }
        if(writeId == null) {
          //so that we only allocate a writeId only if actually adding data
          // (vs. adding a partition w/o data)
          try {
            writeId = getTxnMgr().getTableWriteId(tab.getDbName(),
                tab.getTableName());
          } catch (LockException ex) {
            throw new SemanticException("Failed to allocate the write id", ex);
          }
          stmtId = getTxnMgr().getStmtIdAndIncrement();
        }
        LoadTableDesc loadTableWork = new LoadTableDesc(new Path(desc.getLocation()),
            Utilities.getTableDesc(tab), desc.getPartSpec(),
            LoadTableDesc.LoadFileType.KEEP_EXISTING, //not relevant - creating new partition
            writeId);
        loadTableWork.setStmtId(stmtId);
        loadTableWork.setInheritTableSpecs(true);
        try {
          desc.setLocation(new Path(tab.getDataLocation(),
              Warehouse.makePartPath(desc.getPartSpec())).toString());
        }
        catch (MetaException ex) {
          throw new SemanticException("Could not determine partition path due to: "
              + ex.getMessage(), ex);
        }
        Task moveTask = TaskFactory.get(
            new MoveWork(getInputs(), getOutputs(), loadTableWork, null,
                true,//make sure to check format
                false));//is this right?
        ddlTask.addDependentTask(moveTask);
      }
    }
  }
  /**
   * Rewrite the metadata for one or more partitions in a table. Useful when
   * an external process modifies files on HDFS and you want the pre/post
   * hooks to be fired for the specified partition.
   *
   * @param ast
   *          The parsed command tree.
   * @throws SemanticException
   *           Parsing failed
   */
  private void analyzeAlterTableTouch(String[] qualified, CommonTree ast)
      throws SemanticException {

    Table tab = getTable(qualified);
    validateAlterTableType(tab, AlterTableTypes.TOUCH);
    inputs.add(new ReadEntity(tab));

    // partition name to value
    List> partSpecs = getPartitionSpecs(tab, ast);

    if (partSpecs.size() == 0) {
      AlterTableSimpleDesc touchDesc = new AlterTableSimpleDesc(
          getDotName(qualified), null,
          AlterTableDesc.AlterTableTypes.TOUCH);
      outputs.add(new WriteEntity(tab, WriteEntity.WriteType.DDL_NO_LOCK));
      rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
          touchDesc)));
    } else {
      addTablePartsOutputs(tab, partSpecs, WriteEntity.WriteType.DDL_NO_LOCK);
      for (Map partSpec : partSpecs) {
        AlterTableSimpleDesc touchDesc = new AlterTableSimpleDesc(
            getDotName(qualified), partSpec,
            AlterTableDesc.AlterTableTypes.TOUCH);
        rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
            touchDesc)));
      }
    }
  }

  private void analyzeAlterTableArchive(String[] qualified, CommonTree ast, boolean isUnArchive)
      throws SemanticException {

    if (!conf.getBoolVar(HiveConf.ConfVars.HIVEARCHIVEENABLED)) {
      throw new SemanticException(ErrorMsg.ARCHIVE_METHODS_DISABLED.getMsg());

    }
    Table tab = getTable(qualified);
    // partition name to value
    List> partSpecs = getPartitionSpecs(tab, ast);

    addTablePartsOutputs(tab, partSpecs, true, WriteEntity.WriteType.DDL_NO_LOCK);
    validateAlterTableType(tab, AlterTableTypes.ARCHIVE);
    inputs.add(new ReadEntity(tab));

    if (partSpecs.size() > 1) {
      throw new SemanticException(isUnArchive ?
          ErrorMsg.UNARCHIVE_ON_MULI_PARTS.getMsg() :
          ErrorMsg.ARCHIVE_ON_MULI_PARTS.getMsg());
    }
    if (partSpecs.size() == 0) {
      throw new SemanticException(ErrorMsg.ARCHIVE_ON_TABLE.getMsg());
    }

    Map partSpec = partSpecs.get(0);
    try {
      isValidPrefixSpec(tab, partSpec);
    } catch (HiveException e) {
      throw new SemanticException(e.getMessage(), e);
    }
    AlterTableSimpleDesc archiveDesc = new AlterTableSimpleDesc(
        getDotName(qualified), partSpec,
        (isUnArchive ? AlterTableTypes.UNARCHIVE : AlterTableTypes.ARCHIVE));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        archiveDesc)));

  }

  /**
   * Check if MSCK is called to add partitions.
   *
   * @param keyWord
   *   could be ADD, DROP or SYNC.  ADD or SYNC will indicate that add partition is on.
   *
   * @return true if add is on; false otherwise
   */
  private static boolean isMsckAddPartition(int keyWord) {
    switch (keyWord) {
    case HiveParser.KW_DROP:
      return false;
    case HiveParser.KW_SYNC:
    case HiveParser.KW_ADD:
    default:
      return true;
    }
  }

  /**
   * Check if MSCK is called to drop partitions.
   *
   * @param keyWord
   *   could be ADD, DROP or SYNC.  DROP or SYNC will indicate that drop partition is on.
   *
   * @return true if drop is on; false otherwise
   */
  private static boolean isMsckDropPartition(int keyWord) {
    switch (keyWord) {
    case HiveParser.KW_DROP:
    case HiveParser.KW_SYNC:
      return true;
    case HiveParser.KW_ADD:
    default:
      return false;
    }
  }

  /**
   * Verify that the information in the metastore matches up with the data on
   * the fs.
   *
   * @param ast
   *          Query tree.
   * @throws SemanticException
   */
  private void analyzeMetastoreCheck(CommonTree ast) throws SemanticException {
    String tableName = null;

    boolean addPartitions = true;
    boolean dropPartitions = false;

    boolean repair = false;
    if (ast.getChildCount() > 0) {
      repair = ast.getChild(0).getType() == HiveParser.KW_REPAIR;
      if (!repair) {
        tableName = getUnescapedName((ASTNode) ast.getChild(0));

        if (ast.getChildCount() > 1) {
          addPartitions = isMsckAddPartition(ast.getChild(1).getType());
          dropPartitions = isMsckDropPartition(ast.getChild(1).getType());
        }
      } else if (ast.getChildCount() > 1) {
        tableName = getUnescapedName((ASTNode) ast.getChild(1));

        if (ast.getChildCount() > 2) {
          addPartitions = isMsckAddPartition(ast.getChild(2).getType());
          dropPartitions = isMsckDropPartition(ast.getChild(2).getType());
        }
      }
    }
    Table tab = getTable(tableName);
    List> specs = getPartitionSpecs(tab, ast);
    outputs.add(new WriteEntity(tab, WriteEntity.WriteType.DDL_SHARED));
    MsckDesc checkDesc = new MsckDesc(tableName, specs, ctx.getResFile(),
        repair, addPartitions, dropPartitions);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        checkDesc)));
  }

  /**
   * Get the partition specs from the tree.
   *
   * @param ast
   *          Tree to extract partitions from.
   * @return A list of partition name to value mappings.
   * @throws SemanticException
   */
  private List> getPartitionSpecs(Table tbl, CommonTree ast)
      throws SemanticException {
    List> partSpecs = new ArrayList>();
    int childIndex = 0;
    // get partition metadata if partition specified
    for (childIndex = 0; childIndex < ast.getChildCount(); childIndex++) {
      ASTNode partSpecNode = (ASTNode)ast.getChild(childIndex);
      // sanity check
      if (partSpecNode.getType() == HiveParser.TOK_PARTSPEC) {
        Map partSpec = getValidatedPartSpec(tbl, partSpecNode, conf, false);
        partSpecs.add(partSpec);
      }
    }
    return partSpecs;
  }

  /**
   * Get the partition specs from the tree. This stores the full specification
   * with the comparator operator into the output list.
   *
   * @param ast Tree to extract partitions from.
   * @param tab Table.
   * @return    Map of partitions by prefix length. Most of the time prefix length will
   *            be the same for all partition specs, so we can just OR the expressions.
   */
  private Map> getFullPartitionSpecs(
      CommonTree ast, Table tab, boolean canGroupExprs) throws SemanticException {
    String defaultPartitionName = HiveConf.getVar(conf, HiveConf.ConfVars.DEFAULTPARTITIONNAME);
    Map colTypes = new HashMap();
    for (FieldSchema fs : tab.getPartitionKeys()) {
      colTypes.put(fs.getName().toLowerCase(), fs.getType());
    }

    Map> result =
        new HashMap>();
    for (int childIndex = 0; childIndex < ast.getChildCount(); childIndex++) {
      Tree partSpecTree = ast.getChild(childIndex);
      if (partSpecTree.getType() != HiveParser.TOK_PARTSPEC) {
        continue;
      }
      ExprNodeGenericFuncDesc expr = null;
      HashSet names = new HashSet(partSpecTree.getChildCount());
      for (int i = 0; i < partSpecTree.getChildCount(); ++i) {
        CommonTree partSpecSingleKey = (CommonTree) partSpecTree.getChild(i);
        assert (partSpecSingleKey.getType() == HiveParser.TOK_PARTVAL);
        String key = stripIdentifierQuotes(partSpecSingleKey.getChild(0).getText()).toLowerCase();
        String operator = partSpecSingleKey.getChild(1).getText();
        ASTNode partValNode = (ASTNode)partSpecSingleKey.getChild(2);
        TypeCheckCtx typeCheckCtx = new TypeCheckCtx(null);
        ExprNodeConstantDesc valExpr = (ExprNodeConstantDesc)TypeCheckProcFactory
            .genExprNode(partValNode, typeCheckCtx).get(partValNode);
        Object val = valExpr.getValue();

        boolean isDefaultPartitionName =  val.equals(defaultPartitionName);

        String type = colTypes.get(key);
        PrimitiveTypeInfo pti = TypeInfoFactory.getPrimitiveTypeInfo(type);
        if (type == null) {
          throw new SemanticException("Column " + key + " not found");
        }
        // Create the corresponding hive expression to filter on partition columns.
        if (!isDefaultPartitionName) {
          if (!valExpr.getTypeString().equals(type)) {
            Converter converter = ObjectInspectorConverters.getConverter(
              TypeInfoUtils.getStandardJavaObjectInspectorFromTypeInfo(valExpr.getTypeInfo()),
              TypeInfoUtils.getStandardJavaObjectInspectorFromTypeInfo(pti));
            val = converter.convert(valExpr.getValue());
          }
        }

        ExprNodeColumnDesc column = new ExprNodeColumnDesc(pti, key, null, true);
        ExprNodeGenericFuncDesc op;
        if (!isDefaultPartitionName) {
          op = makeBinaryPredicate(operator, column, new ExprNodeConstantDesc(pti, val));
        } else {
          GenericUDF originalOp = FunctionRegistry.getFunctionInfo(operator).getGenericUDF();
          String fnName;
          if (FunctionRegistry.isEq(originalOp)) {
            fnName = "isnull";
          } else if (FunctionRegistry.isNeq(originalOp)) {
            fnName = "isnotnull";
          } else {
            throw new SemanticException("Cannot use " + operator
                + " in a default partition spec; only '=' and '!=' are allowed.");
          }
          op = makeUnaryPredicate(fnName, column);
        }
        // If it's multi-expr filter (e.g. a='5', b='2012-01-02'), AND with previous exprs.
        expr = (expr == null) ? op : makeBinaryPredicate("and", expr, op);
        names.add(key);
      }
      if (expr == null) {
        continue;
      }
      // We got the expr for one full partition spec. Determine the prefix length.
      int prefixLength = calculatePartPrefix(tab, names);
      List orExpr = result.get(prefixLength);
      // We have to tell apart partitions resulting from spec with different prefix lengths.
      // So, if we already have smth for the same prefix length, we can OR the two.
      // If we don't, create a new separate filter. In most cases there will only be one.
      if (orExpr == null) {
        result.put(prefixLength, Lists.newArrayList(expr));
      } else if (canGroupExprs) {
        orExpr.set(0, makeBinaryPredicate("or", expr, orExpr.get(0)));
      } else {
        orExpr.add(expr);
      }
    }
    return result;
  }

  public static ExprNodeGenericFuncDesc makeBinaryPredicate(
      String fn, ExprNodeDesc left, ExprNodeDesc right) throws SemanticException {
      return new ExprNodeGenericFuncDesc(TypeInfoFactory.booleanTypeInfo,
          FunctionRegistry.getFunctionInfo(fn).getGenericUDF(), Lists.newArrayList(left, right));
  }
  public static ExprNodeGenericFuncDesc makeUnaryPredicate(
      String fn, ExprNodeDesc arg) throws SemanticException {
      return new ExprNodeGenericFuncDesc(TypeInfoFactory.booleanTypeInfo,
          FunctionRegistry.getFunctionInfo(fn).getGenericUDF(), Lists.newArrayList(arg));
  }
  /**
   * Calculates the partition prefix length based on the drop spec.
   * This is used to avoid deleting archived partitions with lower level.
   * For example, if, for A and B key cols, drop spec is A=5, B=6, we shouldn't drop
   * archived A=5/, because it can contain B-s other than 6.
   * @param tbl Table
   * @param partSpecKeys Keys present in drop partition spec.
   */
  private int calculatePartPrefix(Table tbl, HashSet partSpecKeys) {
    int partPrefixToDrop = 0;
    for (FieldSchema fs : tbl.getPartCols()) {
      if (!partSpecKeys.contains(fs.getName())) {
        break;
      }
      ++partPrefixToDrop;
    }
    return partPrefixToDrop;
  }

  /**
   * Certain partition values are are used by hive. e.g. the default partition
   * in dynamic partitioning and the intermediate partition values used in the
   * archiving process. Naturally, prohibit the user from creating partitions
   * with these reserved values. The check that this function is more
   * restrictive than the actual limitation, but it's simpler. Should be okay
   * since the reserved names are fairly long and uncommon.
   */
  private void validatePartitionValues(Map partSpec)
      throws SemanticException {

    for (Entry e : partSpec.entrySet()) {
      for (String s : reservedPartitionValues) {
        String value = e.getValue();
        if (value != null && value.contains(s)) {
          throw new SemanticException(ErrorMsg.RESERVED_PART_VAL.getMsg(
              "(User value: " + e.getValue() + " Reserved substring: " + s + ")"));
        }
      }
    }
  }

  /**
   * Add the table partitions to be modified in the output, so that it is available for the
   * pre-execution hook. If the partition does not exist, no error is thrown.
   */
  private void addTablePartsOutputs(Table table, List> partSpecs,
                                    WriteEntity.WriteType writeType)
      throws SemanticException {
    addTablePartsOutputs(table, partSpecs, false, false, null, writeType);
  }

  /**
   * Add the table partitions to be modified in the output, so that it is available for the
   * pre-execution hook. If the partition does not exist, no error is thrown.
   */
  private void addTablePartsOutputs(Table table, List> partSpecs,
      boolean allowMany, WriteEntity.WriteType writeType)
      throws SemanticException {
    addTablePartsOutputs(table, partSpecs, false, allowMany, null, writeType);
  }

  /**
   * Add the table partitions to be modified in the output, so that it is available for the
   * pre-execution hook. If the partition does not exist, throw an error if
   * throwIfNonExistent is true, otherwise ignore it.
   */
  private void addTablePartsOutputs(Table table, List> partSpecs,
      boolean throwIfNonExistent, boolean allowMany, ASTNode ast, WriteEntity.WriteType writeType)
      throws SemanticException {

    Iterator> i;
    int index;
    for (i = partSpecs.iterator(), index = 1; i.hasNext(); ++index) {
      Map partSpec = i.next();
      List parts = null;
      if (allowMany) {
        try {
          parts = db.getPartitions(table, partSpec);
        } catch (HiveException e) {
          LOG.error("Got HiveException during obtaining list of partitions"
              + StringUtils.stringifyException(e));
          throw new SemanticException(e.getMessage(), e);
        }
      } else {
        parts = new ArrayList();
        try {
          Partition p = db.getPartition(table, partSpec, false);
          if (p != null) {
            parts.add(p);
          }
        } catch (HiveException e) {
          LOG.debug("Wrong specification" + StringUtils.stringifyException(e));
          throw new SemanticException(e.getMessage(), e);
        }
      }
      if (parts.isEmpty()) {
        if (throwIfNonExistent) {
          throw new SemanticException(ErrorMsg.INVALID_PARTITION.getMsg(ast.getChild(index)));
        }
      }
      for (Partition p : parts) {
        // Don't request any locks here, as the table has already been locked.
        outputs.add(new WriteEntity(p, writeType));
      }
    }
  }

  /**
   * Add the table partitions to be modified in the output, so that it is available for the
   * pre-execution hook. If the partition does not exist, throw an error if
   * throwIfNonExistent is true, otherwise ignore it.
   */
  private void addTableDropPartsOutputs(Table tab,
                                        Collection> partSpecs,
                                        boolean throwIfNonExistent) throws SemanticException {
    for (List specs : partSpecs) {
      for (ExprNodeGenericFuncDesc partSpec : specs) {
        List parts = new ArrayList();
        boolean hasUnknown = false;
        try {
          hasUnknown = db.getPartitionsByExpr(tab, partSpec, conf, parts);
        } catch (Exception e) {
          throw new SemanticException(
              ErrorMsg.INVALID_PARTITION.getMsg(partSpec.getExprString()), e);
        }
        if (hasUnknown) {
          throw new SemanticException(
              "Unexpected unknown partitions for " + partSpec.getExprString());
        }

        // TODO: ifExists could be moved to metastore. In fact it already supports that. Check it
        //       for now since we get parts for output anyway, so we can get the error message
        //       earlier... If we get rid of output, we can get rid of this.
        if (parts.isEmpty()) {
          if (throwIfNonExistent) {
            throw new SemanticException(
                ErrorMsg.INVALID_PARTITION.getMsg(partSpec.getExprString()));
          }
        }
        for (Partition p : parts) {
          outputs.add(new WriteEntity(p, WriteEntity.WriteType.DDL_EXCLUSIVE));
        }
      }
    }
  }

  /**
   * Analyze alter table's skewed table
   *
   * @param ast
   *          node
   * @throws SemanticException
   */
  private void analyzeAltertableSkewedby(String[] qualified, ASTNode ast) throws SemanticException {
    /**
     * Throw an error if the user tries to use the DDL with
     * hive.internal.ddl.list.bucketing.enable set to false.
     */
    HiveConf hiveConf = SessionState.get().getConf();

    Table tab = getTable(qualified);

    inputs.add(new ReadEntity(tab));
    outputs.add(new WriteEntity(tab, WriteEntity.WriteType.DDL_EXCLUSIVE));

    validateAlterTableType(tab, AlterTableTypes.ADDSKEWEDBY);

    String tableName = getDotName(qualified);
    if (ast.getChildCount() == 0) {
      /* Convert a skewed table to non-skewed table. */
      AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, true,
          new ArrayList(), new ArrayList>());
      alterTblDesc.setStoredAsSubDirectories(false);
      rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
          alterTblDesc)));
    } else {
      switch (((ASTNode) ast.getChild(0)).getToken().getType()) {
      case HiveParser.TOK_TABLESKEWED:
        handleAlterTableSkewedBy(ast, tableName, tab);
        break;
      case HiveParser.TOK_STOREDASDIRS:
        handleAlterTableDisableStoredAsDirs(tableName, tab);
        break;
      default:
        assert false;
      }
    }
  }

  /**
   * Handle alter table  not stored as directories
   *
   * @param tableName
   * @param tab
   * @throws SemanticException
   */
  private void handleAlterTableDisableStoredAsDirs(String tableName, Table tab)
      throws SemanticException {
  List skewedColNames = tab.getSkewedColNames();
    List> skewedColValues = tab.getSkewedColValues();
    if ((skewedColNames == null) || (skewedColNames.size() == 0) || (skewedColValues == null)
        || (skewedColValues.size() == 0)) {
      throw new SemanticException(ErrorMsg.ALTER_TBL_STOREDASDIR_NOT_SKEWED.getMsg(tableName));
    }
    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, false,
        skewedColNames, skewedColValues);
    alterTblDesc.setStoredAsSubDirectories(false);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  /**
   * Process "alter table  skewed by .. on .. stored as directories
   * @param ast
   * @param tableName
   * @param tab
   * @throws SemanticException
   */
  private void handleAlterTableSkewedBy(ASTNode ast, String tableName, Table tab)
      throws SemanticException {
    List skewedColNames = new ArrayList();
    List> skewedValues = new ArrayList>();
    /* skewed column names. */
    ASTNode skewedNode = (ASTNode) ast.getChild(0);
    skewedColNames = analyzeSkewedTablDDLColNames(skewedColNames, skewedNode);
    /* skewed value. */
    analyzeDDLSkewedValues(skewedValues, skewedNode);
    // stored as directories
    boolean storedAsDirs = analyzeStoredAdDirs(skewedNode);


    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, false,
        skewedColNames, skewedValues);
    alterTblDesc.setStoredAsSubDirectories(storedAsDirs);
    /**
     * Validate information about skewed table
     */
    alterTblDesc.setTable(tab);
    alterTblDesc.validate();
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  /**
   * Analyze skewed column names
   *
   * @param skewedColNames
   * @param child
   * @return
   * @throws SemanticException
   */
  private List analyzeAlterTableSkewedColNames(List skewedColNames,
      ASTNode child) throws SemanticException {
    Tree nNode = child.getChild(0);
    if (nNode == null) {
      throw new SemanticException(ErrorMsg.SKEWED_TABLE_NO_COLUMN_NAME.getMsg());
    } else {
      ASTNode nAstNode = (ASTNode) nNode;
      if (nAstNode.getToken().getType() != HiveParser.TOK_TABCOLNAME) {
        throw new SemanticException(ErrorMsg.SKEWED_TABLE_NO_COLUMN_NAME.getMsg());
      } else {
        skewedColNames = getColumnNames(nAstNode);
      }
    }
    return skewedColNames;
  }

  /**
   * Given a ASTNode, return list of values.
   *
   * use case:
   * create table xyz list bucketed (col1) with skew (1,2,5)
   * AST Node is for (1,2,5)
   *
   * @param ast
   * @return
   */
  private List getColumnValues(ASTNode ast) {
    List colList = new ArrayList();
    int numCh = ast.getChildCount();
    for (int i = 0; i < numCh; i++) {
      ASTNode child = (ASTNode) ast.getChild(i);
      colList.add(stripQuotes(child.getText()).toLowerCase());
    }
    return colList;
  }


  /**
   * Analyze alter table's skewed location
   *
   * @param ast
   * @param tableName
   * @param partSpec
   * @throws SemanticException
   */
  private void analyzeAlterTableSkewedLocation(ASTNode ast, String tableName,
      HashMap partSpec) throws SemanticException {
    /**
     * Throw an error if the user tries to use the DDL with
     * hive.internal.ddl.list.bucketing.enable set to false.
     */
    HiveConf hiveConf = SessionState.get().getConf();
    /**
     * Retrieve mappings from parser
     */
    Map, String> locations = new HashMap, String>();
    ArrayList locNodes = ast.getChildren();
    if (null == locNodes) {
      throw new SemanticException(ErrorMsg.ALTER_TBL_SKEWED_LOC_NO_LOC.getMsg());
    } else {
      for (Node locNode : locNodes) {
        // TOK_SKEWED_LOCATIONS
        ASTNode locAstNode = (ASTNode) locNode;
        ArrayList locListNodes = locAstNode.getChildren();
        if (null == locListNodes) {
          throw new SemanticException(ErrorMsg.ALTER_TBL_SKEWED_LOC_NO_LOC.getMsg());
        } else {
          for (Node locListNode : locListNodes) {
            // TOK_SKEWED_LOCATION_LIST
            ASTNode locListAstNode = (ASTNode) locListNode;
            ArrayList locMapNodes = locListAstNode.getChildren();
            if (null == locMapNodes) {
              throw new SemanticException(ErrorMsg.ALTER_TBL_SKEWED_LOC_NO_LOC.getMsg());
            } else {
              for (Node locMapNode : locMapNodes) {
                // TOK_SKEWED_LOCATION_MAP
                ASTNode locMapAstNode = (ASTNode) locMapNode;
                ArrayList locMapAstNodeMaps = locMapAstNode.getChildren();
                if ((null == locMapAstNodeMaps) || (locMapAstNodeMaps.size() != 2)) {
                  throw new SemanticException(ErrorMsg.ALTER_TBL_SKEWED_LOC_NO_MAP.getMsg());
                } else {
                  List keyList = new LinkedList();
                  ASTNode node = (ASTNode) locMapAstNodeMaps.get(0);
                  if (node.getToken().getType() == HiveParser.TOK_TABCOLVALUES) {
                    keyList = getSkewedValuesFromASTNode(node);
                  } else if (isConstant(node)) {
                    keyList.add(PlanUtils
                        .stripQuotes(node.getText()));
                  } else {
                    throw new SemanticException(ErrorMsg.SKEWED_TABLE_NO_COLUMN_VALUE.getMsg());
                  }
                  String newLocation = PlanUtils
                      .stripQuotes(unescapeSQLString(((ASTNode) locMapAstNodeMaps.get(1))
                          .getText()));
                  validateSkewedLocationString(newLocation);
                  locations.put(keyList, newLocation);
                  addLocationToOutputs(newLocation);
                }
              }
            }
          }
        }
      }
    }
    AlterTableDesc alterTblDesc = new AlterTableDesc(tableName, locations, partSpec);
    addInputsOutputsAlterTable(tableName, partSpec, alterTblDesc);
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterTblDesc)));
  }

  private void addLocationToOutputs(String newLocation) throws SemanticException {
    outputs.add(toWriteEntity(newLocation));
  }

  /**
   * Check if the node is constant.
   *
   * @param node
   * @return
   */
  private boolean isConstant(ASTNode node) {
    switch(node.getToken().getType()) {
      case HiveParser.Number:
      case HiveParser.StringLiteral:
      case HiveParser.IntegralLiteral:
      case HiveParser.NumberLiteral:
      case HiveParser.CharSetName:
      case HiveParser.KW_TRUE:
      case HiveParser.KW_FALSE:
        return true;
      default:
        return false;
    }
  }

  private void validateSkewedLocationString(String newLocation) throws SemanticException {
    /* Validate location string. */
    try {
      URI locUri = new URI(newLocation);
      if (!locUri.isAbsolute() || locUri.getScheme() == null
          || locUri.getScheme().trim().equals("")) {
        throw new SemanticException(
            newLocation
                + " is not absolute or has no scheme information. "
                + "Please specify a complete absolute uri with scheme information.");
      }
    } catch (URISyntaxException e) {
      throw new SemanticException(e);
    }
  }

  private HiveAuthorizationTaskFactory createAuthorizationTaskFactory(HiveConf conf, Hive db) {
    Class authProviderClass = conf.
        getClass(HiveConf.ConfVars.HIVE_AUTHORIZATION_TASK_FACTORY.varname,
            HiveAuthorizationTaskFactoryImpl.class,
            HiveAuthorizationTaskFactory.class);
    String msg = "Unable to create instance of " + authProviderClass.getName() + ": ";
    try {
      Constructor constructor =
          authProviderClass.getConstructor(HiveConf.class, Hive.class);
      return constructor.newInstance(conf, db);
    } catch (NoSuchMethodException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    } catch (SecurityException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    } catch (InstantiationException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    } catch (IllegalAccessException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    } catch (IllegalArgumentException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    } catch (InvocationTargetException e) {
      throw new IllegalStateException(msg + e.getMessage(), e);
    }
  }

  private void analyzeAlterMaterializedViewRewrite(String mvName, ASTNode ast) throws SemanticException {
    // Value for the flag
    boolean enableFlag;
    switch (ast.getChild(0).getType()) {
      case HiveParser.TOK_REWRITE_ENABLED:
        enableFlag = true;
        break;
      case HiveParser.TOK_REWRITE_DISABLED:
        enableFlag = false;
        break;
      default:
        throw new SemanticException("Invalid alter materialized view expression");
    }

    AlterMaterializedViewDesc alterMVDesc =
        new AlterMaterializedViewDesc(AlterMaterializedViewTypes.UPDATE_REWRITE_FLAG);
    alterMVDesc.setMaterializedViewName(mvName);
    alterMVDesc.setRewriteEnableFlag(enableFlag);

    // It can be fully qualified name or use default database
    Table materializedViewTable = getTable(mvName, true);

    // One last test: if we are enabling the rewrite, we need to check that query
    // only uses transactional (MM and ACID) tables
    if (enableFlag) {
      for (String tableName : materializedViewTable.getCreationMetadata().getTablesUsed()) {
        Table table = getTable(tableName, true);
        if (!AcidUtils.isTransactionalTable(table)) {
          throw new SemanticException("Automatic rewriting for materialized view cannot "
              + "be enabled if the materialized view uses non-transactional tables");
        }
      }
    }

    inputs.add(new ReadEntity(materializedViewTable));
    outputs.add(new WriteEntity(materializedViewTable, WriteEntity.WriteType.DDL_EXCLUSIVE));
    rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(),
        alterMVDesc)));
  }

}




© 2015 - 2024 Weber Informatics LLC | Privacy Policy