Examples with FieldSchema org.apache.hadoop.hive.metastore.api.FieldSchema used on opensource projects

Example 96 with FieldSchema

use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.

the class SemanticAnalyzer method genFileSinkPlan.

@SuppressWarnings("nls")
protected Operator genFileSinkPlan(String dest, QB qb, Operator input) throws SemanticException {
    RowResolver inputRR = opParseCtx.get(input).getRowResolver();
    QBMetaData qbm = qb.getMetaData();
    Integer dest_type = qbm.getDestTypeForAlias(dest);
    // destination table if any
    Table dest_tab = null;
    // should the destination table be written to using ACID
    boolean destTableIsAcid = false;
    boolean destTableIsTemporary = false;
    boolean destTableIsMaterialization = false;
    // destination partition if any
    Partition dest_part = null;
    // the intermediate destination directory
    Path queryTmpdir = null;
    // the final destination directory
    Path dest_path = null;
    TableDesc table_desc = null;
    int currentTableId = 0;
    boolean isLocal = false;
    SortBucketRSCtx rsCtx = new SortBucketRSCtx();
    DynamicPartitionCtx dpCtx = null;
    LoadTableDesc ltd = null;
    ListBucketingCtx lbCtx = null;
    Map<String, String> partSpec = null;
    switch(dest_type.intValue()) {
        case QBMetaData.DEST_TABLE:
            {
                dest_tab = qbm.getDestTableForAlias(dest);
                destTableIsAcid = AcidUtils.isAcidTable(dest_tab);
                destTableIsTemporary = dest_tab.isTemporary();
                // Is the user trying to insert into a external tables
                if ((!conf.getBoolVar(HiveConf.ConfVars.HIVE_INSERT_INTO_EXTERNAL_TABLES)) && (dest_tab.getTableType().equals(TableType.EXTERNAL_TABLE))) {
                    throw new SemanticException(ErrorMsg.INSERT_EXTERNAL_TABLE.getMsg(dest_tab.getTableName()));
                }
                partSpec = qbm.getPartSpecForAlias(dest);
                dest_path = dest_tab.getPath();
                // verify that our destination is empty before proceeding
                if (dest_tab.isImmutable() && qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName())) {
                    try {
                        FileSystem fs = dest_path.getFileSystem(conf);
                        if (!MetaStoreUtils.isDirEmpty(fs, dest_path)) {
                            LOG.warn("Attempted write into an immutable table : " + dest_tab.getTableName() + " : " + dest_path);
                            throw new SemanticException(ErrorMsg.INSERT_INTO_IMMUTABLE_TABLE.getMsg(dest_tab.getTableName()));
                        }
                    } catch (IOException ioe) {
                        LOG.warn("Error while trying to determine if immutable table has any data : " + dest_tab.getTableName() + " : " + dest_path);
                        throw new SemanticException(ErrorMsg.INSERT_INTO_IMMUTABLE_TABLE.getMsg(ioe.getMessage()));
                    }
                }
                // check for partition
                List<FieldSchema> parts = dest_tab.getPartitionKeys();
                if (parts != null && parts.size() > 0) {
                    // table is partitioned
                    if (partSpec == null || partSpec.size() == 0) {
                        // user did NOT specify partition
                        throw new SemanticException(generateErrorMessage(qb.getParseInfo().getDestForClause(dest), ErrorMsg.NEED_PARTITION_ERROR.getMsg()));
                    }
                    dpCtx = qbm.getDPCtx(dest);
                    if (dpCtx == null) {
                        dest_tab.validatePartColumnNames(partSpec, false);
                        dpCtx = new DynamicPartitionCtx(dest_tab, partSpec, conf.getVar(HiveConf.ConfVars.DEFAULTPARTITIONNAME), conf.getIntVar(HiveConf.ConfVars.DYNAMICPARTITIONMAXPARTSPERNODE));
                        qbm.setDPCtx(dest, dpCtx);
                    }
                    if (!HiveConf.getBoolVar(conf, HiveConf.ConfVars.DYNAMICPARTITIONING)) {
                        // allow DP
                        throw new SemanticException(generateErrorMessage(qb.getParseInfo().getDestForClause(dest), ErrorMsg.DYNAMIC_PARTITION_DISABLED.getMsg()));
                    }
                    if (dpCtx.getSPPath() != null) {
                        dest_path = new Path(dest_tab.getPath(), dpCtx.getSPPath());
                    }
                    if ((dest_tab.getNumBuckets() > 0)) {
                        dpCtx.setNumBuckets(dest_tab.getNumBuckets());
                    }
                }
                boolean isNonNativeTable = dest_tab.isNonNative();
                if (isNonNativeTable) {
                    queryTmpdir = dest_path;
                } else {
                    queryTmpdir = ctx.getTempDirForPath(dest_path, true);
                }
                if (dpCtx != null) {
                    // set the root of the temporary path where dynamic partition columns will populate
                    dpCtx.setRootPath(queryTmpdir);
                }
                // this table_desc does not contain the partitioning columns
                table_desc = Utilities.getTableDesc(dest_tab);
                // Add sorting/bucketing if needed
                input = genBucketingSortingDest(dest, input, qb, table_desc, dest_tab, rsCtx);
                idToTableNameMap.put(String.valueOf(destTableId), dest_tab.getTableName());
                currentTableId = destTableId;
                destTableId++;
                lbCtx = constructListBucketingCtx(dest_tab.getSkewedColNames(), dest_tab.getSkewedColValues(), dest_tab.getSkewedColValueLocationMaps(), dest_tab.isStoredAsSubDirectories(), conf);
                // NOTE: specify Dynamic partitions in dest_tab for WriteEntity
                if (!isNonNativeTable) {
                    AcidUtils.Operation acidOp = AcidUtils.Operation.NOT_ACID;
                    if (destTableIsAcid) {
                        acidOp = getAcidType(table_desc.getOutputFileFormatClass(), dest);
                        checkAcidConstraints(qb, table_desc, dest_tab);
                    }
                    ltd = new LoadTableDesc(queryTmpdir, table_desc, dpCtx, acidOp);
                    ltd.setReplace(!qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName()));
                    ltd.setLbCtx(lbCtx);
                    loadTableWork.add(ltd);
                } else {
                    // This is a non-native table.
                    // We need to set stats as inaccurate.
                    setStatsForNonNativeTable(dest_tab);
                    // true if it is insert overwrite.
                    boolean overwrite = !qb.getParseInfo().isInsertIntoTable(String.format("%s.%s", dest_tab.getDbName(), dest_tab.getTableName()));
                    createInsertDesc(dest_tab, overwrite);
                }
                WriteEntity output = null;
                // list of dynamically created partitions are known.
                if ((dpCtx == null || dpCtx.getNumDPCols() == 0)) {
                    output = new WriteEntity(dest_tab, determineWriteType(ltd, isNonNativeTable, dest));
                    if (!outputs.add(output)) {
                        throw new SemanticException(ErrorMsg.OUTPUT_SPECIFIED_MULTIPLE_TIMES.getMsg(dest_tab.getTableName()));
                    }
                }
                if ((dpCtx != null) && (dpCtx.getNumDPCols() >= 0)) {
                    // No static partition specified
                    if (dpCtx.getNumSPCols() == 0) {
                        output = new WriteEntity(dest_tab, determineWriteType(ltd, isNonNativeTable, dest), false);
                        outputs.add(output);
                        output.setDynamicPartitionWrite(true);
                    } else // part of the partition specified
                    // Create a DummyPartition in this case. Since, the metastore does not store partial
                    // partitions currently, we need to store dummy partitions
                    {
                        try {
                            String ppath = dpCtx.getSPPath();
                            ppath = ppath.substring(0, ppath.length() - 1);
                            DummyPartition p = new DummyPartition(dest_tab, dest_tab.getDbName() + "@" + dest_tab.getTableName() + "@" + ppath, partSpec);
                            output = new WriteEntity(p, getWriteType(dest), false);
                            output.setDynamicPartitionWrite(true);
                            outputs.add(output);
                        } catch (HiveException e) {
                            throw new SemanticException(e.getMessage(), e);
                        }
                    }
                }
                ctx.getLoadTableOutputMap().put(ltd, output);
                break;
            }
        case QBMetaData.DEST_PARTITION:
            {
                dest_part = qbm.getDestPartitionForAlias(dest);
                dest_tab = dest_part.getTable();
                destTableIsAcid = AcidUtils.isAcidTable(dest_tab);
                if ((!conf.getBoolVar(HiveConf.ConfVars.HIVE_INSERT_INTO_EXTERNAL_TABLES)) && dest_tab.getTableType().equals(TableType.EXTERNAL_TABLE)) {
                    throw new SemanticException(ErrorMsg.INSERT_EXTERNAL_TABLE.getMsg(dest_tab.getTableName()));
                }
                Path tabPath = dest_tab.getPath();
                Path partPath = dest_part.getDataLocation();
                // verify that our destination is empty before proceeding
                if (dest_tab.isImmutable() && qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName())) {
                    try {
                        FileSystem fs = partPath.getFileSystem(conf);
                        if (!MetaStoreUtils.isDirEmpty(fs, partPath)) {
                            LOG.warn("Attempted write into an immutable table partition : " + dest_tab.getTableName() + " : " + partPath);
                            throw new SemanticException(ErrorMsg.INSERT_INTO_IMMUTABLE_TABLE.getMsg(dest_tab.getTableName()));
                        }
                    } catch (IOException ioe) {
                        LOG.warn("Error while trying to determine if immutable table partition has any data : " + dest_tab.getTableName() + " : " + partPath);
                        throw new SemanticException(ErrorMsg.INSERT_INTO_IMMUTABLE_TABLE.getMsg(ioe.getMessage()));
                    }
                }
                // if the table is in a different dfs than the partition,
                // replace the partition's dfs with the table's dfs.
                dest_path = new Path(tabPath.toUri().getScheme(), tabPath.toUri().getAuthority(), partPath.toUri().getPath());
                queryTmpdir = ctx.getTempDirForPath(dest_path, true);
                table_desc = Utilities.getTableDesc(dest_tab);
                // Add sorting/bucketing if needed
                input = genBucketingSortingDest(dest, input, qb, table_desc, dest_tab, rsCtx);
                idToTableNameMap.put(String.valueOf(destTableId), dest_tab.getTableName());
                currentTableId = destTableId;
                destTableId++;
                lbCtx = constructListBucketingCtx(dest_part.getSkewedColNames(), dest_part.getSkewedColValues(), dest_part.getSkewedColValueLocationMaps(), dest_part.isStoredAsSubDirectories(), conf);
                AcidUtils.Operation acidOp = AcidUtils.Operation.NOT_ACID;
                if (destTableIsAcid) {
                    acidOp = getAcidType(table_desc.getOutputFileFormatClass(), dest);
                    checkAcidConstraints(qb, table_desc, dest_tab);
                }
                ltd = new LoadTableDesc(queryTmpdir, table_desc, dest_part.getSpec(), acidOp);
                ltd.setReplace(!qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName()));
                ltd.setLbCtx(lbCtx);
                loadTableWork.add(ltd);
                if (!outputs.add(new WriteEntity(dest_part, determineWriteType(ltd, dest_tab.isNonNative(), dest)))) {
                    throw new SemanticException(ErrorMsg.OUTPUT_SPECIFIED_MULTIPLE_TIMES.getMsg(dest_tab.getTableName() + "@" + dest_part.getName()));
                }
                break;
            }
        case QBMetaData.DEST_LOCAL_FILE:
            isLocal = true;
        // fall through
        case QBMetaData.DEST_DFS_FILE:
            {
                dest_path = new Path(qbm.getDestFileForAlias(dest));
                if (isLocal) {
                    // for local directory - we always write to map-red intermediate
                    // store and then copy to local fs
                    queryTmpdir = ctx.getMRTmpPath();
                } else {
                    try {
                        Path qPath = FileUtils.makeQualified(dest_path, conf);
                        queryTmpdir = ctx.getTempDirForPath(qPath, true);
                    } catch (Exception e) {
                        throw new SemanticException("Error creating temporary folder on: " + dest_path, e);
                    }
                }
                String cols = "";
                String colTypes = "";
                ArrayList<ColumnInfo> colInfos = inputRR.getColumnInfos();
                // CTAS case: the file output format and serde are defined by the create
                // table command rather than taking the default value
                List<FieldSchema> field_schemas = null;
                CreateTableDesc tblDesc = qb.getTableDesc();
                CreateViewDesc viewDesc = qb.getViewDesc();
                if (tblDesc != null) {
                    field_schemas = new ArrayList<FieldSchema>();
                    destTableIsTemporary = tblDesc.isTemporary();
                    destTableIsMaterialization = tblDesc.isMaterialization();
                } else if (viewDesc != null) {
                    field_schemas = new ArrayList<FieldSchema>();
                    destTableIsTemporary = false;
                }
                boolean first = true;
                for (ColumnInfo colInfo : colInfos) {
                    String[] nm = inputRR.reverseLookup(colInfo.getInternalName());
                    if (nm[1] != null) {
                        // non-null column alias
                        colInfo.setAlias(nm[1]);
                    }
                    //default column name
                    String colName = colInfo.getInternalName();
                    if (field_schemas != null) {
                        FieldSchema col = new FieldSchema();
                        if (!("".equals(nm[0])) && nm[1] != null) {
                            // remove ``
                            colName = unescapeIdentifier(colInfo.getAlias()).toLowerCase();
                        }
                        colName = fixCtasColumnName(colName);
                        col.setName(colName);
                        String typeName = colInfo.getType().getTypeName();
                        // CTAS should NOT create a VOID type
                        if (typeName.equals(serdeConstants.VOID_TYPE_NAME)) {
                            throw new SemanticException(ErrorMsg.CTAS_CREATES_VOID_TYPE.getMsg(colName));
                        }
                        col.setType(typeName);
                        field_schemas.add(col);
                    }
                    if (!first) {
                        cols = cols.concat(",");
                        colTypes = colTypes.concat(":");
                    }
                    first = false;
                    cols = cols.concat(colName);
                    // Replace VOID type with string when the output is a temp table or
                    // local files.
                    // A VOID type can be generated under the query:
                    //
                    // select NULL from tt;
                    // or
                    // insert overwrite local directory "abc" select NULL from tt;
                    //
                    // where there is no column type to which the NULL value should be
                    // converted.
                    //
                    String tName = colInfo.getType().getTypeName();
                    if (tName.equals(serdeConstants.VOID_TYPE_NAME)) {
                        colTypes = colTypes.concat(serdeConstants.STRING_TYPE_NAME);
                    } else {
                        colTypes = colTypes.concat(tName);
                    }
                }
                // update the create table descriptor with the resulting schema.
                if (tblDesc != null) {
                    tblDesc.setCols(new ArrayList<FieldSchema>(field_schemas));
                } else if (viewDesc != null) {
                    viewDesc.setSchema(new ArrayList<FieldSchema>(field_schemas));
                }
                boolean isDestTempFile = true;
                if (!ctx.isMRTmpFileURI(dest_path.toUri().toString())) {
                    idToTableNameMap.put(String.valueOf(destTableId), dest_path.toUri().toString());
                    currentTableId = destTableId;
                    destTableId++;
                    isDestTempFile = false;
                }
                boolean isDfsDir = (dest_type.intValue() == QBMetaData.DEST_DFS_FILE);
                loadFileWork.add(new LoadFileDesc(tblDesc, viewDesc, queryTmpdir, dest_path, isDfsDir, cols, colTypes));
                if (tblDesc == null) {
                    if (viewDesc != null) {
                        table_desc = PlanUtils.getTableDesc(viewDesc, cols, colTypes);
                    } else if (qb.getIsQuery()) {
                        String fileFormat;
                        if (SessionState.get().getIsUsingThriftJDBCBinarySerDe()) {
                            fileFormat = "SequenceFile";
                            HiveConf.setVar(conf, HiveConf.ConfVars.HIVEQUERYRESULTFILEFORMAT, fileFormat);
                            table_desc = PlanUtils.getDefaultQueryOutputTableDesc(cols, colTypes, fileFormat, ThriftJDBCBinarySerDe.class);
                            // Set the fetch formatter to be a no-op for the ListSinkOperator, since we'll
                            // write out formatted thrift objects to SequenceFile
                            conf.set(SerDeUtils.LIST_SINK_OUTPUT_FORMATTER, NoOpFetchFormatter.class.getName());
                        } else {
                            fileFormat = HiveConf.getVar(conf, HiveConf.ConfVars.HIVEQUERYRESULTFILEFORMAT);
                            table_desc = PlanUtils.getDefaultQueryOutputTableDesc(cols, colTypes, fileFormat, LazySimpleSerDe.class);
                        }
                    } else {
                        table_desc = PlanUtils.getDefaultTableDesc(qb.getDirectoryDesc(), cols, colTypes);
                    }
                } else {
                    table_desc = PlanUtils.getTableDesc(tblDesc, cols, colTypes);
                }
                if (!outputs.add(new WriteEntity(dest_path, !isDfsDir, isDestTempFile))) {
                    throw new SemanticException(ErrorMsg.OUTPUT_SPECIFIED_MULTIPLE_TIMES.getMsg(dest_path.toUri().toString()));
                }
                break;
            }
        default:
            throw new SemanticException("Unknown destination type: " + dest_type);
    }
    input = genConversionSelectOperator(dest, qb, input, table_desc, dpCtx);
    inputRR = opParseCtx.get(input).getRowResolver();
    ArrayList<ColumnInfo> vecCol = new ArrayList<ColumnInfo>();
    if (updating(dest) || deleting(dest)) {
        vecCol.add(new ColumnInfo(VirtualColumn.ROWID.getName(), VirtualColumn.ROWID.getTypeInfo(), "", true));
    } else {
        try {
            StructObjectInspector rowObjectInspector = (StructObjectInspector) table_desc.getDeserializer(conf).getObjectInspector();
            List<? extends StructField> fields = rowObjectInspector.getAllStructFieldRefs();
            for (int i = 0; i < fields.size(); i++) {
                vecCol.add(new ColumnInfo(fields.get(i).getFieldName(), TypeInfoUtils.getTypeInfoFromObjectInspector(fields.get(i).getFieldObjectInspector()), "", false));
            }
        } catch (Exception e) {
            throw new SemanticException(e.getMessage(), e);
        }
    }
    RowSchema fsRS = new RowSchema(vecCol);
    // The output files of a FileSink can be merged if they are either not being written to a table
    // or are being written to a table which is not bucketed
    // and table the table is not sorted
    boolean canBeMerged = (dest_tab == null || !((dest_tab.getNumBuckets() > 0) || (dest_tab.getSortCols() != null && dest_tab.getSortCols().size() > 0)));
    // If this table is working with ACID semantics, turn off merging
    canBeMerged &= !destTableIsAcid;
    // Generate the partition columns from the parent input
    if (dest_type.intValue() == QBMetaData.DEST_TABLE || dest_type.intValue() == QBMetaData.DEST_PARTITION) {
        genPartnCols(dest, input, qb, table_desc, dest_tab, rsCtx);
    }
    FileSinkDesc fileSinkDesc = new FileSinkDesc(queryTmpdir, table_desc, conf.getBoolVar(HiveConf.ConfVars.COMPRESSRESULT), currentTableId, rsCtx.isMultiFileSpray(), canBeMerged, rsCtx.getNumFiles(), rsCtx.getTotalFiles(), rsCtx.getPartnCols(), dpCtx, dest_path);
    boolean isHiveServerQuery = SessionState.get().isHiveServerQuery();
    fileSinkDesc.setHiveServerQuery(isHiveServerQuery);
    // FileSinkOperator knows how to properly write to it.
    if (destTableIsAcid) {
        AcidUtils.Operation wt = updating(dest) ? AcidUtils.Operation.UPDATE : (deleting(dest) ? AcidUtils.Operation.DELETE : AcidUtils.Operation.INSERT);
        fileSinkDesc.setWriteType(wt);
        acidFileSinks.add(fileSinkDesc);
    }
    fileSinkDesc.setTemporary(destTableIsTemporary);
    fileSinkDesc.setMaterialization(destTableIsMaterialization);
    /* Set List Bucketing context. */
    if (lbCtx != null) {
        lbCtx.processRowSkewedIndex(fsRS);
        lbCtx.calculateSkewedValueSubDirList();
    }
    fileSinkDesc.setLbCtx(lbCtx);
    // set the stats publishing/aggregating key prefix
    // the same as directory name. The directory name
    // can be changed in the optimizer but the key should not be changed
    // it should be the same as the MoveWork's sourceDir.
    fileSinkDesc.setStatsAggPrefix(fileSinkDesc.getDirName().toString());
    if (!destTableIsMaterialization && HiveConf.getVar(conf, HIVESTATSDBCLASS).equalsIgnoreCase(StatDB.fs.name())) {
        String statsTmpLoc = ctx.getTempDirForPath(dest_path).toString();
        fileSinkDesc.setStatsTmpDir(statsTmpLoc);
        LOG.debug("Set stats collection dir : " + statsTmpLoc);
    }
    if (dest_part != null) {
        try {
            String staticSpec = Warehouse.makePartPath(dest_part.getSpec());
            fileSinkDesc.setStaticSpec(staticSpec);
        } catch (MetaException e) {
            throw new SemanticException(e);
        }
    } else if (dpCtx != null) {
        fileSinkDesc.setStaticSpec(dpCtx.getSPPath());
    }
    if (isHiveServerQuery && null != table_desc && table_desc.getSerdeClassName().equalsIgnoreCase(ThriftJDBCBinarySerDe.class.getName()) && HiveConf.getBoolVar(conf, HiveConf.ConfVars.HIVE_SERVER2_THRIFT_RESULTSET_SERIALIZE_IN_TASKS)) {
        fileSinkDesc.setIsUsingThriftJDBCBinarySerDe(true);
    } else {
        fileSinkDesc.setIsUsingThriftJDBCBinarySerDe(false);
    }
    Operator output = putOpInsertMap(OperatorFactory.getAndMakeChild(fileSinkDesc, fsRS, input), inputRR);
    if (ltd != null && SessionState.get() != null) {
        SessionState.get().getLineageState().mapDirToOp(ltd.getSourcePath(), (FileSinkOperator) output);
    } else if (queryState.getCommandType().equals(HiveOperation.CREATETABLE_AS_SELECT.getOperationName())) {
        Path tlocation = null;
        String tName = Utilities.getDbTableName(tableDesc.getTableName())[1];
        try {
            Warehouse wh = new Warehouse(conf);
            tlocation = wh.getTablePath(db.getDatabase(tableDesc.getDatabaseName()), tName);
        } catch (MetaException | HiveException e) {
            throw new SemanticException(e);
        }
        SessionState.get().getLineageState().mapDirToOp(tlocation, (FileSinkOperator) output);
    }
    if (LOG.isDebugEnabled()) {
        LOG.debug("Created FileSink Plan for clause: " + dest + "dest_path: " + dest_path + " row schema: " + inputRR.toString());
    }
    FileSinkOperator fso = (FileSinkOperator) output;
    fso.getConf().setTable(dest_tab);
    fsopToTable.put(fso, dest_tab);
    // and it is an insert overwrite or insert into table
    if (dest_tab != null && conf.getBoolVar(ConfVars.HIVESTATSAUTOGATHER) && conf.getBoolVar(ConfVars.HIVESTATSCOLAUTOGATHER) && ColumnStatsAutoGatherContext.canRunAutogatherStats(fso)) {
        if (dest_type.intValue() == QBMetaData.DEST_TABLE) {
            genAutoColumnStatsGatheringPipeline(qb, table_desc, partSpec, input, qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName()));
        } else if (dest_type.intValue() == QBMetaData.DEST_PARTITION) {
            genAutoColumnStatsGatheringPipeline(qb, table_desc, dest_part.getSpec(), input, qb.getParseInfo().isInsertIntoTable(dest_tab.getDbName(), dest_tab.getTableName()));
        }
    }
    return output;
}

Example 97 with FieldSchema

use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.

the class SemanticAnalyzer method analyzeCreateTable.

/**
   * Analyze the create table command. If it is a regular create-table or
   * create-table-like statements, we create a DDLWork and return true. If it is
   * a create-table-as-select, we get the necessary info such as the SerDe and
   * Storage Format and put it in QB, and return false, indicating the rest of
   * the semantic analyzer need to deal with the select statement with respect
   * to the SerDe and Storage Format.
   */
ASTNode analyzeCreateTable(ASTNode ast, QB qb, PlannerContext plannerCtx) throws SemanticException {
    String[] qualifiedTabName = getQualifiedTableName((ASTNode) ast.getChild(0));
    String dbDotTab = getDotName(qualifiedTabName);
    String likeTableName = null;
    List<FieldSchema> cols = new ArrayList<FieldSchema>();
    List<FieldSchema> partCols = new ArrayList<FieldSchema>();
    List<String> bucketCols = new ArrayList<String>();
    List<SQLPrimaryKey> primaryKeys = new ArrayList<SQLPrimaryKey>();
    List<SQLForeignKey> foreignKeys = new ArrayList<SQLForeignKey>();
    List<Order> sortCols = new ArrayList<Order>();
    int numBuckets = -1;
    String comment = null;
    String location = null;
    Map<String, String> tblProps = null;
    boolean ifNotExists = false;
    boolean isExt = false;
    boolean isTemporary = false;
    boolean isMaterialization = false;
    ASTNode selectStmt = null;
    // regular CREATE TABLE
    final int CREATE_TABLE = 0;
    // CREATE TABLE LIKE ... (CTLT)
    final int CTLT = 1;
    // CREATE TABLE AS SELECT ... (CTAS)
    final int CTAS = 2;
    int command_type = CREATE_TABLE;
    List<String> skewedColNames = new ArrayList<String>();
    List<List<String>> skewedValues = new ArrayList<List<String>>();
    Map<List<String>, String> listBucketColValuesMapping = new HashMap<List<String>, String>();
    boolean storedAsDirs = false;
    boolean isUserStorageFormat = false;
    RowFormatParams rowFormatParams = new RowFormatParams();
    StorageFormat storageFormat = new StorageFormat(conf);
    LOG.info("Creating table " + dbDotTab + " position=" + ast.getCharPositionInLine());
    int numCh = ast.getChildCount();
    /*
     * Check the 1st-level children and do simple semantic checks: 1) CTLT and
     * CTAS should not coexists. 2) CTLT or CTAS should not coexists with column
     * list (target table schema). 3) CTAS does not support partitioning (for
     * now).
     */
    for (int num = 1; num < numCh; num++) {
        ASTNode child = (ASTNode) ast.getChild(num);
        if (storageFormat.fillStorageFormat(child)) {
            isUserStorageFormat = true;
            continue;
        }
        switch(child.getToken().getType()) {
            case HiveParser.TOK_IFNOTEXISTS:
                ifNotExists = true;
                break;
            case HiveParser.KW_EXTERNAL:
                isExt = true;
                break;
            case HiveParser.KW_TEMPORARY:
                isTemporary = true;
                isMaterialization = MATERIALIZATION_MARKER.equals(child.getText());
                break;
            case HiveParser.TOK_LIKETABLE:
                if (child.getChildCount() > 0) {
                    likeTableName = getUnescapedName((ASTNode) child.getChild(0));
                    if (likeTableName != null) {
                        if (command_type == CTAS) {
                            throw new SemanticException(ErrorMsg.CTAS_CTLT_COEXISTENCE.getMsg());
                        }
                        if (cols.size() != 0) {
                            throw new SemanticException(ErrorMsg.CTLT_COLLST_COEXISTENCE.getMsg());
                        }
                    }
                    command_type = CTLT;
                }
                break;
            case // CTAS
            HiveParser.TOK_QUERY:
                if (command_type == CTLT) {
                    throw new SemanticException(ErrorMsg.CTAS_CTLT_COEXISTENCE.getMsg());
                }
                if (cols.size() != 0) {
                    throw new SemanticException(ErrorMsg.CTAS_COLLST_COEXISTENCE.getMsg());
                }
                if (partCols.size() != 0 || bucketCols.size() != 0) {
                    boolean dynPart = HiveConf.getBoolVar(conf, HiveConf.ConfVars.DYNAMICPARTITIONING);
                    if (dynPart == false) {
                        throw new SemanticException(ErrorMsg.CTAS_PARCOL_COEXISTENCE.getMsg());
                    } else {
                        // TODO: support dynamic partition for CTAS
                        throw new SemanticException(ErrorMsg.CTAS_PARCOL_COEXISTENCE.getMsg());
                    }
                }
                if (isExt) {
                    throw new SemanticException(ErrorMsg.CTAS_EXTTBL_COEXISTENCE.getMsg());
                }
                command_type = CTAS;
                if (plannerCtx != null) {
                    plannerCtx.setCTASToken(child);
                }
                selectStmt = child;
                break;
            case HiveParser.TOK_TABCOLLIST:
                cols = getColumns(child, true, primaryKeys, foreignKeys);
                break;
            case HiveParser.TOK_TABLECOMMENT:
                comment = unescapeSQLString(child.getChild(0).getText());
                break;
            case HiveParser.TOK_TABLEPARTCOLS:
                partCols = getColumns((ASTNode) child.getChild(0), false);
                break;
            case HiveParser.TOK_ALTERTABLE_BUCKETS:
                bucketCols = getColumnNames((ASTNode) child.getChild(0));
                if (child.getChildCount() == 2) {
                    numBuckets = Integer.parseInt(child.getChild(1).getText());
                } else {
                    sortCols = getColumnNamesOrder((ASTNode) child.getChild(1));
                    numBuckets = Integer.parseInt(child.getChild(2).getText());
                }
                break;
            case HiveParser.TOK_TABLEROWFORMAT:
                rowFormatParams.analyzeRowFormat(child);
                break;
            case HiveParser.TOK_TABLELOCATION:
                location = unescapeSQLString(child.getChild(0).getText());
                location = EximUtil.relativeToAbsolutePath(conf, location);
                inputs.add(toReadEntity(location));
                break;
            case HiveParser.TOK_TABLEPROPERTIES:
                tblProps = DDLSemanticAnalyzer.getProps((ASTNode) child.getChild(0));
                break;
            case HiveParser.TOK_TABLESERIALIZER:
                child = (ASTNode) child.getChild(0);
                storageFormat.setSerde(unescapeSQLString(child.getChild(0).getText()));
                if (child.getChildCount() == 2) {
                    readProps((ASTNode) (child.getChild(1).getChild(0)), storageFormat.getSerdeProps());
                }
                break;
            case HiveParser.TOK_TABLESKEWED:
                /**
         * 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();
                // skewed column names
                skewedColNames = analyzeSkewedTablDDLColNames(skewedColNames, child);
                // skewed value
                analyzeDDLSkewedValues(skewedValues, child);
                // stored as directories
                storedAsDirs = analyzeStoredAdDirs(child);
                break;
            default:
                throw new AssertionError("Unknown token: " + child.getToken());
        }
    }
    if (command_type == CREATE_TABLE || command_type == CTLT) {
        queryState.setCommandType(HiveOperation.CREATETABLE);
    } else if (command_type == CTAS) {
        queryState.setCommandType(HiveOperation.CREATETABLE_AS_SELECT);
    } else {
        throw new SemanticException("Unrecognized command.");
    }
    storageFormat.fillDefaultStorageFormat(isExt, false);
    // check for existence of table
    if (ifNotExists) {
        try {
            Table table = getTable(qualifiedTabName, false);
            if (table != null) {
                // table exists
                return null;
            }
        } catch (HiveException e) {
            // should not occur since second parameter to getTableWithQN is false
            throw new IllegalStateException("Unexpected Exception thrown: " + e.getMessage(), e);
        }
    }
    addDbAndTabToOutputs(qualifiedTabName, TableType.MANAGED_TABLE);
    if (isTemporary) {
        if (partCols.size() > 0) {
            throw new SemanticException("Partition columns are not supported on temporary tables");
        }
        if (location == null) {
            // it has the same life cycle as the tmp table
            try {
                // Generate a unique ID for temp table path.
                // This path will be fixed for the life of the temp table.
                Path path = new Path(SessionState.getTempTableSpace(conf), UUID.randomUUID().toString());
                path = Warehouse.getDnsPath(path, conf);
                location = path.toString();
            } catch (MetaException err) {
                throw new SemanticException("Error while generating temp table path:", err);
            }
        }
    }
    // Handle different types of CREATE TABLE command
    switch(command_type) {
        case // REGULAR CREATE TABLE DDL
        CREATE_TABLE:
            tblProps = addDefaultProperties(tblProps);
            CreateTableDesc crtTblDesc = new CreateTableDesc(dbDotTab, isExt, isTemporary, cols, partCols, bucketCols, sortCols, numBuckets, rowFormatParams.fieldDelim, rowFormatParams.fieldEscape, rowFormatParams.collItemDelim, rowFormatParams.mapKeyDelim, rowFormatParams.lineDelim, comment, storageFormat.getInputFormat(), storageFormat.getOutputFormat(), location, storageFormat.getSerde(), storageFormat.getStorageHandler(), storageFormat.getSerdeProps(), tblProps, ifNotExists, skewedColNames, skewedValues, primaryKeys, foreignKeys);
            crtTblDesc.setStoredAsSubDirectories(storedAsDirs);
            crtTblDesc.setNullFormat(rowFormatParams.nullFormat);
            crtTblDesc.validate(conf);
            // outputs is empty, which means this create table happens in the current
            // database.
            rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), crtTblDesc), conf));
            break;
        case // create table like <tbl_name>
        CTLT:
            tblProps = addDefaultProperties(tblProps);
            if (isTemporary) {
                Table likeTable = getTable(likeTableName, false);
                if (likeTable != null && likeTable.getPartCols().size() > 0) {
                    throw new SemanticException("Partition columns are not supported on temporary tables " + "and source table in CREATE TABLE LIKE is partitioned.");
                }
            }
            CreateTableLikeDesc crtTblLikeDesc = new CreateTableLikeDesc(dbDotTab, isExt, isTemporary, storageFormat.getInputFormat(), storageFormat.getOutputFormat(), location, storageFormat.getSerde(), storageFormat.getSerdeProps(), tblProps, ifNotExists, likeTableName, isUserStorageFormat);
            rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), crtTblLikeDesc), conf));
            break;
        case // create table as select
        CTAS:
            if (isTemporary) {
                if (!ctx.isExplainSkipExecution() && !isMaterialization) {
                    String dbName = qualifiedTabName[0];
                    String tblName = qualifiedTabName[1];
                    SessionState ss = SessionState.get();
                    if (ss == null) {
                        throw new SemanticException("No current SessionState, cannot create temporary table " + dbName + "." + tblName);
                    }
                    Map<String, Table> tables = SessionHiveMetaStoreClient.getTempTablesForDatabase(dbName);
                    if (tables != null && tables.containsKey(tblName)) {
                        throw new SemanticException("Temporary table " + dbName + "." + tblName + " already exists");
                    }
                }
            } else {
                // dumpTable is only used to check the conflict for non-temporary tables
                try {
                    Table dumpTable = db.newTable(dbDotTab);
                    if (null != db.getTable(dumpTable.getDbName(), dumpTable.getTableName(), false) && !ctx.isExplainSkipExecution()) {
                        throw new SemanticException(ErrorMsg.TABLE_ALREADY_EXISTS.getMsg(dbDotTab));
                    }
                } catch (HiveException e) {
                    throw new SemanticException(e);
                }
            }
            if (location != null && location.length() != 0) {
                Path locPath = new Path(location);
                FileSystem curFs = null;
                FileStatus locStats = null;
                try {
                    curFs = locPath.getFileSystem(conf);
                    if (curFs != null) {
                        locStats = curFs.getFileStatus(locPath);
                    }
                    if (locStats != null && locStats.isDir()) {
                        FileStatus[] lStats = curFs.listStatus(locPath);
                        if (lStats != null && lStats.length != 0) {
                            // Don't throw an exception if the target location only contains the staging-dirs
                            for (FileStatus lStat : lStats) {
                                if (!lStat.getPath().getName().startsWith(HiveConf.getVar(conf, HiveConf.ConfVars.STAGINGDIR))) {
                                    throw new SemanticException(ErrorMsg.CTAS_LOCATION_NONEMPTY.getMsg(location));
                                }
                            }
                        }
                    }
                } catch (FileNotFoundException nfe) {
                //we will create the folder if it does not exist.
                } catch (IOException ioE) {
                    if (LOG.isDebugEnabled()) {
                        LOG.debug("Exception when validate folder ", ioE);
                    }
                }
            }
            tblProps = addDefaultProperties(tblProps);
            tableDesc = new CreateTableDesc(qualifiedTabName[0], dbDotTab, isExt, isTemporary, cols, partCols, bucketCols, sortCols, numBuckets, rowFormatParams.fieldDelim, rowFormatParams.fieldEscape, rowFormatParams.collItemDelim, rowFormatParams.mapKeyDelim, rowFormatParams.lineDelim, comment, storageFormat.getInputFormat(), storageFormat.getOutputFormat(), location, storageFormat.getSerde(), storageFormat.getStorageHandler(), storageFormat.getSerdeProps(), tblProps, ifNotExists, skewedColNames, skewedValues, true, primaryKeys, foreignKeys);
            tableDesc.setMaterialization(isMaterialization);
            tableDesc.setStoredAsSubDirectories(storedAsDirs);
            tableDesc.setNullFormat(rowFormatParams.nullFormat);
            qb.setTableDesc(tableDesc);
            return selectStmt;
        default:
            throw new SemanticException("Unrecognized command.");
    }
    return null;
}

Example 98 with FieldSchema

use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.

the class SemanticAnalyzer method handleInsertStatementSpec.

/**
   * This modifies the Select projections when the Select is part of an insert statement and
   * the insert statement specifies a column list for the target table, e.g.
   * create table source (a int, b int);
   * create table target (x int, y int, z int);
   * insert into target(z,x) select * from source
   *
   * Once the * is resolved to 'a,b', this list needs to rewritten to 'b,null,a' so that it looks
   * as if the original query was written as
   * insert into target select b, null, a from source
   *
   * if target schema is not specified, this is no-op
   *
   * @see #handleInsertStatementSpecPhase1(ASTNode, QBParseInfo, org.apache.hadoop.hive.ql.parse.SemanticAnalyzer.Phase1Ctx)
   * @throws SemanticException
   */
public RowResolver handleInsertStatementSpec(List<ExprNodeDesc> col_list, String dest, RowResolver outputRR, RowResolver inputRR, QB qb, ASTNode selExprList) throws SemanticException {
    //(z,x)
    //specified in the query
    List<String> targetTableSchema = qb.getParseInfo().getDestSchemaForClause(dest);
    if (targetTableSchema == null) {
        //no insert schema was specified
        return outputRR;
    }
    if (targetTableSchema.size() != col_list.size()) {
        Table target = qb.getMetaData().getDestTableForAlias(dest);
        Partition partition = target == null ? qb.getMetaData().getDestPartitionForAlias(dest) : null;
        throw new SemanticException(generateErrorMessage(selExprList, "Expected " + targetTableSchema.size() + " columns for " + dest + (target != null ? "/" + target.getCompleteName() : (partition != null ? "/" + partition.getCompleteName() : "")) + "; select produces " + col_list.size() + " columns"));
    }
    //e.g. map z->expr for a
    Map<String, ExprNodeDesc> targetCol2Projection = new HashMap<String, ExprNodeDesc>();
    //e.g. map z->ColumnInfo for a
    Map<String, ColumnInfo> targetCol2ColumnInfo = new HashMap<String, ColumnInfo>();
    int colListPos = 0;
    for (String targetCol : targetTableSchema) {
        targetCol2ColumnInfo.put(targetCol, outputRR.getColumnInfos().get(colListPos));
        targetCol2Projection.put(targetCol, col_list.get(colListPos++));
    }
    Table target = qb.getMetaData().getDestTableForAlias(dest);
    Partition partition = target == null ? qb.getMetaData().getDestPartitionForAlias(dest) : null;
    if (target == null && partition == null) {
        throw new SemanticException(generateErrorMessage(selExprList, "No table/partition found in QB metadata for dest='" + dest + "'"));
    }
    ArrayList<ExprNodeDesc> new_col_list = new ArrayList<ExprNodeDesc>();
    colListPos = 0;
    List<FieldSchema> targetTableCols = target != null ? target.getCols() : partition.getCols();
    List<String> targetTableColNames = new ArrayList<String>();
    List<TypeInfo> targetTableColTypes = new ArrayList<TypeInfo>();
    for (FieldSchema fs : targetTableCols) {
        targetTableColNames.add(fs.getName());
        targetTableColTypes.add(TypeInfoUtils.getTypeInfoFromTypeString(fs.getType()));
    }
    Map<String, String> partSpec = qb.getMetaData().getPartSpecForAlias(dest);
    if (partSpec != null) {
        //relies on consistent order via LinkedHashMap
        for (Map.Entry<String, String> partKeyVal : partSpec.entrySet()) {
            if (partKeyVal.getValue() == null) {
                //these must be after non-partition cols
                targetTableColNames.add(partKeyVal.getKey());
                targetTableColTypes.add(TypeInfoFactory.stringTypeInfo);
            }
        }
    }
    RowResolver newOutputRR = new RowResolver();
    //where missing columns are NULL-filled
    for (int i = 0; i < targetTableColNames.size(); i++) {
        String f = targetTableColNames.get(i);
        if (targetCol2Projection.containsKey(f)) {
            //put existing column in new list to make sure it is in the right position
            new_col_list.add(targetCol2Projection.get(f));
            //todo: is this OK?
            ColumnInfo ci = targetCol2ColumnInfo.get(f);
            ci.setInternalName(getColumnInternalName(colListPos));
            newOutputRR.put(ci.getTabAlias(), ci.getInternalName(), ci);
        } else {
            //add new 'synthetic' columns for projections not provided by Select
            ExprNodeDesc exp = new ExprNodeConstantDesc(targetTableColTypes.get(i), null);
            new_col_list.add(exp);
            //this column doesn't come from any table
            final String tableAlias = null;
            ColumnInfo colInfo = new ColumnInfo(getColumnInternalName(colListPos), exp.getWritableObjectInspector(), tableAlias, false);
            newOutputRR.put(colInfo.getTabAlias(), colInfo.getInternalName(), colInfo);
        }
        colListPos++;
    }
    col_list.clear();
    col_list.addAll(new_col_list);
    return newOutputRR;
}

Example 99 with FieldSchema

use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.

the class SemanticAnalyzer method analyzeCreateView.

protected ASTNode analyzeCreateView(ASTNode ast, QB qb, PlannerContext plannerCtx) throws SemanticException {
    String[] qualTabName = getQualifiedTableName((ASTNode) ast.getChild(0));
    String dbDotTable = getDotName(qualTabName);
    List<FieldSchema> cols = null;
    boolean ifNotExists = false;
    boolean rewriteEnabled = false;
    boolean orReplace = false;
    boolean isAlterViewAs = false;
    String comment = null;
    ASTNode selectStmt = null;
    Map<String, String> tblProps = null;
    List<String> partColNames = null;
    boolean isMaterialized = ast.getToken().getType() == HiveParser.TOK_CREATE_MATERIALIZED_VIEW;
    String location = null;
    RowFormatParams rowFormatParams = new RowFormatParams();
    StorageFormat storageFormat = new StorageFormat(conf);
    LOG.info("Creating view " + dbDotTable + " position=" + ast.getCharPositionInLine());
    int numCh = ast.getChildCount();
    for (int num = 1; num < numCh; num++) {
        ASTNode child = (ASTNode) ast.getChild(num);
        if (storageFormat.fillStorageFormat(child)) {
            continue;
        }
        switch(child.getToken().getType()) {
            case HiveParser.TOK_IFNOTEXISTS:
                ifNotExists = true;
                break;
            case HiveParser.TOK_REWRITE_ENABLED:
                rewriteEnabled = true;
                break;
            case HiveParser.TOK_ORREPLACE:
                orReplace = true;
                break;
            case HiveParser.TOK_QUERY:
                // For CBO
                if (plannerCtx != null) {
                    plannerCtx.setViewToken(child);
                }
                selectStmt = child;
                break;
            case HiveParser.TOK_TABCOLNAME:
                cols = getColumns(child);
                break;
            case HiveParser.TOK_TABLECOMMENT:
                comment = unescapeSQLString(child.getChild(0).getText());
                break;
            case HiveParser.TOK_TABLEPROPERTIES:
                tblProps = DDLSemanticAnalyzer.getProps((ASTNode) child.getChild(0));
                break;
            case HiveParser.TOK_VIEWPARTCOLS:
                partColNames = getColumnNames((ASTNode) child.getChild(0));
                break;
            case HiveParser.TOK_TABLEROWFORMAT:
                rowFormatParams.analyzeRowFormat(child);
                break;
            case HiveParser.TOK_TABLELOCATION:
                location = unescapeSQLString(child.getChild(0).getText());
                location = EximUtil.relativeToAbsolutePath(conf, location);
                inputs.add(toReadEntity(location));
                break;
            case HiveParser.TOK_TABLESERIALIZER:
                child = (ASTNode) child.getChild(0);
                storageFormat.setSerde(unescapeSQLString(child.getChild(0).getText()));
                if (child.getChildCount() == 2) {
                    readProps((ASTNode) (child.getChild(1).getChild(0)), storageFormat.getSerdeProps());
                }
                break;
            default:
                assert false;
        }
    }
    storageFormat.fillDefaultStorageFormat(false, isMaterialized);
    if (ifNotExists && orReplace) {
        throw new SemanticException("Can't combine IF NOT EXISTS and OR REPLACE.");
    }
    if (ast.getToken().getType() == HiveParser.TOK_ALTERVIEW && ast.getChild(1).getType() == HiveParser.TOK_QUERY) {
        isAlterViewAs = true;
        orReplace = true;
    }
    unparseTranslator.enable();
    if (isMaterialized) {
        createVwDesc = new CreateViewDesc(dbDotTable, cols, comment, tblProps, partColNames, ifNotExists, orReplace, rewriteEnabled, isAlterViewAs, storageFormat.getInputFormat(), storageFormat.getOutputFormat(), location, storageFormat.getSerde(), storageFormat.getStorageHandler(), storageFormat.getSerdeProps());
        addDbAndTabToOutputs(qualTabName, TableType.MATERIALIZED_VIEW);
        queryState.setCommandType(HiveOperation.CREATE_MATERIALIZED_VIEW);
    } else {
        createVwDesc = new CreateViewDesc(dbDotTable, cols, comment, tblProps, partColNames, ifNotExists, orReplace, isAlterViewAs, storageFormat.getInputFormat(), storageFormat.getOutputFormat(), storageFormat.getSerde());
        rootTasks.add(TaskFactory.get(new DDLWork(getInputs(), getOutputs(), createVwDesc), conf));
        addDbAndTabToOutputs(qualTabName, TableType.VIRTUAL_VIEW);
        queryState.setCommandType(HiveOperation.CREATEVIEW);
    }
    qb.setViewDesc(createVwDesc);
    return selectStmt;
}

Example 100 with FieldSchema

use of org.apache.hadoop.hive.metastore.api.FieldSchema in project hive by apache.

the class SemanticAnalyzer method convertRowSchemaToResultSetSchema.

static List<FieldSchema> convertRowSchemaToResultSetSchema(RowResolver rr, boolean useTabAliasIfAvailable) {
    List<FieldSchema> fieldSchemas = new ArrayList<FieldSchema>();
    String[] qualifiedColName;
    String colName;
    for (ColumnInfo colInfo : rr.getColumnInfos()) {
        if (colInfo.isHiddenVirtualCol()) {
            continue;
        }
        qualifiedColName = rr.reverseLookup(colInfo.getInternalName());
        if (useTabAliasIfAvailable && qualifiedColName[0] != null && !qualifiedColName[0].isEmpty()) {
            colName = qualifiedColName[0] + "." + qualifiedColName[1];
        } else {
            colName = qualifiedColName[1];
        }
        fieldSchemas.add(new FieldSchema(colName, colInfo.getType().getTypeName(), null));
    }
    return fieldSchemas;
}