Examples with FunctionIdentifier org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier used on opensource projects

Example 11 with FunctionIdentifier

use of org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier in project asterixdb by apache.

the class SetAsterixPhysicalOperatorsRule method computeDefaultPhysicalOp.

private static void computeDefaultPhysicalOp(AbstractLogicalOperator op, IOptimizationContext context) throws AlgebricksException {
    PhysicalOptimizationConfig physicalOptimizationConfig = context.getPhysicalOptimizationConfig();
    if (op.getOperatorTag().equals(LogicalOperatorTag.GROUP)) {
        GroupByOperator gby = (GroupByOperator) op;
        if (gby.getNestedPlans().size() == 1) {
            ILogicalPlan p0 = gby.getNestedPlans().get(0);
            if (p0.getRoots().size() == 1) {
                Mutable<ILogicalOperator> r0 = p0.getRoots().get(0);
                if (((AbstractLogicalOperator) (r0.getValue())).getOperatorTag().equals(LogicalOperatorTag.AGGREGATE)) {
                    AggregateOperator aggOp = (AggregateOperator) r0.getValue();
                    boolean serializable = true;
                    for (Mutable<ILogicalExpression> exprRef : aggOp.getExpressions()) {
                        AbstractFunctionCallExpression expr = (AbstractFunctionCallExpression) exprRef.getValue();
                        if (!BuiltinFunctions.isAggregateFunctionSerializable(expr.getFunctionIdentifier())) {
                            serializable = false;
                            break;
                        }
                    }
                    if ((gby.getAnnotations().get(OperatorAnnotations.USE_HASH_GROUP_BY) == Boolean.TRUE || gby.getAnnotations().get(OperatorAnnotations.USE_EXTERNAL_GROUP_BY) == Boolean.TRUE)) {
                        boolean setToExternalGby = false;
                        if (serializable) {
                            // if serializable, use external group-by
                            // now check whether the serialized version aggregation function has corresponding intermediate agg
                            boolean hasIntermediateAgg = true;
                            IMergeAggregationExpressionFactory mergeAggregationExpressionFactory = context.getMergeAggregationExpressionFactory();
                            List<LogicalVariable> originalVariables = aggOp.getVariables();
                            List<Mutable<ILogicalExpression>> aggExprs = aggOp.getExpressions();
                            int aggNum = aggExprs.size();
                            for (int i = 0; i < aggNum; i++) {
                                AbstractFunctionCallExpression expr = (AbstractFunctionCallExpression) aggExprs.get(i).getValue();
                                AggregateFunctionCallExpression serialAggExpr = BuiltinFunctions.makeSerializableAggregateFunctionExpression(expr.getFunctionIdentifier(), expr.getArguments());
                                if (mergeAggregationExpressionFactory.createMergeAggregation(originalVariables.get(i), serialAggExpr, context) == null) {
                                    hasIntermediateAgg = false;
                                    break;
                                }
                            }
                            // Check whether there are multiple aggregates in the sub plan.
                            // Currently, we don't support multiple aggregates in one external group-by.
                            boolean multipleAggOpsFound = false;
                            ILogicalOperator r1Logical = aggOp;
                            while (r1Logical.hasInputs()) {
                                r1Logical = r1Logical.getInputs().get(0).getValue();
                                if (r1Logical.getOperatorTag() == LogicalOperatorTag.AGGREGATE) {
                                    multipleAggOpsFound = true;
                                    break;
                                }
                            }
                            if (hasIntermediateAgg && !multipleAggOpsFound) {
                                for (int i = 0; i < aggNum; i++) {
                                    AbstractFunctionCallExpression expr = (AbstractFunctionCallExpression) aggExprs.get(i).getValue();
                                    AggregateFunctionCallExpression serialAggExpr = BuiltinFunctions.makeSerializableAggregateFunctionExpression(expr.getFunctionIdentifier(), expr.getArguments());
                                    aggOp.getExpressions().get(i).setValue(serialAggExpr);
                                }
                                ExternalGroupByPOperator externalGby = new ExternalGroupByPOperator(gby.getGroupByList(), physicalOptimizationConfig.getMaxFramesExternalGroupBy(), (long) physicalOptimizationConfig.getMaxFramesExternalGroupBy() * physicalOptimizationConfig.getFrameSize());
                                generateMergeAggregationExpressions(gby, context);
                                op.setPhysicalOperator(externalGby);
                                setToExternalGby = true;
                            }
                        }
                        if (!setToExternalGby) {
                            // if not serializable or no intermediate agg, use pre-clustered group-by
                            List<Pair<LogicalVariable, Mutable<ILogicalExpression>>> gbyList = gby.getGroupByList();
                            List<LogicalVariable> columnList = new ArrayList<LogicalVariable>(gbyList.size());
                            for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : gbyList) {
                                ILogicalExpression expr = p.second.getValue();
                                if (expr.getExpressionTag() == LogicalExpressionTag.VARIABLE) {
                                    VariableReferenceExpression varRef = (VariableReferenceExpression) expr;
                                    columnList.add(varRef.getVariableReference());
                                }
                            }
                            op.setPhysicalOperator(new PreclusteredGroupByPOperator(columnList, gby.isGroupAll()));
                        }
                    }
                } else if (((AbstractLogicalOperator) (r0.getValue())).getOperatorTag().equals(LogicalOperatorTag.RUNNINGAGGREGATE)) {
                    List<Pair<LogicalVariable, Mutable<ILogicalExpression>>> gbyList = gby.getGroupByList();
                    List<LogicalVariable> columnList = new ArrayList<LogicalVariable>(gbyList.size());
                    for (Pair<LogicalVariable, Mutable<ILogicalExpression>> p : gbyList) {
                        ILogicalExpression expr = p.second.getValue();
                        if (expr.getExpressionTag() == LogicalExpressionTag.VARIABLE) {
                            VariableReferenceExpression varRef = (VariableReferenceExpression) expr;
                            columnList.add(varRef.getVariableReference());
                        }
                    }
                    op.setPhysicalOperator(new PreclusteredGroupByPOperator(columnList, gby.isGroupAll()));
                } else {
                    throw new AlgebricksException("Unsupported nested operator within a group-by: " + ((AbstractLogicalOperator) (r0.getValue())).getOperatorTag().name());
                }
            }
        }
    }
    if (op.getPhysicalOperator() == null) {
        switch(op.getOperatorTag()) {
            case INNERJOIN:
                {
                    JoinUtils.setJoinAlgorithmAndExchangeAlgo((InnerJoinOperator) op, context);
                    break;
                }
            case LEFTOUTERJOIN:
                {
                    JoinUtils.setJoinAlgorithmAndExchangeAlgo((LeftOuterJoinOperator) op, context);
                    break;
                }
            case UNNEST_MAP:
            case LEFT_OUTER_UNNEST_MAP:
                {
                    ILogicalExpression unnestExpr = null;
                    unnestExpr = ((AbstractUnnestMapOperator) op).getExpressionRef().getValue();
                    if (unnestExpr.getExpressionTag() == LogicalExpressionTag.FUNCTION_CALL) {
                        AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) unnestExpr;
                        FunctionIdentifier fid = f.getFunctionIdentifier();
                        if (!fid.equals(BuiltinFunctions.INDEX_SEARCH)) {
                            throw new IllegalStateException();
                        }
                        AccessMethodJobGenParams jobGenParams = new AccessMethodJobGenParams();
                        jobGenParams.readFromFuncArgs(f.getArguments());
                        MetadataProvider mp = (MetadataProvider) context.getMetadataProvider();
                        DataSourceId dataSourceId = new DataSourceId(jobGenParams.getDataverseName(), jobGenParams.getDatasetName());
                        Dataset dataset = mp.findDataset(jobGenParams.getDataverseName(), jobGenParams.getDatasetName());
                        IDataSourceIndex<String, DataSourceId> dsi = mp.findDataSourceIndex(jobGenParams.getIndexName(), dataSourceId);
                        INodeDomain storageDomain = mp.findNodeDomain(dataset.getNodeGroupName());
                        if (dsi == null) {
                            throw new AlgebricksException("Could not find index " + jobGenParams.getIndexName() + " for dataset " + dataSourceId);
                        }
                        IndexType indexType = jobGenParams.getIndexType();
                        boolean requiresBroadcast = jobGenParams.getRequiresBroadcast();
                        switch(indexType) {
                            case BTREE:
                                {
                                    BTreeJobGenParams btreeJobGenParams = new BTreeJobGenParams();
                                    btreeJobGenParams.readFromFuncArgs(f.getArguments());
                                    op.setPhysicalOperator(new BTreeSearchPOperator(dsi, storageDomain, requiresBroadcast, btreeJobGenParams.isPrimaryIndex(), btreeJobGenParams.isEqCondition(), btreeJobGenParams.getLowKeyVarList(), btreeJobGenParams.getHighKeyVarList()));
                                    break;
                                }
                            case RTREE:
                                {
                                    op.setPhysicalOperator(new RTreeSearchPOperator(dsi, storageDomain, requiresBroadcast));
                                    break;
                                }
                            case SINGLE_PARTITION_WORD_INVIX:
                            case SINGLE_PARTITION_NGRAM_INVIX:
                                {
                                    op.setPhysicalOperator(new InvertedIndexPOperator(dsi, storageDomain, requiresBroadcast, false));
                                    break;
                                }
                            case LENGTH_PARTITIONED_WORD_INVIX:
                            case LENGTH_PARTITIONED_NGRAM_INVIX:
                                {
                                    op.setPhysicalOperator(new InvertedIndexPOperator(dsi, storageDomain, requiresBroadcast, true));
                                    break;
                                }
                            default:
                                {
                                    throw new NotImplementedException(indexType + " indexes are not implemented.");
                                }
                        }
                    }
                    break;
                }
        }
    }
    if (op.hasNestedPlans()) {
        AbstractOperatorWithNestedPlans nested = (AbstractOperatorWithNestedPlans) op;
        for (ILogicalPlan p : nested.getNestedPlans()) {
            setPhysicalOperators(p, context);
        }
    }
    for (Mutable<ILogicalOperator> opRef : op.getInputs()) {
        computeDefaultPhysicalOp((AbstractLogicalOperator) opRef.getValue(), context);
    }
}

Example 12 with FunctionIdentifier

use of org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier in project asterixdb by apache.

the class IntroduceLSMComponentFilterRule method analyzeCondition.

private AccessMethodAnalysisContext analyzeCondition(ILogicalExpression cond, IOptimizationContext context, IVariableTypeEnvironment typeEnvironment) throws AlgebricksException {
    AccessMethodAnalysisContext analysisCtx = new AccessMethodAnalysisContext();
    AbstractFunctionCallExpression funcExpr = (AbstractFunctionCallExpression) cond;
    FunctionIdentifier funcIdent = funcExpr.getFunctionIdentifier();
    if (funcIdent != AlgebricksBuiltinFunctions.OR) {
        analyzeFunctionExpr(funcExpr, analysisCtx, context, typeEnvironment);
        for (Mutable<ILogicalExpression> arg : funcExpr.getArguments()) {
            ILogicalExpression argExpr = arg.getValue();
            if (argExpr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
                continue;
            }
            analyzeFunctionExpr((AbstractFunctionCallExpression) argExpr, analysisCtx, context, typeEnvironment);
        }
    }
    return analysisCtx;
}

Example 13 with FunctionIdentifier

use of org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier in project asterixdb by apache.

the class IntroduceLSMComponentFilterRule method getFieldNameFromSubAssignTree.

private Pair<ARecordType, List<String>> getFieldNameFromSubAssignTree(IOptimizableFuncExpr optFuncExpr, AbstractLogicalOperator op, int varIndex, ARecordType recType) {
    AbstractLogicalExpression expr = null;
    if (op.getOperatorTag() == LogicalOperatorTag.ASSIGN) {
        AssignOperator assignOp = (AssignOperator) op;
        expr = (AbstractLogicalExpression) assignOp.getExpressions().get(varIndex).getValue();
    }
    if (expr == null || expr.getExpressionTag() != LogicalExpressionTag.FUNCTION_CALL) {
        return null;
    }
    AbstractFunctionCallExpression funcExpr = (AbstractFunctionCallExpression) expr;
    FunctionIdentifier funcIdent = funcExpr.getFunctionIdentifier();
    if (funcIdent == BuiltinFunctions.FIELD_ACCESS_BY_NAME || funcIdent == BuiltinFunctions.FIELD_ACCESS_BY_INDEX) {
        //get the variable from here. Figure out which input it came from. Go to that input!!!
        ArrayList<LogicalVariable> usedVars = new ArrayList<>();
        expr.getUsedVariables(usedVars);
        LogicalVariable usedVar = usedVars.get(0);
        List<String> returnList = new ArrayList<>();
        //Find the input that it came from
        for (int varCheck = 0; varCheck < op.getInputs().size(); varCheck++) {
            AbstractLogicalOperator nestedOp = (AbstractLogicalOperator) op.getInputs().get(varCheck).getValue();
            if (nestedOp.getOperatorTag() != LogicalOperatorTag.ASSIGN) {
                if (varCheck == op.getInputs().size() - 1) {
                }
            } else {
                int nestedAssignVar = ((AssignOperator) nestedOp).getVariables().indexOf(usedVar);
                if (nestedAssignVar == -1) {
                    continue;
                }
                //get the nested info from the lower input
                Pair<ARecordType, List<String>> lowerInfo = getFieldNameFromSubAssignTree(optFuncExpr, (AbstractLogicalOperator) op.getInputs().get(varCheck).getValue(), nestedAssignVar, recType);
                if (lowerInfo != null) {
                    recType = lowerInfo.first;
                    returnList = lowerInfo.second;
                }
            }
        }
        if (funcIdent == BuiltinFunctions.FIELD_ACCESS_BY_NAME) {
            String fieldName = ConstantExpressionUtil.getStringArgument(funcExpr, 1);
            if (fieldName == null) {
                return null;
            }
            returnList.add(fieldName);
            return new Pair<>(recType, returnList);
        } else if (funcIdent == BuiltinFunctions.FIELD_ACCESS_BY_INDEX) {
            Integer fieldIndex = ConstantExpressionUtil.getIntArgument(funcExpr, 1);
            if (fieldIndex == null) {
                return null;
            }
            returnList.add(recType.getFieldNames()[fieldIndex]);
            IAType subType = recType.getFieldTypes()[fieldIndex];
            if (subType.getTypeTag() == ATypeTag.OBJECT) {
                recType = (ARecordType) subType;
            }
            return new Pair<>(recType, returnList);
        }
    }
    ILogicalExpression argExpr = funcExpr.getArguments().get(0).getValue();
    if (argExpr.getExpressionTag() != LogicalExpressionTag.VARIABLE) {
        return null;
    }
    return null;
}

Example 14 with FunctionIdentifier

use of org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier in project asterixdb by apache.

the class IntroduceLSMComponentFilterRule method findMacthedExprFieldName.

private boolean findMacthedExprFieldName(IOptimizableFuncExpr optFuncExpr, AbstractLogicalOperator op, Dataset dataset, ARecordType recType, List<Index> datasetIndexes, IOptimizationContext context) throws AlgebricksException {
    AbstractLogicalOperator descendantOp = (AbstractLogicalOperator) op.getInputs().get(0).getValue();
    while (descendantOp != null) {
        if (descendantOp.getOperatorTag() == LogicalOperatorTag.ASSIGN) {
            AssignOperator assignOp = (AssignOperator) descendantOp;
            List<LogicalVariable> varList = assignOp.getVariables();
            for (int varIndex = 0; varIndex < varList.size(); varIndex++) {
                LogicalVariable var = varList.get(varIndex);
                int funcVarIndex = optFuncExpr.findLogicalVar(var);
                if (funcVarIndex == -1) {
                    continue;
                }
                List<String> fieldName = getFieldNameFromSubAssignTree(optFuncExpr, descendantOp, varIndex, recType).second;
                if (fieldName == null) {
                    return false;
                }
                optFuncExpr.setFieldName(funcVarIndex, fieldName);
                return true;
            }
        } else if (descendantOp.getOperatorTag() == LogicalOperatorTag.DATASOURCESCAN) {
            DataSourceScanOperator scanOp = (DataSourceScanOperator) descendantOp;
            List<LogicalVariable> varList = scanOp.getVariables();
            for (int varIndex = 0; varIndex < varList.size(); varIndex++) {
                LogicalVariable var = varList.get(varIndex);
                int funcVarIndex = optFuncExpr.findLogicalVar(var);
                if (funcVarIndex == -1) {
                    continue;
                }
                // The variable value is one of the partitioning fields.
                List<String> fieldName = dataset.getPrimaryKeys().get(varIndex);
                if (fieldName == null) {
                    return false;
                }
                optFuncExpr.setFieldName(funcVarIndex, fieldName);
                return true;
            }
        } else if (descendantOp.getOperatorTag() == LogicalOperatorTag.UNNEST_MAP) {
            UnnestMapOperator unnestMapOp = (UnnestMapOperator) descendantOp;
            List<LogicalVariable> varList = unnestMapOp.getVariables();
            for (int varIndex = 0; varIndex < varList.size(); varIndex++) {
                LogicalVariable var = varList.get(varIndex);
                int funcVarIndex = optFuncExpr.findLogicalVar(var);
                if (funcVarIndex == -1) {
                    continue;
                }
                String indexName;
                Index index = null;
                ILogicalExpression unnestExpr = unnestMapOp.getExpressionRef().getValue();
                if (unnestExpr.getExpressionTag() == LogicalExpressionTag.FUNCTION_CALL) {
                    AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) unnestExpr;
                    FunctionIdentifier fid = f.getFunctionIdentifier();
                    if (!fid.equals(BuiltinFunctions.INDEX_SEARCH)) {
                        throw new IllegalStateException();
                    }
                    AccessMethodJobGenParams jobGenParams = new AccessMethodJobGenParams();
                    jobGenParams.readFromFuncArgs(f.getArguments());
                    indexName = jobGenParams.indexName;
                    for (Index idx : datasetIndexes) {
                        if (idx.getIndexName().compareTo(indexName) == 0) {
                            index = idx;
                            break;
                        }
                    }
                }
                IAType metaItemType = ((MetadataProvider) context.getMetadataProvider()).findType(dataset.getMetaItemTypeDataverseName(), dataset.getMetaItemTypeName());
                ARecordType metaRecType = (ARecordType) metaItemType;
                int numSecondaryKeys = KeyFieldTypeUtil.getNumSecondaryKeys(index, recType, metaRecType);
                List<String> fieldName;
                if (varIndex >= numSecondaryKeys) {
                    fieldName = dataset.getPrimaryKeys().get(varIndex - numSecondaryKeys);
                } else {
                    fieldName = index.getKeyFieldNames().get(varIndex);
                }
                if (fieldName == null) {
                    return false;
                }
                optFuncExpr.setFieldName(funcVarIndex, fieldName);
                return true;
            }
        }
        if (descendantOp.getInputs().isEmpty()) {
            break;
        }
        descendantOp = (AbstractLogicalOperator) descendantOp.getInputs().get(0).getValue();
    }
    return false;
}

Example 15 with FunctionIdentifier

use of org.apache.hyracks.algebricks.core.algebra.functions.FunctionIdentifier in project asterixdb by apache.

the class IntroduceLSMComponentFilterRule method getDataset.

private Dataset getDataset(AbstractLogicalOperator op, IOptimizationContext context) throws AlgebricksException {
    AbstractLogicalOperator descendantOp = (AbstractLogicalOperator) op.getInputs().get(0).getValue();
    while (descendantOp != null) {
        if (descendantOp.getOperatorTag() == LogicalOperatorTag.DATASOURCESCAN) {
            DataSourceScanOperator dataSourceScanOp = (DataSourceScanOperator) descendantOp;
            DataSource ds = (DataSource) dataSourceScanOp.getDataSource();
            if (ds.getDatasourceType() != DataSource.Type.INTERNAL_DATASET) {
                return null;
            }
            return ((DatasetDataSource) ds).getDataset();
        } else if (descendantOp.getOperatorTag() == LogicalOperatorTag.UNNEST_MAP) {
            UnnestMapOperator unnestMapOp = (UnnestMapOperator) descendantOp;
            ILogicalExpression unnestExpr = unnestMapOp.getExpressionRef().getValue();
            if (unnestExpr.getExpressionTag() == LogicalExpressionTag.FUNCTION_CALL) {
                AbstractFunctionCallExpression f = (AbstractFunctionCallExpression) unnestExpr;
                FunctionIdentifier fid = f.getFunctionIdentifier();
                String dataverseName;
                String datasetName;
                if (BuiltinFunctions.EXTERNAL_LOOKUP.equals(fid)) {
                    dataverseName = AccessMethodUtils.getStringConstant(f.getArguments().get(0));
                    datasetName = AccessMethodUtils.getStringConstant(f.getArguments().get(1));
                } else if (fid.equals(BuiltinFunctions.INDEX_SEARCH)) {
                    AccessMethodJobGenParams jobGenParams = new AccessMethodJobGenParams();
                    jobGenParams.readFromFuncArgs(f.getArguments());
                    dataverseName = jobGenParams.dataverseName;
                    datasetName = jobGenParams.datasetName;
                } else {
                    throw new AlgebricksException("Unexpected function for Unnest Map: " + fid);
                }
                return ((MetadataProvider) context.getMetadataProvider()).findDataset(dataverseName, datasetName);
            }
        }
        if (descendantOp.getInputs().isEmpty()) {
            break;
        }
        descendantOp = (AbstractLogicalOperator) descendantOp.getInputs().get(0).getValue();
    }
    return null;
}