Search in sources :

Example 1 with PTFTranslator

use of org.apache.hadoop.hive.ql.parse.PTFTranslator in project hive by apache.

the class HiveProjectVisitor method genPTF.

private OpAttr genPTF(OpAttr inputOpAf, WindowingSpec wSpec) throws SemanticException {
    Operator<?> input = inputOpAf.inputs.get(0);
    wSpec.validateAndMakeEffective();
    WindowingComponentizer groups = new WindowingComponentizer(wSpec);
    RowResolver rr = new RowResolver();
    for (ColumnInfo ci : input.getSchema().getSignature()) {
        rr.put(inputOpAf.tabAlias, ci.getInternalName(), ci);
    }
    while (groups.hasNext()) {
        wSpec = groups.next(hiveOpConverter.getHiveConf(), hiveOpConverter.getSemanticAnalyzer(), hiveOpConverter.getUnparseTranslator(), rr);
        // 1. Create RS and backtrack Select operator on top
        ArrayList<ExprNodeDesc> keyCols = new ArrayList<ExprNodeDesc>();
        ArrayList<ExprNodeDesc> partCols = new ArrayList<ExprNodeDesc>();
        StringBuilder order = new StringBuilder();
        StringBuilder nullOrder = new StringBuilder();
        for (PartitionExpression partCol : wSpec.getQueryPartitionSpec().getExpressions()) {
            ExprNodeDesc partExpr = hiveOpConverter.getSemanticAnalyzer().genExprNodeDesc(partCol.getExpression(), rr);
            if (ExprNodeDescUtils.indexOf(partExpr, partCols) < 0) {
                keyCols.add(partExpr);
                partCols.add(partExpr);
                order.append('+');
                nullOrder.append('a');
            }
        }
        if (wSpec.getQueryOrderSpec() != null) {
            for (OrderExpression orderCol : wSpec.getQueryOrderSpec().getExpressions()) {
                ExprNodeDesc orderExpr = hiveOpConverter.getSemanticAnalyzer().genExprNodeDesc(orderCol.getExpression(), rr);
                char orderChar = orderCol.getOrder() == PTFInvocationSpec.Order.ASC ? '+' : '-';
                char nullOrderChar = orderCol.getNullOrder() == PTFInvocationSpec.NullOrder.NULLS_FIRST ? 'a' : 'z';
                int index = ExprNodeDescUtils.indexOf(orderExpr, keyCols);
                if (index >= 0) {
                    order.setCharAt(index, orderChar);
                    nullOrder.setCharAt(index, nullOrderChar);
                    continue;
                }
                keyCols.add(orderExpr);
                order.append(orderChar);
                nullOrder.append(nullOrderChar);
            }
        }
        SelectOperator selectOp = genReduceSinkAndBacktrackSelect(input, keyCols.toArray(new ExprNodeDesc[keyCols.size()]), 0, partCols, order.toString(), nullOrder.toString(), -1, Operation.NOT_ACID, hiveOpConverter.getHiveConf());
        // 2. Finally create PTF
        PTFTranslator translator = new PTFTranslator();
        PTFDesc ptfDesc = translator.translate(wSpec, hiveOpConverter.getSemanticAnalyzer(), hiveOpConverter.getHiveConf(), rr, hiveOpConverter.getUnparseTranslator());
        RowResolver ptfOpRR = ptfDesc.getFuncDef().getOutputShape().getRr();
        Operator<?> ptfOp = OperatorFactory.getAndMakeChild(ptfDesc, new RowSchema(ptfOpRR.getColumnInfos()), selectOp);
        if (LOG.isDebugEnabled()) {
            LOG.debug("Generated " + ptfOp + " with row schema: [" + ptfOp.getSchema() + "]");
        }
        // 3. Prepare for next iteration (if any)
        rr = ptfOpRR;
        input = ptfOp;
    }
    return inputOpAf.clone(input);
}
Also used : RowSchema(org.apache.hadoop.hive.ql.exec.RowSchema) OrderExpression(org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.OrderExpression) ArrayList(java.util.ArrayList) PTFTranslator(org.apache.hadoop.hive.ql.parse.PTFTranslator) ColumnInfo(org.apache.hadoop.hive.ql.exec.ColumnInfo) RowResolver(org.apache.hadoop.hive.ql.parse.RowResolver) SelectOperator(org.apache.hadoop.hive.ql.exec.SelectOperator) PartitionExpression(org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.PartitionExpression) PTFDesc(org.apache.hadoop.hive.ql.plan.PTFDesc) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) WindowingComponentizer(org.apache.hadoop.hive.ql.parse.WindowingComponentizer)

Example 2 with PTFTranslator

use of org.apache.hadoop.hive.ql.parse.PTFTranslator in project hive by apache.

the class HiveOpConverter method genPTF.

private OpAttr genPTF(OpAttr inputOpAf, WindowingSpec wSpec) throws SemanticException {
    Operator<?> input = inputOpAf.inputs.get(0);
    wSpec.validateAndMakeEffective();
    WindowingComponentizer groups = new WindowingComponentizer(wSpec);
    RowResolver rr = new RowResolver();
    for (ColumnInfo ci : input.getSchema().getSignature()) {
        rr.put(inputOpAf.tabAlias, ci.getInternalName(), ci);
    }
    while (groups.hasNext()) {
        wSpec = groups.next(hiveConf, semanticAnalyzer, unparseTranslator, rr);
        // 1. Create RS and backtrack Select operator on top
        ArrayList<ExprNodeDesc> keyCols = new ArrayList<ExprNodeDesc>();
        ArrayList<ExprNodeDesc> partCols = new ArrayList<ExprNodeDesc>();
        StringBuilder order = new StringBuilder();
        StringBuilder nullOrder = new StringBuilder();
        for (PartitionExpression partCol : wSpec.getQueryPartitionSpec().getExpressions()) {
            ExprNodeDesc partExpr = semanticAnalyzer.genExprNodeDesc(partCol.getExpression(), rr);
            if (ExprNodeDescUtils.indexOf(partExpr, partCols) < 0) {
                keyCols.add(partExpr);
                partCols.add(partExpr);
                order.append('+');
                nullOrder.append('a');
            }
        }
        if (wSpec.getQueryOrderSpec() != null) {
            for (OrderExpression orderCol : wSpec.getQueryOrderSpec().getExpressions()) {
                ExprNodeDesc orderExpr = semanticAnalyzer.genExprNodeDesc(orderCol.getExpression(), rr);
                char orderChar = orderCol.getOrder() == PTFInvocationSpec.Order.ASC ? '+' : '-';
                char nullOrderChar = orderCol.getNullOrder() == PTFInvocationSpec.NullOrder.NULLS_FIRST ? 'a' : 'z';
                int index = ExprNodeDescUtils.indexOf(orderExpr, keyCols);
                if (index >= 0) {
                    order.setCharAt(index, orderChar);
                    nullOrder.setCharAt(index, nullOrderChar);
                    continue;
                }
                keyCols.add(orderExpr);
                order.append(orderChar);
                nullOrder.append(nullOrderChar);
            }
        }
        SelectOperator selectOp = genReduceSinkAndBacktrackSelect(input, keyCols.toArray(new ExprNodeDesc[keyCols.size()]), 0, partCols, order.toString(), nullOrder.toString(), -1, Operation.NOT_ACID, hiveConf);
        // 2. Finally create PTF
        PTFTranslator translator = new PTFTranslator();
        PTFDesc ptfDesc = translator.translate(wSpec, semanticAnalyzer, hiveConf, rr, unparseTranslator);
        RowResolver ptfOpRR = ptfDesc.getFuncDef().getOutputShape().getRr();
        Operator<?> ptfOp = OperatorFactory.getAndMakeChild(ptfDesc, new RowSchema(ptfOpRR.getColumnInfos()), selectOp);
        if (LOG.isDebugEnabled()) {
            LOG.debug("Generated " + ptfOp + " with row schema: [" + ptfOp.getSchema() + "]");
        }
        // 3. Prepare for next iteration (if any)
        rr = ptfOpRR;
        input = ptfOp;
    }
    return inputOpAf.clone(input);
}
Also used : RowSchema(org.apache.hadoop.hive.ql.exec.RowSchema) OrderExpression(org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.OrderExpression) ArrayList(java.util.ArrayList) PTFTranslator(org.apache.hadoop.hive.ql.parse.PTFTranslator) ColumnInfo(org.apache.hadoop.hive.ql.exec.ColumnInfo) RowResolver(org.apache.hadoop.hive.ql.parse.RowResolver) SelectOperator(org.apache.hadoop.hive.ql.exec.SelectOperator) PartitionExpression(org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.PartitionExpression) PTFDesc(org.apache.hadoop.hive.ql.plan.PTFDesc) ExprNodeDesc(org.apache.hadoop.hive.ql.plan.ExprNodeDesc) WindowingComponentizer(org.apache.hadoop.hive.ql.parse.WindowingComponentizer)

Aggregations

ArrayList (java.util.ArrayList)2 ColumnInfo (org.apache.hadoop.hive.ql.exec.ColumnInfo)2 RowSchema (org.apache.hadoop.hive.ql.exec.RowSchema)2 SelectOperator (org.apache.hadoop.hive.ql.exec.SelectOperator)2 OrderExpression (org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.OrderExpression)2 PartitionExpression (org.apache.hadoop.hive.ql.parse.PTFInvocationSpec.PartitionExpression)2 PTFTranslator (org.apache.hadoop.hive.ql.parse.PTFTranslator)2 RowResolver (org.apache.hadoop.hive.ql.parse.RowResolver)2 WindowingComponentizer (org.apache.hadoop.hive.ql.parse.WindowingComponentizer)2 ExprNodeDesc (org.apache.hadoop.hive.ql.plan.ExprNodeDesc)2 PTFDesc (org.apache.hadoop.hive.ql.plan.PTFDesc)2