Examples of org.apache.hadoop.hive.ql.exec.GroupByOperator

org.apache.hadoop.hive.ql.exec.GroupByOperator
GroupBy operator implementation.

   * Node Processor for Column Pruning on Group By Operators.
   */
  public static class ColumnPrunerGroupByProc implements NodeProcessor {
    public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx ctx,
        Object... nodeOutputs) throws SemanticException {
      GroupByOperator op = (GroupByOperator) nd;
      ColumnPrunerProcCtx cppCtx = (ColumnPrunerProcCtx) ctx;
      List<String> colLists = new ArrayList<String>();
      GroupByDesc conf = op.getConf();
      ArrayList<ExprNodeDesc> keys = conf.getKeys();
      for (ExprNodeDesc key : keys) {
        colLists = Utilities.mergeUniqElems(colLists, key.getCols());
      }

View Full Code Here

    @Override
    public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
        Object... nodeOutputs) throws SemanticException {


      LineageCtx lctx = (LineageCtx)procCtx;
      GroupByOperator gop = (GroupByOperator)nd;
      ArrayList<ColumnInfo> col_infos = gop.getSchema().getSignature();
      Operator<? extends OperatorDesc> inpOp = getParent(stack);
      int cnt = 0;


      for(ExprNodeDesc expr : gop.getConf().getKeys()) {
        lctx.getIndex().putDependency(gop, col_infos.get(cnt++),
            ExprProcFactory.getExprDependency(lctx, inpOp, expr));
      }


      for(AggregationDesc agg : gop.getConf().getAggregators()) {
        // Concatenate the dependencies of all the parameters to
        // create the new dependency
        Dependency dep = new Dependency();
        DependencyType new_type = LineageInfo.DependencyType.EXPRESSION;
        // TODO: Get the actual string here.

View Full Code Here

      Operator<?> child = CorrelationUtilities.getSingleChild(cRS);
      if (child instanceof JoinOperator) {
        return false; // not supported
      }
      if (child instanceof GroupByOperator) {
        GroupByOperator cGBY = (GroupByOperator) child;
        if (!CorrelationUtilities.hasGroupingSet(cRS) && !cGBY.getConf().isGroupingSetsPresent()) {
          return process(cRS, cGBY, dedupCtx);
        }
        return false;
      }
      if (child instanceof ExtractOperator) {

View Full Code Here


    // pRS-pGBY-cRS
    @Override
    public Object process(ReduceSinkOperator cRS, ReduceSinkDeduplicateProcCtx dedupCtx)
        throws SemanticException {
      GroupByOperator pGBY =
          CorrelationUtilities.findPossibleParent(
              cRS, GroupByOperator.class, dedupCtx.trustScript());
      if (pGBY == null) {
        return false;
      }

View Full Code Here

    @Override
    public Object process(ReduceSinkOperator cRS, GroupByOperator cGBY,
        ReduceSinkDeduplicateProcCtx dedupCtx)
        throws SemanticException {
      Operator<?> start = CorrelationUtilities.getStartForGroupBy(cRS);
      GroupByOperator pGBY =
          CorrelationUtilities.findPossibleParent(
              start, GroupByOperator.class, dedupCtx.trustScript());
      if (pGBY == null) {
        return false;
      }

View Full Code Here

   */
  private static class CheckGroupByProc implements NodeProcessor {


     public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx ctx,
         Object... nodeOutputs) throws SemanticException {
       GroupByOperator operator = (GroupByOperator)nd;
       canApplyCtx = (RewriteCanApplyCtx)ctx;
       //for each group-by clause in query, only one GroupByOperator of the
       //GBY-RS-GBY sequence is stored in  getGroupOpToInputTables
       //we need to process only this operator
       //Also, we do not rewrite for cases when same query branch has multiple group-by constructs
       if(canApplyCtx.getParseContext().getGroupOpToInputTables().containsKey(operator) &&
           !canApplyCtx.isQueryHasGroupBy()){


         canApplyCtx.setQueryHasGroupBy(true);
         GroupByDesc conf = (GroupByDesc) operator.getConf();
         List<AggregationDesc> aggrList = conf.getAggregators();
         if(aggrList != null && aggrList.size() > 0){
             for (AggregationDesc aggregationDesc : aggrList) {
               canApplyCtx.setAggFuncCnt(canApplyCtx.getAggFuncCnt() + 1);
               //In the current implementation, we do not support more than 1 agg funcs in group-by

View Full Code Here

        // distinct columns will be added to the key columns.
        boolean isCorrelated = sameKeys(rsKeyCols, backtrackedKeyCols) &&
            sameOrder(rsop.getConf().getOrder(), childRSOrder) &&
            sameKeys(backtrackedPartitionCols, rsPartitionCols) &&
            correlation.adjustNumReducers(rsop.getConf().getNumReducers());
        GroupByOperator cGBY =
            CorrelationUtilities.getSingleChild(rsop, GroupByOperator.class);
        if (cGBY != null) {
          if (CorrelationUtilities.hasGroupingSet(rsop) ||
              cGBY.getConf().isGroupingSetsPresent()) {
            // Do not support grouping set right now
            isCorrelated = false;
          }
        }

View Full Code Here

      for (ExprNodeDesc key : partitionCols) {
        if (!(key instanceof ExprNodeColumnDesc)) {
          shouldDetect = false;
        }
      }
      GroupByOperator cGBY =
          CorrelationUtilities.getSingleChild(op, GroupByOperator.class);
      if (cGBY != null) {
        if (CorrelationUtilities.hasGroupingSet(op) ||
            cGBY.getConf().isGroupingSetsPresent()) {
          // Do not support grouping set right now
          shouldDetect = false;
        }
      }

View Full Code Here

  public static class MultiGroupByInferrer extends GroupByInferrer implements NodeProcessor {
    @Override
    public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
        Object... nodeOutputs) throws SemanticException {
      BucketingSortingCtx bctx = (BucketingSortingCtx)procCtx;
      GroupByOperator gop = (GroupByOperator)nd;


      if (gop.getParentOperators().size() != 1) {
        return null;
      }


      Operator<? extends OperatorDesc> fop =  gop.getParentOperators().get(0);


      // The caller of this method should guarantee this
      assert(fop instanceof ForwardOperator);


      if (fop.getParentOperators().size() != 1) {

View Full Code Here

    @Override
    public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
        Object... nodeOutputs) throws SemanticException {


      BucketingSortingCtx bctx = (BucketingSortingCtx)procCtx;
      GroupByOperator gop = (GroupByOperator)nd;


      // As of writing this, there is no case where this could be false, this is just protection
      // from possible future changes
      if (gop.getParentOperators().size() != 1) {
        return null;
      }


      Operator<? extends OperatorDesc> rop = gop.getParentOperators().get(0);


      // The caller of this method should guarantee this
      assert(rop instanceof ReduceSinkOperator);


      processGroupByReduceSink((ReduceSinkOperator) rop, gop, bctx);

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of org.apache.hadoop.hive.ql.exec.GroupByOperator

javolution.util.FastBitSet

org.apache.hadoop.hive.ql.optimizer.ColumnPrunerProcFactory$ColumnPrunerGroupByProc

org.apache.hadoop.hive.ql.optimizer.ConstantPropagateProcFactory$ConstantPropagateGroupByProc

org.apache.hadoop.hive.ql.optimizer.correlation.CorrelationOptimizer$CorrelationNodeProc

org.apache.hadoop.hive.ql.optimizer.correlation.CorrelationUtilities

org.apache.hadoop.hive.ql.optimizer.correlation.QueryPlanTreeTransformation

org.apache.hadoop.hive.ql.optimizer.correlation.ReduceSinkDeDuplication$AbsctractReducerReducerProc

org.apache.hadoop.hive.ql.optimizer.correlation.ReduceSinkDeDuplication$GroupbyReducerProc

org.apache.hadoop.hive.ql.optimizer.DynamicPartitionPruningOptimization

org.apache.hadoop.hive.ql.optimizer.GroupByOptimizer$BucketGroupByProcessor

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.