Package org.apache.hadoop.hive.ql.exec

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

3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
          }
        }
      }       
    }

    ReduceSinkOperator rsOp = (ReduceSinkOperator)  putOpInsertMap(
      OperatorFactory.getAndMakeChild(PlanUtils.getReduceSinkDesc(reduceKeys, reduceValues, outputColumnNames, true, -1, reduceKeys.size(), -1),
                                      new RowSchema(reduceSinkOutputRowResolver.getColumnInfos()), input),
      reduceSinkOutputRowResolver);
   
    rsOp.setColumnExprMap(colExprMap);
    return rsOp;
  }
View Full Code Here
196
197
198
199
200
201
202
203
204
205
206
207
      LineageCtx lCtx = (LineageCtx) procCtx;
      JoinOperator op = (JoinOperator)nd;
      JoinDesc jd = op.getConf();

      // The input operator to the join is always a reduce sink operator
      ReduceSinkOperator inpOp = (ReduceSinkOperator)getParent(stack);
      ReduceSinkDesc rd = inpOp.getConf();
      int tag = rd.getTag();

      // Iterate over the outputs of the join operator and merge the
      // dependencies of the columns that corresponding to the tag.
      int cnt = 0;
View Full Code Here
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
      // be called for leafs.
      assert(!stack.isEmpty());

      // LineageCtx
      LineageCtx lCtx = (LineageCtx) procCtx;
      ReduceSinkOperator rop = (ReduceSinkOperator)nd;

      ArrayList<ColumnInfo> col_infos = rop.getSchema().getSignature();
      Operator<? extends Serializable> inpOp = getParent(stack);
      int cnt = 0;

      // The keys are included only in case the reduce sink feeds into
      // a group by operator through a chain of forward operators
      Operator<? extends Serializable> op = rop.getChildOperators().get(0);
      while (op instanceof ForwardOperator) {
        op = op.getChildOperators().get(0);
      }

      if (op instanceof GroupByOperator) {
        for(ExprNodeDesc expr : rop.getConf().getKeyCols()) {
          lCtx.getIndex().putDependency(rop, col_infos.get(cnt++),
              ExprProcFactory.getExprDependency(lCtx, inpOp, expr));
        }
      }

      for(ExprNodeDesc expr : rop.getConf().getValueCols()) {
        lCtx.getIndex().putDependency(rop, col_infos.get(cnt++),
            ExprProcFactory.getExprDependency(lCtx, inpOp, expr));
      }

      return null;
View Full Code Here
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
      pos++;
    }

    // get the join keys from old parent ReduceSink operators
    for (pos = 0; pos < newParentOps.size(); pos++) {
      ReduceSinkOperator oldPar = (ReduceSinkOperator) oldReduceSinkParentOps.get(pos);
      ReduceSinkDesc rsconf = oldPar.getConf();
      Byte tag = (byte) rsconf.getTag();
      List<ExprNodeDesc> keys = rsconf.getKeyCols();
      keyExprMap.put(tag, keys);

      // set column transfer
      HashMap<String, ExprNodeDesc> map = (HashMap<String, ExprNodeDesc>) oldPar.getColumnExprMap();
      columnTransfer.put(tag, map);
    }

    // create the map-join operator
    for (pos = 0; pos < newParentOps.size(); pos++) {
View Full Code Here
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
      @Override
      public Object process(Node nd, Stack<Node> stack,
          NodeProcessorCtx procCtx, Object... nodeOutputs)
          throws SemanticException {
        ReduceSinkDeduplicateProcCtx ctx = (ReduceSinkDeduplicateProcCtx) procCtx;
        ReduceSinkOperator childReduceSink = (ReduceSinkOperator)nd;
       
        if(ctx.contains(childReduceSink)) {
          return null;
        }
       
        ParseContext pGraphContext = ctx.getPctx();
        HashMap<String, String> childColumnMapping = getPartitionAndKeyColumnMapping(childReduceSink);
        ReduceSinkOperator parentRS = null;
        parentRS = findSingleParentReduceSink(childReduceSink, pGraphContext);
        if (parentRS == null) {
          ctx.addRejectedReduceSinkOperator(childReduceSink);
          return null;
        }
        HashMap<String, String> parentColumnMapping = getPartitionAndKeyColumnMapping(parentRS);
        Operator<? extends Serializable> stopBacktrackFlagOp = null;
        if (parentRS.getParentOperators() == null
            || parentRS.getParentOperators().size() == 0) {
          stopBacktrackFlagOp =  parentRS;
        } else if (parentRS.getParentOperators().size() != 1) {
          return null;
        } else {
          stopBacktrackFlagOp = parentRS.getParentOperators().get(0);
        }
       
        boolean succeed = backTrackColumnNames(childColumnMapping, childReduceSink, stopBacktrackFlagOp, pGraphContext);
        if (!succeed) {
          return null;
View Full Code Here
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
   * @param opProcCtx
   *          context
   */
  public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx opProcCtx,
      Object... nodeOutputs) throws SemanticException {
    ReduceSinkOperator op = (ReduceSinkOperator) nd;
    GenMRProcContext ctx = (GenMRProcContext) opProcCtx;

    ctx.getParseCtx();

    // map-join consisted on a bunch of map-only jobs, and it has been split
    // after the mapjoin
    Operator<? extends Serializable> reducer = op.getChildOperators().get(0);
    Map<Operator<? extends Serializable>, GenMapRedCtx> mapCurrCtx = ctx
        .getMapCurrCtx();
    GenMapRedCtx mapredCtx = mapCurrCtx.get(op.getParentOperators().get(0));
    Task<? extends Serializable> currTask = mapredCtx.getCurrTask();
    MapredWork plan = (MapredWork) currTask.getWork();
    HashMap<Operator<? extends Serializable>, Task<? extends Serializable>> opTaskMap = ctx
        .getOpTaskMap();
    Task<? extends Serializable> opMapTask = opTaskMap.get(reducer);
View Full Code Here
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
   * @param opProcCtx
   *          context
   */
  public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx opProcCtx,
      Object... nodeOutputs) throws SemanticException {
    ReduceSinkOperator op = (ReduceSinkOperator) nd;
    GenMRProcContext ctx = (GenMRProcContext) opProcCtx;

    Map<Operator<? extends Serializable>, GenMapRedCtx> mapCurrCtx = ctx
        .getMapCurrCtx();
    GenMapRedCtx mapredCtx = mapCurrCtx.get(op.getParentOperators().get(0));
    Task<? extends Serializable> currTask = mapredCtx.getCurrTask();
    Operator<? extends Serializable> currTopOp = mapredCtx.getCurrTopOp();
    String currAliasId = mapredCtx.getCurrAliasId();
    Operator<? extends Serializable> reducer = op.getChildOperators().get(0);
    Map<Operator<? extends Serializable>, Task<? extends Serializable>> opTaskMap = ctx
        .getOpTaskMap();
    Task<? extends Serializable> opMapTask = opTaskMap.get(reducer);

    ctx.setCurrTopOp(currTopOp);
View Full Code Here
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
   * The Node Processor for Column Pruning on Reduce Sink Operators.
   */
  public static class ColumnPrunerReduceSinkProc implements NodeProcessor {
    public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx ctx,
        Object... nodeOutputs) throws SemanticException {
      ReduceSinkOperator op = (ReduceSinkOperator) nd;
      ColumnPrunerProcCtx cppCtx = (ColumnPrunerProcCtx) ctx;
      HashMap<Operator<? extends Serializable>, OpParseContext> opToParseCtxMap = cppCtx
          .getOpToParseCtxMap();
      RowResolver redSinkRR = opToParseCtxMap.get(op).getRowResolver();
      ReduceSinkDesc conf = op.getConf();
      List<Operator<? extends Serializable>> childOperators = op
          .getChildOperators();
      List<Operator<? extends Serializable>> parentOperators = op
          .getParentOperators();

      List<String> colLists = new ArrayList<String>();
      ArrayList<ExprNodeDesc> keys = conf.getKeyCols();
      for (ExprNodeDesc key : keys) {
        colLists = Utilities.mergeUniqElems(colLists, key.getCols());
      }

      if ((childOperators.size() == 1)
          && (childOperators.get(0) instanceof JoinOperator)) {
        assert parentOperators.size() == 1;
        Operator<? extends Serializable> par = parentOperators.get(0);
        JoinOperator childJoin = (JoinOperator) childOperators.get(0);
        RowResolver parRR = opToParseCtxMap.get(par).getRowResolver();
        List<String> childJoinCols = cppCtx.getJoinPrunedColLists().get(
            childJoin).get((byte) conf.getTag());
        boolean[] flags = new boolean[conf.getValueCols().size()];
        for (int i = 0; i < flags.length; i++) {
          flags[i] = false;
        }
        if (childJoinCols != null && childJoinCols.size() > 0) {
          Map<String, ExprNodeDesc> exprMap = op.getColumnExprMap();
          for (String childCol : childJoinCols) {
            ExprNodeDesc desc = exprMap.get(childCol);
            int index = conf.getValueCols().indexOf(desc);
            flags[index] = true;
            String[] nm = redSinkRR.reverseLookup(childCol);
View Full Code Here
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
   * @param opProcCtx
   *          context
   */
  public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx opProcCtx,
      Object... nodeOutputs) throws SemanticException {
    ReduceSinkOperator op = (ReduceSinkOperator) nd;
    GenMRProcContext ctx = (GenMRProcContext) opProcCtx;

    Map<Operator<? extends Serializable>, GenMapRedCtx> mapCurrCtx = ctx
        .getMapCurrCtx();
    GenMapRedCtx mapredCtx = mapCurrCtx.get(stack.get(stack.size() - 2));
    Task<? extends Serializable> currTask = mapredCtx.getCurrTask();
    MapredWork currPlan = (MapredWork) currTask.getWork();
    Operator<? extends Serializable> currTopOp = mapredCtx.getCurrTopOp();
    String currAliasId = mapredCtx.getCurrAliasId();
    Operator<? extends Serializable> reducer = op.getChildOperators().get(0);
    HashMap<Operator<? extends Serializable>, Task<? extends Serializable>> opTaskMap = ctx
        .getOpTaskMap();
    Task<? extends Serializable> opMapTask = opTaskMap.get(reducer);

    ctx.setCurrTopOp(currTopOp);
View Full Code Here
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
   * @param opProcCtx
   *          context
   */
  public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx opProcCtx,
      Object... nodeOutputs) throws SemanticException {
    ReduceSinkOperator op = (ReduceSinkOperator) nd;
    GenMRProcContext ctx = (GenMRProcContext) opProcCtx;

    ParseContext parseCtx = ctx.getParseCtx();
    UnionProcContext uCtx = parseCtx.getUCtx();

    // union was map only - no special processing needed
    if (uCtx.isMapOnlySubq()) {
      return (new GenMRRedSink1()).process(nd, stack, opProcCtx, nodeOutputs);
    }

    // union consisted on a bunch of map-reduce jobs, and it has been split at
    // the union
    Operator<? extends Serializable> reducer = op.getChildOperators().get(0);
    Map<Operator<? extends Serializable>, GenMapRedCtx> mapCurrCtx = ctx
        .getMapCurrCtx();
    GenMapRedCtx mapredCtx = mapCurrCtx.get(op.getParentOperators().get(0));
    Task<? extends Serializable> currTask = mapredCtx.getCurrTask();
    MapredWork plan = (MapredWork) currTask.getWork();
    HashMap<Operator<? extends Serializable>, Task<? extends Serializable>> opTaskMap = ctx
        .getOpTaskMap();
    Task<? extends Serializable> opMapTask = opTaskMap.get(reducer);
View Full Code Here
TOP

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

  • org.apache.hadoop.hive.ql.io.HiveKey
  • org.apache.hadoop.hive.ql.optimizer.BucketingSortingReduceSinkOptimizer$BucketSortReduceSinkProcessor
  • org.apache.hadoop.hive.ql.optimizer.ColumnPrunerProcFactory$ColumnPrunerReduceSinkProc
  • org.apache.hadoop.hive.ql.optimizer.ConstantPropagateProcFactory$ConstantPropagateReduceSinkProc
  • org.apache.hadoop.hive.ql.optimizer.ConvertJoinMapJoin
  • org.apache.hadoop.hive.ql.optimizer.correlation.CorrelationOptimizer$CorrelationNodeProc
  • 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.correlation.ReduceSinkDeDuplication$ReducerReducerProc
    • Copyright © 2018 www.massapicom. All rights reserved.
    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.