org.apache.calcite.rel.core.Join类的使用及代码示例

x33g5p2x  于2022-01-22 转载在 其他  
字(13.5k)|赞(0)|评价(0)|浏览(121)

本文整理了Java中org.apache.calcite.rel.core.Join类的一些代码示例，展示了Join类的具体用法。这些代码示例主要来源于Github/Stackoverflow/Maven等平台，是从一些精选项目中提取出来的代码，具有较强的参考意义，能在一定程度帮忙到你。Join类的具体详情如下：
包路径：org.apache.calcite.rel.core.Join
类名称：Join

Join介绍

[英]Relational expression that combines two relational expressions according to some condition.

Each output row has columns from the left and right inputs. The set of output rows is a subset of the cartesian product of the two inputs; precisely which subset depends on the join condition.
[中]根据某种条件组合两个关系表达式的关系表达式。
每个输出行都有来自左侧和右侧输入的列。输出行集是两个输入的笛卡尔乘积的子集；确切地说，哪个子集取决于连接条件。

代码示例

代码示例来源：origin: apache/hive

@Override
 public void onMatch(RelOptRuleCall call) {
  final Join join = call.rel(0);
  final RexBuilder rexBuilder = join.getCluster().getRexBuilder();
  final RexNode condition = RexUtil.pullFactors(rexBuilder, join.getCondition());
  RexNode newCondition = analyzeRexNode(rexBuilder, condition);
  // If we could not transform anything, we bail out
  if (newCondition.toString().equals(condition.toString())) {
   return;
  }
  RelNode newNode = join.copy(join.getTraitSet(),
    newCondition,
    join.getLeft(),
    join.getRight(),
    join.getJoinType(),
    join.isSemiJoinDone());
  call.transformTo(newNode);
 }
}

代码示例来源：origin: apache/hive

private Project swapInputs(Join join, Project topProject, RelBuilder builder) {
 RexBuilder rexBuilder = join.getCluster().getRexBuilder();
 int rightInputSize = join.getRight().getRowType().getFieldCount();
 int leftInputSize = join.getLeft().getRowType().getFieldCount();
 List<RelDataTypeField> joinFields = join.getRowType().getFieldList();
 RexNode newJoinCond = join.getCondition().accept(new RelOptUtil.RexInputConverter(rexBuilder, joinFields,
                                          joinFields, condAdjustments));
 Join swappedJoin = (Join)builder.push(join.getRight()).push(join.getLeft()).join(join.getJoinType(),
                                          newJoinCond).build();
 List<RelDataTypeField> swappedJoinFeilds = swappedJoin.getRowType().getFieldList();
 for(RexNode project:topProject.getProjects()) {
  RexNode newProject = project.accept(new RelOptUtil.RexInputConverter(rexBuilder, swappedJoinFeilds,

代码示例来源：origin: apache/hive

private static boolean validJoinParent(RelNode joinNode, RelNode parent) {
 boolean validParent = true;
 if (parent instanceof Join) {
  // In Hive AST, right child of join cannot be another join,
  // thus we need to introduce a project on top of it.
  // But we only need the additional project if the left child
  // is another join too; if it is not, ASTConverter will swap
  // the join inputs, leaving the join operator on the left.
  // we also do it if parent is HiveSemiJoin since ASTConverter won't
  // swap inputs then
  // This will help triggering multijoin recognition methods that
  // are embedded in SemanticAnalyzer.
  if (((Join) parent).getRight() == joinNode &&
     (((Join) parent).getLeft() instanceof Join || parent instanceof HiveSemiJoin) ) {
   validParent = false;
  }
 } else if (parent instanceof SetOp) {
  validParent = false;
 }
 return validParent;
}

代码示例来源：origin: apache/hive

/**
 *
 * @param j
 * @param additionalPredicate
 * @return if predicate is the join condition return (true, joinCond)
 * else return (false, minusPred)
 */
private Pair<Boolean,RexNode> getCombinedPredicateForJoin(Join j, RexNode additionalPredicate) {
 RexNode minusPred = RelMdUtil.minusPreds(j.getCluster().getRexBuilder(), additionalPredicate,
   j.getCondition());
 if (minusPred != null) {
  List<RexNode> minusList = new ArrayList<RexNode>();
  minusList.add(j.getCondition());
  minusList.add(minusPred);
  return new Pair<Boolean,RexNode>(false, minusPred);
 }
 return new Pair<Boolean,RexNode>(true,j.getCondition());
}

代码示例来源：origin: apache/hive

} else if (call.rel(0) instanceof Join) {
 Join join = call.rel(0);
 newOp = join.copy(join.getTraitSet(), join.getCondition(),
   newInputs.get(0), newInputs.get(1), join.getJoinType(),
   join.isSemiJoinDone());
} else {
 return;

代码示例来源：origin: org.apache.calcite/calcite-core

public void onMatch(RelOptRuleCall call) {
  final Delta delta = call.rel(0);
  Util.discard(delta);
  final Join join = call.rel(1);
  final RelNode left = join.getLeft();
  final RelNode right = join.getRight();
  final LogicalDelta rightWithDelta = LogicalDelta.create(right);
  final LogicalJoin joinL = LogicalJoin.create(left, rightWithDelta,
    join.getCondition(), join.getVariablesSet(), join.getJoinType(),
    join.isSemiJoinDone(),
    ImmutableList.copyOf(join.getSystemFieldList()));
  final LogicalDelta leftWithDelta = LogicalDelta.create(left);
  final LogicalJoin joinR = LogicalJoin.create(leftWithDelta, right,
    join.getCondition(), join.getVariablesSet(), join.getJoinType(),
    join.isSemiJoinDone(),
    ImmutableList.copyOf(join.getSystemFieldList()));
  List<RelNode> inputsToUnion = new ArrayList<>();
  inputsToUnion.add(joinL);
  inputsToUnion.add(joinR);
  final LogicalUnion newNode = LogicalUnion.create(inputsToUnion, true);
  call.transformTo(newNode);
 }
}

代码示例来源：origin: apache/hive

Join join, List<RexNode> joinFilters, boolean leftInputPotentialFK, RelMetadataQuery mq) {
final List<RexNode> residualPreds = new ArrayList<>();
final JoinRelType joinType = join.getJoinType();
final RelNode fkInput = leftInputPotentialFK ? join.getLeft() : join.getRight();
final PKFKJoinInfo cannotExtract =
  PKFKJoinInfo.of(false, null, null);
 join = join.copy(join.getTraitSet(), join.getCluster().getRexBuilder().makeLiteral(true),
   join.getLeft(), join.getRight(), JoinRelType.INNER, false);
final Set<RelTableRef> leftTables = mq.getTableReferences(join.getLeft());
final Set<RelTableRef> rightTables =
  Sets.difference(mq.getTableReferences(join), mq.getTableReferences(join.getLeft()));
final Set<RelTableRef> fkTables = join.getLeft() == fkInput ? leftTables : rightTables;
final Set<RelTableRef> nonFkTables = join.getLeft() == fkInput ? rightTables : leftTables;
       int nFieldsLeft = join.getLeft().getRowType().getFieldList().size();
       int nFieldsRight = join.getRight().getRowType().getFieldList().size();
       int nSysFields = join.getSystemFieldList().size();
       ImmutableBitSet rightFieldsBitSet = ImmutableBitSet.range(nSysFields + nFieldsLeft,
         nSysFields + nFieldsLeft + nFieldsRight);

代码示例来源：origin: Qihoo360/Quicksql

public Set<ImmutableBitSet> getUniqueKeys(Join rel, RelMetadataQuery mq,
  boolean ignoreNulls) {
 final RelNode left = rel.getLeft();
 final RelNode right = rel.getRight();
 final JoinInfo joinInfo = rel.analyzeCondition();
   && rightUnique
   && (leftSet != null)
   && !(rel.getJoinType().generatesNullsOnLeft())) {
  retSet.addAll(leftSet);
   && leftUnique
   && (rightSet != null)
   && !(rel.getJoinType().generatesNullsOnRight())) {
  retSet.addAll(rightSet);

代码示例来源：origin: apache/drill

if (join.getJoinType() != JoinRelType.INNER) {
 return;
  mq.getPulledUpPredicates(join).pulledUpPredicates);
final ImmutableBitSet joinColumns =
  RelOptUtil.InputFinder.bits(join.getCondition());
final boolean allColumnsInAggregate =
  keyColumns.contains(joinColumns);
final List<Boolean> filterNulls = Lists.newArrayList();
RexNode nonEquiConj =
  RelOptUtil.splitJoinCondition(join.getLeft(), join.getRight(),
    join.getCondition(), leftKeys, rightKeys, filterNulls);
for (int s = 0; s < 2; s++) {
 final Side side = new Side();
 final RelNode joinInput = join.getInput(s);
 int fieldCount = joinInput.getRowType().getFieldCount();
 final ImmutableBitSet fieldSet =
  join.getRowType().getFieldCount(),
  belowOffset);
final RexNode newCondition =
  RexUtil.apply(mapping, join.getCondition());
  sides.get(1).newInput, newCondition, join.getJoinType(),
  join.getVariablesStopped(), join.isSemiJoinDone());

代码示例来源：origin: apache/hive

this.joinRel = joinRel;
this.isSemiJoin = isSemiJoin;
nFieldsLeft = joinRel.getLeft().getRowType().getFieldList().size();
nFieldsRight = joinRel.getRight().getRowType().getFieldList().size();
nSysFields = joinRel.getSystemFieldList().size();
leftFieldsBitSet = ImmutableBitSet.range(nSysFields,
  nSysFields + nFieldsLeft);
   nSysFields + nFieldsLeft, nSysFields, 0, nFieldsLeft);
 leftChildPredicates = lPreds.accept(
   new RexPermuteInputsShuttle(leftMapping, joinRel.getInput(0)));
   nSysFields + nFieldsLeft, 0, nFieldsRight);
 rightChildPredicates = rPreds.accept(
   new RexPermuteInputsShuttle(rightMapping, joinRel.getInput(1)));
RexBuilder rexBuilder = joinRel.getCluster().getRexBuilder();
List<RexNode> exprs =
  RelOptUtil.conjunctions(
    compose(rexBuilder, ImmutableList.of(joinRel.getCondition())));

代码示例来源：origin: apache/hive

protected void perform(RelOptRuleCall call, ImmutableBitSet topRefs,
            RelNode topOperator, Join join, RelNode left, Aggregate aggregate) {
 LOG.debug("Matched HiveSemiJoinRule");
 final RelOptCluster cluster = join.getCluster();
 final RexBuilder rexBuilder = cluster.getRexBuilder();
 final ImmutableBitSet rightBits =
   ImmutableBitSet.range(left.getRowType().getFieldCount(),
              join.getRowType().getFieldCount());
 if (topRefs.intersects(rightBits)) {
  return;
 final JoinInfo joinInfo = join.analyzeCondition();
 if (!joinInfo.rightSet().equals(
   ImmutableBitSet.range(aggregate.getGroupCount()))) {
 if(join.getJoinType() == JoinRelType.LEFT) {
 if (join.getJoinType() != JoinRelType.INNER) {
  return;
  Join rightJoin = (Join)(((HepRelVertex)aggregate.getInput()).getCurrentRel());
  List<RexNode> projects = new ArrayList<>();
  for(int i=0; i<rightJoin.getRowType().getFieldCount(); i++){
   projects.add(rexBuilder.makeInputRef(rightJoin, i));
  RelNode topProject =  call.builder().push(rightJoin).project(projects, rightJoin.getRowType().getFieldNames(),
                                 true).build();
  semi = call.builder().push(left).push(topProject).semiJoin(newCondition).build();

代码示例来源：origin: apache/hive

final RelNode oldLeft = rel.getInput(0);
final RelNode oldRight = rel.getInput(1);
final RelNode newJoin = HiveJoin.getJoin(rel.getCluster(), leftFrame.r,
    rightFrame.r, decorrelateExpr(rel.getCondition()), rel.getJoinType());
assert rel.getRowType().getFieldCount()
    == oldLeftFieldCount + oldRightFieldCount;

代码示例来源：origin: org.apache.calcite/calcite-core

private List<Double> averageJoinColumnSizes(Join rel, RelMetadataQuery mq,
  boolean semijoin) {
 final RelNode left = rel.getLeft();
 final RelNode right = rel.getRight();
 final List<Double> lefts = mq.getAverageColumnSizes(left);
 final List<Double> rights =
   semijoin ? null : mq.getAverageColumnSizes(right);
 if (lefts == null && rights == null) {
  return null;
 }
 final int fieldCount = rel.getRowType().getFieldCount();
 Double[] sizes = new Double[fieldCount];
 if (lefts != null) {
  lefts.toArray(sizes);
 }
 if (rights != null) {
  final int leftCount = left.getRowType().getFieldCount();
  for (int i = 0; i < rights.size(); i++) {
   sizes[leftCount + i] = rights.get(i);
  }
 }
 return ImmutableNullableList.copyOf(sizes);
}

代码示例来源：origin: apache/hive

Project project,
   int nullIndicatorPos) {
final RelDataTypeFactory typeFactory = join.getCluster().getTypeFactory();
final RelNode left = join.getLeft();
final JoinRelType joinType = join.getJoinType();
        nullIndicatorPos,
        typeFactory.createTypeWithNullability(
            join.getRowType().getFieldList().get(nullIndicatorPos)
                .getType(),
            true));

代码示例来源：origin: apache/hive

final RelNode left = join.getInputs().get(0);
final RelNode right = join.getInputs().get(1);
  join.getCondition());
final List<RexNode> leftFilters = new ArrayList<>();
final List<RexNode> rightFilters = new ArrayList<>();
RelOptUtil.classifyFilters(join, joinFilters, join.getJoinType(),false,
  !join.getJoinType().generatesNullsOnRight(), !join.getJoinType().generatesNullsOnLeft(),
  joinFilters, leftFilters, rightFilters);
boolean leftIsKey = (join.getJoinType() == JoinRelType.INNER || join.getJoinType() == JoinRelType.RIGHT)
  && leftInputResult.isPkFkJoin;
boolean rightIsKey = (join.getJoinType() == JoinRelType.INNER || join.getJoinType() == JoinRelType.LEFT)
  && rightInputResult.isPkFkJoin;
if (!leftIsKey && !rightIsKey) {
boolean isPKSideSimpleTree = simpleTree.left;
boolean isNoFilteringPKSideTree = simpleTree.right;
RexBuilder rexBuilder = join.getCluster().getRexBuilder();
RexNode leftPred = RexUtil.composeConjunction(
  rexBuilder, leftFilters, true);
 PKSideInfo pkInfo = new PKSideInfo(leftRowCount,
   leftNDV,
   join.getJoinType().generatesNullsOnRight() ? 1.0 :
     pkSelectivity);
 double ndvScalingFactor = isPKSideSimpleTree ? leftNDV/rightNDV : 1.0;

代码示例来源：origin: apache/hive

public Double getSelectivity(Join j, RelMetadataQuery mq, RexNode predicate) {
 if (j.getJoinType().equals(JoinRelType.INNER)) {
  return computeInnerJoinSelectivity(j, mq, predicate);
 } else if (j.getJoinType().equals(JoinRelType.LEFT) ||
     j.getJoinType().equals(JoinRelType.RIGHT)) {
  double left = mq.getRowCount(j.getLeft());
  double right = mq.getRowCount(j.getRight());
  double product = left * right;
  double innerJoinSelectivity = computeInnerJoinSelectivity(j, mq, predicate);
  if (j.getJoinType().equals(JoinRelType.LEFT)) {
   return Math.max(innerJoinSelectivity, left/product);
  }
  return Math.max(innerJoinSelectivity, right/product);
 }
 return 1.0;
}

代码示例来源：origin: Qihoo360/Quicksql

for (RelNode input : join.getInputs()) {
 if (convertInputTraits && input.getConvention() != getOutTrait()) {
  input =
if (convertInputTraits && !canJoinOnCondition(join.getCondition())) {
 return null;
   join.getCluster(),
   join.getTraitSet().replace(out),
   newInputs.get(0),
   newInputs.get(1),
   join.getCondition(),
   join.getVariablesSet(),
   join.getJoinType());
} catch (InvalidRelException e) {
 LOGGER.debug(e.toString());

代码示例来源：origin: apache/hive

final JoinInfo joinInfo = join.analyzeCondition();
if(!joinInfo.isEqui()) {
 return;
  ImmutableBitSet.range(0, join.getLeft().getRowType().getFieldCount());

代码示例来源：origin: apache/hive

int rightOffSet = j.getLeft().getRowType().getFieldCount();
   HiveRelMdDistinctRowCount.getDistinctRowCount(j.getLeft(), mq, ljk));
   HiveRelMdDistinctRowCount.getDistinctRowCount(j.getRight(), mq, rjk));
double ndvEstimate = 1;
if (noOfPE > 0) {
 boolean isCorrelatedColumns = j.getCluster().getPlanner().getContext().
   unwrap(HiveConfPlannerContext.class).getIsCorrelatedColumns();
 if (noOfPE > 1 && isCorrelatedColumns ){
  ndvEstimate = Math.min(mq.getRowCount(j.getLeft()),
    ndvEstimate);
 }else if (j instanceof HiveJoin){
  ndvEstimate = Math.min(mq.getRowCount(j.getLeft())
    * mq.getRowCount(j.getRight()), ndvEstimate);
 } else {
  throw new RuntimeException("Unexpected Join type: " + j.getClass().getName());

代码示例来源：origin: apache/hive

int nFieldsLeft = joinRel.getLeft().getRowType().getFieldList().size();
int nFieldsRight = joinRel.getRight().getRowType().getFieldList().size();
int nSysFields = joinRel.getSystemFieldList().size();
ImmutableBitSet rightFieldsBitSet = ImmutableBitSet.range(nSysFields + nFieldsLeft,
  nSysFields + nFieldsLeft + nFieldsRight);

内容来源于网络，如有侵权，请联系作者删除！

Join

相关文章

    查看更多
    微信公众号

    热门标签

    更多
    JavaquerypythonNode开发语言requestUtil数据库Table后端算法LoggerMessageElementParser

    最新文章

    更多
    • 蜀ICP备13028337号-1 大数据知识库 https://www.saoniuhuo.com © All rights reserved
    • 本站内容来源互联网,如果侵犯您的权益请联系我们删除, 联系方式：448109455@qq.com