[CALCITE-5195] ArrayIndexOutOfBoundsException when inferring more equal conditions from join condition for semi join

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.31.0
• Labels: pull-request-available
• Assignee: Chunwei Lei
• Reporter: Chunwei Lei
• Created: 22/Jun/22 03:38
• Updated: 03/Aug/22 16:21
• Resolved: 27/Jun/22 02:27

Description

The following test can reproduce the exception.

  @Test void testJoinConditionPushdown4() {
    final Function<RelBuilder, RelNode> relFn = b -> {
      RelNode left = b.scan("EMP")
          .project(
              b.field("DEPTNO"),
              b.field("ENAME"))
          .build();
      RelNode right = b.scan("DEPT")
          .project(
              b.field("DEPTNO"),
              b.field("DNAME"))
          .build();

      b.push(left).push(right);

      RexInputRef ref1 = b.field(2, 0, "DEPTNO");
      RexInputRef ref2 = b.field(2, 1, "DEPTNO");
      RexInputRef ref3 = b.field(2, 1, "DNAME");

      RexCall cond1 = (RexCall) b.equals(ref1, ref2);
      RexCall cond2 = (RexCall) b.equals(ref1, ref3);

      RexNode cond = b.and(cond1, cond2);
      return b.semiJoin(cond)
          .project(b.field(0))
          .build();
    };

    relFn(relFn)
        .withRule(
            CoreRules.JOIN_PUSH_EXPRESSIONS,
            CoreRules.JOIN_CONDITION_PUSH,
            CoreRules.SEMI_JOIN_PROJECT_TRANSPOSE,
            CoreRules.JOIN_REDUCE_EXPRESSIONS,
            CoreRules.FILTER_REDUCE_EXPRESSIONS)
        .check();
  }

Issue Links

• is caused by: CALCITE-5073 JoinConditionPushRule cannot infer 'LHS.C1 = LHS.C2' from 'LHS.C1 = RHS.C1 AND LHS.C2 = RHS.C1'
• is related to: CALCITE-5201 Improve SemiJoinRule to match Join's right input which is unique for Join keys
• links to: GitHub Pull Request #2837