Add test for Simplify redundant predicates

[src255]

Write rules to simplify both a OR a and a AND a to a.

Hello! I wanted to help with issue #3895. This is my attempt, but I'm not sure if this is the desired comparison of the left and right fields of the BinaryExpr struct. I hope this helps and I would appreciate any feedback!

Which issue does this PR close?

Closes #3895 .

[isidentical]

Thank you for the PR @src255, I think it looks pretty good as is (an example comparison is available inside expr_contains). I think all you have to do next is add some tests (a very similiar example is test_simplify_optimized_plan). One thing to note is, I think we might have a AND a already so might make sense to check if the test fails first and gets fixed with your revision. But a OR a should be still good.

[src255]

Thanks for the feedback. Here is the new test I wrote with test_simplify_optimized_plan as a guide:

#[test]
fn test_simplify_optimized_plan_with_or() {
    let table_scan = test_table_scan();
    let plan = LogicalPlanBuilder::from(table_scan)
        .project(vec![col("a")])
        .unwrap()
        .filter(or(col("b").gt(lit(1)), col("b").gt(lit(1))))  // use `or` instead of `and`
        .unwrap()
        .build()
        .unwrap();

    assert_optimized_plan_eq(
        &plan,
        "\
        Filter: test.b > Int32(1)\
        \n  Projection: test.a\
        \n    TableScan: test",
    );
}

After running this test, I noticed that both test_simplify_optimized_plan_with_or and test_simplify_optimized_plan pass without my changes (OR and AND changes commented out)! Perhaps the optimizer is already making both simplifications:

• a OR a --> a
• a AND a --> a.

Also, I looked at the feedback from clippy and removed the needless references of left and right from the comparisons I wrote, but I'm not sure the changes I made are even necessary given the passing tests.

[isidentical]

Thanks for checking it out. That seems to be true (the simplification below) 🤔 (which probably means somewhere in the #3859 we are missing something else, cc: @alamb @Ted-Jiang).

https://github.com/apache/arrow-datafusion/blob/e534c2536858cf18aac219c33b0bddef54c7f214/datafusion/optimizer/src/simplify_expressions.rs#L768-L779

[alamb]

Thanks for the test and investigation @src255 and @isidentical -- clearly I got something wrong.