Fix unnest conjunction with selecting wildcard expression (#12760)

* fix unnest statement with wildcard expression

* add commnets

Author: goldmedal

datafusion/expr/src/utils.rs

@@ -19,6 +19,7 @@
 
 use std::cmp::Ordering;
 use std::collections::{HashMap, HashSet};
+use std::ops::Deref;
 use std::sync::Arc;
 
 use crate::expr::{Alias, Sort, WildcardOptions, WindowFunction};
@@ -754,6 +755,15 @@ pub fn find_base_plan(input: &LogicalPlan) -> &LogicalPlan {
     match input {
         LogicalPlan::Window(window) => find_base_plan(&window.input),
         LogicalPlan::Aggregate(agg) => find_base_plan(&agg.input),
+        // [SqlToRel::try_process_unnest] will convert Expr(Unnest(Expr)) to Projection/Unnest/Projection
+        // We should expand the wildcard expression based on the input plan of the inner Projection.
+        LogicalPlan::Unnest(unnest) => {
+            if let LogicalPlan::Projection(projection) = unnest.input.deref() {
+                find_base_plan(&projection.input)
+            } else {
+                input
+            }
+        }
         LogicalPlan::Filter(filter) => {
             if filter.having {
                 // If a filter is used for a having clause, its input plan is an aggregation.

datafusion/sql/src/utils.rs

@@ -619,7 +619,9 @@ pub(crate) fn rewrite_recursive_unnest_bottom_up(
     } = original_expr.clone().rewrite(&mut rewriter)?;
 
     if !transformed {
-        if matches!(&transformed_expr, Expr::Column(_)) {
+        if matches!(&transformed_expr, Expr::Column(_))
+            || matches!(&transformed_expr, Expr::Wildcard { .. })
+        {
             push_projection_dedupl(inner_projection_exprs, transformed_expr.clone());
             Ok(vec![transformed_expr])
         } else {

datafusion/sqllogictest/test_files/unnest.slt

@@ -33,7 +33,7 @@ AS VALUES
 statement ok
 CREATE TABLE nested_unnest_table
 AS VALUES 
-    (struct('a', 'b', struct('c')), (struct('a', 'b', [10,20])), [struct('a', 'b')]), 
+    (struct('a', 'b', struct('c')), (struct('a', 'b', [10,20])), [struct('a', 'b')]),
     (struct('d', 'e', struct('f')), (struct('x', 'y', [30,40, 50])), null)
 ;
 
@@ -780,3 +780,55 @@ NULL 1
 ### TODO: group by unnest struct
 query error DataFusion error: Error during planning: Projection references non\-aggregate values
 select unnest(column1) c1 from nested_unnest_table group by c1.c0;
+
+query II??I??
+select unnest(column5), * from unnest_table;
+----
+1 2 [1, 2, 3] [7] 1 [13, 14] {c0: 1, c1: 2}
+3 4 [4, 5] [8, 9, 10] 2 [15, 16] {c0: 3, c1: 4}
+NULL NULL [6] [11, 12] 3 NULL NULL
+7 8 [12] [, 42, ] NULL NULL {c0: 7, c1: 8}
+NULL NULL NULL NULL 4 [17, 18] NULL
+
+query TT????
+select unnest(column1), * from nested_unnest_table
+----
+a b {c0: c} {c0: a, c1: b, c2: {c0: c}} {c0: a, c1: b, c2: [10, 20]} [{c0: a, c1: b}]
+d e {c0: f} {c0: d, c1: e, c2: {c0: f}} {c0: x, c1: y, c2: [30, 40, 50]} NULL
+
+query ?????
+select unnest(unnest(column3)), * from recursive_unnest_table
+----
+[1] [[1, 2]] {c0: [1], c1: a} [[[1], [2]], [[1, 1]]] [{c0: [1], c1: [[1, 2]]}]
+[2] [[3], [4]] {c0: [2], c1: b} [[[3, 4], [5]], [[, 6], , [7, 8]]] [{c0: [2], c1: [[3], [4]]}]
+
+statement ok
+CREATE TABLE join_table
+AS VALUES
+    (1, 2, 3),
+    (2, 3, 4),
+    (4, 5, 6)
+;
+
+query IIIII
+select unnest(u.column5), j.* from unnest_table u join join_table j on u.column3 = j.column1
+----
+1 2 1 2 3
+3 4 2 3 4
+NULL NULL 4 5 6
+
+query II?I?
+select unnest(column5), * except (column5, column1) from unnest_table;
+----
+1 2 [7] 1 [13, 14]
+3 4 [8, 9, 10] 2 [15, 16]
+NULL NULL [11, 12] 3 NULL
+7 8 [, 42, ] NULL NULL
+NULL NULL NULL 4 [17, 18]
+
+query III
+select unnest(u.column5), j.* except(column2, column3) from unnest_table u join join_table j on u.column3 = j.column1
+----
+1 2 1
+3 4 2
+NULL NULL 4