From 1e000384ca8b19d39b96da7fa7b5f569062d333b Mon Sep 17 00:00:00 2001
From: Clemens Winter <clemenswinter1@gmail.com>
Date: Tue, 30 Jul 2024 19:58:14 -0700
Subject: [PATCH] Fix mixed integer float aggregation

---
 src/engine/execution/batch_merging.rs |  3 ++
 test_data/edge_cases.csv              | 22 +++++++-------
 tests/query_tests.rs                  | 42 +++++++++++++++++----------
 3 files changed, 40 insertions(+), 27 deletions(-)

diff --git a/src/engine/execution/batch_merging.rs b/src/engine/execution/batch_merging.rs
index 6ce08303..11d9daaa 100644
--- a/src/engine/execution/batch_merging.rs
+++ b/src/engine/execution/batch_merging.rs
@@ -175,6 +175,9 @@ pub fn combine<'a>(
         {
             let left = left[ileft];
             let right = right[iright];
+            let (left, right) = unify_types(&mut qp, left, right);
+            let left = null_to_val(&mut qp, left);
+            let right = null_to_val(&mut qp, right);
             let aggregated = qp.merge_aggregate(ops, left, right, aggregator);
             aggregates.push((aggregated.any(), aggregator));
         }
diff --git a/test_data/edge_cases.csv b/test_data/edge_cases.csv
index 2adf09a0..c18b6117 100644
--- a/test_data/edge_cases.csv
+++ b/test_data/edge_cases.csv
@@ -1,11 +1,11 @@
-u8_offset_encoded,non_dense_ints,enum,string_packed,constant0,constant0_2,negative,id,nullable_int,nullable_int2,country,largenum,float,nullable_float,float01
-256,0,aa,xyz,0,0,-199,0,-1,,Germany,-9223372036854775808,0.123412,,0.3
-258,2,aa,abc,0,0,39,1,-40,-40,USA,9223372036854775806,3e-4,,-0.4
-259,3,aa,axz,0,0,-100,2,,,France,9223372036854775806,-124.0,0.4,0.421231
-257,1,bb,AXY,0,0,34,3,,0,,9223372036854775806,3.15159,,0.9482
-275,4,bb,azy,0,0,4031,4,10,9,France,-9223372036854775808,0.1234e30,,0.1
-500,0,aa,$sss,0,0,32,5,,6,,9223372036854775806,1e-6,,0.2
-343,2,cc,asd,0,0,-130,6,,,Turkey,-9223372036854775808,0.0,1e-32,0.5
-432,1,aa,_f,0,0,-120,7,20,,,9223372036854775806,0.000001,,0.23
-511,2,cc,t,0,0,4010,8,,1,,-9223372036854775808,-1.0,,0.742
-500,3,bb,😈,0,0,-40,9,13,14,Germany,9223372036854775806,1234124.51325,1.123124e30,-0.2
\ No newline at end of file
+u8_offset_encoded,non_dense_ints,enum,string_packed,constant0,constant0_2,negative,id,nullable_int,nullable_int2,country,largenum,float,nullable_float,float01,mixed_float_int_null
+256,0,aa,xyz,0,0,-199,0,-1,,Germany,-9223372036854775808,0.123412,,0.3,1
+258,2,aa,abc,0,0,39,1,-40,-40,USA,9223372036854775806,3e-4,,-0.4,10
+259,3,aa,axz,0,0,-100,2,,,France,9223372036854775806,-124.0,0.4,0.421231,3
+257,1,bb,AXY,0,0,34,3,,0,,9223372036854775806,3.15159,,0.9482,0.21
+275,4,bb,azy,0,0,4031,4,10,9,France,-9223372036854775808,0.1234e30,,0.1,0.12
+500,0,aa,$sss,0,0,32,5,,6,,9223372036854775806,1e-6,,0.2,
+343,2,cc,asd,0,0,-130,6,,,Turkey,-9223372036854775808,0.0,1e-32,0.5,
+432,1,aa,_f,0,0,-120,7,20,,,9223372036854775806,0.000001,,0.23,0.1
+511,2,cc,t,0,0,4010,8,,1,,-9223372036854775808,-1.0,,0.742,0.1
+500,3,bb,😈,0,0,-40,9,13,14,Germany,9223372036854775806,1234124.51325,1.123124e30,-0.2,0.5
\ No newline at end of file
diff --git a/tests/query_tests.rs b/tests/query_tests.rs
index 80424a06..ab50dddf 100644
--- a/tests/query_tests.rs
+++ b/tests/query_tests.rs
@@ -891,6 +891,17 @@ fn test_divide_multiply_null() {
     );
 }
 
+#[test]
+fn test_aggregate_mixed_int_float_null() {
+    test_query_ec(
+        "SELECT SUM(mixed_float_int_null), COUNT(mixed_float_int_null), MIN(mixed_float_int_null), MAX(mixed_float_int_null), id / 5 FROM default",
+        &[
+            vec![Float(14.33), Int(5), Float(0.12), Float(10.0), Int(0)],
+            vec![Float(0.7), Int(3), Float(0.1), Float(0.5), Int(1)]
+        ]
+    );
+}
+
 #[test]
 fn test_sort_by_nullable1() {
     test_query_ec(
@@ -1395,7 +1406,8 @@ fn test_long_nullable() {
             "nullable_int".to_string(),
             locustdb::colgen::nullable_ints(vec![None, Some(1), Some(-10)], vec![0.9, 0.05, 0.05]),
         )],
-    })).unwrap();
+    }))
+    .unwrap();
     let query = "SELECT nullable_int FROM test LIMIT 0;";
     let expected_rows: Vec<[Value; 1]> = vec![];
     let result = block_on(locustdb.run_query(query, true, true, vec![])).unwrap();
@@ -1877,34 +1889,32 @@ fn test_float_greater_than_int() {
 //     );
 // }
 
-
 #[test]
 fn test_floor1() {
     test_query_ec(
         "SELECT MAX(id), MIN(id), FLOOR(float01 * 10) FROM default",
         &[
-              vec![Int(1), Int(1), Int(-4)],
-              vec![Int(9), Int(9), Int(-2)],
-              vec![Int(4), Int(4), Int(1)],
-              vec![Int(7), Int(5), Int(2)],
-              vec![Int(0), Int(0), Int(3)],
-              vec![Int(2), Int(2), Int(4)],
-              vec![Int(6), Int(6), Int(5)],
-              vec![Int(8), Int(8), Int(7)],
-              vec![Int(3), Int(3), Int(9)]
+            vec![Int(1), Int(1), Int(-4)],
+            vec![Int(9), Int(9), Int(-2)],
+            vec![Int(4), Int(4), Int(1)],
+            vec![Int(7), Int(5), Int(2)],
+            vec![Int(0), Int(0), Int(3)],
+            vec![Int(2), Int(2), Int(4)],
+            vec![Int(6), Int(6), Int(5)],
+            vec![Int(8), Int(8), Int(7)],
+            vec![Int(3), Int(3), Int(9)],
         ],
     );
 }
 
-
 #[test]
 fn test_floor2() {
     test_query_ec(
         "SELECT MIN(id), MAX(id), FLOOR(id * 0.23) FROM default",
         &[
-              vec![Int(0), Int(4), Int(0)],
-              vec![Int(5), Int(8), Int(1)],
-              vec![Int(9), Int(9), Int(2)],
+            vec![Int(0), Int(4), Int(0)],
+            vec![Int(5), Int(8), Int(1)],
+            vec![Int(9), Int(9), Int(2)],
         ],
     );
-}
\ No newline at end of file
+}