feat(sql): support null semantics in static IN-list expressions (#1659)

dozer-sql/expression/src/in_list.rs

@@ -12,17 +12,44 @@ pub(crate) fn evaluate_in_list(
     record: &Record,
 ) -> Result<Field, Error> {
     let field = expr.evaluate(record, schema)?;
-    let mut result = false;
+    let mut match_found = false;
+    let mut list_contains_null = false;
+
+    if field == Field::Null {
+        return Ok(Field::Null);
+    }
+
     for item in list {
         let item = item.evaluate(record, schema)?;
+        if item == Field::Null {
+            list_contains_null = true;
+            continue;
+        }
+
         if field == item {
-            result = true;
+            match_found = true;
             break;
         }
     }
+
+    let result_field = if match_found {
+        Field::Boolean(true)
+    } else if list_contains_null {
+        // No exact match, but NULL was present in the list -> UNKNOWN
+        Field::Null
+    } else {
+        // No exact match, and no NULLs in the list -> FALSE
+        Field::Boolean(false)
+    };
+
     // Negate the result if the IN list was negated.
     if negated {
-        result = !result;
+        match result_field {
+            Field::Boolean(b) => Ok(Field::Boolean(!b)),
+            Field::Null => Ok(Field::Null), // NOT NULL is NULL
+            _ => unreachable!(),
+        }
+    } else {
+        Ok(result_field)
     }
-    Ok(Field::Boolean(result))
 }

dozer-sql/src/expression/tests/in_list.rs

@@ -84,3 +84,75 @@ fn test_not_in_list() {
     );
     assert_eq!(f, Field::Boolean(false));
 }
+
+#[test]
+fn test_in_list_with_nulls() {
+    let null_field = Field::Null;
+
+    // 1. 42 IN (NULL) -> Field::Null
+    let f = run_fct("SELECT 42 IN (NULL)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 2. 42 IN (1, NULL) -> Field::Null
+    let f = run_fct("SELECT 42 IN (1, NULL)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 3. 42 IN (1, NULL, 42) -> Field::Boolean(true) (Exact match found)
+    let f = run_fct("SELECT 42 IN (1, NULL, 42)", Schema::default(), vec![]);
+    assert_eq!(f, Field::Boolean(true));
+
+    // 4. 42 NOT IN (1, NULL) -> Field::Null
+    let f = run_fct("SELECT 42 NOT IN (1, NULL)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 5. NULL IN (1, 2, 3) -> Field::Null
+    let f = run_fct("SELECT NULL IN (1, 2, 3)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 6. No exact match, list contains NULL -> Field::Null
+    let f = run_fct("SELECT 99 IN (1, 2, NULL)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 7. NOT IN (1, 2, 3) -> True (42 is not in list, no NULLs)
+    let f = run_fct("SELECT 42 NOT IN (1, 2, 3)", Schema::default(), vec![]);
+    assert_eq!(f, Field::Boolean(true));
+
+    // 8. NOT IN (1, 2, 3, 42) -> False (42 is in list)
+    let f = run_fct("SELECT 42 NOT IN (1, 2, 3, 42)", Schema::default(), vec![]);
+    assert_eq!(f, Field::Boolean(false));
+
+    // 9. NOT IN (1, 2, 3, 42, NULL) -> False (42 matches, so NOT(true) == false)
+    let f = run_fct("SELECT 42 NOT IN (1, 2, 3, 42, NULL)", Schema::default(), vec![]);
+    assert_eq!(f, Field::Boolean(false));
+
+    // 10. NOT IN (1, 2, 3, NULL) with no match -> Null
+    let f = run_fct("SELECT 42 NOT IN (1, 2, 3, NULL)", Schema::default(), vec![]);
+    assert_eq!(f, null_field);
+
+    // 11. Column expression evaluates to NULL in record
+    let schema = Schema::default()
+        .field(
+            FieldDefinition::new(
+                String::from("age"),
+                FieldType::Int,
+                true,
+                SourceDefinition::Dynamic,
+            ),
+            false,
+        )
+        .clone();
+    let f = run_fct(
+        "SELECT age IN (1, 2, 3) FROM users",
+        schema.clone(),
+        vec![Field::Null],
+    );
+    assert_eq!(f, null_field);
+
+    // 12. Column expression matches with NULL in list
+    let f = run_fct(
+        "SELECT age IN (1, NULL, 42) FROM users",
+        schema,
+        vec![Field::Int(42)],
+    );
+    assert_eq!(f, Field::Boolean(true));
+}