Validity taken into account when writing StructArray to json

src/io/json/write/serialize.rs

@@ -18,6 +18,7 @@
 use serde_json::map::Map;
 use serde_json::{Number, Value};
 
+use crate::bitmap::utils::zip_validity;
 use crate::{array::*, datatypes::*, record_batch::RecordBatch, types::NativeType};
 
 trait JsonSerializable {
@@ -117,7 +118,6 @@ fn struct_array_to_jsonmap_array(array: &StructArray, row_count: usize) -> Vec<M
         .take(row_count)
         .collect::<Vec<Map<String, Value>>>();
 
-    // todo: use validity...
     array
         .values()
         .iter()
@@ -202,7 +202,9 @@ fn write_array(array: &dyn Array) -> Value {
                 array.as_any().downcast_ref::<StructArray>().unwrap(),
                 array.len(),
             );
-            jsonmaps.into_iter().map(Value::Object).collect()
+            zip_validity(jsonmaps.into_iter(), array.validity().map(|v| v.iter()))
+                .map(|m| m.map(|o| Value::Object(o)).unwrap_or(Value::Null))
+                .collect()
         }
         _ => {
             panic!(
@@ -295,9 +297,15 @@ fn set_column_for_json_rows(
             let inner_objs = struct_array_to_jsonmap_array(array, row_count);
             rows.iter_mut()
                 .take(row_count)
-                .zip(inner_objs.into_iter())
+                .zip(zip_validity(
+                    inner_objs.into_iter(),
+                    array.validity().map(|v| v.iter()),
+                ))
                 .for_each(|(row, obj)| {
-                    row.insert(col_name.to_string(), Value::Object(obj));
+                    row.insert(
+                        col_name.to_string(),
+                        obj.map(|o| Value::Object(o)).unwrap_or(Value::Null),
+                    );
                 });
         }
         DataType::List(_) => {

tests/it/io/json/write.rs

@@ -38,6 +38,55 @@ fn write_simple_rows() {
     );
 }
 
+#[test]
+fn write_nested_struct_with_validity() {
+    let inner = vec![
+        Field::new("c121", DataType::Utf8, false),
+        Field::new("c122", DataType::Int32, false),
+    ];
+    let fields = vec![
+        Field::new("c11", DataType::Int32, false),
+        Field::new("c12", DataType::Struct(inner.clone()), false),
+    ];
+    let schema = Schema::new(vec![
+        Field::new("c1", DataType::Struct(fields.clone()), false),
+        Field::new("c2", DataType::Utf8, false),
+    ]);
+
+    let c1 = StructArray::from_data(
+        DataType::Struct(fields),
+        vec![
+            Arc::new(Int32Array::from(&[Some(1), None, Some(5)])),
+            Arc::new(StructArray::from_data(
+                DataType::Struct(inner),
+                vec![
+                    Arc::new(Utf8Array::<i32>::from(&vec![None, Some("f"), Some("g")])),
+                    Arc::new(Int32Array::from(&[Some(20), None, Some(43)])),
+                ],
+                Some(Bitmap::from([false, true, true])),
+            )),
+        ],
+        Some(Bitmap::from([true, true, false])),
+    );
+    let c2 = Utf8Array::<i32>::from(&vec![Some("a"), Some("b"), Some("c")]);
+
+    let batch = RecordBatch::try_new(Arc::new(schema), vec![Arc::new(c1), Arc::new(c2)]).unwrap();
+
+    let mut buf = Vec::new();
+    {
+        let mut writer = LineDelimitedWriter::new(&mut buf);
+        writer.write_batches(&[batch]).unwrap();
+    }
+
+    assert_eq!(
+        String::from_utf8(buf).unwrap(),
+        r#"{"c1":{"c11":1,"c12":null},"c2":"a"}
+{"c1":{"c11":null,"c12":{"c121":"f","c122":null}},"c2":"b"}
+{"c1":null,"c2":"c"}
+"#
+    );
+}
+
 #[test]
 fn write_nested_structs() {
     let c121 = Field::new("c121", DataType::Utf8, false);
@@ -202,10 +251,10 @@ fn write_list_of_struct() {
                     Some("f"),
                     Some("g"),
                 ]))],
-                None,
+                Some(Bitmap::from([false, true, true])),
             )),
         ],
-        None,
+        Some(Bitmap::from([true, true, false])),
     );
 
     // list column rows (c1):
@@ -231,9 +280,9 @@ fn write_list_of_struct() {
 
     assert_eq!(
         String::from_utf8(buf).unwrap(),
-        r#"{"c1":[{"c11":1,"c12":{"c121":"e"}},{"c11":null,"c12":{"c121":"f"}}],"c2":1}
+        r#"{"c1":[{"c11":1,"c12":null},{"c11":null,"c12":{"c121":"f"}}],"c2":1}
 {"c1":null,"c2":2}
-{"c1":[{"c11":5,"c12":{"c121":"g"}}],"c2":3}
+{"c1":[null],"c2":3}
 "#
     );
 }