Added support for json binary (#304)

src/io/json/read/deserialize.rs

@@ -25,10 +25,7 @@ use serde_json::Value;
 
 use crate::types::NaturalDataType;
 use crate::{
-    array::{
-        Array, BooleanArray, DictionaryArray, DictionaryKey, ListArray, NullArray, Offset,
-        PrimitiveArray, StructArray, Utf8Array,
-    },
+    array::*,
     bitmap::MutableBitmap,
     buffer::MutableBuffer,
     datatypes::{DataType, IntervalUnit},
@@ -90,6 +87,14 @@ fn read_primitive<T: NativeType + NaturalDataType + NumCast>(
     PrimitiveArray::from_trusted_len_iter(iter).to(data_type)
 }
 
+fn read_binary<O: Offset>(rows: &[&Value]) -> BinaryArray<O> {
+    let iter = rows.iter().map(|row| match row {
+        Value::String(v) => Some(v.as_bytes()),
+        _ => None,
+    });
+    BinaryArray::from_trusted_len_iter(iter)
+}
+
 fn read_boolean(rows: &[&Value]) -> BooleanArray {
     let iter = rows.iter().map(|row| match row {
         Value::Bool(v) => Some(v),
@@ -236,9 +241,8 @@ pub fn read(rows: &[&Value], data_type: DataType) -> Arc<dyn Array> {
         DataType::LargeUtf8 => Arc::new(read_utf8::<i64>(rows)),
         DataType::List(_) => Arc::new(read_list::<i32>(rows, data_type)),
         DataType::LargeList(_) => Arc::new(read_list::<i64>(rows, data_type)),
-        // unsupported
-        //DataType::Binary => Box::new(BinaryArray::<i32>::new_empty()),
-        //DataType::LargeBinary => Box::new(BinaryArray::<i64>::new_empty()),
+        DataType::Binary => Arc::new(read_binary::<i32>(rows)),
+        DataType::LargeBinary => Arc::new(read_binary::<i64>(rows)),
         DataType::Struct(_) => Arc::new(read_struct(rows, data_type)),
         DataType::Dictionary(key_type, _) => match key_type.as_ref() {
             DataType::Int8 => Arc::new(read_dictionary::<i8>(rows, data_type)),

tests/it/io/json/mod.rs

@@ -167,20 +167,26 @@ fn case_basics() -> (String, Schema, Vec<Box<dyn Array>>) {
 }
 
 fn case_basics_schema() -> (String, Schema, Vec<Box<dyn Array>>) {
-    let data = r#"{"a":1, "b":2.0, "c":false, "d":"4"}
-    {"a":10, "b":-3.5, "c":true, "d":null}
+    let data = r#"{"a":1, "b":2.0, "c":false, "d":"4", "e":"4"}
+    {"a":10, "b":-3.5, "c":true, "d":null, "e":"text"}
     {"a":100000000, "b":0.6, "d":"text"}"#
         .to_string();
     let schema = Schema::new(vec![
         Field::new("a", DataType::UInt32, true),
         Field::new("b", DataType::Float32, true),
         Field::new("c", DataType::Boolean, true),
         // note how "d" is not here
+        Field::new("e", DataType::Binary, true),
     ]);
     let columns = vec![
         Box::new(UInt32Array::from_slice(&[1, 10, 100000000])) as Box<dyn Array>,
         Box::new(Float32Array::from_slice(&[2.0, -3.5, 0.6])),
         Box::new(BooleanArray::from(&[Some(false), Some(true), None])),
+        Box::new(BinaryArray::<i32>::from(&[
+            Some(b"4".as_ref()),
+            Some(b"text".as_ref()),
+            None,
+        ])),
     ];
     (data, schema, columns)
 }