From 2812f285334dc1c76d0bde1dfc4c9d8915b8b01e Mon Sep 17 00:00:00 2001
From: SimonSchneider <dev@simonschneider.org>
Date: Mon, 16 May 2022 10:21:37 +0200
Subject: [PATCH 1/2] fix: encode float::NAN as null in json

---
 src/io/json/write/serialize.rs | 44 ++++++++++++++++++++++++++++++++--
 tests/it/io/json/write.rs      | 18 ++++++++++++++
 2 files changed, 60 insertions(+), 2 deletions(-)

diff --git a/src/io/json/write/serialize.rs b/src/io/json/write/serialize.rs
index 3b1f9e087b5..41c756586df 100644
--- a/src/io/json/write/serialize.rs
+++ b/src/io/json/write/serialize.rs
@@ -43,6 +43,46 @@ fn primitive_serializer<'a, T: NativeType + ToLexical>(
     ))
 }
 
+fn f64_serializer<'a>(
+    array: &'a PrimitiveArray<f64>,
+) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync> {
+    Box::new(BufStreamingIterator::new(
+        array.iter(),
+        |x, buf| {
+            if let Some(x) = x {
+                if f64::is_nan(*x) {
+                    buf.extend(b"null")
+                } else {
+                    lexical_to_bytes_mut(*x, buf)
+                }
+            } else {
+                buf.extend(b"null")
+            }
+        },
+        vec![],
+    ))
+}
+
+fn f32_serializer<'a>(
+    array: &'a PrimitiveArray<f32>,
+) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync> {
+    Box::new(BufStreamingIterator::new(
+        array.iter(),
+        |x, buf| {
+            if let Some(x) = x {
+                if f32::is_nan(*x) {
+                    buf.extend(b"null")
+                } else {
+                    lexical_to_bytes_mut(*x, buf)
+                }
+            } else {
+                buf.extend(b"null")
+            }
+        },
+        vec![],
+    ))
+}
+
 fn utf8_serializer<'a, O: Offset>(
     array: &'a Utf8Array<O>,
 ) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync> {
@@ -196,8 +236,8 @@ pub(crate) fn new_serializer<'a>(
         DataType::UInt16 => primitive_serializer::<u16>(array.as_any().downcast_ref().unwrap()),
         DataType::UInt32 => primitive_serializer::<u32>(array.as_any().downcast_ref().unwrap()),
         DataType::UInt64 => primitive_serializer::<u64>(array.as_any().downcast_ref().unwrap()),
-        DataType::Float32 => primitive_serializer::<f32>(array.as_any().downcast_ref().unwrap()),
-        DataType::Float64 => primitive_serializer::<f64>(array.as_any().downcast_ref().unwrap()),
+        DataType::Float32 => f32_serializer(array.as_any().downcast_ref().unwrap()),
+        DataType::Float64 => f64_serializer(array.as_any().downcast_ref().unwrap()),
         DataType::Utf8 => utf8_serializer::<i32>(array.as_any().downcast_ref().unwrap()),
         DataType::LargeUtf8 => utf8_serializer::<i64>(array.as_any().downcast_ref().unwrap()),
         DataType::Struct(_) => struct_serializer(array.as_any().downcast_ref().unwrap()),
diff --git a/tests/it/io/json/write.rs b/tests/it/io/json/write.rs
index 69e4bfa9d94..d79a5310f06 100644
--- a/tests/it/io/json/write.rs
+++ b/tests/it/io/json/write.rs
@@ -28,6 +28,24 @@ fn int32() -> Result<()> {
     test!(array, expected)
 }
 
+#[test]
+fn f32() -> Result<()> {
+    let array = Float32Array::from([Some(1.5), Some(2.5), Some(f32::NAN), None, Some(5.5)]);
+
+    let expected = r#"[1.5,2.5,null,null,5.5]"#;
+
+    test!(array, expected)
+}
+
+#[test]
+fn f64() -> Result<()> {
+    let array = Float64Array::from([Some(1.5), Some(2.5), Some(f64::NAN), None, Some(5.5)]);
+
+    let expected = r#"[1.5,2.5,null,null,5.5]"#;
+
+    test!(array, expected)
+}
+
 #[test]
 fn utf8() -> Result<()> {
     let array = Utf8Array::<i32>::from(&vec![Some("a"), Some("b"), Some("c"), Some("d"), None]);

From 19968a34d77df07af084bbe87c7c16da1b632b36 Mon Sep 17 00:00:00 2001
From: SimonSchneider <dev@simonschneider.org>
Date: Mon, 16 May 2022 18:01:16 +0200
Subject: [PATCH 2/2] generic float serializer

---
 src/io/json/write/serialize.rs | 35 +++++++++-------------------------
 1 file changed, 9 insertions(+), 26 deletions(-)

diff --git a/src/io/json/write/serialize.rs b/src/io/json/write/serialize.rs
index 41c756586df..fb90c17ad0f 100644
--- a/src/io/json/write/serialize.rs
+++ b/src/io/json/write/serialize.rs
@@ -43,34 +43,17 @@ fn primitive_serializer<'a, T: NativeType + ToLexical>(
     ))
 }
 
-fn f64_serializer<'a>(
-    array: &'a PrimitiveArray<f64>,
-) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync> {
-    Box::new(BufStreamingIterator::new(
-        array.iter(),
-        |x, buf| {
-            if let Some(x) = x {
-                if f64::is_nan(*x) {
-                    buf.extend(b"null")
-                } else {
-                    lexical_to_bytes_mut(*x, buf)
-                }
-            } else {
-                buf.extend(b"null")
-            }
-        },
-        vec![],
-    ))
-}
-
-fn f32_serializer<'a>(
-    array: &'a PrimitiveArray<f32>,
-) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync> {
+fn float_serializer<'a, T>(
+    array: &'a PrimitiveArray<T>,
+) -> Box<dyn StreamingIterator<Item = [u8]> + 'a + Send + Sync>
+where
+    T: num_traits::Float + NativeType + ToLexical,
+{
     Box::new(BufStreamingIterator::new(
         array.iter(),
         |x, buf| {
             if let Some(x) = x {
-                if f32::is_nan(*x) {
+                if T::is_nan(*x) {
                     buf.extend(b"null")
                 } else {
                     lexical_to_bytes_mut(*x, buf)
@@ -236,8 +219,8 @@ pub(crate) fn new_serializer<'a>(
         DataType::UInt16 => primitive_serializer::<u16>(array.as_any().downcast_ref().unwrap()),
         DataType::UInt32 => primitive_serializer::<u32>(array.as_any().downcast_ref().unwrap()),
         DataType::UInt64 => primitive_serializer::<u64>(array.as_any().downcast_ref().unwrap()),
-        DataType::Float32 => f32_serializer(array.as_any().downcast_ref().unwrap()),
-        DataType::Float64 => f64_serializer(array.as_any().downcast_ref().unwrap()),
+        DataType::Float32 => float_serializer::<f32>(array.as_any().downcast_ref().unwrap()),
+        DataType::Float64 => float_serializer::<f64>(array.as_any().downcast_ref().unwrap()),
         DataType::Utf8 => utf8_serializer::<i32>(array.as_any().downcast_ref().unwrap()),
         DataType::LargeUtf8 => utf8_serializer::<i64>(array.as_any().downcast_ref().unwrap()),
         DataType::Struct(_) => struct_serializer(array.as_any().downcast_ref().unwrap()),