AVRO-3747: [Rust] Set is_human_readable hint to false for Value (…

…apache#2202)

* Set `is_human_readable` hint to `false` for `Value`

* Include for `de` as well

* Fix windows path error

* Include test

* Fix clippy lint and add util for setting whether it's human readable

* AVRO-3747: [Rust] Make serde `is_human_readable` configurable

Signed-off-by: Martin Tzvetanov Grigorov <mgrigorov@apache.org>

* AVRO-3747: Get rid of clippy::unnecessary_mut_passed

Signed-off-by: Martin Tzvetanov Grigorov <mgrigorov@apache.org>

* AVRO-3747: Fix test comments

Signed-off-by: Martin Tzvetanov Grigorov <mgrigorov@apache.org>

---------

Signed-off-by: Martin Tzvetanov Grigorov <mgrigorov@apache.org>
Co-authored-by: Martin Tzvetanov Grigorov <mgrigorov@apache.org>

lang/rust/avro/src/de.rs

@@ -537,6 +537,10 @@ impl<'a, 'de> de::Deserializer<'de> for &'a Deserializer<'de> {
     {
         self.deserialize_any(visitor)
     }
+
+    fn is_human_readable(&self) -> bool {
+        crate::util::is_human_readable()
+    }
 }
 
 impl<'de> de::SeqAccess<'de> for SeqDeserializer<'de> {
@@ -1220,4 +1224,36 @@ mod tests {
         assert_eq!(deserialized, reference);
         Ok(())
     }
+
+    #[test]
+    fn avro_3747_human_readable_false() -> TestResult<()> {
+        // AVRO-3747: set serde's is_human_readable to false
+        use serde::de::Deserializer as SerdeDeserializer;
+
+        unsafe {
+            crate::util::SERDE_HUMAN_READABLE = false;
+        }
+
+        let deser = Deserializer::new(&Value::Null);
+
+        assert_eq!((&deser).is_human_readable(), false);
+
+        Ok(())
+    }
+
+    #[test]
+    fn avro_3747_human_readable_true() -> TestResult<()> {
+        // AVRO-3747: set serde's is_human_readable to true
+        use serde::de::Deserializer as SerdeDeserializer;
+
+        unsafe {
+            crate::util::SERDE_HUMAN_READABLE = true;
+        }
+
+        let deser = Deserializer::new(&Value::Null);
+
+        assert!((&deser).is_human_readable());
+
+        Ok(())
+    }
 }

lang/rust/avro/src/lib.rs

@@ -748,7 +748,7 @@ pub use reader::{
 };
 pub use schema::{AvroSchema, Schema};
 pub use ser::to_value;
-pub use util::max_allocation_bytes;
+pub use util::{max_allocation_bytes, set_serde_human_readable};
 pub use writer::{
     to_avro_datum, to_avro_datum_schemata, GenericSingleObjectWriter, SpecificSingleObjectWriter,
     Writer,

lang/rust/avro/src/ser.rs

@@ -283,6 +283,10 @@ impl<'b> ser::Serializer for &'b mut Serializer {
     ) -> Result<Self::SerializeStructVariant, Self::Error> {
         Ok(StructVariantSerializer::new(index, variant, len))
     }
+
+    fn is_human_readable(&self) -> bool {
+        crate::util::is_human_readable()
+    }
 }
 
 impl ser::SerializeSeq for SeqSerializer {
@@ -998,4 +1002,30 @@ mod tests {
             "error serializing tuple untagged enum"
         );
     }
+
+    #[test]
+    fn avro_3747_human_readable_false() {
+        use serde::ser::Serializer as SerdeSerializer;
+
+        unsafe {
+            crate::util::SERDE_HUMAN_READABLE = false;
+        }
+
+        let ser = &mut Serializer {};
+
+        assert_eq!(ser.is_human_readable(), false);
+    }
+
+    #[test]
+    fn avro_3747_human_readable_true() {
+        use serde::ser::Serializer as SerdeSerializer;
+
+        unsafe {
+            crate::util::SERDE_HUMAN_READABLE = true;
+        }
+
+        let ser = &mut Serializer {};
+
+        assert!(ser.is_human_readable());
+    }
 }

lang/rust/avro/src/util.rs

@@ -27,6 +27,14 @@ pub const DEFAULT_MAX_ALLOCATION_BYTES: usize = 512 * 1024 * 1024;
 static mut MAX_ALLOCATION_BYTES: usize = DEFAULT_MAX_ALLOCATION_BYTES;
 static MAX_ALLOCATION_BYTES_ONCE: Once = Once::new();
 
+/// Whether to set serialization & deserialization traits
+/// as `human_readable` or not.
+/// See [set_serde_human_readable] to change this value.
+pub const DEFAULT_SERDE_HUMAN_READABLE: bool = true;
+// crate visible for testing
+pub(crate) static mut SERDE_HUMAN_READABLE: bool = DEFAULT_SERDE_HUMAN_READABLE;
+static SERDE_HUMAN_READABLE_ONCE: Once = Once::new();
+
 pub trait MapHelper {
     fn string(&self, key: &str) -> Option<String>;
 
@@ -153,6 +161,27 @@ pub fn safe_len(len: usize) -> AvroResult<usize> {
     }
 }
 
+/// Set whether serializing/deserializing is marked as human readable in serde traits.
+/// This will adjust the return value of `is_human_readable()` for both.
+/// Once called, the value cannot be changed.
+///
+/// **NOTE** This function must be called before serializing/deserializing **any** data. The
+/// library leverages [`std::sync::Once`](https://doc.rust-lang.org/std/sync/struct.Once.html)
+/// to set the limit either when calling this method, or when decoding for
+/// the first time.
+pub fn set_serde_human_readable(human_readable: bool) -> bool {
+    unsafe {
+        SERDE_HUMAN_READABLE_ONCE.call_once(|| {
+            SERDE_HUMAN_READABLE = human_readable;
+        });
+        SERDE_HUMAN_READABLE
+    }
+}
+
+pub(crate) fn is_human_readable() -> bool {
+    unsafe { SERDE_HUMAN_READABLE }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;