[rust] Fixed nullable byte arrays

[ranger-ross]

Fixes one of the issues reported in #20500.

The issue is types with nullable string of bytes

properties:
  myByteArray:
    type: string
    format: byte
    nullable: true

will generate invalid code

struct Foo {
    #[serde_as(as = "Option<serde_with::base64::Base64>")]
    #[serde(rename = "myByteArray", default, with = "::serde_with::rust::double_option", skip_serializing_if = "Option::is_none")]
    pub my_byte_array: Option<Option<Vec<u8>>>,
}

this does not compile due to serde_as not being compatible with serde_with.

error: Cannot combine `serde_as` with serde's `with`, `deserialize_with`, or `serialize_with`.
  --> src/models/credentials.rs:22:5
   |
22 |     #[serde_as(as = "Option<serde_with::base64::Base64>")]
   |     

This PR removes the with and adds 2 Options if the type is a nullable byte array

#[serde_as(as = "Option<Option<serde_with::base64::Base64>>")]
#[serde(rename = "myByteArray", default, skip_serializing_if = "Option::is_none")]
pub my_byte_array: Option<Option<Vec<u8>>>,

PR checklist

• Read the contribution guidelines.
• Pull Request title clearly describes the work in the pull request and Pull Request description provides details about how to validate the work. Missing information here may result in delayed response from the community.
• Run the following to build the project and update samples:

  ./mvnw clean package || exit
  ./bin/generate-samples.sh ./bin/configs/*.yaml || exit
  ./bin/utils/export_docs_generators.sh || exit
  

  (For Windows users, please run the script in Git BASH)
  Commit all changed files.
  This is important, as CI jobs will verify all generator outputs of your HEAD commit as it would merge with master.
  These must match the expectations made by your contribution.
  You may regenerate an individual generator by passing the relevant config(s) as an argument to the script, for example ./bin/generate-samples.sh bin/configs/java*.
  IMPORTANT: Do NOT purge/delete any folders/files (e.g. tests) when regenerating the samples as manually written tests may be removed.
• File the PR against the correct branch: master (upcoming 7.x.0 minor release - breaking changes with fallbacks), 8.0.x (breaking changes without fallbacks)
• If your PR is targeting a particular programming language, @mention the technical committee members, so they are more likely to review the pull request.

cc: @frol @farcaller @richardwhiuk @paladinzh @jacob-pro

[jonasbb]

This change will deserialize a null into None. double_option converts a null into Some(None).

[ranger-ross]

This change will deserialize a null into None. double_option converts a null into Some(None).

Oh thanks, good catch. I suppose I need to take another approach.
Let me convert this PR to a draft for the time being.

[ranger-ross]

@jonasbb by the way, since you are the maintainer of serde_with perhaps you can give some guidance on the best way to handle this?

I suppose one option would be to always generate Option<Vec<u8>> as the type for byte array fields that need #[serde_as(as = "Option<serde_with::base64::Base64>")].

[jonasbb]

In general the type within serde_as should mirror the type of the field.

• Vec<u8> -> serde_with::base64::Base64
• Option<Vec<u8>> -> Option<serde_with::base64::Base64>
• Option<Option<Vec<u8>>> -> Option<Option<serde_with::base64::Base64>>

Now the challenge here is capturing the JSON/Yaml semantics of missing/null/value into Rust. Since there is no such three state type in std, the Option<Option<T>> workaround is used with:

• missing -> None
• null -> Some(None)
• value -> Some(Some(value))

You don't have to use ::serde_with::rust::double_option and instead can use something that works with serde_as. Something like this should work (untested). You will still need the default and skip_serializing_if attributes to get the correct behavior.

struct Foo {
    #[serde_as(as = "DoubleOption<serde_with::base64::Base64>")]
    #[serde(rename = "myByteArray", default, skip_serializing_if = "Option::is_none")]
    pub my_byte_array: Option<Option<Vec<u8>>>,
}

struct DoubleOption<T>(PhantomData<T>);

impl<T, TAs> SerializeAs<Option<Option<T>>> for DoubleOption<TAs>
where
    TAs: SerializeAs<T>,
{
    fn serialize_as<S>(values: &Option<Option<T>>, serializer: S) -> Result<S::Ok, S::Error>
    where
        S: Serializer,
    {
        match values {
            None => serializer.serialize_unit(),
            Some(None) => serializer.serialize_none(),
            Some(Some(v)) => serializer.serialize_some(&SerializeAsWrap::<T, TAs>::new(v)),
        }
    }
}

impl<'de, T, TAs> DeserializeAs<'de, Option<Option<T>>> for DoubleOption<TAs>
where
    TAs: DeserializeAs<'de, T>,
{
    fn deserialize_as<D>(deserializer: D) -> Result<Option<Option<T>>, D::Error>
    where
        D: Deserializer<'de>,
    {
        Ok(
            DeserializeAsWrap::<Option<T>, Option<TAs>>::deserialize(deserializer)?
                .into_inner()
                .map(Some),
        )
    }
}

[ranger-ross]

Thanks for the details response!
I think your solution should work.

I tested and noticed a minor issue where null would result in None when deserializing because .map(Some) would only map to Some when there inner value was also Some. (so getting Some(None) was not possible).
But this was pretty easily fixed by always returning

Some(
  DeserializeAsWrap::<Option<T>, Option<TAs>>::deserialize(deserializer)?
                .into_inner(),
)

since the deserializer for this will only be called when the value is included.