DynamoDB avoid to insert values when they are None

[ymwjbxxq]

Hello,

imagine this struct:

struct Person {
    name: String,

    #[serde(skip_serializing_if = "Option::is_none")]
    surname: Option<String>,
}

If I POST my JSON:

{
  name: daniele
}

I would like to see it in DynamoDB as:

{
 "name": {
  "S": "daniele"
 }
}

At the moment, I have to default always the value because:

 .expression_attribute_values(":name",AttributeValue::S(request.name.clone()))
.expression_attribute_values(":surname",AttributeValue::S(request.surname.clone().unwrap_or("".to_string())))

And I end up with:

{
 "name": {
  "S": "daniele"
 },
"surname": {
  "S": ""
 }
}

Is there any way to do that? Like an option to configure to declare to skip the None?

Of course, I can do this:

 let mut update_expression: String = "SET name = :name".to_owned();
 if request.surname.is_some() {
        query = query
                    .expression_attribute_values(":surname",AttributeValue::S(request.surname.clone().unwrap()));
       update_expression.push_str(", surname = :surname ");
      }

But imagine a big struct. I just wish or would like to know if there is a way to instruct the SDK to skip None and have much less code around.

Thanks,
Dan

[jdisanti]

I don't think there is a great non-verbose solution for this right now since we haven't implemented any kind of high level DynamoDB library (see discussion on #70). In the meantime, I know of at least one library in crates.io that does this for you. Otherwise, you may be able to move the bulk of this conditional logic into a macro.

[ymwjbxxq]

Thanks for the feedback. I would like something aws official and not another library even if it comes from serde people.
I wish to have the SDK more helpful and follow the official way.

Could you point me to some macro examples that I can use? I am not very familiar with that.