Partial document struct for updates

[RoDmitry]

Auto-generates a partial document struct for updates. To avoid serde_json::Value as an input.

#[derive(typesense::Typesense, Serialize, Deserialize, Debug, PartialEq, Clone)]
struct Book {
    id: String,
    title: String,
    author: String,
    publication_year: i32,
    #[serde(default, skip_serializing_if = "Option::is_none")]
    in_stock: Option<bool>,
}
let partial_update_struct = BookPartial {
    publication_year: Some(1966),
    in_stock: Some(Some(false)),
    ..Default::default()
};
let updated_book = typed_collection
    .document(&book_1.id)
    .update(&partial_update_struct, None)
    .await
    .expect("Failed to update typed document");

Derive macro creates struct BookPartial with all optional fields.

@haydenhoang what do you think? Need help with deciding about collection_schemaless tests (in typesense/tests/client/documents_test.rs). Do you need collection_schemaless updates and creates? As you know I hate serde_json::Value. Is there any point in implementing it?

Also I don't know if id is updatable.

[haydenhoang]

The partial struct derive is neat!

I couldn't think of any good use cases for collection_schemaless .update() and .create() other than consistency, the user might expect collection_schemaless to have those two methods. All in all, I think it's not worth implementing.

Also I don't know if id is updatable.

id is not updatable

[RoDmitry]

@haydenhoang check it out. Does it look good?

[RoDmitry]

Added a trick to comment an auto-generated code in mustache, if allParams is empty. Otherwise vendorExtensions was not accessible. Did not find any better solution to avoid empty structs.

[haydenhoang]

This looks good! But we need to re-add tests for the bulk documents operations: import, export, bulk update and bulk delete tests

[RoDmitry]

I did not notice that strongly-typed representation tests are different. I thought it's a copy of schemaless. Can you check them? I'm not that good at writing tests. I have left them commented in typesense/tests/client/documents_test.rs.

Also I'm thinking that maybe jsonl methods (import_jsonl, export) are not for everybody. I can see them useful in exporting to file, and then importing it back without parsing. But I think it would be also useful to have methods that take &[Document] as input, and join them with '\n' internally. Also I guess jsonl export must be called export_jsonl.

[haydenhoang]

But I think it would be also useful to have methods that take &[Document] as input, and join them with '\n' internally. Also I guess jsonl export must be called export_jsonl.

Yes, let's implement this in another PR.

Will add those missing tests shortly.

[haydenhoang]

This is ready for merge!

[RoDmitry]

Can't merge my own PR using the button.