Dynarust

An opinionated DynamoDB ODM library for Rust that maps structs to native Dynamo items using serde and serde_json.

It only works with tables with a HASH key and a RANGE key, referred in the code as pk and sk. This setup can be useful for a very wide variety of use cases.

Usage

You will need the following deps to your Cargo.toml, tweak the versions to your needs:

[dependencies]
dynarust = "*"
serde = "*"
tokio = { version = "*", features = ["full"] }

// Spawn a local dynamo: `docker run -p 8000:8000 amazon/dynamodb-local`
use dynarust::serde::{Deserialize, Serialize};
use dynarust::serde_json::json;

#[derive(Serialize, Deserialize)]
struct Car {
    brand: String,
    model: String,
    horse_power: i64,
}

impl dynarust::Resource for Car {
    fn table() -> String { "Cars".into() }
    fn pk_sk(&self) -> (String, String) { (self.brand.clone(), self.model.clone()) }
}

#[tokio::main]
async fn main() {
    let client = dynarust::Client::local().await;
    client.create_table::<Car>(None).await.unwrap();

    let car = Car {
        brand: "Tesla".into(),
        model: "Y".into(),
        horse_power: 347,
    };

    client.create(&car).await.unwrap();

    client.update(&car, json!({ "horse_power": 534 })).await.unwrap();

    let tesla = client.get::<Car>(("Tesla".into(), "Y".into())).await.unwrap();

    assert_eq!(tesla.unwrap().horse_power, 534)
}

Transactions

Dynarust's api is very focused on transaction ergonomics, and offers a way or doing transactional operations almost as if they were just normal operations

// Spawn a local dynamo: `docker run -p 8000:8000 amazon/dynamodb-local`
use dynarust::serde::{Deserialize, Serialize};
use dynarust::serde_json::json;

#[derive(Serialize, Deserialize)]
struct Car {
    brand: String,
    model: String,
    n_modifications: i64,
}

impl dynarust::Resource for Car {
    fn table() -> String { "Cars".into() }
    fn pk_sk(&self) -> (String, String) { (self.brand.clone(), self.model.clone()) }
}

#[derive(Serialize, Deserialize)]
struct CarModification {
    car_brand: String,
    car_model: String,
    modification: String,
    price: f64,
}

impl dynarust::Resource for CarModification {
    fn table() -> String { "CarModifications".into() }
    fn pk_sk(&self) -> (String, String) {(
        format!("{}-{}", self.car_brand, self.car_model),
        self.modification.clone(),
    )}
}

#[tokio::main]
async fn main() {
    let client = dynarust::Client::local().await;
    client.create_table::<Car>(None).await.unwrap();
    client.create_table::<CarModification>(None).await.unwrap();

    let car = Car {
        brand: "Tesla".into(),
        model: "Y".into(),
        n_modifications: 0,
    };

    client.create(&car).await.unwrap();

    let car_modification = CarModification {
        car_brand: "Tesla".into(),
        car_model: "Y".into(),
        modification: "Loud exhaust".into(),
        price: 14999.95,
    };

    let mut context = dynarust::begin_transaction();
    dynarust::transact_create(&car_modification, &mut context).unwrap();
    dynarust::transact_update(&car, json!({ "n_modifications": car.n_modifications + 1 }), &mut context).unwrap();
    client.execute_transaction(context).await.unwrap();

    let tesla = client.get::<Car>(("Tesla".into(), "Y".into())).await.unwrap();

    assert_eq!(tesla.unwrap().n_modifications, 1)
}

Setup

In order to use this library, the table setup would need to look like this equivalent CFN template:

DynamoDBTable:
  Type: AWS::DynamoDB::Table
  Properties:
    TableName: my-table
    AttributeDefinitions:
      - AttributeName: PrimaryKey
        AttributeType: S
      - AttributeName: SecondaryKey
        AttributeType: S
    KeySchema:
      - AttributeName: PrimaryKey
        KeyType: HASH
      - AttributeName: SecondaryKey
        KeyType: RANGE

dynarust expects that attributes PrimaryKey and SecondaryKey are declared in the key schema as HASH and RANGE values, and that both are of type S (string).

Coming next

• Support for global secondary indexes, with the same API that exists right now for the primary ones.

Current Limitations

• Only works with the HASH and RANGE keys setup.
• It is not compatible with global or local secondary indexes.
• Missing some batch operations, right now only batch_get is implemented.