#embedded-database #document-database #store #async #db #serde

reddb

Minimalistic in-memory embedded database with persistance

4 releases

0.2.3 Jan 3, 2021
0.2.2 Jan 2, 2021
0.2.1 Dec 30, 2020
0.2.0 Dec 28, 2020

#2414 in Database interfaces

MIT/Apache

40KB
893 lines

RedDb

Actions Status Crates.io

RedDb is an async, fast, lightweight and embedded in-memory document database with persistance in different serde-compatible formats (ron and json at the moment and bindcode and cbor soon). RedDb uses Tokio fort its easy to use async API for inserting, finding, updating and deleting data.

Quickstart

Add RedDb to your Cargo.toml specifing what serializer you want to use:

[dependencies.RedDb]
version = "0.2.3"
features = ["ron_ser"] # Ron serialization / deserialization
features = ["json_ser"] # Json serialization / deserialization

use reddb::{Document, RonDb};

#[derive(Clone, Serialize, PartialEq, Deserialize)]
struct MyStruct {
  foo: String,
}

#[tokio::main]
async fn main() -> Result<()> {
  // RedDb with RON persistance for MyStruct structs
  let db = RonDb::new::<MyStruct>("my.db").unwrap();
  let my_struct = MyStruct {
    foo: String::from("hello")
  };

  // Insert data
  let doc = db.insert_one(my_struct).await?;
  // Find by id
  let my_doc: Document<MyStruct> = db.find_one(&doc._id).await?;
  // Find all records equal to my_struct
  let my_docs : Vec<Document<MyStruct>> = db.find(&my_struct).await?;
  Ok(())
}

Why

RedDb is the migration of a side project originally written in NodeJs that was designed to store objects in memory (with hd persistance) and do searchs on them.

When

If you are looking for a classic Key/Value storage you will find better options since RedDb is not a Key/Value (RedDb uses autogeneratd Uuids). You can store any kind of data since data will be handled as a generic but RedDb was designed to store Objects/Structs and peform basic search operations in those Structs. Said that, if yo if you are looking for an lightweight and easy to use in-memory data store with persistance, RedDb could be a good choice!

API

Data

Data is serialized and deserialized in different serde-compatible formats (json, ron, yaml) and wrapped into the Document struct as follows:

pub struct Document<T> {
  pub _id: Uuid,
  pub data: T,
  pub _st: Status,
}

Since data field is a generic you can store any kind of data you want. As you will see on the API, Document<T> is the default return type for most operations.

Persistance

RedDb's persistence uses an append-only format (AOF) so all write operations (Insert, Update, Delete) are added to to the end of the database file. The database is automatically compacted in just one line per object/record everytime you start the database in your application.

The API provides bulk-like write operations (insert, update and delete) for vectors of data that are faster to persist due to hd sync operations. Use them instead iterate over the *_one() methods you'll see on the API.

Inserting Data

Insert data is pretty straightforward. If you want to insert just one document use insert_one method:

Insert one

#[derive(Clone, Serialize, PartialEq, Deserialize)]
struct MyStruct {
  foo: String,
}

let my_struct = MyStruct {
  foo: String::from("hello")
};

let doc: Document<TestStruct> = store.insert_one(my_struct).await?;
println!("{:?}", doc._id);
// 94d69737-4b2e-4985-aaa1-e28bbff2e6d0

Insert

If you want to insert a vector of data insert() is more suitable and faster to persists than iterate over insert_one() method due to the nature of the AOF persistance.

let my_docs = vec![MyStruct {
  foo: String::from("one"),
},
MyStruct {
  foo: String::from("two"),
}];

let docs: Vec<Document<MyStruct>> = db.insert(my_docs).await?;

Finding Data

There are two ways to find your data. By it's id or looking into the database what data matches your query.

Find one

Performs a search by id.

let my_struct = MyStruct {
  foo: String::from("hello")
};

let inserted_doc = db.insert_one(my_struct).await?;
let doc: Document<MyStruct> = db.find_one(&inserted_doc._id).await?;

Find

Look into the database for data matching your query.

let one = MyStruct {
  foo: String::from("Hello"),
};

let two = MyStruct {
  foo: String::from("Hello"),
};

let three = MyStruct {
  foo: String::from("Bye"),
};


let many = vec![one.clone(), two.clone(), three.clone()];
let inserted_doc : Document<MyStruct> = db.insert(many).await?;
let docs: Vec<Document<MyStruct>> = db.find(&one).await?;

Updating Data

Update data is pretty straightforward. You can update data

Update one

Update one record, using it's id as search param.

let my_struct = MyStruct {
  foo: String::from("hello")
};

let new_value = MyStruct {
  foo: String::from("bye"),
};

let inserted_doc = db.insert_one(my_struct).await?;
let updated: bool = db.update_one(&inserted_doc._id, new_value)).await?;

Update

You can update all data in the databas that matches your query param. Update will return the number of updated documents.

let search = MyStruct {
  foo: String::from("hello")
};

let new_value = MyStruct {
  foo: String::from("bye"),
};

let updated: usize = store.update(&search, &new_value).await?;

Deleting Data

Delete one

Delete a record by it's id.

let my_struct = MyStruct {
  foo: String::from("hello")
};

let doc = db.insert_one(my_struct).await?;
let deleted : bool = db.delete_one(&doc._id)).await?;

Delete

Like in update method, this method will lookup into the database which data matches your query and then delete it.

let search = MyStruct {
  foo: String::from("hello")
};

let deleted = store.delete(&search).await?;
println!("{:?}", updated);
// 1

License

This library is licensed under

Dependencies

~6.5MB
~110K SLoC