filipton / planetscale-driver

Planetscale driver for wasm written in rust

PlanetScale Serverless Driver for Rust™

Rust "version" of database-js. As stated in database-js it uses PlanetScale HTTP api for database queries. It will perfectly run on Cloudflare Workers Or Vercel Edge Functions.

Usage

cargo add planetscale-driver
# also "cargo add anyhow"

Then proceed like in examples!

Examples

Connection and simple SQL execution

use planetscale_driver::PSConnection;

let conn = PSConnection::new(
  "<host>",
  "<user>",
  "<password>",
);
    
let res = conn.execute("SELECT 1").await.unwrap();

fetch_one/fetch_all/fetch_scalar

As you can see, deserialization doesn't use field names (MAYBE IN FUTURE) so remember to write your structs correctly!

use planetscale_driver::{Database, query};

#[derive(Database, Debug)]
struct TestD {
  val: u32
}

// ...

let res: TestD = query("SELECT 1").fetch_one(&conn).await?;
println!("{:?}", res);

let res: Vec<TestD> = query("SELECT val FROM testds").fetch_all(&conn).await?;
println!("{:?}", res);

let res: bool = query("SELECT true").fetch_scalar(&conn).await?;
println!("{:?}", res);

QueryBuilder

If you want to bind safely values into your query, you should use QueryBuilder

Note: now query method is wrapper around QueryBuilder

// ...

// note: values passed to .bind function must have trait ToString 
let id = 69;
let name = "420";

// res there will be empty result, if you want to get reponse data use "execute_raw"
let res = query("INSERT INTO test(id, name) VALUES($0, \"$1\")")
  .bind(id)
  .bind(name)
  .execute(&conn)
  .await?;

Json

use planetscale_driver::{Database, DatabaseJSON, query};

#[derive(Database, Debug)]
pub struct TestD {
    pub val: u32,
    pub test: TestJSON,
}

#[derive(DatabaseJSON, serde::Deserialize, serde::Serialize, Debug)]
pub struct TestJSON {
    pub text: String,
    pub test: u32,
}

// ...

let json: TestJSON = TestJSON {
    text: "test1234321".to_string(),
    test: 1234,
};

let res: TestD = query("SELECT 1010, '$0'").bind(json).fetch_one(&conn).await?;
println!("{:?}", res);

Transactions

// ...

conn.transaction(|conn| async move {
    conn.execute("QUERY")
        .await?;

    conn.execute("OTHER QUERY")
        .await?;

    //     ^- use question mark to propagate errors "up"
    //        it's required if you want to rollback changes after error

    Ok(())
}).await?;

Features

This crate uses reqwest for making http requests. By default, this crate will use the default-tls feature of reqwest which may not build in your environment (e.g. netlify serverless functions). You can override this by disabling the default features of this crate and enabling a different reqwest tls feature like so:

planetscale-driver = {version="0.5.0", default-features=false}
reqwest= {version= "0.11.17", default-features=false, features=["rustls-tls"]}

Warning If you simply disable default features and do not enable a different tls feature, reqwest will panic at runtime.

More examples in the examples folder

If you want to run them:

PS_HOST=<host> PS_USER=<username> PS_PASS=<pscale_password> cargo run --example <example_name>

planetscale-driver's Issues

This Vs Sqlx?

I know I can just use sqlx to query planetscale db ,

What is the difference of using this package?

aside it uses HTTP api of planetscale.... and the query builder you created?

Why would i use this? what use case? would it fit...
i see this is a database-js counter part in rust
which is built for serverless websites

Any performance, benchmark comparison vs http and direct db connection with sqlx?

on query builder whats difference on sqlx?

Also , why not just extend on sqlx , and build on top of it? like error handling on Vitest...

I hope I can have broader info , im highly interested on using planetscale on my next project...

and I wanna know if just sqlx if enough and do i need to use this package (if recommended)

thanks