salvo: A Rust repository from nyxtom

English 简体中文繁體中文

Salvo is an extremely simple and powerful Rust web backend framework. Only basic Rust knowledge is required to develop backend services.

🎯 Features

Built with Hyper and Tokio;
Unified middleware and handle interface;
Routing supports multi-level nesting, and middleware can be added at any level;
Integrated Multipart form processing;
Support Websocket;
Acme support, automatically get TLS certificate from let's encrypt.

⚡️ Quick start

You can view samples here, or view offical website.

Create a new rust project:

cargo new hello_salvo --bin

Add this to Cargo.toml

[dependencies]
salvo = "0.26"
tokio = { version = "1", features = ["macros"] }

Create a simple function handler in the main.rs file, we call it hello_world, this function just render plain text "Hello World".

use salvo::prelude::*;

#[fn_handler]
async fn hello_world(res: &mut Response) {
    res.render(Text::Plain("Hello World"));
}

In the main function, we need to create a root Router first, and then create a server and call it's bind function:

use salvo::prelude::*;

#[fn_handler]
async fn hello_world() -> &'static str {
    "Hello World"
}
#[tokio::main]
async fn main() {
    let router = Router::new().get(hello_world);
    Server::new(TcpListener::bind("127.0.0.1:7878")).serve(router).await;
}

Middleware

There is no difference between Handler and Middleware, Middleware is just Handler. So you can write middlewares without to know concpets like associated type, generic type. You can write middleware if you can write function!!!

use salvo::http::header::{self, HeaderValue};
use salvo::prelude::*;

#[fn_handler]
async fn add_header(res: &mut Response) {
    res.headers_mut()
        .insert(header::SERVER, HeaderValue::from_static("Salvo"));
}

Then add it to router:

Router::new().hoop(add_header).get(hello_world)

This is a very simple middleware, it add Header to Response, view full source code.

Chainable tree routing system

Normally we write routing like this：

Router::with_path("articles").get(list_articles).post(create_article);
Router::with_path("articles/<id>")
    .get(show_article)
    .patch(edit_article)
    .delete(delete_article);

Often viewing articles and article lists does not require user login, but creating, editing, deleting articles, etc. require user login authentication permissions. The tree-like routing system in Salvo can meet this demand. We can write routers without user login together:

Router::with_path("articles")
    .get(list_articles)
    .push(Router::with_path("<id>").get(show_article));

Then write the routers that require the user to login together, and use the corresponding middleware to verify whether the user is logged in:

Router::with_path("articles")
    .hoop(auth_check)
    .post(list_articles)
    .push(Router::with_path("<id>").patch(edit_article).delete(delete_article));

Although these two routes have the same path("articles"), they can still be added to the same parent route at the same time, so the final route looks like this:

Router::new()
    .push(
        Router::with_path("articles")
            .get(list_articles)
            .push(Router::with_path("<id>").get(show_article)),
    )
    .push(
        Router::with_path("articles")
            .hoop(auth_check)
            .post(list_articles)
            .push(Router::with_path("<id>").patch(edit_article).delete(delete_article)),
    );

<id> matches a fragment in the path, under normal circumstances, the article id is just a number, which we can use regular expressions to restrict id matching rules, r"<id:/\d+/>".

You can also use <*> or <**> to match all remaining path fragments. In order to make the code more readable, you can also add appropriate name to make the path semantics more clear, for example: <**file_path>.

Some regular expressions for matching paths need to be used frequently, and it can be registered in advance, such as GUID:

PathFilter::register_part_regex(
    "guid",
    Regex::new("[0-9a-fA-F]{8}-([0-9a-fA-F]{4}-){3}[0-9a-fA-F]{12}").unwrap(),
);

This makes it more concise when path matching is required:

Router::with_path("<id:guid>").get(index)

View full source code

File upload

We can get file async by the function file in Request:

#[fn_handler]
async fn upload(req: &mut Request, res: &mut Response) {
    let file = req.file("file").await;
    if let Some(file) = file {
        let dest = format!("temp/{}", file.name().unwrap_or_else(|| "file".into()));
        if let Err(e) = tokio::fs::copy(&file.path, Path::new(&dest)).await {
            res.set_status_code(StatusCode::INTERNAL_SERVER_ERROR);
        } else {
            res.render("Ok");
        }
    } else {
        res.set_status_code(StatusCode::BAD_REQUEST);
    }
}

Extract data from request

You can easily get data from multiple different data sources and assemble it into the type you want. You can define a custom type first, for example:

#[derive(Serialize, Deserialize, Extractible, Debug)]
/// Get the data field value from the body by default.
#[extract(default_source(from = "body"))]
struct GoodMan<'a> {
    /// The id number is obtained from the request path parameter, and the data is automatically parsed as i64 type.
    #[extract(source(from = "param"))]
    id: i64,
    /// Reference types can be used to avoid memory copying.
    username: &'a str,
    first_name: String,
    last_name: String,
}

Then in Handler you can get the data like this:

#[fn_handler]
async fn edit(req: &mut Request) -> String {
    let good_man: GoodMan<'_> = req.extract().await.unwrap();
}

You can even pass the type directly to the function as a parameter, like this:

#[fn_handler]
async fn edit<'a>(good_man: GoodMan<'a>) -> String {
    res.render(Json(good_man));
}

There is considerable flexibility in the definition of data types, and can even be resolved into nested structures as needed:

#[derive(Serialize, Deserialize, Extractible, Debug)]
#[extract(default_source(from = "body", format = "json"))]
struct GoodMan<'a> {
    #[extract(source(from = "param"))]
    id: i64,
    #[extract(source(from = "query"))]
    username: &'a str,
    first_name: String,
    last_name: String,
    lovers: Vec<String>,
    /// The nested field is completely reparsed from Request.
    #[extract(source(from = "request"))]
    nested: Nested<'a>,
}

#[derive(Serialize, Deserialize, Extractible, Debug)]
#[extract(default_source(from = "body", format = "json"))]
struct Nested<'a> {
    #[extract(source(from = "param"))]
    id: i64,
    #[extract(source(from = "query"))]
    username: &'a str,
    first_name: String,
    last_name: String,
    #[extract(rename = "lovers")]
    #[serde(default)]
    pets: Vec<String>,
}

View full source code

More Examples

Your can find more examples in examples folder. You can run these examples with the following command:

cargo run --bin example-basic-auth

You can use any example name you want to run instead of basic-auth here.

There is a real and open source project use Salvo: https://github.com/driftluo/myblog.

🚀 Performance

Benchmark testing result can be found from here:

https://web-frameworks-benchmark.netlify.app/result?l=rust

https://tfb-status.techempower.com/

🩸 Contributing

Contributions are absolutely, positively welcome and encouraged! Contributions come in many forms. You could:

Submit a feature request or bug report as an issue;
Comment on issues that require feedback;
Contribute code via pull requests;
Publish Salvo-related technical articles on blogs or technical platforms。

All pull requests are code reviewed and tested by the CI. Note that unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in Salvo by you shall be dual licensed under the MIT License, without any additional terms or conditions.

☕ Supporters

Salvo is an open source project. If you want to support Salvo, you can ☕ buy a coffee here.

⚠️ License

Salvo is licensed under either of

Apache License, Version 2.0, (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

nyxtom/salvo