Easy boilerplate utilities for Rust http services which use async-std, Tide, Surf, and friends.
Allows for service setup with feature-configured built-ins for maximum service consistency with low developer overhead, and for easily integration testing the service without using a live network.
Scroll to the bottom for API Reference
use std::sync::Arc;
use tide::{Request, Route};
struct AppState {
greeting: &'static str,
}
type AppRequest = Request<Arc<AppState>>;
async fn setup_app_state() -> preroll::SetupResult<AppState> {
Ok(AppState {
greeting: "Hello World!",
})
}
fn setup_routes(mut server: Route<'_, Arc<AppState>>) {
server
.at("hello-world")
.get(|req: AppRequest| async move {
Ok(req.state().greeting)
});
}
// The "magic" happens here!
preroll::main!("hello-world", setup_app_state, setup_routes);- Boilerplate
mainsetup viapreroll::main!, with optional features automatically configured. - A
preroll::prelude::*;with all extension traits. - Response logging with many details.
- Automatic JSON responses for errors in the form of
JsonError. - Test utils with easy mock client setup.
Add-on features must be enabled via cargo features, e.g.
[dependencies.preroll]
version = "0.5"
features = ["honeycomb", "postgres"]"honeycomb": Enables tracing to honeycomb.io.- Env variable
HONEYCOMBIO_WRITE_KEY(required). - Env variable
TRACELEVEL, sets the tracing level filter, defaults toinfo. - Writes to a dataset named
{service_name}-{environment}.service_nameis frompreroll::main!("service_name", ...).environmentis fromENVIRONMENT, or defaults to"development".
- Env variable
"lambda-http": Changes the HTTP listener to connect to an AWS Lambda execution environment.- Is no longer reachable as a regular http server, but accepts http lambda requests as if it were one.
- Some environment variables, such as
PORT, are disregarded. - If the
"honeycomb"feature is enabled, trace events are written to stdout, and must be collected via a layer provided by Honeycomb. See: https://docs.honeycomb.io/getting-data-in/integrations/aws/aws-lambda/
"postgres": Enables a postgres connection pool with transactions.- Env variable
PGURL, which should be a properly formattedpostgres://database url.- Defaults to
"postgres://localhost/{service_name}"(default postgres port). service_nameis frompreroll::main!("service_name", ...).
- Defaults to
- Env variable
PGMAXCONNECTIONS, default 5 connections. - Env variable
PGMAXLIFETIME, default30(minutes). - Enables [
PostgresRequestExt][prelude::PostgresRequestExt] and [test_utils::create_client_and_postgres][].
- Env variable
"panic-on-error": Makes the response logger [panic][] on error rather than log.- Do not use in production. Prevents
--releasecompilation.
- Do not use in production. Prevents
The following environment variables are read during preroll::main!:
ENVIRONMENT: If this starts withprod, load the production-mode JSON logger, avoid.env.FORCE_DOTENV: Override production-mode, force-load environment from.env.HOST: Sets the hostname that this service will listen on. Defaults to"127.0.0.1".LOGLEVEL: Set the logger's level filter, defaults toinfoin production-mode,debugin development-mode.PORT: Sets the port that this service will listen on. Defaults to8080.
This crate is intentionally somewhat prescriptive in how it templates a service and the interaction with add-on features such as Postgres (via SQLx).
Licensed under the BlueOak Model License 1.0.0 — Contributions via DCO 1.1