path: root/corgi/src/main.rs

use std::{
	net::{IpAddr, SocketAddr},
	pin::Pin,
	process::Stdio,
	time::Instant,
};

use caller::HttpRequest;
use confindent::{Confindent, Value, ValueParseError};
use http_body_util::{BodyExt, Full};
use hyper::{
	HeaderMap, Request, Response, StatusCode,
	body::{Bytes, Incoming},
	header::HeaderValue,
	server::conn::http1,
	service::Service,
};
use hyper_util::rt::TokioIo;
use regex_lite::Regex;
use tokio::{io::AsyncWriteExt, net::TcpListener, process::Command, runtime::Runtime};

mod caller;

#[derive(Clone, Debug)]
pub struct Settings {
	port: u16,
	scripts: Vec<Script>,
}

#[derive(Clone, Debug, PartialEq)]
pub enum ScriptKind {
	Executable,
	Object,
}

#[derive(Clone, Debug)]
pub struct Script {
	name: String,
	kind: ScriptKind,
	regex: Option<Regex>,
	filename: String,
	env: Vec<(String, String)>,
}

const CONF_DEFAULT: &str = "/etc/corgi.conf";

fn main() {
	let conf_path = std::env::args().nth(1).unwrap_or(String::from(CONF_DEFAULT));
	let conf = Confindent::from_file(conf_path).expect("failed to open conf");

	let mut settings = Settings {
		port: 26744,
		scripts: conf.children("Script").into_iter().map(parse_script_conf).collect(),
	};

	if let Some(server) = conf.child("Server") {
		match server.child_parse("Port") {
			Err(ValueParseError::NoValue) => (),
			Err(err) => {
				eprintln!("Server.Port is malformed: {err}");
				std::process::exit(1);
			}
			Ok(port) => settings.port = port,
		}
	}

	let rt = Runtime::new().unwrap();
	rt.block_on(async { run(settings).await });
}

fn parse_script_conf(conf: &Value) -> Script {
	let name = conf.value_owned().expect("Missing value for 'Script' key");
	let filename = conf.child_owned("Path").expect("Missing 'Path' key");
	let environment = conf.child("Environment");
	let env = environment
		.map(|e| e.values().map(|v| (v.key_owned(), v.value_owned().unwrap())).collect());

	let regex = match conf.get("Match/Regex") {
		None => None,
		Some(restr) => match Regex::new(restr) {
			Err(err) => {
				eprintln!("Failed to compile regex: {restr}\nerror: {err}");
				std::process::exit(1);
			}
			Ok(re) => Some(re),
		},
	};

	let kind = match conf.get("Type") {
		None => ScriptKind::Executable,
		Some("executable") => ScriptKind::Executable,
		Some("object") => ScriptKind::Object,
		Some(kind) => {
			eprintln!("'{kind}' is not a valid script type");
			std::process::exit(1)
		}
	};

	Script {
		name,
		kind,
		regex,
		filename,
		env: env.unwrap_or_default(),
	}
}

// We have tokio::main at home :)
async fn run(settings: Settings) {
	let addr = SocketAddr::from(([0, 0, 0, 0], settings.port));
	let listen = TcpListener::bind(addr).await.unwrap();

	let svc = Svc {
		settings,
		client_addr: addr,
	};

	loop {
		let (stream, caddr) = listen.accept().await.unwrap();
		let io = TokioIo::new(stream);
		let mut svc_clone = svc.clone();
		svc_clone.client_addr = caddr;
		tokio::task::spawn(
			async move { http1::Builder::new().serve_connection(io, svc_clone).await },
		);
	}
}

#[derive(Clone, Debug)]
struct Svc {
	settings: Settings,
	client_addr: SocketAddr,
}

impl Service<Request<Incoming>> for Svc {
	type Response = Response<Full<Bytes>>;
	type Error = hyper::Error;
	type Future = Pin<Box<dyn Future<Output = Result<Self::Response, Self::Error>> + Send>>;

	fn call(&self, req: Request<Incoming>) -> Self::Future {
		let settings = self.settings.clone();
		let caddr = self.client_addr;

		Box::pin(async move { Ok(Self::handle(settings, caddr, req).await) })
	}
}

impl Svc {
	async fn handle(
		settings: Settings,
		caddr: SocketAddr,
		req: Request<Incoming>,
	) -> Response<Full<Bytes>> {
		let start = Instant::now();

		// Collect things we need from the request before we eat it's body
		let method = req.method().as_str().to_ascii_uppercase();
		let version = req.version();
		let path = req.uri().path().to_owned();
		let query = req.uri().query().unwrap_or_default().to_owned();
		let headers = req.headers().clone();

		let body = req.into_body().collect().await.unwrap().to_bytes().to_vec();
		let content_length = body.len();

		let mut maybe_script = None;
		for set_script in settings.scripts {
			if let Some(regex) = set_script.regex.as_ref() {
				if regex.is_match(&path) {
					maybe_script = Some(set_script);
					break;
				}
			} else {
				maybe_script = Some(set_script);
				break;
			}
		}

		let script = match maybe_script {
			Some(script) => script,
			None => {
				eprintln!("path didn't match any script");
				panic!("TODO recover?");
			}
		};

		let content_type = headers
			.get("content-type")
			.map(|s| s.to_str().ok())
			.flatten()
			.unwrap_or_default()
			.to_owned();

		let uagent = headers
			.get("user-agent")
			.map(|s| s.to_str().ok())
			.flatten()
			.unwrap_or_default()
			.to_owned();

		// Find the client address
		let client_addr = {
			let x_forward = Self::parse_addr_from_header(headers.get("x-forwarded-for"));
			let forward = Self::parse_addr_from_header(headers.get("forwarded-for"));

			forward.unwrap_or(x_forward.unwrap_or(caddr.ip()))
		};

		let server_name = headers
			.get("Host")
			.expect("no http host header set")
			.to_str()
			.expect("failed to decode http host as string");

		let http_request = HttpRequest {
			content_type,
			path_info: path.clone(),
			query_string: query,
			remote_addr: client_addr,
			request_method: method,
			script_name: script.filename.to_owned(),
			server_name: server_name.to_owned(),
			server_port: settings.port,
			server_protocol: format!("{:?}", version),
			http_headers: Self::build_http_vec(headers),
			body: if content_length > 0 { Some(body) } else { None },
		};

		let start_cgi = Instant::now();
		let cgi_response = caller::call_and_parse_cgi(script.clone(), http_request).await;
		let cgi_time = start_cgi.elapsed();

		let status = StatusCode::from_u16(cgi_response.status).unwrap();
		let mut response = Response::builder().status(status);

		for (key, value) in cgi_response.headers {
			response = response.header(key, value);
		}

		println!(
			"served to [{client_addr}]\n\tscript: {}\n\tpath: {path}\n\tcgi took {}ms. total time {}ms\n\tUA: {uagent}",
			&script.name,
			cgi_time.as_millis(),
			start.elapsed().as_millis()
		);

		let response_body = cgi_response.body.map(|v| Bytes::from(v)).unwrap_or(Bytes::new());
		response.body(Full::new(response_body)).unwrap()
	}

	fn parse_addr_from_header(maybe_hval: Option<&HeaderValue>) -> Option<IpAddr> {
		maybe_hval.map(|h| h.to_str().ok()).flatten().map(|s| s.parse().ok()).flatten()
	}

	fn build_http_vec(headers: HeaderMap) -> Vec<(String, String)> {
		let mut ret = vec![];

		for (key, value) in headers.iter() {
			let key_str = key.as_str();

			let mut key_upper = String::with_capacity(key_str.len() + 5);
			key_upper.push_str("HTTP_");

			for ch in key_str.chars() {
				match ch {
					_ if ch as u8 > 0x60 && ch as u8 <= 0x7A => {
						key_upper.push((ch as u8 - 0x20) as char);
					}
					'-' => key_upper.push('_'),
					ch => key_upper.push(ch),
				}
			}

			match value.to_str() {
				Ok(val_str) => {
					ret.push((key_upper, val_str.to_owned()));
				}
				Err(err) => {
					eprintln!("value for header {key_str} is not a string: {err}")
				}
			}
		}

		ret
	}
}

fn path_to_name(path: &str) -> String {
	let mut ret = String::with_capacity(path.len());
	for ch in path.chars() {
		match ch {
			'/' => ret.push('-'),
			ch => ret.push(ch),
		}
	}
	ret
}