garage/src/web/web_server.rs

235 lines
6.5 KiB
Rust
Raw Normal View History

2020-11-21 16:50:19 +00:00
use std::{borrow::Cow, convert::Infallible, net::SocketAddr, sync::Arc};
2020-11-02 14:48:39 +00:00
use futures::future::Future;
2020-11-21 14:15:25 +00:00
use hyper::{
header::HOST,
server::conn::AddrStream,
service::{make_service_fn, service_fn},
2020-11-21 16:50:19 +00:00
Body, Request, Response, Server,
};
2020-11-02 14:48:39 +00:00
2020-11-20 20:23:32 +00:00
use idna::domain_to_unicode;
2020-11-21 16:50:19 +00:00
use crate::error::*;
use garage_api::s3_get::handle_get;
2020-11-02 14:48:39 +00:00
use garage_model::garage::Garage;
2020-11-19 13:56:00 +00:00
use garage_util::error::Error as GarageError;
2020-11-02 14:48:39 +00:00
pub async fn run_web_server(
garage: Arc<Garage>,
shutdown_signal: impl Future<Output = ()>,
2020-11-19 13:56:00 +00:00
) -> Result<(), GarageError> {
2020-11-10 08:57:07 +00:00
let addr = &garage.config.s3_web.bind_addr;
2020-11-02 14:48:39 +00:00
let service = make_service_fn(|conn: &AddrStream| {
let garage = garage.clone();
let client_addr = conn.remote_addr();
2020-11-10 08:59:52 +00:00
async move {
2020-11-02 14:48:39 +00:00
Ok::<_, Error>(service_fn(move |req: Request<Body>| {
let garage = garage.clone();
2020-11-11 20:17:34 +00:00
handle_request(garage, req, client_addr)
2020-11-02 14:48:39 +00:00
}))
}
});
let server = Server::bind(&addr).serve(service);
let graceful = server.with_graceful_shutdown(shutdown_signal);
info!("Web server listening on http://{}", addr);
graceful.await?;
Ok(())
}
2020-11-08 14:47:25 +00:00
2020-11-11 20:17:34 +00:00
async fn handle_request(
2020-11-08 14:47:25 +00:00
garage: Arc<Garage>,
req: Request<Body>,
addr: SocketAddr,
2020-11-11 20:17:34 +00:00
) -> Result<Response<Body>, Infallible> {
info!("{} {} {}", addr, req.method(), req.uri());
let res = serve_file(garage, req).await;
match &res {
Ok(r) => debug!("{} {:?}", r.status(), r.headers()),
Err(e) => warn!("Response: error {}, {}", e.http_status_code(), e),
}
Ok(res.unwrap_or_else(error_to_res))
}
fn error_to_res(e: Error) -> Response<Body> {
let body: Body = Body::from(format!("{}\n", e));
let mut http_error = Response::new(body);
*http_error.status_mut() = e.http_status_code();
http_error
}
async fn serve_file(garage: Arc<Garage>, req: Request<Body>) -> Result<Response<Body>, Error> {
2020-11-10 08:57:07 +00:00
// Get http authority string (eg. [::1]:3902 or garage.tld:80)
2020-11-08 14:47:25 +00:00
let authority = req
.headers()
.get(HOST)
.ok_or(Error::BadRequest(format!("HOST header required")))?
.to_str()?;
2020-11-10 08:57:07 +00:00
// Get bucket
2020-11-20 20:23:32 +00:00
let (host, _) = domain_to_unicode(authority_to_host(authority)?);
2020-11-10 08:59:52 +00:00
let root = &garage.config.s3_web.root_domain;
let bucket = host_to_bucket(&host, root);
2020-11-10 08:57:07 +00:00
// Get path
let path = req.uri().path().to_string();
2020-11-11 18:48:01 +00:00
let index = &garage.config.s3_web.index;
let key = path_to_key(&path, &index)?;
2020-11-10 08:59:52 +00:00
2020-11-11 18:48:01 +00:00
info!("Selected bucket: \"{}\", selected key: \"{}\"", bucket, key);
2020-11-21 16:50:19 +00:00
let r = handle_get(garage, &req, bucket, &key).await?;
2020-11-21 14:15:25 +00:00
2020-11-21 16:50:19 +00:00
Ok(r)
2020-11-08 14:47:25 +00:00
}
2020-11-08 15:02:16 +00:00
/// Extract host from the authority section given by the HTTP host header
///
/// The HTTP host contains both a host and a port.
/// Extracting the port is more complex than just finding the colon (:) symbol due to IPv6
2020-11-10 14:48:40 +00:00
/// We do not use the collect pattern as there is no way in std rust to collect over a stack allocated value
/// check here: https://docs.rs/collect_slice/1.2.0/collect_slice/
fn authority_to_host(authority: &str) -> Result<&str, Error> {
let mut iter = authority.chars().enumerate();
2020-11-10 14:48:40 +00:00
let (_, first_char) = iter
.next()
.ok_or(Error::BadRequest(format!("Authority is empty")))?;
let split = match first_char {
'[' => {
let mut iter = iter.skip_while(|(_, c)| c != &']');
2020-11-10 14:52:20 +00:00
iter.next().expect("Authority parsing logic error");
2020-11-10 14:48:40 +00:00
iter.next()
2020-11-10 14:40:33 +00:00
}
2020-11-10 14:48:40 +00:00
_ => iter.skip_while(|(_, c)| c != &':').next(),
};
2020-11-10 14:40:33 +00:00
match split {
Some((i, ':')) => Ok(&authority[..i]),
None => Ok(authority),
2020-11-10 14:40:33 +00:00
Some((_, _)) => Err(Error::BadRequest(format!(
"Authority {} has an illegal format",
authority
))),
2020-11-08 14:47:25 +00:00
}
}
2020-11-10 16:05:10 +00:00
/// Host to bucket
///
/// Convert a host, like "bucket.garage-site.tld" or "john.doe.com"
/// to the corresponding bucket, resp. "bucket" and "john.doe.com"
/// considering that ".garage-site.tld" is the "root domain".
/// This behavior has been chosen to follow AWS S3 semantic.
2020-11-10 08:57:07 +00:00
fn host_to_bucket<'a>(host: &'a str, root: &str) -> &'a str {
if root.len() >= host.len() || !host.ends_with(root) {
return host;
}
let len_diff = host.len() - root.len();
let missing_starting_dot = root.chars().next() != Some('.');
2020-11-10 08:59:52 +00:00
let cursor = if missing_starting_dot {
len_diff - 1
} else {
len_diff
};
&host[..cursor]
2020-11-10 08:57:07 +00:00
}
2020-11-11 18:48:01 +00:00
/// Path to key
///
/// Convert the provided path to the internal key
/// When a path ends with "/", we append the index name to match traditional web server behavior
/// which is also AWS S3 behavior.
fn path_to_key<'a>(path: &'a str, index: &str) -> Result<Cow<'a, str>, Error> {
let path_utf8 = percent_encoding::percent_decode_str(&path).decode_utf8()?;
2020-11-21 16:50:19 +00:00
2020-11-21 11:01:02 +00:00
if path_utf8.chars().next() != Some('/') {
return Err(Error::BadRequest(format!(
"Path must start with a / (slash)"
2020-11-21 16:50:19 +00:00
)));
2020-11-21 11:01:02 +00:00
}
2020-11-11 18:48:01 +00:00
match path_utf8.chars().last() {
None => Err(Error::BadRequest(format!(
"Path must have at least a character"
))),
Some('/') => {
let mut key = String::with_capacity(path_utf8.len() + index.len());
2020-11-21 11:01:02 +00:00
key.push_str(&path_utf8[1..]);
2020-11-11 18:48:01 +00:00
key.push_str(index);
Ok(key.into())
}
2020-11-21 16:50:19 +00:00
Some(_) => match path_utf8 {
Cow::Borrowed(pu8) => Ok((&pu8[1..]).into()),
Cow::Owned(pu8) => Ok((&pu8[1..]).to_string().into()),
},
2020-11-11 18:48:01 +00:00
}
}
2020-11-08 14:47:25 +00:00
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn authority_to_host_with_port() -> Result<(), Error> {
let domain = authority_to_host("[::1]:3902")?;
assert_eq!(domain, "[::1]");
let domain2 = authority_to_host("garage.tld:65200")?;
assert_eq!(domain2, "garage.tld");
let domain3 = authority_to_host("127.0.0.1:80")?;
assert_eq!(domain3, "127.0.0.1");
Ok(())
}
#[test]
fn authority_to_host_without_port() -> Result<(), Error> {
let domain = authority_to_host("[::1]")?;
assert_eq!(domain, "[::1]");
let domain2 = authority_to_host("garage.tld")?;
assert_eq!(domain2, "garage.tld");
let domain3 = authority_to_host("127.0.0.1")?;
assert_eq!(domain3, "127.0.0.1");
Ok(())
}
2020-11-10 08:57:07 +00:00
#[test]
fn host_to_bucket_test() {
2020-11-10 08:59:52 +00:00
assert_eq!(
2020-11-10 08:57:07 +00:00
host_to_bucket("john.doe.garage.tld", ".garage.tld"),
2020-11-10 08:59:52 +00:00
"john.doe"
);
2020-11-10 08:57:07 +00:00
assert_eq!(
2020-11-10 08:59:52 +00:00
host_to_bucket("john.doe.garage.tld", "garage.tld"),
"john.doe"
);
assert_eq!(host_to_bucket("john.doe.com", "garage.tld"), "john.doe.com");
2020-11-10 08:57:07 +00:00
assert_eq!(
host_to_bucket("john.doe.com", ".garage.tld"),
2020-11-10 08:59:52 +00:00
"john.doe.com"
);
assert_eq!(host_to_bucket("garage.tld", "garage.tld"), "garage.tld");
assert_eq!(host_to_bucket("garage.tld", ".garage.tld"), "garage.tld");
2020-11-10 08:57:07 +00:00
}
2020-11-11 18:48:01 +00:00
#[test]
fn path_to_key_test() -> Result<(), Error> {
2020-11-21 11:01:02 +00:00
assert_eq!(path_to_key("/file%20.jpg", "index.html")?, "file .jpg");
assert_eq!(path_to_key("/%20t/", "index.html")?, " t/index.html");
assert_eq!(path_to_key("/", "index.html")?, "index.html");
assert_eq!(path_to_key("/hello", "index.html")?, "hello");
2020-11-11 18:48:01 +00:00
assert!(path_to_key("", "index.html").is_err());
2020-11-21 11:01:02 +00:00
assert!(path_to_key("i/am/relative", "index.html").is_err());
2020-11-11 18:48:01 +00:00
Ok(())
}
2020-11-08 14:47:25 +00:00
}