Deuxfleurs/garage
42
39
Fork 55
Code Issues 73 Pull requests 7 Releases 29 Wiki Activity
garage/src/web/web_server.rs

343 lines
9.3 KiB
Rust
Raw Normal View History

Build path correctly
2020-11-11 18:48:01 +00:00
use std::borrow::Cow;
Build error
2020-11-11 20:17:34 +00:00
use std::convert::Infallible;
Parse host header
2020-11-08 14:47:25 +00:00
use std::net::SocketAddr;
Fix formatting
2020-11-10 08:59:52 +00:00
use std::sync::Arc;
We are able to serve a file
2020-11-21 14:15:25 +00:00
use std::time::{Duration, UNIX_EPOCH};
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
use futures::future::Future;
We are able to serve a file
2020-11-21 14:15:25 +00:00
use futures::stream::*;
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
We are able to serve a file
2020-11-21 14:15:25 +00:00
use hyper::{
header::HOST,
body::Bytes,
server::conn::AddrStream,
service::{make_service_fn, service_fn},
Body, Request, Response, Server, StatusCode};
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
Support punnycode
2020-11-20 20:23:32 +00:00
use idna::domain_to_unicode;
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
use garage_model::garage::Garage;
We are able to serve a file
2020-11-21 14:15:25 +00:00
use garage_model::object_table::*;
use garage_table::EmptyKey;
Make it compile again
2020-11-19 13:56:00 +00:00
use garage_util::error::Error as GarageError;
use crate::error::*;
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
pub async fn run_web_server(
garage: Arc<Garage>,
shutdown_signal: impl Future<Output = ()>,
Make it compile again
2020-11-19 13:56:00 +00:00
) -> Result<(), GarageError> {
Extract bucket
2020-11-10 08:57:07 +00:00
let addr = &garage.config.s3_web.bind_addr;
Skeleton to the new web API
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();
Fix formatting
2020-11-10 08:59:52 +00:00
async move {
Skeleton to the new web API
2020-11-02 14:48:39 +00:00
Ok::<_, Error>(service_fn(move |req: Request<Body>| {
let garage = garage.clone();
Build error
2020-11-11 20:17:34 +00:00
handle_request(garage, req, client_addr)
Skeleton to the new web API
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(())
}
Parse host header
2020-11-08 14:47:25 +00:00
Build error
2020-11-11 20:17:34 +00:00
async fn handle_request(
Parse host header
2020-11-08 14:47:25 +00:00
garage: Arc<Garage>,
req: Request<Body>,
addr: SocketAddr,
Build error
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> {
Extract bucket
2020-11-10 08:57:07 +00:00
// Get http authority string (eg. [::1]:3902 or garage.tld:80)
Parse host header
2020-11-08 14:47:25 +00:00
let authority = req
.headers()
.get(HOST)
.ok_or(Error::BadRequest(format!("HOST header required")))?
.to_str()?;
Extract bucket
2020-11-10 08:57:07 +00:00
// Get bucket
Support punnycode
2020-11-20 20:23:32 +00:00
let (host, _) = domain_to_unicode(authority_to_host(authority)?);
Fix formatting
2020-11-10 08:59:52 +00:00
let root = &garage.config.s3_web.root_domain;
let bucket = host_to_bucket(&host, root);
Extract bucket
2020-11-10 08:57:07 +00:00
// Get path
let path = req.uri().path().to_string();
Build path correctly
2020-11-11 18:48:01 +00:00
let index = &garage.config.s3_web.index;
let key = path_to_key(&path, &index)?;
Fix formatting
2020-11-10 08:59:52 +00:00
Build path correctly
2020-11-11 18:48:01 +00:00
info!("Selected bucket: \"{}\", selected key: \"{}\"", bucket, key);
// Get bucket descriptor
WIP fetch object
2020-11-11 14:24:25 +00:00
let object = garage
.object_table
.get(&bucket.to_string(), &key.to_string())
.await?
.ok_or(Error::NotFound)?;
We are able to serve a file
2020-11-21 14:15:25 +00:00
// Get last complete version descriptor
let last_v = object
.versions()
.iter()
.rev()
.filter(|v| v.is_complete())
.next()
.ok_or(Error::NotFound)?;
// Unwrap version
let last_v_data = match &last_v.state {
ObjectVersionState::Complete(x) => x,
_ => unreachable!(),
};
// Get metadata from version
let last_v_meta = match last_v_data {
ObjectVersionData::DeleteMarker => return Err(Error::NotFound),
ObjectVersionData::Inline(meta, _) => meta,
ObjectVersionData::FirstBlock(meta, _) => meta,
};
// @FIXME Support range
// Set headers
let resp_builder = object_headers(&last_v, last_v_meta).status(StatusCode::OK);
// Stream body
match &last_v_data {
ObjectVersionData::DeleteMarker => unreachable!(),
ObjectVersionData::Inline(_, bytes) => {
let body: Body = Body::from(bytes.to_vec());
Ok(resp_builder.body(body)?)
}
ObjectVersionData::FirstBlock(_, first_block_hash) => {
let read_first_block = garage.block_manager.rpc_get_block(&first_block_hash);
let get_next_blocks = garage.version_table.get(&last_v.uuid, &EmptyKey);
let (first_block, version) = futures::try_join!(read_first_block, get_next_blocks)?;
let version = version.ok_or(Error::NotFound)?;
let mut blocks = version
.blocks()
.iter()
.map(|vb| (vb.hash, None))
.collect::<Vec<_>>();
blocks[0].1 = Some(first_block);
let body_stream = futures::stream::iter(blocks)
.map(move |(hash, data_opt)| {
let garage = garage.clone();
async move {
if let Some(data) = data_opt {
Ok(Bytes::from(data))
} else {
garage
.block_manager
.rpc_get_block(&hash)
.await
.map(Bytes::from)
}
}
})
.buffered(2);
//let body: Body = Box::new(StreamBody::new(Box::pin(body_stream)));
let body = hyper::body::Body::wrap_stream(body_stream);
Ok(resp_builder.body(body)?)
}
}
}
// Copied from api/s3_get.rs
fn object_headers(
version: &ObjectVersion,
version_meta: &ObjectVersionMeta,
) -> http::response::Builder {
let date = UNIX_EPOCH + Duration::from_millis(version.timestamp);
let date_str = httpdate::fmt_http_date(date);
let mut resp = Response::builder()
.header(
"Content-Type",
version_meta.headers.content_type.to_string(),
)
.header("Content-Length", format!("{}", version_meta.size))
.header("ETag", version_meta.etag.to_string())
.header("Last-Modified", date_str)
.header("Accept-Ranges", format!("bytes"));
for (k, v) in version_meta.headers.other.iter() {
resp = resp.header(k, v.to_string());
}
resp
Parse host header
2020-11-08 14:47:25 +00:00
}
Add some documentation
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.
Rewrite authority to host while staying on stack
2020-11-10 14:26:48 +00:00
/// Extracting the port is more complex than just finding the colon (:) symbol due to IPv6
Rewrite for clarity
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
Rewrite authority to host while staying on stack
2020-11-10 14:26:48 +00:00
/// 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();
Rewrite for clarity
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 != &']');
Panic when it is a logical error
2020-11-10 14:52:20 +00:00
iter.next().expect("Authority parsing logic error");
Rewrite for clarity
2020-11-10 14:48:40 +00:00
iter.next()
Fix indent again
2020-11-10 14:40:33 +00:00
}
Rewrite for clarity
2020-11-10 14:48:40 +00:00
_ => iter.skip_while(|(_, c)| c != &':').next(),
Rewrite authority to host while staying on stack
2020-11-10 14:26:48 +00:00
};
Fix indent again
2020-11-10 14:40:33 +00:00
match split {
Rewrite authority to host while staying on stack
2020-11-10 14:26:48 +00:00
Some((i, ':')) => Ok(&authority[..i]),
None => Ok(authority),
Fix indent again
2020-11-10 14:40:33 +00:00
Some((_, _)) => Err(Error::BadRequest(format!(
"Authority {} has an illegal format",
authority
))),
Parse host header
2020-11-08 14:47:25 +00:00
}
}
Add documentation to host_to_bucket
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.
Extract bucket
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('.');
Fix formatting
2020-11-10 08:59:52 +00:00
let cursor = if missing_starting_dot {
len_diff - 1
} else {
len_diff
};
&host[..cursor]
Extract bucket
2020-11-10 08:57:07 +00:00
}
Build path correctly
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()?;
Fix host to key
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)"
)))
}
Build path correctly
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());
Fix host to key
2020-11-21 11:01:02 +00:00
key.push_str(&path_utf8[1..]);
Build path correctly
2020-11-11 18:48:01 +00:00
key.push_str(index);
Ok(key.into())
}
Fix host to key
2020-11-21 11:01:02 +00:00
Some(_) => {
match path_utf8 {
Cow::Borrowed(pu8) => Ok((&pu8[1..]).into()),
Cow::Owned(pu8) => Ok((&pu8[1..]).to_string().into()),
}
}
Build path correctly
2020-11-11 18:48:01 +00:00
}
}
Parse host header
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(())
}
Extract bucket
2020-11-10 08:57:07 +00:00
#[test]
fn host_to_bucket_test() {
Fix formatting
2020-11-10 08:59:52 +00:00
assert_eq!(
Extract bucket
2020-11-10 08:57:07 +00:00
host_to_bucket("john.doe.garage.tld", ".garage.tld"),
Fix formatting
2020-11-10 08:59:52 +00:00
"john.doe"
);
Extract bucket
2020-11-10 08:57:07 +00:00
assert_eq!(
Fix formatting
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");
Extract bucket
2020-11-10 08:57:07 +00:00
assert_eq!(
host_to_bucket("john.doe.com", ".garage.tld"),
Fix formatting
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");
Extract bucket
2020-11-10 08:57:07 +00:00
}
Build path correctly
2020-11-11 18:48:01 +00:00
#[test]
fn path_to_key_test() -> Result<(), Error> {
Fix host to key
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");
Build path correctly
2020-11-11 18:48:01 +00:00
assert!(path_to_key("", "index.html").is_err());
Fix host to key
2020-11-21 11:01:02 +00:00
assert!(path_to_key("i/am/relative", "index.html").is_err());
Build path correctly
2020-11-11 18:48:01 +00:00
Ok(())
}
Parse host header
2020-11-08 14:47:25 +00:00
}
Reference in a new issue Copy permalink