feat: update http/1.1 compliance

README.md

@@ -21,46 +21,46 @@ test-bed][bandit-tests], [h2spec][h2spec], and [Autobahn][autobahn]
 
 ### HTTP/1.1
 
-- [ ] invalid requests
-  - [ ] returns a 400 if the request cannot be parsed
-  - [ ] returns a 400 if the request has an invalid http version
-- [ ] keepalive requests
-  - [ ] closes connection after max_requests is reached
-  - [ ] idle keepalive connections are closed after read_timeout
-  - [ ] unread content length bodies are read before starting a new request
-  - [ ] unread chunked bodies are read before starting a new request
-- [ ] origin-form request target (RFC9112§3.2.1)
-  - [ ] derives scheme from underlying transport
-  - [ ] derives host from host header
-  - [ ] returns 400 if no host header set in HTTP/1.1
-  - [ ] sets a blank host if no host header set in HTTP/1.0
-  - [ ] derives port from host header
-  - [ ] derives host from host header with ipv6 host
-  - [ ] derives host and port from host header with ipv6 host
+- [x] invalid requests
+  - [x] returns a 400 if the request cannot be parsed
+  - [x] returns a 400 if the request has an invalid http version
+- [x] keepalive requests
+  - [x] closes connection after max_requests is reached
+  - [x] idle keepalive connections are closed after read_timeout
+  - [x] unread content length bodies are read before starting a new request
+  - [x] unread chunked bodies are read before starting a new request
+- [x] origin-form request target (RFC9112§3.2.1)
+  - [x] derives scheme from underlying transport
+  - [x] derives host from host header
+  - [x] returns 400 if no host header set in HTTP/1.1
+  - [x] sets a blank host if no host header set in HTTP/1.0
+  - [x] derives port from host header
+  - [x] derives host from host header with ipv6 host
+  - [x] derives host and port from host header with ipv6 host
   - [ ] returns 400 if port cannot be parsed from host header
-  - [ ] derives port from schema default if no port specified in host header
-  - [ ] derives port from schema default if no host header set in HTTP/1.0
-  - [ ] sets path and query string properly when no query string is present
-  - [ ] sets path and query string properly when query string is present
-  - [ ] ignores fragment when no query string is present
-  - [ ] ignores fragment when query string is present
-  - [ ] handles query strings with question mark characters in them
-  - [ ] returns 400 if a non-absolute path is send
-  - [ ] returns 400 if path has no leading slash
-- [ ] absolute-form request target (RFC9112§3.2.2)
-  - [ ] uses request-line scheme even if it does not match the transport
-  - [ ] derives host from the URI, even if it differs from host header
-  - [ ] derives ipv6 host from the URI, even if it differs from host header
-  - [ ] does not require a host header set in HTTP/1.1 (RFC9112§3.2.2)
-  - [ ] derives port from the URI, even if it differs from host header
-  - [ ] derives port from schema default if no port specified in the URI
-  - [ ] sets path and query string properly when no query string is present
-  - [ ] sets path and query string properly when query string is present
-  - [ ] ignores fragment when no query string is present
-  - [ ] ignores fragment when query string is present
-  - [ ] handles query strings with question mark characters in them
-- [ ] authority-form request target (RFC9112§3.2.3)
-  - [ ] returns 400 for authority-form / CONNECT requests
+  - [x] derives port from schema default if no port specified in host header
+  - [x] derives port from schema default if no host header set in HTTP/1.0
+  - [x] sets path and query string properly when no query string is present
+  - [x] sets path and query string properly when query string is present
+  - [x] ignores fragment when no query string is present
+  - [x] ignores fragment when query string is present
+  - [x] handles query strings with question mark characters in them
+  - [x] returns 400 if a non-absolute path is send
+  - [x] returns 400 if path has no leading slash
+- [x] absolute-form request target (RFC9112§3.2.2)
+  - [x] uses request-line scheme even if it does not match the transport
+  - [x] derives host from the URI, even if it differs from host header
+  - [x] derives ipv6 host from the URI, even if it differs from host header
+  - [x] does not require a host header set in HTTP/1.1 (RFC9112§3.2.2)
+  - [x] derives port from the URI, even if it differs from host header
+  - [x] derives port from schema default if no port specified in the URI
+  - [x] sets path and query string properly when no query string is present
+  - [x] sets path and query string properly when query string is present
+  - [x] ignores fragment when no query string is present
+  - [x] ignores fragment when query string is present
+  - [x] handles query strings with question mark characters in them
+- [x] authority-form request target (RFC9112§3.2.3)
+  - [x] returns 400 for authority-form / CONNECT requests
 - [ ] asterisk-form request target (RFC9112§3.2.4)
   - [ ] parse global OPTIONS path correctly
 - [ ] request line limits
@@ -96,8 +96,8 @@ test-bed][bandit-tests], [h2spec][h2spec], and [Autobahn][autobahn]
   - [ ] writes out a response with a valid date header
   - [ ] returns user-defined date header instead of internal version
 - [ ] response body
-  - [ ] writes out a response with deflate encoding if so negotiated
-  - [ ] writes out a response with gzip encoding if so negotiated
+  - [x] writes out a response with deflate encoding if so negotiated
+  - [x] writes out a response with gzip encoding if so negotiated
   - [ ] writes out a response with x-gzip encoding if so negotiated
   - [x] uses the first matching encoding in accept-encoding
   - [x] falls back to no encoding if no encodings provided
@@ -114,18 +114,17 @@ test-bed][bandit-tests], [h2spec][h2spec], and [Autobahn][autobahn]
   - [x] writes out a response with zero content-length for 200 responses
   - [x] writes out a response with zero content-length for 301 responses
   - [x] writes out a response with zero content-length for 401 responses
-  - [ ] writes out a chunked response
-  - [ ] does not write out a body for a chunked response to a HEAD request
-  - [ ] writes out a chunked iolist response
-  - [ ] returns socket errors on chunk calls
-  - [ ] writes out a sent file for the entire file with content length
-  - [ ] writes out headers but not body for files requested via HEAD request
-  - [ ] does not write out a content-length header or body for files on a 204
-  - [ ] does not write out a content-length header or body for files on a 304
-  - [ ] writes out a sent file for parts of a file with content length
-- [ ] sending informational responses
-- [ ] does not send informational responses to HTTP/1.0 clients
-- [ ] reading HTTP version
+  - [x] writes out a chunked response
+  - [x] does not write out a body for a chunked response to a HEAD request
+  - [x] returns socket errors on chunk calls
+  - [x] writes out a sent file for the entire file with content length
+  - [x] writes out headers but not body for files requested via HEAD request
+  - [x] does not write out a content-length header or body for files on a 204
+  - [x] does not write out a content-length header or body for files on a 304
+  - [x] writes out a sent file for parts of a file with content length
+- [x] sending informational responses
+- [x] does not send informational responses to HTTP/1.0 clients
+- [x] reading HTTP version
 - [ ] reading peer data
 
 ### HTTP/2

nomad/adapter.ml

@@ -22,10 +22,10 @@ let deflate_string str =
   Buffer.contents r
 
 let gzip_string str =
-  let time () = Int32.of_float (Unix.gettimeofday ()) in
+  let time () = 2112l in
   let i = De.bigstring_create De.io_buffer_size in
   let o = De.bigstring_create De.io_buffer_size in
-  let w = De.Lz77.make_window ~bits:15 in
+  let w = De.Lz77.make_window ~bits:16 in
   let q = De.Queue.create 0x1000 in
   let r = Buffer.create 0x1000 in
   let p = ref 0 in
@@ -78,49 +78,91 @@ let maybe_compress (req : Request.t) buf =
       |> Option.value ~default:[] |> List.map String.trim
     in
     let accepts_deflate = List.mem "deflate" accepted_encodings in
-    let accepts_gzip =
-      List.mem "gzip" accepted_encodings || List.mem "x-gzip" accepted_encodings
-    in
+    let accepts_gzip = List.mem "gzip" accepted_encodings in
+    let accepts_x_gzip = List.mem "x-gzip" accepted_encodings in
     if accepts_deflate then (Some (deflate buf), Some "deflate")
     else if accepts_gzip then (Some (gzip buf), Some "gzip")
+    else if accepts_x_gzip then (Some (gzip buf), Some "x-gzip")
     else (Some buf, None))
 
 let send conn (req : Request.t) (res : Response.t) =
-  let body, encoding =
-    if
-      has_custom_content_encoding res
-      || has_strong_etag res || has_no_transform res
-    then (res.body, None)
-    else
-      match res.body with
-      | Some body -> maybe_compress req body
-      | None -> (None, None)
-  in
-  let headers =
-    match encoding with
-    | Some encoding -> Http.Header.add res.headers "content-encoding" encoding
-    | None -> res.headers
-  in
-  let headers = Http.Header.add headers "vary" "accept-encoding" in
+  if req.version = `HTTP_1_0 && res.status = `Continue then ()
+  else
+    let body, encoding =
+      if
+        has_custom_content_encoding res
+        || has_strong_etag res || has_no_transform res
+      then (res.body, None)
+      else
+        match res.body with
+        | Some body -> maybe_compress req body
+        | None -> (None, None)
+    in
+    let headers =
+      match encoding with
+      | Some encoding -> Http.Header.add res.headers "content-encoding" encoding
+      | None -> res.headers
+    in
+    let headers = Http.Header.add headers "vary" "accept-encoding" in
+
+    let content_length =
+      Option.map IO.Buffer.filled body |> Option.value ~default:0
+    in
+    let headers =
+      if res.status = `No_content then
+        Http.Header.remove headers "content-length"
+      else if res.status != `Not_modified && content_length > 0 then
+        Http.Header.replace headers "content-length"
+          (Int.to_string content_length)
+      else headers
+    in
+    let body =
+      if
+        req.meth = `HEAD || res.status = `No_content
+        || res.status = `Not_modified
+      then None
+      else body
+    in
+
+    let buf = Response.to_buffer { res with headers; body } in
+    Logger.debug (fun f -> f "res: %S" (IO.Buffer.to_string buf));
+    let _ = Atacama.Connection.send conn buf in
+    ()
 
-  let content_length =
-    Option.map IO.Buffer.filled body |> Option.value ~default:0
+let send_chunk conn (req : Request.t) buf =
+  if req.meth != `HEAD then (
+    let chunk =
+      Format.sprintf "%x\r\n%s\r\n" (IO.Buffer.filled buf)
+        (IO.Buffer.to_string buf)
+    in
+    Logger.debug (fun f -> f "sending chunk: %S" chunk);
+    let chunk = IO.Buffer.of_string chunk in
+    let _ = Atacama.Connection.send conn chunk in
+    ())
+
+let send_file conn (req : Request.t) (res : Response.t) ?off ?len ~path () =
+  let len =
+    match len with
+    | Some len -> len
+    | None ->
+        let stat = File.stat path in
+        stat.st_size
   in
   let headers =
-    if res.status != `No_content && res.status != `Not_modified && content_length > 0 then
-      Http.Header.replace headers "content-length" (Int.to_string content_length)
-    else headers
+    Http.Header.replace res.headers "content-length" (Int.to_string len)
   in
-  let body = if req.meth = `HEAD || res.status = `No_content || res.status = `Not_modified  then None else body in
-
-  let buf = Response.to_buffer { res with headers; body } in
-  Logger.debug (fun f -> f "res: %S" (IO.Buffer.to_string buf));
-  let _ = Atacama.Connection.send conn buf in
-  ()
-
-let send_chunk conn (_req : Request.t) buf =
-  let chunk = Format.sprintf "%d\r\n%s\r\n" (IO.Buffer.filled buf) (IO.Buffer.to_string buf) in
-  Logger.debug (fun f-> f "sending chunk: %S" chunk);
-  let chunk = IO.Buffer.of_string chunk in
-  let _ = Atacama.Connection.send conn chunk in
-  ()
+  let res = { res with headers; body = None } in
+  let _ = send conn req res in
+  if res.status != `No_content then
+    let _ = Atacama.Connection.send_file conn ?off ~len (File.open_read path) in
+    ()
+
+let close conn (req: Request.t) (res: Response.t) =
+(if req.meth = `HEAD then ()
+else if res.status = `No_content then ()
+else
+         let _ =
+           Atacama.Connection.send conn (IO.Buffer.of_string "0\r\n\r\n")
+         in
+         ());
+

nomad/http1.ml

@@ -167,10 +167,12 @@ let make_uri state (req : Trail.Request.t) =
   if Uri.port uri |> Option.value ~default:0 < 0 then raise_notrace Bad_port;
 
   let path = Uri.path req.uri in
+  let query = Uri.query req.uri in
   if not (String.starts_with ~prefix:"/" path) then
     raise_notrace Path_missing_leading_slash;
 
   let uri = Uri.with_path uri path in
+  let uri = Uri.with_query uri query in
   Logger.error (fun f -> f "parse uri: %a" Uri.pp uri);
   uri
 
@@ -194,9 +196,7 @@ let handle_request state conn req body =
   | Handler.Close _conn when is_keep_alive ->
       Logger.debug (fun f -> f "connection is keep alive, continuing");
       Continue { state with sniffed_data = None }
-  | Handler.Close conn ->
-      let _ = Atacama.Connection.send conn (IO.Buffer.of_string "0\r\n\r\n") in
-      Close state
+  | Handler.Close _conn -> Close state
   | Handler.Upgrade (`websocket (upgrade_opts, handler)) ->
       let state = Ws.make ~upgrade_opts ~handler ~req ~conn () in
       let state = Ws.handshake conn state in
@@ -218,7 +218,7 @@ let run_handler state conn req body =
       handle_request state conn { req with uri } body
   | `HTTP_1_1, Some _host, uri ->
       handle_request state conn { req with uri } body
-  | `HTTP_1_0, None, uri -> handle_request state conn { req with uri } body
+  | `HTTP_1_0, _, uri -> handle_request state conn { req with uri } body
   | _ -> bad_request conn state
 
 let handle_data data conn state =