test: add httparse tests and make errors identical

http/src/conn.rs

@@ -713,27 +713,22 @@ where
             .find(&buffer)
             .ok_or(Error::InvalidHead)?;
 
-        let (method, path, version) = match &memchr::memchr_iter(b' ', &buffer[..first_line_index])
-            .collect::<Vec<usize>>()[..]
-        {
-            [first, second] => {
-                let (first, second) = (*first, *second);
-                let method = Method::parse(&buffer[0..first])?;
-                let path = str::from_utf8(&buffer[first + 1..second])
-                    .map_err(|_| Error::RequestPathMissing)?
-                    .to_string();
-                let version = Version::parse(&buffer[second + 1..first_line_index])?;
-
-                if !matches!(version, Version::Http1_1 | Version::Http1_0) {
-                    return Err(Error::UnsupportedVersion(version));
-                }
-
-                (method, path, version)
-            }
-            _ => return Err(Error::InvalidHead),
-        };
+        let mut spaces = memchr::memchr_iter(b' ', &buffer[..first_line_index]);
+        let first_space = spaces.next().ok_or(Error::MissingMethod)?;
+        let method = Method::parse(&buffer[0..first_space])?;
+        let second_space = spaces.next().ok_or(Error::RequestPathMissing)?;
+        let path = str::from_utf8(&buffer[first_space + 1..second_space])
+            .map_err(|_| Error::RequestPathMissing)?
+            .to_string();
+        if path.is_empty() {
+            return Err(Error::InvalidHead);
+        }
+        let version = Version::parse(&buffer[second_space + 1..first_line_index])?;
+        if !matches!(version, Version::Http1_1 | Version::Http1_0) {
+            return Err(Error::UnsupportedVersion(version));
+        }
 
-        let request_headers = Headers::parse(&buffer[first_line_index + 2..])?;
+        let request_headers = Headers::parse(&buffer[first_line_index + 2..head_size])?;
 
         Self::validate_headers(&request_headers)?;
 

http/src/headers.rs

@@ -114,7 +114,11 @@ impl Headers {
             }
 
             let mut value_start = token_end + 1;
-            while (bytes[value_start] as char).is_whitespace() {
+
+            while bytes
+                .get(value_start)
+                .is_some_and(|b| matches!(b, b'\t' | b' '))
+            {
                 value_start += 1;
             }
 

http/tests/corpus.rs

@@ -1,5 +1,6 @@
 use indoc::formatdoc;
-use pretty_assertions::assert_eq;
+use pretty_assertions::assert_str_eq;
+use std::{env, net::Shutdown, path::PathBuf};
 use test_harness::test;
 use trillium_http::{Conn, KnownHeaderName, Swansong};
 use trillium_testing::{harness, RuntimeTrait, TestTransport};
@@ -45,8 +46,8 @@ async fn handler(mut conn: Conn<TestTransport>) -> Conn<TestTransport> {
 async fn corpus_test() {
     env_logger::init();
     let runtime = trillium_testing::runtime();
-    let dir = std::path::PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("tests/corpus");
-    let filter = std::env::var("CORPUS_TEST_FILTER").unwrap_or_default();
+    let dir = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("tests/corpus");
+    let filter = env::var("CORPUS_TEST_FILTER").unwrap_or_default();
     let corpus_request_files = std::fs::read_dir(dir)
         .unwrap()
         .filter_map(|f| {
@@ -74,6 +75,7 @@ async fn corpus_test() {
         });
 
         client.write_all(request);
+        client.shutdown(Shutdown::Write);
         let (response, extension) = match res.await.unwrap() {
             Ok(None) => (client.read_available_string().await, "response"),
             Err(e) => (e.to_string(), "error"),
@@ -94,7 +96,7 @@ async fn corpus_test() {
                 .replace(['\n', '\r'], "")
                 .replace("\\r", "\r")
                 .replace("\\n", "\n");
-            assert_eq!(expected_response, response, "{file:?}");
+            assert_str_eq!(expected_response, response, "\n\n{file:?}");
         }
 
         swansong.shut_down();

http/tests/corpus/cl-and-tec.error

@@ -0,0 +1 @@
+Unexpected header: Content-Length

http/tests/corpus/cl-and-tec.request

@@ -0,0 +1,8 @@
+POST http://example.com:8080/ogdlqgxkbepd HTTP/1.1\r\n
+Host: example.com:8080\r\n
+Transfer-Encoding: chunked\r\n
+Content-Length: 100\r\n
+Connection: close\r\n
+\r\n
+0; name=value\r\n
+\r\n

http/tests/corpus/httparse-001.request

@@ -0,0 +1,3 @@
+GET /bar;par?b HTTP/1.1\r\n
+Host: foo\r\n
+\r\n

http/tests/corpus/httparse-001.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 97\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /bar;par\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: foo\r\n
+\n
+===body===\n

http/tests/corpus/httparse-002.request

@@ -0,0 +1,3 @@
+GET /x HTTP/1.1\r\n
+Host: test\r\n
+\r\n

http/tests/corpus/httparse-002.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 92\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /x\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: test\r\n
+\n
+===body===\n

http/tests/corpus/httparse-004.request

@@ -0,0 +1,3 @@
+GET /foo/foo.com HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-004.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 109\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/foo.com\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n

http/tests/corpus/httparse-005.request

@@ -0,0 +1,3 @@
+GET /foo/:foo.com HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-005.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 110\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/:foo.com\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n

http/tests/corpus/httparse-006.request

@@ -0,0 +1,3 @@
+GET /foo/foo.com HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-006.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 109\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/foo.com\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n

http/tests/corpus/httparse-007.request

@@ -0,0 +1,3 @@
+GET  foo.com HTTP/1.1\r\n
+Host: \r\n
+\r\n

http/tests/corpus/httparse-008.request

@@ -0,0 +1,3 @@
+GET /%20b%20?%20d%20 HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-008.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 95\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /%20b%20\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-009.error

@@ -0,0 +1 @@
+Invalid or missing version

http/tests/corpus/httparse-009.request

@@ -0,0 +1,3 @@
+GET x x HTTP/1.1\r\n
+Host: \r\n
+\r\n

http/tests/corpus/httparse-010.request

@@ -0,0 +1,3 @@
+GET /c HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-010.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 89\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /c\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-011.request

@@ -0,0 +1,3 @@
+GET /c HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-011.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 89\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /c\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-012.request

@@ -0,0 +1,3 @@
+GET /c HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-012.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 89\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /c\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-013.request

@@ -0,0 +1,3 @@
+GET /c HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-013.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 89\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /c\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-014.request

@@ -0,0 +1,3 @@
+GET /c HTTP/1.1\r\n
+Host: f\r\n
+\r\n

http/tests/corpus/httparse-014.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 89\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /c\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: f\r\n
+\n
+===body===\n

http/tests/corpus/httparse-015.request

@@ -0,0 +1,3 @@
+GET /foo/bar HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-015.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 105\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/bar\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n

http/tests/corpus/httparse-016.request

@@ -0,0 +1,3 @@
+GET /foo/bar HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-016.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 105\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/bar\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n

http/tests/corpus/httparse-017.request

@@ -0,0 +1,3 @@
+GET /foo/:foo.com/ HTTP/1.1\r\n
+Host: example.org\r\n
+\r\n

http/tests/corpus/httparse-017.response

@@ -0,0 +1,14 @@
+HTTP/1.1 200 OK\r\n
+Date: Tue, 21 Nov 2023 21:27:21 GMT\r\n
+Content-Length: 111\r\n
+Server: corpus-test\r\n
+\r\n
+===request===\n
+method: GET\n
+path: /foo/:foo.com/\n
+version: HTTP/1.1\n
+\n
+===headers===\n
+Host: example.org\r\n
+\n
+===body===\n