Fix parsing of long host with no scheme (#351)

This change fixes an issue where an URI with no scheme, a host
part with a length greater than the max for a scheme, while only
consisting of valid characters for a scheme would fail to parse.

Rather than assuming the string being parsed is meant to contain
a scheme and returning an error if the max length for a scheme is
met, we instead extract the scheme part if it exists and validate
its length.

The trade-off is worse performance in the error case of a scheme
that is too long.

Author: avitex

src/uri/scheme.rs

@@ -336,10 +336,6 @@ impl Scheme2<usize> {
             for i in 0..s.len() {
                 let b = s[i];
 
-                if i == MAX_SCHEME_LEN {
-                    return Err(ErrorKind::SchemeTooLong.into());
-                }
-
                 match SCHEME_CHARS[b as usize] {
                     b':' => {
                         // Not enough data remaining
@@ -352,6 +348,10 @@ impl Scheme2<usize> {
                             break;
                         }
 
+                        if i > MAX_SCHEME_LEN {
+                            return Err(ErrorKind::SchemeTooLong.into());
+                        }
+
                         // Return scheme
                         return Ok(Scheme2::Other(i));
                     }

src/uri/tests.rs

@@ -231,6 +231,30 @@ test_parse! {
     port_part = None,
 }
 
+test_parse! {
+    test_uri_parse_long_host_with_no_scheme,
+    "thequickbrownfoxjumpedoverthelazydogtofindthelargedangerousdragon.localhost",
+    [],
+
+    scheme_part = None,
+    authority_part = part!("thequickbrownfoxjumpedoverthelazydogtofindthelargedangerousdragon.localhost"),
+    path = "",
+    query = None,
+    port_part = None,
+}
+
+test_parse! {
+    test_uri_parse_long_host_with_port_and_no_scheme,
+    "thequickbrownfoxjumpedoverthelazydogtofindthelargedangerousdragon.localhost:1234",
+    [],
+
+    scheme_part = None,
+    authority_part = part!("thequickbrownfoxjumpedoverthelazydogtofindthelargedangerousdragon.localhost:1234"),
+    path = "",
+    query = None,
+    port_part = Port::from_str("1234").ok(),
+}
+
 test_parse! {
     test_userinfo1,
     "http://a:[email protected]:1234/",
@@ -430,7 +454,7 @@ fn test_max_uri_len() {
 }
 
 #[test]
-fn test_long_scheme() {
+fn test_overflowing_scheme() {
     let mut uri = vec![];
     uri.extend(vec![b'a'; 256]);
     uri.extend(b"://localhost/");
@@ -441,6 +465,18 @@ fn test_long_scheme() {
     assert_eq!(res.unwrap_err().0, ErrorKind::SchemeTooLong);
 }
 
+#[test]
+fn test_max_length_scheme() {
+    let mut uri = vec![];
+    uri.extend(vec![b'a'; 64]);
+    uri.extend(b"://localhost/");
+
+    let uri = String::from_utf8(uri).unwrap();
+    let uri: Uri = uri.parse().unwrap();
+
+    assert_eq!(uri.scheme_str().unwrap().len(), 64);
+}
+
 #[test]
 fn test_uri_to_path_and_query() {
     let cases = vec![