More strict parsing rules for the setters.

Author: o0Ignition0o

src/parser.rs

@@ -201,8 +201,18 @@ impl<'i> Input<'i> {
         Input::with_log(input, None)
     }
 
-    pub fn with_log(input: &'i str, vfn: Option<&dyn Fn(SyntaxViolation)>) -> Self {
+    pub fn no_trim(input: &'i str) -> Self {
+        Input {
+            chars: input.chars(),
+        }
+    }
+
+    pub fn with_log(original_input: &'i str, vfn: Option<&dyn Fn(SyntaxViolation)>) -> Self {
+        let input = original_input.trim_matches(c0_control_or_space);
         if let Some(vfn) = vfn {
+            if input.len() != original_input.len() {
+                vfn(SyntaxViolation::C0SpaceIgnored)
+            }
             if input.chars().any(|c| matches!(c, '\t' | '\n' | '\r')) {
                 vfn(SyntaxViolation::TabOrNewlineIgnored)
             }
@@ -1463,6 +1473,12 @@ fn c0_control_or_space(ch: char) -> bool {
     ch <= ' ' // U+0000 to U+0020
 }
 
+/// https://infra.spec.whatwg.org/#ascii-tab-or-newline
+#[inline]
+fn ascii_tab_or_new_line(ch: char) -> bool {
+    matches!(ch, '\t' | '\r' | '\n')
+}
+
 /// https://url.spec.whatwg.org/#ascii-alpha
 #[inline]
 pub fn ascii_alpha(ch: char) -> bool {

src/quirks.rs

@@ -12,6 +12,8 @@
 //! you probably want to use `Url` method instead.
 
 use parser::{default_port, Context, Input, Parser, SchemeType};
+use std::cell::RefCell;
+use SyntaxViolation;
 use {idna, Host, ParseError, Position, Url};
 
 /// https://url.spec.whatwg.org/#dom-url-domaintoascii
@@ -103,15 +105,18 @@ pub fn set_host(url: &mut Url, new_host: &str) -> Result<(), ()> {
     if url.cannot_be_a_base() {
         return Err(());
     }
+    // Host parsing rules are strict,
+    // We don't want to trim the input
+    let input = Input::no_trim(new_host);
     let host;
     let opt_port;
     {
         let scheme = url.scheme();
         let scheme_type = SchemeType::from(scheme);
         let result = if scheme_type == SchemeType::File {
-            Parser::get_file_host(Input::new(new_host))
+            Parser::get_file_host(input)
         } else {
-            Parser::parse_host(Input::new(new_host), scheme_type)
+            Parser::parse_host(input, scheme_type)
         };
         match result {
             Ok((h, remaining)) => {
@@ -160,11 +165,14 @@ pub fn set_hostname(url: &mut Url, new_hostname: &str) -> Result<(), ()> {
     if url.cannot_be_a_base() {
         return Err(());
     }
+    // Host parsing rules are strict,
+    // We don't want to trim the input
+    let input = Input::no_trim(new_hostname);
     let scheme_type = SchemeType::from(url.scheme());
     let result = if scheme_type == SchemeType::File {
-        Parser::get_file_host(Input::new(new_hostname))
+        Parser::get_file_host(input)
     } else {
-        Parser::parse_host(Input::new(new_hostname), scheme_type)
+        Parser::parse_host(input, scheme_type)
     };
     if let Ok((host, _remaining)) = result {
         if let Host::Domain(h) = &host {