handle IDN conversion better, and log expected failures

Author: micolous

url/Cargo.toml

@@ -33,7 +33,7 @@ serde = { version = "1.0", optional = true, features = ["derive"] }
 idna = { version = "0.5.0", path = "../idna" }
 
 [target.'cfg(target_arch = "wasm32")'.dependencies]
-web-sys = { version = "0.3.65", features = ["Url"] }
+web-sys = { version = "0.3.65", features = ["Navigator", "Url", "Window"] }
 
 [features]
 default = []

url/src/host.rs

@@ -145,20 +145,36 @@ impl Host<String> {
         idna::domain_to_ascii(domain).map_err(Into::into)
     }
 
+    #[cfg(all(target_arch = "wasm32", target_os = "unknown"))]
+    const SENTINEL_HOSTNAME: &'static str = "url-host-web-sys-sentinel";
+    #[cfg(all(target_arch = "wasm32", target_os = "unknown"))]
+    const SENTINEL_URL: &'static str = "http://url-host-web-sys-sentinel";
+
     #[cfg(all(target_arch = "wasm32", target_os = "unknown"))]
     /// Convert IDN domain to ASCII form with [web_sys::Url]
     fn domain_to_ascii(domain: &str) -> Result<String, ParseError> {
+        debug_assert!(Self::SENTINEL_URL.ends_with(Self::SENTINEL_HOSTNAME));
         // Url throws an error on empty hostnames
         if domain.is_empty() {
-            return Err(ParseError::EmptyHost);
+            return Ok(domain.to_string());
         }
         // Url returns strange results for invalid domain chars
         if domain.contains(Self::is_invalid_domain_char) {
             return Err(ParseError::InvalidDomainCharacter);
         }
-        let u =
-            web_sys::Url::new(&format!("http://{domain}")).map_err(|_| ParseError::IdnaError)?;
-        Ok(u.host())
+
+        // Create a new Url with a sentinel value.
+        let u = web_sys::Url::new(Self::SENTINEL_URL).map_err(|_| ParseError::IdnaError)?;
+        debug_assert_eq!(u.hostname(), Self::SENTINEL_HOSTNAME);
+        // Whenever set_hostname fails, it doesn't update the Url.
+        u.set_hostname(domain);
+        let h = u.hostname();
+        if h.eq_ignore_ascii_case(Self::SENTINEL_HOSTNAME) || h.is_empty() {
+            // It's probably invalid
+            Err(ParseError::IdnaError)
+        } else {
+            Ok(h)
+        }
     }
 
     fn is_invalid_domain_char(c: char) -> bool {

url/tests/expected_failures_chromium.txt

@@ -0,0 +1,14 @@
+<http://\u{ff05}\u{ff14}\u{ff11}.com> against <http://other.com/>
+<http://%ef%bc%85%ef%bc%94%ef%bc%91.com> against <http://other.com/>
+<http://!\"$&\'()*+,-.;=_`{}~/>
+<wss://!\"$&\'()*+,-.;=_`{}~/>
+<https://example.com/> set host to <xn-->
+<https://example.com/> set hostname to <xn-->
+<http://a.b.c.xn--pokxncvks>
+<http://10.0.0.xn--pokxncvks>
+<http://a.b.c.XN--pokxncvks>
+<http://a.b.c.Xn--pokxncvks>
+<http://10.0.0.XN--pokxncvks>
+<http://10.0.0.xN--pokxncvks>
+<file://xn--/p>
+<https://xn--/>

url/tests/expected_failures_firefox.txt

@@ -0,0 +1,2 @@
+<http://!\"$&\'()*+,-.;=_`{}~/>
+<wss://!\"$&\'()*+,-.;=_`{}~/>

url/tests/expected_failures_safari.txt

@@ -0,0 +1,2 @@
+<http://\u{ff05}\u{ff14}\u{ff11}.com> against <http://other.com/>
+<http://%ef%bc%85%ef%bc%94%ef%bc%91.com> against <http://other.com/>

url/tests/wpt.rs

@@ -139,6 +139,20 @@ fn main() {
     let mut expected_failures = include_str!("expected_failures.txt")
         .lines()
         .collect::<Vec<_>>();
+    #[cfg(all(target_arch = "wasm32", target_os = "unknown"))]
+    {
+        // Every browser has its quirks.
+        let user_agent = web_sys::window().unwrap().navigator().user_agent().unwrap().to_ascii_lowercase();
+        if user_agent.contains("chrom") {
+            expected_failures.extend(include_str!("expected_failures_chromium.txt").lines());
+        }
+        if user_agent.contains("gecko/20") {
+            expected_failures.extend(include_str!("expected_failures_firefox.txt").lines());
+        }
+        if user_agent.contains("safari") && !user_agent.contains("chrom") {
+            expected_failures.extend(include_str!("expected_failures_safari.txt").lines());
+        }
+    }
 
     let mut errors = vec![];